Commit b09cac05 authored by Philip Gladstone's avatar Philip Gladstone Committed by Johny Mattsson
Browse files

Add support for streaming JSON encoder/decoder (#1755)

Replaces the problematic cjson module.
parent b6ef1ffe
These JSON examples were taken from the JSON website
(http://json.org/example.html) and RFC 4627.
Used with permission.
#!/usr/bin/env lua
-- This benchmark script measures wall clock time and should be
-- run on an unloaded system.
--
-- Your Mileage May Vary.
--
-- Mark Pulford <mark@kyne.com.au>
local json_module = os.getenv("JSON_MODULE") or "cjson"
require "socket"
local json = require(json_module)
local util = require "cjson.util"
local function find_func(mod, funcnames)
for _, v in ipairs(funcnames) do
if mod[v] then
return mod[v]
end
end
return nil
end
local json_encode = find_func(json, { "encode", "Encode", "to_string", "stringify", "json" })
local json_decode = find_func(json, { "decode", "Decode", "to_value", "parse" })
local function average(t)
local total = 0
for _, v in ipairs(t) do
total = total + v
end
return total / #t
end
function benchmark(tests, seconds, rep)
local function bench(func, iter)
-- Use socket.gettime() to measure microsecond resolution
-- wall clock time.
local t = socket.gettime()
for i = 1, iter do
func(i)
end
t = socket.gettime() - t
-- Don't trust any results when the run lasted for less than a
-- millisecond - return nil.
if t < 0.001 then
return nil
end
return (iter / t)
end
-- Roughly calculate the number of interations required
-- to obtain a particular time period.
local function calc_iter(func, seconds)
local iter = 1
local rate
-- Warm up the bench function first.
func()
while not rate do
rate = bench(func, iter)
iter = iter * 10
end
return math.ceil(seconds * rate)
end
local test_results = {}
for name, func in pairs(tests) do
-- k(number), v(string)
-- k(string), v(function)
-- k(number), v(function)
if type(func) == "string" then
name = func
func = _G[name]
end
local iter = calc_iter(func, seconds)
local result = {}
for i = 1, rep do
result[i] = bench(func, iter)
end
-- Remove the slowest half (round down) of the result set
table.sort(result)
for i = 1, math.floor(#result / 2) do
table.remove(result, 1)
end
test_results[name] = average(result)
end
return test_results
end
function bench_file(filename)
local data_json = util.file_load(filename)
local data_obj = json_decode(data_json)
local function test_encode()
json_encode(data_obj)
end
local function test_decode()
json_decode(data_json)
end
local tests = {}
if json_encode then tests.encode = test_encode end
if json_decode then tests.decode = test_decode end
return benchmark(tests, 0.1, 5)
end
-- Optionally load any custom configuration required for this module
local success, data = pcall(util.file_load, ("bench-%s.lua"):format(json_module))
if success then
util.run_script(data, _G)
configure(json)
end
for i = 1, #arg do
local results = bench_file(arg[i])
for k, v in pairs(results) do
print(("%s\t%s\t%d"):format(arg[i], k, v))
end
end
-- vi:ai et sw=4 ts=4:
{
"glossary": {
"title": "example glossary",
"GlossDiv": {
"title": "S",
"GlossList": {
"GlossEntry": {
"ID": "SGML",
"SortAs": "SGML",
"GlossTerm": "Standard Generalized Mark up Language",
"Acronym": "SGML",
"Abbrev": "ISO 8879:1986",
"GlossDef": {
"para": "A meta-markup language, used to create markup languages such as DocBook.",
"GlossSeeAlso": ["GML", "XML"]
},
"GlossSee": "markup"
}
}
}
}
}
{"menu": {
"id": "file",
"value": "File",
"popup": {
"menuitem": [
{"value": "New", "onclick": "CreateNewDoc()"},
{"value": "Open", "onclick": "OpenDoc()"},
{"value": "Close", "onclick": "CloseDoc()"}
]
}
}}
{"widget": {
"debug": "on",
"window": {
"title": "Sample Konfabulator Widget",
"name": "main_window",
"width": 500,
"height": 500
},
"image": {
"src": "Images/Sun.png",
"name": "sun1",
"hOffset": 250,
"vOffset": 250,
"alignment": "center"
},
"text": {
"data": "Click Here",
"size": 36,
"style": "bold",
"name": "text1",
"hOffset": 250,
"vOffset": 100,
"alignment": "center",
"onMouseUp": "sun1.opacity = (sun1.opacity / 100) * 90;"
}
}}
{"web-app": {
"servlet": [
{
"servlet-name": "cofaxCDS",
"servlet-class": "org.cofax.cds.CDSServlet",
"init-param": {
"configGlossary:installationAt": "Philadelphia, PA",
"configGlossary:adminEmail": "ksm@pobox.com",
"configGlossary:poweredBy": "Cofax",
"configGlossary:poweredByIcon": "/images/cofax.gif",
"configGlossary:staticPath": "/content/static",
"templateProcessorClass": "org.cofax.WysiwygTemplate",
"templateLoaderClass": "org.cofax.FilesTemplateLoader",
"templatePath": "templates",
"templateOverridePath": "",
"defaultListTemplate": "listTemplate.htm",
"defaultFileTemplate": "articleTemplate.htm",
"useJSP": false,
"jspListTemplate": "listTemplate.jsp",
"jspFileTemplate": "articleTemplate.jsp",
"cachePackageTagsTrack": 200,
"cachePackageTagsStore": 200,
"cachePackageTagsRefresh": 60,
"cacheTemplatesTrack": 100,
"cacheTemplatesStore": 50,
"cacheTemplatesRefresh": 15,
"cachePagesTrack": 200,
"cachePagesStore": 100,
"cachePagesRefresh": 10,
"cachePagesDirtyRead": 10,
"searchEngineListTemplate": "forSearchEnginesList.htm",
"searchEngineFileTemplate": "forSearchEngines.htm",
"searchEngineRobotsDb": "WEB-INF/robots.db",
"useDataStore": true,
"dataStoreClass": "org.cofax.SqlDataStore",
"redirectionClass": "org.cofax.SqlRedirection",
"dataStoreName": "cofax",
"dataStoreDriver": "com.microsoft.jdbc.sqlserver.SQLServerDriver",
"dataStoreUrl": "jdbc:microsoft:sqlserver://LOCALHOST:1433;DatabaseName=goon",
"dataStoreUser": "sa",
"dataStorePassword": "dataStoreTestQuery",
"dataStoreTestQuery": "SET NOCOUNT ON;select test='test';",
"dataStoreLogFile": "/usr/local/tomcat/logs/datastore.log",
"dataStoreInitConns": 10,
"dataStoreMaxConns": 100,
"dataStoreConnUsageLimit": 100,
"dataStoreLogLevel": "debug",
"maxUrlLength": 500}},
{
"servlet-name": "cofaxEmail",
"servlet-class": "org.cofax.cds.EmailServlet",
"init-param": {
"mailHost": "mail1",
"mailHostOverride": "mail2"}},
{
"servlet-name": "cofaxAdmin",
"servlet-class": "org.cofax.cds.AdminServlet"},
{
"servlet-name": "fileServlet",
"servlet-class": "org.cofax.cds.FileServlet"},
{
"servlet-name": "cofaxTools",
"servlet-class": "org.cofax.cms.CofaxToolsServlet",
"init-param": {
"templatePath": "toolstemplates/",
"log": 1,
"logLocation": "/usr/local/tomcat/logs/CofaxTools.log",
"logMaxSize": "",
"dataLog": 1,
"dataLogLocation": "/usr/local/tomcat/logs/dataLog.log",
"dataLogMaxSize": "",
"removePageCache": "/content/admin/remove?cache=pages&id=",
"removeTemplateCache": "/content/admin/remove?cache=templates&id=",
"fileTransferFolder": "/usr/local/tomcat/webapps/content/fileTransferFolder",
"lookInContext": 1,
"adminGroupID": 4,
"betaServer": true}}],
"servlet-mapping": {
"cofaxCDS": "/",
"cofaxEmail": "/cofaxutil/aemail/*",
"cofaxAdmin": "/admin/*",
"fileServlet": "/static/*",
"cofaxTools": "/tools/*"},
"taglib": {
"taglib-uri": "cofax.tld",
"taglib-location": "/WEB-INF/tlds/cofax.tld"}}}
{"menu": {
"header": "SVG Viewer",
"items": [
{"id": "Open"},
{"id": "OpenNew", "label": "Open New"},
null,
{"id": "ZoomIn", "label": "Zoom In"},
{"id": "ZoomOut", "label": "Zoom Out"},
{"id": "OriginalView", "label": "Original View"},
null,
{"id": "Quality"},
{"id": "Pause"},
{"id": "Mute"},
null,
{"id": "Find", "label": "Find..."},
{"id": "FindAgain", "label": "Find Again"},
{"id": "Copy"},
{"id": "CopyAgain", "label": "Copy Again"},
{"id": "CopySVG", "label": "Copy SVG"},
{"id": "ViewSVG", "label": "View SVG"},
{"id": "ViewSource", "label": "View Source"},
{"id": "SaveAs", "label": "Save As"},
null,
{"id": "Help"},
{"id": "About", "label": "About Adobe CVG Viewer..."}
]
}}
#!/usr/bin/env perl
# Create test comparison data using a different UTF-8 implementation.
# The generated utf8.dat file must have the following MD5 sum:
# cff03b039d850f370a7362f3313e5268
use strict;
# 0xD800 - 0xDFFF are used to encode supplementary codepoints
# 0x10000 - 0x10FFFF are supplementary codepoints
my (@codepoints) = (0 .. 0xD7FF, 0xE000 .. 0x10FFFF);
my $utf8 = pack("U*", @codepoints);
defined($utf8) or die "Unable create UTF-8 string\n";
open(FH, ">:utf8", "utf8.dat")
or die "Unable to open utf8.dat: $!\n";
print FH $utf8
or die "Unable to write utf8.dat\n";
close(FH);
# vi:ai et sw=4 ts=4:
[ 0.110001,
0.12345678910111,
0.412454033640,
2.6651441426902,
2.718281828459,
3.1415926535898,
2.1406926327793 ]
"\u0000\u0001\u0002\u0003\u0004\u0005\u0006\u0007\b\t\n\u000b\f\r\u000e\u000f\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f !\"#$%&'()*+,-.\/0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\u007f"
\ No newline at end of file
{
"Image": {
"Width": 800,
"Height": 600,
"Title": "View from 15th Floor",
"Thumbnail": {
"Url": "http://www.example.com/image/481989943",
"Height": 125,
"Width": "100"
},
"IDs": [116, 943, 234, 38793]
}
}
[
{
"precision": "zip",
"Latitude": 37.7668,
"Longitude": -122.3959,
"Address": "",
"City": "SAN FRANCISCO",
"State": "CA",
"Zip": "94107",
"Country": "US"
},
{
"precision": "zip",
"Latitude": 37.371991,
"Longitude": -122.026020,
"Address": "",
"City": "SUNNYVALE",
"State": "CA",
"Zip": "94085",
"Country": "US"
}
]
#!/usr/bin/env lua
-- Lua CJSON tests
--
-- Mark Pulford <mark@kyne.com.au>
--
-- Note: The output of this script is easier to read with "less -S"
local json = require "cjson"
local json_safe = require "cjson.safe"
local util = require "cjson.util"
local function gen_raw_octets()
local chars = {}
for i = 0, 255 do chars[i + 1] = string.char(i) end
return table.concat(chars)
end
-- Generate every UTF-16 codepoint, including supplementary codes
local function gen_utf16_escaped()
-- Create raw table escapes
local utf16_escaped = {}
local count = 0
local function append_escape(code)
local esc = ('\\u%04X'):format(code)
table.insert(utf16_escaped, esc)
end
table.insert(utf16_escaped, '"')
for i = 0, 0xD7FF do
append_escape(i)
end
-- Skip 0xD800 - 0xDFFF since they are used to encode supplementary
-- codepoints
for i = 0xE000, 0xFFFF do
append_escape(i)
end
-- Append surrogate pair for each supplementary codepoint
for high = 0xD800, 0xDBFF do
for low = 0xDC00, 0xDFFF do
append_escape(high)
append_escape(low)
end
end
table.insert(utf16_escaped, '"')
return table.concat(utf16_escaped)
end
function load_testdata()
local data = {}
-- Data for 8bit raw <-> escaped octets tests
data.octets_raw = gen_raw_octets()
data.octets_escaped = util.file_load("octets-escaped.dat")
-- Data for \uXXXX -> UTF-8 test
data.utf16_escaped = gen_utf16_escaped()
-- Load matching data for utf16_escaped
local utf8_loaded
utf8_loaded, data.utf8_raw = pcall(util.file_load, "utf8.dat")
if not utf8_loaded then
data.utf8_raw = "Failed to load utf8.dat - please run genutf8.pl"
end
data.table_cycle = {}
data.table_cycle[1] = data.table_cycle
local big = {}
for i = 1, 1100 do
big = { { 10, false, true, json.null }, "string", a = big }
end
data.deeply_nested_data = big
return data
end
function test_decode_cycle(filename)
local obj1 = json.decode(util.file_load(filename))
local obj2 = json.decode(json.encode(obj1))
return util.compare_values(obj1, obj2)
end
-- Set up data used in tests
local Inf = math.huge;
local NaN = math.huge * 0;
local testdata = load_testdata()
local cjson_tests = {
-- Test API variables
{ "Check module name, version",
function () return json._NAME, json._VERSION end, { },
true, { "cjson", "2.1devel" } },
-- Test decoding simple types
{ "Decode string",
json.decode, { '"test string"' }, true, { "test string" } },
{ "Decode numbers",
json.decode, { '[ 0.0, -5e3, -1, 0.3e-3, 1023.2, 0e10 ]' },
true, { { 0.0, -5000, -1, 0.0003, 1023.2, 0 } } },
{ "Decode null",
json.decode, { 'null' }, true, { json.null } },
{ "Decode true",
json.decode, { 'true' }, true, { true } },
{ "Decode false",
json.decode, { 'false' }, true, { false } },
{ "Decode object with numeric keys",
json.decode, { '{ "1": "one", "3": "three" }' },
true, { { ["1"] = "one", ["3"] = "three" } } },
{ "Decode object with string keys",
json.decode, { '{ "a": "a", "b": "b" }' },
true, { { a = "a", b = "b" } } },
{ "Decode array",
json.decode, { '[ "one", null, "three" ]' },
true, { { "one", json.null, "three" } } },
-- Test decoding errors
{ "Decode UTF-16BE [throw error]",
json.decode, { '\0"\0"' },
false, { "JSON parser does not support UTF-16 or UTF-32" } },
{ "Decode UTF-16LE [throw error]",
json.decode, { '"\0"\0' },
false, { "JSON parser does not support UTF-16 or UTF-32" } },
{ "Decode UTF-32BE [throw error]",
json.decode, { '\0\0\0"' },
false, { "JSON parser does not support UTF-16 or UTF-32" } },
{ "Decode UTF-32LE [throw error]",
json.decode, { '"\0\0\0' },
false, { "JSON parser does not support UTF-16 or UTF-32" } },
{ "Decode partial JSON [throw error]",
json.decode, { '{ "unexpected eof": ' },
false, { "Expected value but found T_END at character 21" } },
{ "Decode with extra comma [throw error]",
json.decode, { '{ "extra data": true }, false' },
false, { "Expected the end but found T_COMMA at character 23" } },
{ "Decode invalid escape code [throw error]",
json.decode, { [[ { "bad escape \q code" } ]] },
false, { "Expected object key string but found invalid escape code at character 16" } },
{ "Decode invalid unicode escape [throw error]",
json.decode, { [[ { "bad unicode \u0f6 escape" } ]] },
false, { "Expected object key string but found invalid unicode escape code at character 17" } },
{ "Decode invalid keyword [throw error]",
json.decode, { ' [ "bad barewood", test ] ' },
false, { "Expected value but found invalid token at character 20" } },
{ "Decode invalid number #1 [throw error]",
json.decode, { '[ -+12 ]' },
false, { "Expected value but found invalid number at character 3" } },
{ "Decode invalid number #2 [throw error]",
json.decode, { '-v' },
false, { "Expected value but found invalid number at character 1" } },
{ "Decode invalid number exponent [throw error]",
json.decode, { '[ 0.4eg10 ]' },
false, { "Expected comma or array end but found invalid token at character 6" } },
-- Test decoding nested arrays / objects
{ "Set decode_max_depth(5)",
json.decode_max_depth, { 5 }, true, { 5 } },
{ "Decode array at nested limit",
json.decode, { '[[[[[ "nested" ]]]]]' },
true, { {{{{{ "nested" }}}}} } },
{ "Decode array over nested limit [throw error]",
json.decode, { '[[[[[[ "nested" ]]]]]]' },
false, { "Found too many nested data structures (6) at character 6" } },
{ "Decode object at nested limit",
json.decode, { '{"a":{"b":{"c":{"d":{"e":"nested"}}}}}' },
true, { {a={b={c={d={e="nested"}}}}} } },
{ "Decode object over nested limit [throw error]",
json.decode, { '{"a":{"b":{"c":{"d":{"e":{"f":"nested"}}}}}}' },
false, { "Found too many nested data structures (6) at character 26" } },
{ "Set decode_max_depth(1000)",
json.decode_max_depth, { 1000 }, true, { 1000 } },
{ "Decode deeply nested array [throw error]",
json.decode, { string.rep("[", 1100) .. '1100' .. string.rep("]", 1100)},
false, { "Found too many nested data structures (1001) at character 1001" } },
-- Test encoding nested tables
{ "Set encode_max_depth(5)",
json.encode_max_depth, { 5 }, true, { 5 } },
{ "Encode nested table as array at nested limit",
json.encode, { {{{{{"nested"}}}}} }, true, { '[[[[["nested"]]]]]' } },
{ "Encode nested table as array after nested limit [throw error]",
json.encode, { { {{{{{"nested"}}}}} } },
false, { "Cannot serialise, excessive nesting (6)" } },
{ "Encode nested table as object at nested limit",
json.encode, { {a={b={c={d={e="nested"}}}}} },
true, { '{"a":{"b":{"c":{"d":{"e":"nested"}}}}}' } },
{ "Encode nested table as object over nested limit [throw error]",
json.encode, { {a={b={c={d={e={f="nested"}}}}}} },
false, { "Cannot serialise, excessive nesting (6)" } },
{ "Encode table with cycle [throw error]",
json.encode, { testdata.table_cycle },
false, { "Cannot serialise, excessive nesting (6)" } },
{ "Set encode_max_depth(1000)",
json.encode_max_depth, { 1000 }, true, { 1000 } },
{ "Encode deeply nested data [throw error]",
json.encode, { testdata.deeply_nested_data },
false, { "Cannot serialise, excessive nesting (1001)" } },
-- Test encoding simple types
{ "Encode null",
json.encode, { json.null }, true, { 'null' } },
{ "Encode true",
json.encode, { true }, true, { 'true' } },
{ "Encode false",
json.encode, { false }, true, { 'false' } },
{ "Encode empty object",
json.encode, { { } }, true, { '{}' } },
{ "Encode integer",
json.encode, { 10 }, true, { '10' } },
{ "Encode string",
json.encode, { "hello" }, true, { '"hello"' } },
{ "Encode Lua function [throw error]",
json.encode, { function () end },
false, { "Cannot serialise function: type not supported" } },
-- Test decoding invalid numbers
{ "Set decode_invalid_numbers(true)",
json.decode_invalid_numbers, { true }, true, { true } },
{ "Decode hexadecimal",
json.decode, { '0x6.ffp1' }, true, { 13.9921875 } },
{ "Decode numbers with leading zero",
json.decode, { '[ 0123, 00.33 ]' }, true, { { 123, 0.33 } } },
{ "Decode +-Inf",
json.decode, { '[ +Inf, Inf, -Inf ]' }, true, { { Inf, Inf, -Inf } } },
{ "Decode +-Infinity",
json.decode, { '[ +Infinity, Infinity, -Infinity ]' },
true, { { Inf, Inf, -Inf } } },
{ "Decode +-NaN",
json.decode, { '[ +NaN, NaN, -NaN ]' }, true, { { NaN, NaN, NaN } } },
{ "Decode Infrared (not infinity) [throw error]",
json.decode, { 'Infrared' },
false, { "Expected the end but found invalid token at character 4" } },
{ "Decode Noodle (not NaN) [throw error]",
json.decode, { 'Noodle' },
false, { "Expected value but found invalid token at character 1" } },
{ "Set decode_invalid_numbers(false)",
json.decode_invalid_numbers, { false }, true, { false } },
{ "Decode hexadecimal [throw error]",
json.decode, { '0x6' },
false, { "Expected value but found invalid number at character 1" } },
{ "Decode numbers with leading zero [throw error]",
json.decode, { '[ 0123, 00.33 ]' },
false, { "Expected value but found invalid number at character 3" } },
{ "Decode +-Inf [throw error]",
json.decode, { '[ +Inf, Inf, -Inf ]' },
false, { "Expected value but found invalid token at character 3" } },
{ "Decode +-Infinity [throw error]",
json.decode, { '[ +Infinity, Infinity, -Infinity ]' },
false, { "Expected value but found invalid token at character 3" } },
{ "Decode +-NaN [throw error]",
json.decode, { '[ +NaN, NaN, -NaN ]' },
false, { "Expected value but found invalid token at character 3" } },
{ 'Set decode_invalid_numbers("on")',
json.decode_invalid_numbers, { "on" }, true, { true } },
-- Test encoding invalid numbers
{ "Set encode_invalid_numbers(false)",
json.encode_invalid_numbers, { false }, true, { false } },
{ "Encode NaN [throw error]",
json.encode, { NaN },
false, { "Cannot serialise number: must not be NaN or Infinity" } },
{ "Encode Infinity [throw error]",
json.encode, { Inf },
false, { "Cannot serialise number: must not be NaN or Infinity" } },
{ "Set encode_invalid_numbers(\"null\")",
json.encode_invalid_numbers, { "null" }, true, { "null" } },
{ "Encode NaN as null",
json.encode, { NaN }, true, { "null" } },
{ "Encode Infinity as null",
json.encode, { Inf }, true, { "null" } },
{ "Set encode_invalid_numbers(true)",
json.encode_invalid_numbers, { true }, true, { true } },
{ "Encode NaN",
json.encode, { NaN }, true, { "NaN" } },
{ "Encode +Infinity",
json.encode, { Inf }, true, { "Infinity" } },
{ "Encode -Infinity",
json.encode, { -Inf }, true, { "-Infinity" } },
{ 'Set encode_invalid_numbers("off")',
json.encode_invalid_numbers, { "off" }, true, { false } },
-- Test encoding tables
{ "Set encode_sparse_array(true, 2, 3)",
json.encode_sparse_array, { true, 2, 3 }, true, { true, 2, 3 } },
{ "Encode sparse table as array #1",
json.encode, { { [3] = "sparse test" } },
true, { '[null,null,"sparse test"]' } },
{ "Encode sparse table as array #2",
json.encode, { { [1] = "one", [4] = "sparse test" } },
true, { '["one",null,null,"sparse test"]' } },
{ "Encode sparse array as object",
json.encode, { { [1] = "one", [5] = "sparse test" } },
true, { '{"1":"one","5":"sparse test"}' } },
{ "Encode table with numeric string key as object",
json.encode, { { ["2"] = "numeric string key test" } },
true, { '{"2":"numeric string key test"}' } },
{ "Set encode_sparse_array(false)",
json.encode_sparse_array, { false }, true, { false, 2, 3 } },
{ "Encode table with incompatible key [throw error]",
json.encode, { { [false] = "wrong" } },
false, { "Cannot serialise boolean: table key must be a number or string" } },
-- Test escaping
{ "Encode all octets (8-bit clean)",
json.encode, { testdata.octets_raw }, true, { testdata.octets_escaped } },
{ "Decode all escaped octets",
json.decode, { testdata.octets_escaped }, true, { testdata.octets_raw } },
{ "Decode single UTF-16 escape",
json.decode, { [["\uF800"]] }, true, { "\239\160\128" } },
{ "Decode all UTF-16 escapes (including surrogate combinations)",
json.decode, { testdata.utf16_escaped }, true, { testdata.utf8_raw } },
{ "Decode swapped surrogate pair [throw error]",
json.decode, { [["\uDC00\uD800"]] },
false, { "Expected value but found invalid unicode escape code at character 2" } },
{ "Decode duplicate high surrogate [throw error]",
json.decode, { [["\uDB00\uDB00"]] },
false, { "Expected value but found invalid unicode escape code at character 2" } },
{ "Decode duplicate low surrogate [throw error]",
json.decode, { [["\uDB00\uDB00"]] },
false, { "Expected value but found invalid unicode escape code at character 2" } },
{ "Decode missing low surrogate [throw error]",
json.decode, { [["\uDB00"]] },
false, { "Expected value but found invalid unicode escape code at character 2" } },
{ "Decode invalid low surrogate [throw error]",
json.decode, { [["\uDB00\uD"]] },
false, { "Expected value but found invalid unicode escape code at character 2" } },
-- Test locale support
--
-- The standard Lua interpreter is ANSI C online doesn't support locales
-- by default. Force a known problematic locale to test strtod()/sprintf().
{ "Set locale to cs_CZ (comma separator)", function ()
os.setlocale("cs_CZ")
json.new()
end },
{ "Encode number under comma locale",
json.encode, { 1.5 }, true, { '1.5' } },
{ "Decode number in array under comma locale",
json.decode, { '[ 10, "test" ]' }, true, { { 10, "test" } } },
{ "Revert locale to POSIX", function ()
os.setlocale("C")
json.new()
end },
-- Test encode_keep_buffer() and enable_number_precision()
{ "Set encode_keep_buffer(false)",
json.encode_keep_buffer, { false }, true, { false } },
{ "Set encode_number_precision(3)",
json.encode_number_precision, { 3 }, true, { 3 } },
{ "Encode number with precision 3",
json.encode, { 1/3 }, true, { "0.333" } },
{ "Set encode_number_precision(14)",
json.encode_number_precision, { 14 }, true, { 14 } },
{ "Set encode_keep_buffer(true)",
json.encode_keep_buffer, { true }, true, { true } },
-- Test config API errors
-- Function is listed as '?' due to pcall
{ "Set encode_number_precision(0) [throw error]",
json.encode_number_precision, { 0 },
false, { "bad argument #1 to '?' (expected integer between 1 and 14)" } },
{ "Set encode_number_precision(\"five\") [throw error]",
json.encode_number_precision, { "five" },
false, { "bad argument #1 to '?' (number expected, got string)" } },
{ "Set encode_keep_buffer(nil, true) [throw error]",
json.encode_keep_buffer, { nil, true },
false, { "bad argument #2 to '?' (found too many arguments)" } },
{ "Set encode_max_depth(\"wrong\") [throw error]",
json.encode_max_depth, { "wrong" },
false, { "bad argument #1 to '?' (number expected, got string)" } },
{ "Set decode_max_depth(0) [throw error]",
json.decode_max_depth, { "0" },
false, { "bad argument #1 to '?' (expected integer between 1 and 2147483647)" } },
{ "Set encode_invalid_numbers(-2) [throw error]",
json.encode_invalid_numbers, { -2 },
false, { "bad argument #1 to '?' (invalid option '-2')" } },
{ "Set decode_invalid_numbers(true, false) [throw error]",
json.decode_invalid_numbers, { true, false },
false, { "bad argument #2 to '?' (found too many arguments)" } },
{ "Set encode_sparse_array(\"not quite on\") [throw error]",
json.encode_sparse_array, { "not quite on" },
false, { "bad argument #1 to '?' (invalid option 'not quite on')" } },
{ "Reset Lua CJSON configuration", function () json = json.new() end },
-- Wrap in a function to ensure the table returned by json.new() is used
{ "Check encode_sparse_array()",
function (...) return json.encode_sparse_array(...) end, { },
true, { false, 2, 10 } },
{ "Encode (safe) simple value",
json_safe.encode, { true },
true, { "true" } },
{ "Encode (safe) argument validation [throw error]",
json_safe.encode, { "arg1", "arg2" },
false, { "bad argument #1 to '?' (expected 1 argument)" } },
{ "Decode (safe) error generation",
json_safe.decode, { "Oops" },
true, { nil, "Expected value but found invalid token at character 1" } },
{ "Decode (safe) error generation after new()",
function(...) return json_safe.new().decode(...) end, { "Oops" },
true, { nil, "Expected value but found invalid token at character 1" } },
}
print(("==> Testing Lua CJSON version %s\n"):format(json._VERSION))
util.run_test_group(cjson_tests)
for _, filename in ipairs(arg) do
util.run_test("Decode cycle " .. filename, test_decode_cycle, { filename },
true, { true })
end
local pass, total = util.run_test_summary()
if pass == total then
print("==> Summary: all tests succeeded")
else
print(("==> Summary: %d/%d tests failed"):format(total - pass, total))
os.exit(1)
end
-- vi:ai et sw=4 ts=4:
{ "array": [ 10, true, null ] }
...@@ -24,7 +24,6 @@ ...@@ -24,7 +24,6 @@
#define LUA_USE_MODULES_BIT #define LUA_USE_MODULES_BIT
//#define LUA_USE_MODULES_BMP085 //#define LUA_USE_MODULES_BMP085
//#define LUA_USE_MODULES_BME280 //#define LUA_USE_MODULES_BME280
//#define LUA_USE_MODULES_CJSON
//#define LUA_USE_MODULES_COAP //#define LUA_USE_MODULES_COAP
//#define LUA_USE_MODULES_CRON //#define LUA_USE_MODULES_CRON
//#define LUA_USE_MODULES_CRYPTO //#define LUA_USE_MODULES_CRYPTO
...@@ -54,6 +53,7 @@ ...@@ -54,6 +53,7 @@
//#define LUA_USE_MODULES_RTCMEM //#define LUA_USE_MODULES_RTCMEM
//#define LUA_USE_MODULES_RTCTIME //#define LUA_USE_MODULES_RTCTIME
//#define LUA_USE_MODULES_SIGMA_DELTA //#define LUA_USE_MODULES_SIGMA_DELTA
//#define LUA_USE_MODULES_SJSON
//#define LUA_USE_MODULES_SNTP //#define LUA_USE_MODULES_SNTP
//#define LUA_USE_MODULES_SOMFY //#define LUA_USE_MODULES_SOMFY
#define LUA_USE_MODULES_SPI #define LUA_USE_MODULES_SPI
......
...@@ -50,10 +50,10 @@ INCLUDES += -I ../pcm ...@@ -50,10 +50,10 @@ INCLUDES += -I ../pcm
INCLUDES += -I ../platform INCLUDES += -I ../platform
INCLUDES += -I ../spiffs INCLUDES += -I ../spiffs
INCLUDES += -I ../smart INCLUDES += -I ../smart
INCLUDES += -I ../cjson
INCLUDES += -I ../dhtlib INCLUDES += -I ../dhtlib
INCLUDES += -I ../fatfs INCLUDES += -I ../fatfs
INCLUDES += -I ../http INCLUDES += -I ../http
INCLUDES += -I ../sjson
INCLUDES += -I ../websocket INCLUDES += -I ../websocket
PDIR := ../$(PDIR) PDIR := ../$(PDIR)
sinclude $(PDIR)Makefile sinclude $(PDIR)Makefile
......
/* Lua CJSON - JSON support for Lua
*
* Copyright (c) 2010-2012 Mark Pulford <mark@kyne.com.au>
*
* Permission is hereby granted, free of charge, to any person obtaining
* a copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sublicense, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
/* Caveats:
* - JSON "null" values are represented as lightuserdata since Lua
* tables cannot contain "nil". Compare with cjson.null.
* - Invalid UTF-8 characters are not detected and will be passed
* untouched. If required, UTF-8 error checking should be done
* outside this library.
* - Javascript comments are not part of the JSON spec, and are not
* currently supported.
*
* Note: Decoding is slower than encoding. Lua spends significant
* time (30%) managing tables when parsing JSON since it is
* difficult to know object/array sizes ahead of time.
*/
// #include <assert.h>
#include "module.h"
#include "c_string.h"
#include "c_math.h"
#include "c_limits.h"
#include "lauxlib.h"
#include "flash_api.h"
#include "ctype.h"
#include "strbuf.h"
#include "cjson_mem.h"
#define FPCONV_G_FMT_BUFSIZE 32
#define fpconv_strtod c_strtod
#define fpconv_init() ((void)0)
#ifndef CJSON_MODNAME
#define CJSON_MODNAME "cjson"
#endif
#ifndef CJSON_VERSION
#define CJSON_VERSION "2.1devel"
#endif
/* Workaround for Solaris platforms missing isinf() */
#if !defined(isinf) && (defined(USE_INTERNAL_ISINF) || defined(MISSING_ISINF))
#define isinf(x) (!isnan(x) && isnan((x) - (x)))
#endif
#define DEFAULT_SPARSE_CONVERT 0
#define DEFAULT_SPARSE_RATIO 2
#define DEFAULT_SPARSE_SAFE 10
#define DEFAULT_ENCODE_MAX_DEPTH 1000
#define DEFAULT_DECODE_MAX_DEPTH 1000
#define DEFAULT_ENCODE_INVALID_NUMBERS 0
#define DEFAULT_DECODE_INVALID_NUMBERS 1
#define DEFAULT_ENCODE_KEEP_BUFFER 0
#define DEFAULT_ENCODE_NUMBER_PRECISION 14
#ifdef DISABLE_INVALID_NUMBERS
#undef DEFAULT_DECODE_INVALID_NUMBERS
#define DEFAULT_DECODE_INVALID_NUMBERS 0
#endif
typedef enum {
T_OBJ_BEGIN,
T_OBJ_END,
T_ARR_BEGIN,
T_ARR_END,
T_STRING,
T_NUMBER,
T_BOOLEAN,
T_NULL,
T_COLON,
T_COMMA,
T_END,
T_WHITESPACE,
T_ERROR,
T_UNKNOWN
} json_token_type_t;
#if 0
static const char *json_token_type_name[] = {
"T_OBJ_BEGIN",
"T_OBJ_END",
"T_ARR_BEGIN",
"T_ARR_END",
"T_STRING",
"T_NUMBER",
"T_BOOLEAN",
"T_NULL",
"T_COLON",
"T_COMMA",
"T_END",
"T_WHITESPACE",
"T_ERROR",
"T_UNKNOWN",
NULL
};
#endif
static const char json_token_type_name[14][16] ICACHE_STORE_ATTR ICACHE_RODATA_ATTR = {
{'T','_','O','B','J','_','B','E','G','I','N',0},
{'T','_','O','B','J','_','E','N','D',0},
{'T','_','A','R','R','_','B','E','G','I','N',0},
{'T','_','A','R','R','_','E','N','D',0},
{'T','_','S','T','R','I','N','G',0},
{'T','_','N','U','M','B','E','R',0},
{'T','_','B','O','O','L','E','A','N',0},
{'T','_','N','U','L','L',0},
{'T','_','C','O','L','O','N',0},
{'T','_','C','O','M','M','A',0},
{'T','_','E','N','D',0},
{'T','_','W','H','I','T','E','S','P','A','C','E',0},
{'T','_','E','R','R','O','R',0},
{'T','_','U','N','K','N','O','W','N',0}
};
typedef struct {
// json_token_type_t ch2token[256]; // 256*4 = 1024 byte
// char escape2char[256]; /* Decoding */
/* encode_buf is only allocated and used when
* encode_keep_buffer is set */
strbuf_t encode_buf;
int encode_sparse_convert;
int encode_sparse_ratio;
int encode_sparse_safe;
int encode_max_depth;
int encode_invalid_numbers; /* 2 => Encode as "null" */
int encode_number_precision;
int encode_keep_buffer;
int decode_invalid_numbers;
int decode_max_depth;
} json_config_t;
typedef struct {
const char *data;
const char *ptr;
strbuf_t *tmp; /* Temporary storage for strings */
json_config_t *cfg;
int current_depth;
} json_parse_t;
typedef struct {
json_token_type_t type;
int index;
union {
const char *string;
double number;
int boolean;
} value;
int string_len;
} json_token_t;
#if 0
static const char *char2escape[256] = {
"\\u0000", "\\u0001", "\\u0002", "\\u0003",
"\\u0004", "\\u0005", "\\u0006", "\\u0007",
"\\b", "\\t", "\\n", "\\u000b",
"\\f", "\\r", "\\u000e", "\\u000f",
"\\u0010", "\\u0011", "\\u0012", "\\u0013",
"\\u0014", "\\u0015", "\\u0016", "\\u0017",
"\\u0018", "\\u0019", "\\u001a", "\\u001b",
"\\u001c", "\\u001d", "\\u001e", "\\u001f",
NULL, NULL, "\\\"", NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, "\\/",
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, "\\\\", NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, "\\u007f",
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
};
#endif
/* ===== HELPER FUNCTION ===== */
static const char escape_array[36][8] ICACHE_STORE_ATTR ICACHE_RODATA_ATTR = {
{'\\','u','0','0','0','0','\0','\0'},
{'\\','u','0','0','0','1','\0','\0'},
{'\\','u','0','0','0','2','\0','\0'},
{'\\','u','0','0','0','3','\0','\0'},
{'\\','u','0','0','0','4','\0','\0'},
{'\\','u','0','0','0','5','\0','\0'},
{'\\','u','0','0','0','6','\0','\0'},
{'\\','u','0','0','0','7','\0','\0'},
{'\\','b','\0','\0','\0','\0','\0','\0'},
{'\\','t','\0','\0','\0','\0','\0','\0'},
{'\\','n','\0','\0','\0','\0','\0','\0'},
{'\\','u','0','0','0','b','\0','\0'},
{'\\','f','\0','\0','\0','\0','\0','\0'},
{'\\','r','\0','\0','\0','\0','\0','\0'},
{'\\','u','0','0','0','e','\0','\0'},
{'\\','u','0','0','0','f','\0','\0'},
{'\\','u','0','0','1','0','\0','\0'},
{'\\','u','0','0','1','1','\0','\0'},
{'\\','u','0','0','1','2','\0','\0'},
{'\\','u','0','0','1','3','\0','\0'},
{'\\','u','0','0','1','4','\0','\0'},
{'\\','u','0','0','1','5','\0','\0'},
{'\\','u','0','0','1','6','\0','\0'},
{'\\','u','0','0','1','7','\0','\0'},
{'\\','u','0','0','1','8','\0','\0'},
{'\\','u','0','0','1','9','\0','\0'},
{'\\','u','0','0','1','a','\0','\0'},
{'\\','u','0','0','1','b','\0','\0'},
{'\\','u','0','0','1','c','\0','\0'},
{'\\','u','0','0','1','d','\0','\0'},
{'\\','u','0','0','1','e','\0','\0'},
{'\\','u','0','0','1','f','\0','\0'},
{'\\','\"','\0','\0','\0','\0','\0','\0'},
{'\\','/','\0','\0','\0','\0','\0','\0'},
{'\\','\\','\0','\0','\0','\0','\0','\0'},
{'\\','u','0','0','7','f','\0','\0'}
};
static const char *char2escape(unsigned char c){
if(c<32) return escape_array[c];
switch(c){
case 34: return escape_array[32];
case 47: return escape_array[33];
case 92: return escape_array[34];
case 127: return escape_array[35];
default:
return NULL;
}
}
static json_token_type_t ch2token(unsigned char c){
switch(c){
case '{': return T_OBJ_BEGIN;
case '}': return T_OBJ_END;
case '[': return T_ARR_BEGIN;
case ']': return T_ARR_END;
case ',': return T_COMMA;
case ':': return T_COLON;
case '\0': return T_END;
case ' ': return T_WHITESPACE;
case '\t': return T_WHITESPACE;
case '\n': return T_WHITESPACE;
case '\r': return T_WHITESPACE;
/* Update characters that require further processing */
case 'f': case 'i': case 'I': case 'n': case 'N': case 't': case '"': case '+': case '-':
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9':
return T_UNKNOWN;
default:
return T_ERROR;
}
}
static char escape2char(unsigned char c){
switch(c){
case '"': return '"';
case '\\': return '\\';
case '/': return '/';
case 'b': return '\b';
case 't': return '\t';
case 'n': return '\n';
case 'f': return '\f';
case 'r': return '\r';
case 'u': return 'u';
default:
return 0;
}
}
/* ===== CONFIGURATION ===== */
#if 0
static json_config_t *json_fetch_config(lua_State *l)
{
json_config_t *cfg;
cfg = lua_touserdata(l, lua_upvalueindex(1));
if (!cfg)
luaL_error(l, "BUG: Unable to fetch CJSON configuration");
return cfg;
}
/* Ensure the correct number of arguments have been provided.
* Pad with nil to allow other functions to simply check arg[i]
* to find whether an argument was provided */
static json_config_t *json_arg_init(lua_State *l, int args)
{
luaL_argcheck(l, lua_gettop(l) <= args, args + 1,
"found too many arguments");
while (lua_gettop(l) < args)
lua_pushnil(l);
return json_fetch_config(l);
}
/* Process integer options for configuration functions */
static int json_integer_option(lua_State *l, int optindex, int *setting,
int min, int max)
{
char errmsg[64];
int value;
if (!lua_isnil(l, optindex)) {
value = luaL_checkinteger(l, optindex);
c_sprintf(errmsg, "expected integer between %d and %d", min, max);
luaL_argcheck(l, min <= value && value <= max, 1, errmsg);
*setting = value;
}
lua_pushinteger(l, *setting);
return 1;
}
/* Process enumerated arguments for a configuration function */
static int json_enum_option(lua_State *l, int optindex, int *setting,
const char **options, int bool_true)
{
static const char *bool_options[] = { "off", "on", NULL };
if (!options) {
options = bool_options;
bool_true = 1;
}
if (!lua_isnil(l, optindex)) {
if (bool_true && lua_isboolean(l, optindex))
*setting = lua_toboolean(l, optindex) * bool_true;
else
*setting = luaL_checkoption(l, optindex, NULL, options);
}
if (bool_true && (*setting == 0 || *setting == bool_true))
lua_pushboolean(l, *setting);
else
lua_pushstring(l, options[*setting]);
return 1;
}
/* Configures handling of extremely sparse arrays:
* convert: Convert extremely sparse arrays into objects? Otherwise error.
* ratio: 0: always allow sparse; 1: never allow sparse; >1: use ratio
* safe: Always use an array when the max index <= safe */
static int json_cfg_encode_sparse_array(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 3);
json_enum_option(l, 1, &cfg->encode_sparse_convert, NULL, 1);
json_integer_option(l, 2, &cfg->encode_sparse_ratio, 0, INT_MAX);
json_integer_option(l, 3, &cfg->encode_sparse_safe, 0, INT_MAX);
return 3;
}
/* Configures the maximum number of nested arrays/objects allowed when
* encoding */
static int json_cfg_encode_max_depth(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 1);
return json_integer_option(l, 1, &cfg->encode_max_depth, 1, INT_MAX);
}
/* Configures the maximum number of nested arrays/objects allowed when
* encoding */
static int json_cfg_decode_max_depth(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 1);
return json_integer_option(l, 1, &cfg->decode_max_depth, 1, INT_MAX);
}
/* Configures number precision when converting doubles to text */
static int json_cfg_encode_number_precision(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 1);
return json_integer_option(l, 1, &cfg->encode_number_precision, 1, 14);
}
/* Configures JSON encoding buffer persistence */
static int json_cfg_encode_keep_buffer(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 1);
int old_value;
old_value = cfg->encode_keep_buffer;
json_enum_option(l, 1, &cfg->encode_keep_buffer, NULL, 1);
/* Init / free the buffer if the setting has changed */
if (old_value ^ cfg->encode_keep_buffer) {
if (cfg->encode_keep_buffer){
if(-1==strbuf_init(&cfg->encode_buf, 0))
return luaL_error(l, "not enough memory");
}
else
strbuf_free(&cfg->encode_buf);
}
return 1;
}
#if defined(DISABLE_INVALID_NUMBERS) && !defined(USE_INTERNAL_FPCONV)
void json_verify_invalid_number_setting(lua_State *l, int *setting)
{
if (*setting == 1) {
*setting = 0;
luaL_error(l, "Infinity, NaN, and/or hexadecimal numbers are not supported.");
}
}
#else
#define json_verify_invalid_number_setting(l, s) do { } while(0)
#endif
static int json_cfg_encode_invalid_numbers(lua_State *l)
{
static const char *options[] = { "off", "on", "null", NULL };
json_config_t *cfg = json_arg_init(l, 1);
json_enum_option(l, 1, &cfg->encode_invalid_numbers, options, 1);
json_verify_invalid_number_setting(l, &cfg->encode_invalid_numbers);
return 1;
}
static int json_cfg_decode_invalid_numbers(lua_State *l)
{
json_config_t *cfg = json_arg_init(l, 1);
json_enum_option(l, 1, &cfg->decode_invalid_numbers, NULL, 1);
json_verify_invalid_number_setting(l, &cfg->encode_invalid_numbers);
return 1;
}
static int json_destroy_config(lua_State *l)
{
json_config_t *cfg;
cfg = lua_touserdata(l, 1);
if (cfg)
strbuf_free(&cfg->encode_buf);
cfg = NULL;
return 0;
}
static void json_create_config(lua_State *l)
{
json_config_t *cfg;
int i;
cfg = lua_newuserdata(l, sizeof(*cfg));
/* Create GC method to clean up strbuf */
lua_newtable(l);
lua_pushcfunction(l, json_destroy_config);
lua_setfield(l, -2, "__gc");
lua_setmetatable(l, -2);
cfg->encode_sparse_convert = DEFAULT_SPARSE_CONVERT;
cfg->encode_sparse_ratio = DEFAULT_SPARSE_RATIO;
cfg->encode_sparse_safe = DEFAULT_SPARSE_SAFE;
cfg->encode_max_depth = DEFAULT_ENCODE_MAX_DEPTH;
cfg->decode_max_depth = DEFAULT_DECODE_MAX_DEPTH;
cfg->encode_invalid_numbers = DEFAULT_ENCODE_INVALID_NUMBERS;
cfg->decode_invalid_numbers = DEFAULT_DECODE_INVALID_NUMBERS;
cfg->encode_keep_buffer = DEFAULT_ENCODE_KEEP_BUFFER;
cfg->encode_number_precision = DEFAULT_ENCODE_NUMBER_PRECISION;
#if DEFAULT_ENCODE_KEEP_BUFFER > 0
strbuf_init(&cfg->encode_buf, 0);
#endif
/* Decoding init */
/* Tag all characters as an error */
for (i = 0; i < 256; i++)
cfg->ch2token[i] = T_ERROR;
/* Set tokens that require no further processing */
cfg->ch2token['{'] = T_OBJ_BEGIN;
cfg->ch2token['}'] = T_OBJ_END;
cfg->ch2token['['] = T_ARR_BEGIN;
cfg->ch2token[']'] = T_ARR_END;
cfg->ch2token[','] = T_COMMA;
cfg->ch2token[':'] = T_COLON;
cfg->ch2token['\0'] = T_END;
cfg->ch2token[' '] = T_WHITESPACE;
cfg->ch2token['\t'] = T_WHITESPACE;
cfg->ch2token['\n'] = T_WHITESPACE;
cfg->ch2token['\r'] = T_WHITESPACE;
/* Update characters that require further processing */
cfg->ch2token['f'] = T_UNKNOWN; /* false? */
cfg->ch2token['i'] = T_UNKNOWN; /* inf, ininity? */
cfg->ch2token['I'] = T_UNKNOWN;
cfg->ch2token['n'] = T_UNKNOWN; /* null, nan? */
cfg->ch2token['N'] = T_UNKNOWN;
cfg->ch2token['t'] = T_UNKNOWN; /* true? */
cfg->ch2token['"'] = T_UNKNOWN; /* string? */
cfg->ch2token['+'] = T_UNKNOWN; /* number? */
cfg->ch2token['-'] = T_UNKNOWN;
for (i = 0; i < 10; i++)
cfg->ch2token['0' + i] = T_UNKNOWN;
/* Lookup table for parsing escape characters */
for (i = 0; i < 256; i++)
cfg->escape2char[i] = 0; /* String error */
cfg->escape2char['"'] = '"';
cfg->escape2char['\\'] = '\\';
cfg->escape2char['/'] = '/';
cfg->escape2char['b'] = '\b';
cfg->escape2char['t'] = '\t';
cfg->escape2char['n'] = '\n';
cfg->escape2char['f'] = '\f';
cfg->escape2char['r'] = '\r';
cfg->escape2char['u'] = 'u'; /* Unicode parsing required */
}
#endif
json_config_t _cfg;
static json_config_t *json_fetch_config(lua_State *l)
{
return &_cfg;
}
static int cfg_init(json_config_t *cfg){
cfg->encode_sparse_convert = DEFAULT_SPARSE_CONVERT;
cfg->encode_sparse_ratio = DEFAULT_SPARSE_RATIO;
cfg->encode_sparse_safe = DEFAULT_SPARSE_SAFE;
cfg->encode_max_depth = DEFAULT_ENCODE_MAX_DEPTH;
cfg->decode_max_depth = DEFAULT_DECODE_MAX_DEPTH;
cfg->encode_invalid_numbers = DEFAULT_ENCODE_INVALID_NUMBERS;
cfg->decode_invalid_numbers = DEFAULT_DECODE_INVALID_NUMBERS;
cfg->encode_keep_buffer = DEFAULT_ENCODE_KEEP_BUFFER;
cfg->encode_number_precision = DEFAULT_ENCODE_NUMBER_PRECISION;
#if DEFAULT_ENCODE_KEEP_BUFFER > 0
if(-1==strbuf_init(&cfg->encode_buf, 0)){
NODE_ERR("not enough memory\n");
return -1;
}
#endif
return 0;
}
/* ===== ENCODING ===== */
static void json_encode_exception(lua_State *l, json_config_t *cfg, strbuf_t *json, int lindex,
const char *reason)
{
if (!cfg->encode_keep_buffer)
strbuf_free(json);
luaL_error(l, "Cannot serialise %s: %s",
lua_typename(l, lua_type(l, lindex)), reason);
}
/* json_append_string args:
* - lua_State
* - JSON strbuf
* - String (Lua stack index)
*
* Returns nothing. Doesn't remove string from Lua stack */
static void json_append_string(lua_State *l, strbuf_t *json, int lindex)
{
const char *escstr;
int i;
const char *str;
size_t len;
str = lua_tolstring(l, lindex, &len);
/* Worst case is len * 6 (all unicode escapes).
* This buffer is reused constantly for small strings
* If there are any excess pages, they won't be hit anyway.
* This gains ~5% speedup. */
strbuf_ensure_empty_length(json, len * 6 + 2);
strbuf_append_char_unsafe(json, '\"');
for (i = 0; i < len; i++) {
escstr = char2escape((unsigned char)str[i]);
if (escstr){
int i;
char temp[8]; // for now, 8-bytes is enough.
for (i=0; i < 8; ++i)
{
temp[i] = byte_of_aligned_array(escstr, i);
if(temp[i]==0) break;
}
escstr = temp;
strbuf_append_string(json, escstr);
}
else
strbuf_append_char_unsafe(json, str[i]);
}
strbuf_append_char_unsafe(json, '\"');
}
/* Find the size of the array on the top of the Lua stack
* -1 object (not a pure array)
* >=0 elements in array
*/
static int lua_array_length(lua_State *l, json_config_t *cfg, strbuf_t *json)
{
double k;
int max;
int items;
max = 0;
items = 0;
lua_pushnil(l);
/* table, startkey */
while (lua_next(l, -2) != 0) {
/* table, key, value */
if (lua_type(l, -2) == LUA_TNUMBER &&
(k = lua_tonumber(l, -2))) {
/* Integer >= 1 ? */
if (floor(k) == k && k >= 1) {
if (k > max)
max = k;
items++;
lua_pop(l, 1);
continue;
}
}
/* Must not be an array (non integer key) */
lua_pop(l, 2);
return -1;
}
/* Encode excessively sparse arrays as objects (if enabled) */
if (cfg->encode_sparse_ratio > 0 &&
max > items * cfg->encode_sparse_ratio &&
max > cfg->encode_sparse_safe) {
if (!cfg->encode_sparse_convert)
json_encode_exception(l, cfg, json, -1, "excessively sparse array");
return -1;
}
return max;
}
static void json_check_encode_depth(lua_State *l, json_config_t *cfg,
int current_depth, strbuf_t *json)
{
/* Ensure there are enough slots free to traverse a table (key,
* value) and push a string for a potential error message.
*
* Unlike "decode", the key and value are still on the stack when
* lua_checkstack() is called. Hence an extra slot for luaL_error()
* below is required just in case the next check to lua_checkstack()
* fails.
*
* While this won't cause a crash due to the EXTRA_STACK reserve
* slots, it would still be an improper use of the API. */
if (current_depth <= cfg->encode_max_depth && lua_checkstack(l, 3))
return;
if (!cfg->encode_keep_buffer)
strbuf_free(json);
luaL_error(l, "Cannot serialise, excessive nesting (%d)",
current_depth);
}
static void json_append_data(lua_State *l, json_config_t *cfg,
int current_depth, strbuf_t *json);
/* json_append_array args:
* - lua_State
* - JSON strbuf
* - Size of passwd Lua array (top of stack) */
static void json_append_array(lua_State *l, json_config_t *cfg, int current_depth,
strbuf_t *json, int array_length)
{
int comma, i;
strbuf_append_char(json, '[');
comma = 0;
for (i = 1; i <= array_length; i++) {
if (comma)
strbuf_append_char(json, ',');
else
comma = 1;
lua_rawgeti(l, -1, i);
json_append_data(l, cfg, current_depth, json);
lua_pop(l, 1);
}
strbuf_append_char(json, ']');
}
static void json_append_number(lua_State *l, json_config_t *cfg,
strbuf_t *json, int lindex)
{
double num = lua_tonumber(l, lindex);
int len;
if (cfg->encode_invalid_numbers == 0) {
/* Prevent encoding invalid numbers */
if (isinf(num) || isnan(num))
json_encode_exception(l, cfg, json, lindex,
"must not be NaN or Infinity");
} else if (cfg->encode_invalid_numbers == 1) {
/* Encode NaN/Infinity separately to ensure Javascript compatible
* values are used. */
if (isnan(num)) {
strbuf_append_mem(json, "NaN", 3);
return;
}
if (isinf(num)) {
if (num < 0)
strbuf_append_mem(json, "-Infinity", 9);
else
strbuf_append_mem(json, "Infinity", 8);
return;
}
} else {
/* Encode invalid numbers as "null" */
if (isinf(num) || isnan(num)) {
strbuf_append_mem(json, "null", 4);
return;
}
}
strbuf_ensure_empty_length(json, FPCONV_G_FMT_BUFSIZE);
// len = fpconv_g_fmt(strbuf_empty_ptr(json), num, cfg->encode_number_precision);
c_sprintf(strbuf_empty_ptr(json), LUA_NUMBER_FMT, (LUA_NUMBER)num);
len = c_strlen(strbuf_empty_ptr(json));
strbuf_extend_length(json, len);
}
static void json_append_object(lua_State *l, json_config_t *cfg,
int current_depth, strbuf_t *json)
{
int comma, keytype;
/* Object */
strbuf_append_char(json, '{');
lua_pushnil(l);
/* table, startkey */
comma = 0;
while (lua_next(l, -2) != 0) {
if (comma)
strbuf_append_char(json, ',');
else
comma = 1;
/* table, key, value */
keytype = lua_type(l, -2);
if (keytype == LUA_TNUMBER) {
strbuf_append_char(json, '"');
json_append_number(l, cfg, json, -2);
strbuf_append_mem(json, "\":", 2);
} else if (keytype == LUA_TSTRING) {
json_append_string(l, json, -2);
strbuf_append_char(json, ':');
} else {
json_encode_exception(l, cfg, json, -2,
"table key must be a number or string");
/* never returns */
}
/* table, key, value */
json_append_data(l, cfg, current_depth, json);
lua_pop(l, 1);
/* table, key */
}
strbuf_append_char(json, '}');
}
/* Serialise Lua data into JSON string. */
static void json_append_data(lua_State *l, json_config_t *cfg,
int current_depth, strbuf_t *json)
{
int len;
switch (lua_type(l, -1)) {
case LUA_TSTRING:
json_append_string(l, json, -1);
break;
case LUA_TNUMBER:
json_append_number(l, cfg, json, -1);
break;
case LUA_TBOOLEAN:
if (lua_toboolean(l, -1))
strbuf_append_mem(json, "true", 4);
else
strbuf_append_mem(json, "false", 5);
break;
case LUA_TTABLE:
current_depth++;
json_check_encode_depth(l, cfg, current_depth, json);
len = lua_array_length(l, cfg, json);
if (len > 0)
json_append_array(l, cfg, current_depth, json, len);
else
json_append_object(l, cfg, current_depth, json);
break;
case LUA_TNIL:
strbuf_append_mem(json, "null", 4);
break;
case LUA_TLIGHTUSERDATA:
if (lua_touserdata(l, -1) == NULL) {
strbuf_append_mem(json, "null", 4);
break;
}
default:
/* Remaining types (LUA_TFUNCTION, LUA_TUSERDATA, LUA_TTHREAD,
* and LUA_TLIGHTUSERDATA) cannot be serialised */
json_encode_exception(l, cfg, json, -1, "type not supported");
/* never returns */
}
}
static int json_encode(lua_State *l)
{
json_config_t *cfg = json_fetch_config(l);
strbuf_t local_encode_buf;
strbuf_t *encode_buf;
char *json;
int len;
luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument");
if (!cfg->encode_keep_buffer) {
/* Use private buffer */
encode_buf = &local_encode_buf;
if(-1==strbuf_init(encode_buf, 0))
return luaL_error(l, "not enough memory");
} else {
/* Reuse existing buffer */
encode_buf = &cfg->encode_buf;
strbuf_reset(encode_buf);
}
json_append_data(l, cfg, 0, encode_buf);
json = strbuf_string(encode_buf, &len);
lua_pushlstring(l, json, len);
if (!cfg->encode_keep_buffer)
strbuf_free(encode_buf);
return 1;
}
/* ===== DECODING ===== */
static void json_process_value(lua_State *l, json_parse_t *json,
json_token_t *token);
static int hexdigit2int(char hex)
{
if ('0' <= hex && hex <= '9')
return hex - '0';
/* Force lowercase */
hex |= 0x20;
if ('a' <= hex && hex <= 'f')
return 10 + hex - 'a';
return -1;
}
static int decode_hex4(const char *hex)
{
int digit[4];
int i;
/* Convert ASCII hex digit to numeric digit
* Note: this returns an error for invalid hex digits, including
* NULL */
for (i = 0; i < 4; i++) {
digit[i] = hexdigit2int(hex[i]);
if (digit[i] < 0) {
return -1;
}
}
return (digit[0] << 12) +
(digit[1] << 8) +
(digit[2] << 4) +
digit[3];
}
/* Converts a Unicode codepoint to UTF-8.
* Returns UTF-8 string length, and up to 4 bytes in *utf8 */
static int codepoint_to_utf8(char *utf8, int codepoint)
{
/* 0xxxxxxx */
if (codepoint <= 0x7F) {
utf8[0] = codepoint;
return 1;
}
/* 110xxxxx 10xxxxxx */
if (codepoint <= 0x7FF) {
utf8[0] = (codepoint >> 6) | 0xC0;
utf8[1] = (codepoint & 0x3F) | 0x80;
return 2;
}
/* 1110xxxx 10xxxxxx 10xxxxxx */
if (codepoint <= 0xFFFF) {
utf8[0] = (codepoint >> 12) | 0xE0;
utf8[1] = ((codepoint >> 6) & 0x3F) | 0x80;
utf8[2] = (codepoint & 0x3F) | 0x80;
return 3;
}
/* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
if (codepoint <= 0x1FFFFF) {
utf8[0] = (codepoint >> 18) | 0xF0;
utf8[1] = ((codepoint >> 12) & 0x3F) | 0x80;
utf8[2] = ((codepoint >> 6) & 0x3F) | 0x80;
utf8[3] = (codepoint & 0x3F) | 0x80;
return 4;
}
return 0;
}
/* Called when index pointing to beginning of UTF-16 code escape: \uXXXX
* \u is guaranteed to exist, but the remaining hex characters may be
* missing.
* Translate to UTF-8 and append to temporary token string.
* Must advance index to the next character to be processed.
* Returns: 0 success
* -1 error
*/
static int json_append_unicode_escape(json_parse_t *json)
{
char utf8[4]; /* Surrogate pairs require 4 UTF-8 bytes */
int codepoint;
int surrogate_low;
int len;
int escape_len = 6;
/* Fetch UTF-16 code unit */
codepoint = decode_hex4(json->ptr + 2);
if (codepoint < 0)
return -1;
/* UTF-16 surrogate pairs take the following 2 byte form:
* 11011 x yyyyyyyyyy
* When x = 0: y is the high 10 bits of the codepoint
* x = 1: y is the low 10 bits of the codepoint
*
* Check for a surrogate pair (high or low) */
if ((codepoint & 0xF800) == 0xD800) {
/* Error if the 1st surrogate is not high */
if (codepoint & 0x400)
return -1;
/* Ensure the next code is a unicode escape */
if (*(json->ptr + escape_len) != '\\' ||
*(json->ptr + escape_len + 1) != 'u') {
return -1;
}
/* Fetch the next codepoint */
surrogate_low = decode_hex4(json->ptr + 2 + escape_len);
if (surrogate_low < 0)
return -1;
/* Error if the 2nd code is not a low surrogate */
if ((surrogate_low & 0xFC00) != 0xDC00)
return -1;
/* Calculate Unicode codepoint */
codepoint = (codepoint & 0x3FF) << 10;
surrogate_low &= 0x3FF;
codepoint = (codepoint | surrogate_low) + 0x10000;
escape_len = 12;
}
/* Convert codepoint to UTF-8 */
len = codepoint_to_utf8(utf8, codepoint);
if (!len)
return -1;
/* Append bytes and advance parse index */
strbuf_append_mem_unsafe(json->tmp, utf8, len);
json->ptr += escape_len;
return 0;
}
static void json_set_token_error(json_token_t *token, json_parse_t *json,
const char *errtype)
{
token->type = T_ERROR;
token->index = json->ptr - json->data;
token->value.string = errtype;
}
static void json_next_string_token(json_parse_t *json, json_token_t *token)
{
// char *escape2char = json->cfg->escape2char;
char ch;
/* Caller must ensure a string is next */
if(!(*json->ptr == '"')) return;
/* Skip " */
json->ptr++;
/* json->tmp is the temporary strbuf used to accumulate the
* decoded string value.
* json->tmp is sized to handle JSON containing only a string value.
*/
strbuf_reset(json->tmp);
while ((ch = *json->ptr) != '"') {
if (!ch) {
/* Premature end of the string */
json_set_token_error(token, json, "unexpected end of string");
return;
}
/* Handle escapes */
if (ch == '\\') {
/* Fetch escape character */
ch = *(json->ptr + 1);
/* Translate escape code and append to tmp string */
ch = escape2char((unsigned char)ch);
if (ch == 'u') {
if (json_append_unicode_escape(json) == 0)
continue;
json_set_token_error(token, json,
"invalid unicode escape code");
return;
}
if (!ch) {
json_set_token_error(token, json, "invalid escape code");
return;
}
/* Skip '\' */
json->ptr++;
}
/* Append normal character or translated single character
* Unicode escapes are handled above */
strbuf_append_char_unsafe(json->tmp, ch);
json->ptr++;
}
json->ptr++; /* Eat final quote (") */
strbuf_ensure_null(json->tmp);
token->type = T_STRING;
token->value.string = strbuf_string(json->tmp, &token->string_len);
}
/* JSON numbers should take the following form:
* -?(0|[1-9]|[1-9][0-9]+)(.[0-9]+)?([eE][-+]?[0-9]+)?
*
* json_next_number_token() uses strtod() which allows other forms:
* - numbers starting with '+'
* - NaN, -NaN, infinity, -infinity
* - hexadecimal numbers
* - numbers with leading zeros
*
* json_is_invalid_number() detects "numbers" which may pass strtod()'s
* error checking, but should not be allowed with strict JSON.
*
* json_is_invalid_number() may pass numbers which cause strtod()
* to generate an error.
*/
static int json_is_invalid_number(json_parse_t *json)
{
const char *p = json->ptr;
/* Reject numbers starting with + */
if (*p == '+')
return 1;
/* Skip minus sign if it exists */
if (*p == '-')
p++;
/* Reject numbers starting with 0x, or leading zeros */
if (*p == '0') {
int ch2 = *(p + 1);
if ((ch2 | 0x20) == 'x' || /* Hex */
('0' <= ch2 && ch2 <= '9')) /* Leading zero */
return 1;
return 0;
} else if (*p <= '9') {
return 0; /* Ordinary number */
}
char tmp[4]; // conv to lower. because c_strncasecmp == c_strcmp
int i;
for (i = 0; i < 3; ++i)
{
if(p[i]!=0)
tmp[i] = tolower(p[i]);
else
tmp[i] = 0;
}
tmp[3] = 0;
/* Reject inf/nan */
if (!c_strncasecmp(tmp, "inf", 3))
return 1;
if (!c_strncasecmp(tmp, "nan", 3))
return 1;
/* Pass all other numbers which may still be invalid, but
* strtod() will catch them. */
return 0;
}
static void json_next_number_token(json_parse_t *json, json_token_t *token)
{
char *endptr;
token->type = T_NUMBER;
token->value.number = fpconv_strtod(json->ptr, &endptr);
if (json->ptr == endptr)
json_set_token_error(token, json, "invalid number");
else
json->ptr = endptr; /* Skip the processed number */
return;
}
/* Fills in the token struct.
* T_STRING will return a pointer to the json_parse_t temporary string
* T_ERROR will leave the json->ptr pointer at the error.
*/
static void json_next_token(json_parse_t *json, json_token_t *token)
{
// const json_token_type_t *ch2token = json->cfg->ch2token;
int ch;
/* Eat whitespace. */
while (1) {
ch = (unsigned char)*(json->ptr);
token->type = ch2token(ch);
if (token->type != T_WHITESPACE)
break;
json->ptr++;
}
/* Store location of new token. Required when throwing errors
* for unexpected tokens (syntax errors). */
token->index = json->ptr - json->data;
/* Don't advance the pointer for an error or the end */
if (token->type == T_ERROR) {
json_set_token_error(token, json, "invalid token");
return;
}
if (token->type == T_END) {
return;
}
/* Found a known single character token, advance index and return */
if (token->type != T_UNKNOWN) {
json->ptr++;
return;
}
/* Process characters which triggered T_UNKNOWN
*
* Must use strncmp() to match the front of the JSON string.
* JSON identifier must be lowercase.
* When strict_numbers if disabled, either case is allowed for
* Infinity/NaN (since we are no longer following the spec..) */
if (ch == '"') {
json_next_string_token(json, token);
return;
} else if (ch == '-' || ('0' <= ch && ch <= '9')) {
if (!json->cfg->decode_invalid_numbers && json_is_invalid_number(json)) {
json_set_token_error(token, json, "invalid number");
return;
}
json_next_number_token(json, token);
return;
} else if (!c_strncmp(json->ptr, "true", 4)) {
token->type = T_BOOLEAN;
token->value.boolean = 1;
json->ptr += 4;
return;
} else if (!c_strncmp(json->ptr, "false", 5)) {
token->type = T_BOOLEAN;
token->value.boolean = 0;
json->ptr += 5;
return;
} else if (!c_strncmp(json->ptr, "null", 4)) {
token->type = T_NULL;
json->ptr += 4;
return;
} else if (json->cfg->decode_invalid_numbers &&
json_is_invalid_number(json)) {
/* When decode_invalid_numbers is enabled, only attempt to process
* numbers we know are invalid JSON (Inf, NaN, hex)
* This is required to generate an appropriate token error,
* otherwise all bad tokens will register as "invalid number"
*/
json_next_number_token(json, token);
return;
}
/* Token starts with t/f/n but isn't recognised above. */
json_set_token_error(token, json, "invalid token");
}
/* This function does not return.
* DO NOT CALL WITH DYNAMIC MEMORY ALLOCATED.
* The only supported exception is the temporary parser string
* json->tmp struct.
* json and token should exist on the stack somewhere.
* luaL_error() will long_jmp and release the stack */
static void json_throw_parse_error(lua_State *l, json_parse_t *json,
const char *exp, json_token_t *token)
{
const char *found;
char temp[16]; // for now, 16-bytes is enough.
strbuf_free(json->tmp);
if (token->type == T_ERROR)
found = token->value.string;
else
{
found = json_token_type_name[token->type];
int i;
for (i=0; i < 16; ++i)
{
temp[i] = byte_of_aligned_array(found, i);
if(temp[i]==0) break;
}
found = temp;
}
/* Note: token->index is 0 based, display starting from 1 */
luaL_error(l, "Expected %s but found %s at character %d",
exp, found, token->index + 1);
}
static inline void json_decode_ascend(json_parse_t *json)
{
json->current_depth--;
}
static void json_decode_descend(lua_State *l, json_parse_t *json, int slots)
{
json->current_depth++;
if (json->current_depth <= json->cfg->decode_max_depth &&
lua_checkstack(l, slots)) {
return;
}
strbuf_free(json->tmp);
luaL_error(l, "Found too many nested data structures (%d) at character %d",
json->current_depth, json->ptr - json->data);
}
static void json_parse_object_context(lua_State *l, json_parse_t *json)
{
json_token_t token;
/* 3 slots required:
* .., table, key, value */
json_decode_descend(l, json, 3);
lua_newtable(l);
json_next_token(json, &token);
/* Handle empty objects */
if (token.type == T_OBJ_END) {
json_decode_ascend(json);
return;
}
while (1) {
if (token.type != T_STRING)
json_throw_parse_error(l, json, "object key string", &token);
/* Push key */
lua_pushlstring(l, token.value.string, token.string_len);
json_next_token(json, &token);
if (token.type != T_COLON)
json_throw_parse_error(l, json, "colon", &token);
/* Fetch value */
json_next_token(json, &token);
json_process_value(l, json, &token);
/* Set key = value */
lua_rawset(l, -3);
json_next_token(json, &token);
if (token.type == T_OBJ_END) {
json_decode_ascend(json);
return;
}
if (token.type != T_COMMA)
json_throw_parse_error(l, json, "comma or object end", &token);
json_next_token(json, &token);
}
}
/* Handle the array context */
static void json_parse_array_context(lua_State *l, json_parse_t *json)
{
json_token_t token;
int i;
/* 2 slots required:
* .., table, value */
json_decode_descend(l, json, 2);
lua_newtable(l);
json_next_token(json, &token);
/* Handle empty arrays */
if (token.type == T_ARR_END) {
json_decode_ascend(json);
return;
}
for (i = 1; ; i++) {
json_process_value(l, json, &token);
lua_rawseti(l, -2, i); /* arr[i] = value */
json_next_token(json, &token);
if (token.type == T_ARR_END) {
json_decode_ascend(json);
return;
}
if (token.type != T_COMMA)
json_throw_parse_error(l, json, "comma or array end", &token);
json_next_token(json, &token);
}
}
/* Handle the "value" context */
static void json_process_value(lua_State *l, json_parse_t *json,
json_token_t *token)
{
switch (token->type) {
case T_STRING:
lua_pushlstring(l, token->value.string, token->string_len);
break;;
case T_NUMBER:
lua_pushnumber(l, token->value.number);
break;;
case T_BOOLEAN:
lua_pushboolean(l, token->value.boolean);
break;;
case T_OBJ_BEGIN:
json_parse_object_context(l, json);
break;;
case T_ARR_BEGIN:
json_parse_array_context(l, json);
break;;
case T_NULL:
/* In Lua, setting "t[k] = nil" will delete k from the table.
* Hence a NULL pointer lightuserdata object is used instead */
lua_pushlightuserdata(l, NULL);
break;;
default:
json_throw_parse_error(l, json, "value", token);
}
}
static int json_decode(lua_State *l)
{
json_parse_t json;
json_token_t token;
size_t json_len;
luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument");
json.cfg = json_fetch_config(l);
json.data = luaL_checklstring(l, 1, &json_len);
json.current_depth = 0;
json.ptr = json.data;
/* Detect Unicode other than UTF-8 (see RFC 4627, Sec 3)
*
* CJSON can support any simple data type, hence only the first
* character is guaranteed to be ASCII (at worst: '"'). This is
* still enough to detect whether the wrong encoding is in use. */
if (json_len >= 2 && (!json.data[0] || !json.data[1]))
luaL_error(l, "JSON parser does not support UTF-16 or UTF-32");
/* Ensure the temporary buffer can hold the entire string.
* This means we no longer need to do length checks since the decoded
* string must be smaller than the entire json string */
json.tmp = strbuf_new(json_len);
if(json.tmp == NULL){
return luaL_error(l, "not enough memory");
}
json_next_token(&json, &token);
json_process_value(l, &json, &token);
/* Ensure there is no more input left */
json_next_token(&json, &token);
if (token.type != T_END)
json_throw_parse_error(l, &json, "the end", &token);
strbuf_free(json.tmp);
return 1;
}
/* ===== INITIALISATION ===== */
#if 0
#if !defined(LUA_VERSION_NUM) || LUA_VERSION_NUM < 502
/* Compatibility for Lua 5.1.
*
* luaL_setfuncs() is used to create a module table where the functions have
* json_config_t as their first upvalue. Code borrowed from Lua 5.2 source. */
static void luaL_setfuncs (lua_State *l, const luaL_Reg *reg, int nup)
{
int i;
luaL_checkstack(l, nup, "too many upvalues");
for (; reg->name != NULL; reg++) { /* fill the table with given functions */
for (i = 0; i < nup; i++) /* copy upvalues to the top */
lua_pushvalue(l, -nup);
lua_pushcclosure(l, reg->func, nup); /* closure with those upvalues */
lua_setfield(l, -(nup + 2), reg->name);
}
lua_pop(l, nup); /* remove upvalues */
}
#endif
/* Call target function in protected mode with all supplied args.
* Assumes target function only returns a single non-nil value.
* Convert and return thrown errors as: nil, "error message" */
static int json_protect_conversion(lua_State *l)
{
int err;
/* Deliberately throw an error for invalid arguments */
luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument");
/* pcall() the function stored as upvalue(1) */
lua_pushvalue(l, lua_upvalueindex(1));
lua_insert(l, 1);
err = lua_pcall(l, 1, 1, 0);
if (!err)
return 1;
if (err == LUA_ERRRUN) {
lua_pushnil(l);
lua_insert(l, -2);
return 2;
}
/* Since we are not using a custom error handler, the only remaining
* errors are memory related */
return luaL_error(l, "Memory allocation error in CJSON protected call");
}
/* Return cjson module table */
static int lua_cjson_new(lua_State *l)
{
/* Initialise number conversions */
fpconv_init();
/* cjson module table */
lua_newtable(l);
/* Register functions with config data as upvalue */
json_create_config(l);
luaL_setfuncs(l, reg, 1);
/* Set cjson.null */
lua_pushlightuserdata(l, NULL);
lua_setfield(l, -2, "null");
/* Set module name / version fields */
lua_pushliteral(l, CJSON_MODNAME);
lua_setfield(l, -2, "_NAME");
lua_pushliteral(l, CJSON_VERSION);
lua_setfield(l, -2, "_VERSION");
return 1;
}
/* Return cjson.safe module table */
static int lua_cjson_safe_new(lua_State *l)
{
const char *func[] = { "decode", "encode", NULL };
int i;
lua_cjson_new(l);
/* Fix new() method */
lua_pushcfunction(l, lua_cjson_safe_new);
lua_setfield(l, -2, "new");
for (i = 0; func[i]; i++) {
lua_getfield(l, -1, func[i]);
lua_pushcclosure(l, json_protect_conversion, 1);
lua_setfield(l, -2, func[i]);
}
return 1;
}
int luaopen_cjson(lua_State *l)
{
lua_cjson_new(l);
#ifdef ENABLE_CJSON_GLOBAL
/* Register a global "cjson" table. */
lua_pushvalue(l, -1);
lua_setglobal(l, CJSON_MODNAME);
#endif
/* Return cjson table */
return 1;
}
int luaopen_cjson_safe(lua_State *l)
{
lua_cjson_safe_new(l);
/* Return cjson.safe table */
return 1;
}
#endif
// Module function map
static const LUA_REG_TYPE cjson_map[] = {
{ LSTRKEY( "encode" ), LFUNCVAL( json_encode ) },
{ LSTRKEY( "decode" ), LFUNCVAL( json_decode ) },
//{ LSTRKEY( "encode_sparse_array" ), LFUNCVAL( json_cfg_encode_sparse_array ) },
//{ LSTRKEY( "encode_max_depth" ), LFUNCVAL( json_cfg_encode_max_depth ) },
//{ LSTRKEY( "decode_max_depth" ), LFUNCVAL( json_cfg_decode_max_depth ) },
//{ LSTRKEY( "encode_number_precision" ), LFUNCVAL( json_cfg_encode_number_precision ) },
//{ LSTRKEY( "encode_keep_buffer" ), LFUNCVAL( json_cfg_encode_keep_buffer ) },
//{ LSTRKEY( "encode_invalid_numbers" ), LFUNCVAL( json_cfg_encode_invalid_numbers ) },
//{ LSTRKEY( "decode_invalid_numbers" ), LFUNCVAL( json_cfg_decode_invalid_numbers ) },
//{ LSTRKEY( "new" ), LFUNCVAL( lua_cjson_new ) },
{ LNILKEY, LNILVAL }
};
int luaopen_cjson( lua_State *L )
{
/* Initialise number conversions */
// fpconv_init(); // not needed for a specific cpu.
if(-1==cfg_init(&_cfg)){
return luaL_error(L, "BUG: Unable to init config for cjson");;
}
return 0;
}
NODEMCU_MODULE(CJSON, "cjson", cjson_map, luaopen_cjson);
/* vi:ai et sw=4 ts=4:
*/
#define LUA_LIB
#include "lua.h"
#include "lauxlib.h"
#include "lstring.h"
#ifndef LOCAL_LUA
#include "module.h"
#include "c_string.h"
#include "c_math.h"
#include "c_limits.h"
#endif
#define JSONSL_STATE_USER_FIELDS int lua_object_ref; int used_count;
#define JSONSL_NO_JPR
#include "jsonsl.c"
#define LUA_SJSONLIBNAME "sjson"
#define DEFAULT_DEPTH 20
#define DBG_PRINTF(...)
typedef struct {
jsonsl_t jsn;
int result_ref;
int hkey_ref;
int null_ref;
int metatable;
int pos_ref;
uint8_t complete;
const char *error;
lua_State *L;
size_t min_needed;
size_t min_available;
size_t buffer_len;
const char *buffer; // Points into buffer_ref
int buffer_ref;
} JSN_DATA;
#define get_parent_object_ref() ((state->level == 1) ? data->result_ref : state[-1].lua_object_ref)
#define get_parent_object_used_count_pre_inc() ((state->level == 1) ? 1 : ++state[-1].used_count)
static const char* get_state_buffer(JSN_DATA *ctx, struct jsonsl_state_st *state)
{
size_t offset = state->pos_begin - ctx->min_available;
return ctx->buffer + offset;
}
// The elem data is a ref
static int error_callback(jsonsl_t jsn,
jsonsl_error_t err,
struct jsonsl_state_st *state,
char *at)
{
JSN_DATA *data = (JSN_DATA *) jsn->data;
data->error = jsonsl_strerror(err);
//fprintf(stderr, "Got error at pos %lu: %s\n", jsn->pos, jsonsl_strerror(err));
return 0;
}
static void
create_table(JSN_DATA *data) {
lua_newtable(data->L);
if (data->metatable != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->metatable);
lua_setmetatable(data->L, -2);
}
}
static void
create_new_element(jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st *state,
const char *buf)
{
JSN_DATA *data = jsn->data;
DBG_PRINTF("L%d: new action %d @ %d state->type %s\n", state->level, action, state->pos_begin, jsonsl_strtype(state->type));
DBG_PRINTF("buf: '%s' ('%.10s')\n", buf, get_state_buffer(data, state));
state->lua_object_ref = LUA_NOREF;
switch(state->type) {
case JSONSL_T_SPECIAL:
case JSONSL_T_STRING:
case JSONSL_T_HKEY:
break;
case JSONSL_T_LIST:
case JSONSL_T_OBJECT:
create_table(data);
state->lua_object_ref = lua_ref(data->L, 1);
state->used_count = 0;
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
DBG_PRINTF("Adding array element\n");
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
DBG_PRINTF("Adding hash element\n");
}
if (data->pos_ref != LUA_NOREF && state->level > 1) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_pushnumber(data->L, state->level - 1);
lua_pushvalue(data->L, -3); // get the key
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
// At this point, the stack:
// top: index/hash key
// : table
int want_value = 1;
// Invoke the checkpath method if possible
if (data->pos_ref != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->metatable);
lua_getfield(data->L, -1, "checkpath");
if (lua_type(data->L, -1) != LUA_TNIL) {
// Call with the new table and the path as arguments
lua_rawgeti(data->L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_call(data->L, 2, 1);
want_value = lua_toboolean(data->L, -1);
}
lua_pop(data->L, 2); // Discard the metatable and either the getfield result or retval
}
if (want_value) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_settable(data->L, -3);
lua_pop(data->L, 1); // the table
} else {
lua_pop(data->L, 2); // the index and table
}
break;
default:
DBG_PRINTF("Unhandled type %c\n", state->type);
luaL_error(data->L, "Unhandled type");
break;
}
data->min_needed = state->pos_begin;
}
static void push_number(JSN_DATA *data, struct jsonsl_state_st *state) {
lua_pushlstring(data->L, get_state_buffer(data, state), state->pos_cur - state->pos_begin);
LUA_NUMBER r = lua_tonumber(data->L, -1);
lua_pop(data->L, 1);
lua_pushnumber(data->L, r);
}
static int fromhex(char c) {
if (c <= '9') {
return c & 0xf;
}
return ((c - 'A' + 10) & 0xf);
}
static void output_utf8(luaL_Buffer *buf, int c) {
char space[4];
char *b = space;
if (c<0x80) *b++=c;
else if (c<0x800) *b++=192+c/64, *b++=128+c%64;
else if (c-0xd800u<0x800) *b++ = '?';
else if (c<0x10000) *b++=224+c/4096, *b++=128+c/64%64, *b++=128+c%64;
else if (c<0x110000) *b++=240+c/262144, *b++=128+c/4096%64, *b++=128+c/64%64, *b++=128+c%64;
else *b++ = '?';
luaL_addlstring(buf, space, b - space);
}
static void push_string(JSN_DATA *data, struct jsonsl_state_st *state) {
luaL_Buffer b;
luaL_buffinit(data->L, &b);
int i;
const char *c = get_state_buffer(data, state) + 1;
for (i = 0; i < state->pos_cur - state->pos_begin - 1; i++) {
int nc = c[i];
if (nc == '\\') {
i++;
nc = c[i] & 255;
switch (c[i]) {
case 'b':
nc = '\b';
break;
case 'f':
nc = '\f';
break;
case 'n':
nc = '\n';
break;
case 'r':
nc = '\r';
break;
case 't':
nc = '\t';
break;
case 'u':
nc = fromhex(c[++i]) << 12;
nc += fromhex(c[++i]) << 8;
nc += fromhex(c[++i]) << 4;
nc += fromhex(c[++i]) ;
output_utf8(&b, nc);
continue;
}
}
luaL_putchar(&b, nc);
}
luaL_pushresult(&b);
}
static void
cleanup_closing_element(jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st *state,
const char *at)
{
JSN_DATA *data = (JSN_DATA *) jsn->data;
DBG_PRINTF( "L%d: cc action %d state->type %s\n", state->level, action, jsonsl_strtype(state->type));
DBG_PRINTF( "buf (%d - %d): '%.*s'\n", state->pos_begin, state->pos_cur, state->pos_cur - state->pos_begin, get_state_buffer(data, state));
DBG_PRINTF( "at: '%s'\n", at);
switch (state->type) {
case JSONSL_T_HKEY:
push_string(data, state);
data->hkey_ref = lua_ref(data->L, 1);
break;
case JSONSL_T_STRING:
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
}
push_string(data, state);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
break;
case JSONSL_T_SPECIAL:
DBG_PRINTF("Special flags = 0x%x\n", state->special_flags);
// need to deal with true/false/null
if (state->special_flags & (JSONSL_SPECIALf_TRUE|JSONSL_SPECIALf_FALSE|JSONSL_SPECIALf_NUMERIC|JSONSL_SPECIALf_NULL)) {
if (state->special_flags & JSONSL_SPECIALf_TRUE) {
lua_pushboolean(data->L, 1);
} else if (state->special_flags & JSONSL_SPECIALf_FALSE) {
lua_pushboolean(data->L, 0);
} else if (state->special_flags & JSONSL_SPECIALf_NULL) {
DBG_PRINTF("Outputting null\n");
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->null_ref);
} else if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
push_number(data, state);
}
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
}
lua_pushvalue(data->L, -3);
lua_remove(data->L, -4);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
break;
case JSONSL_T_OBJECT:
case JSONSL_T_LIST:
lua_unref(data->L, state->lua_object_ref);
state->lua_object_ref = LUA_NOREF;
if (data->pos_ref != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_pushnumber(data->L, state->level);
lua_pushnil(data->L);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
if (state->level == 1) {
data->complete = 1;
}
break;
}
}
static int sjson_decoder_int(lua_State *L, int argno) {
int nlevels = DEFAULT_DEPTH;
if (lua_type(L, argno) == LUA_TTABLE) {
lua_getfield(L, argno, "depth");
nlevels = lua_tointeger(L, argno);
if (nlevels == 0) {
nlevels = DEFAULT_DEPTH;
}
if (nlevels < 4) {
nlevels = 4;
}
if (nlevels > 1000) {
nlevels = 1000;
}
lua_pop(L, 1);
}
JSN_DATA *data = (JSN_DATA *) lua_newuserdata(L, sizeof(JSN_DATA) + jsonsl_get_size(nlevels));
//
// Associate its metatable
luaL_getmetatable(L, "sjson.decoder");
lua_setmetatable(L, -2);
jsonsl_t jsn = jsonsl_init((jsonsl_t) (data + 1), nlevels);
int i;
for (i = 0; i < jsn->levels_max; i++) {
jsn->stack[i].lua_object_ref = LUA_NOREF;
}
data->jsn = jsn;
data->result_ref = LUA_NOREF;
data->null_ref = LUA_REFNIL;
data->metatable = LUA_NOREF;
data->hkey_ref = LUA_NOREF;
data->pos_ref = LUA_NOREF;
data->buffer_ref = LUA_NOREF;
data->complete = 0;
data->error = NULL;
data->L = L;
data->buffer_len = 0;
data->min_needed = data->min_available = jsn->pos;
lua_pushlightuserdata(L, 0);
data->null_ref = lua_ref(L, 1);
// This may throw...
lua_newtable(L);
data->result_ref = luaL_ref(L, LUA_REGISTRYINDEX);
if (lua_type(L, argno) == LUA_TTABLE) {
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
lua_getfield(L, argno, "null");
data->null_ref = lua_ref(L, 1);
lua_getfield(L, argno, "metatable");
lua_pushvalue(L, -1);
data->metatable = lua_ref(L, 1);
if (lua_type(L, -1) != LUA_TNIL) {
lua_getfield(L, -1, "checkpath");
if (lua_type(L, -1) != LUA_TNIL) {
lua_newtable(L);
data->pos_ref = lua_ref(L, 1);
}
lua_pop(L, 1); // Throw away the checkpath value
}
lua_pop(L, 1); // Throw away the metatable
}
jsonsl_enable_all_callbacks(data->jsn);
jsn->action_callback = NULL;
jsn->action_callback_PUSH = create_new_element;
jsn->action_callback_POP = cleanup_closing_element;
jsn->error_callback = error_callback;
jsn->data = data;
jsn->max_callback_level = nlevels;
return 1;
}
static int sjson_decoder(lua_State *L) {
return sjson_decoder_int(L, 1);
}
static int sjson_decoder_result_int(lua_State *L, JSN_DATA *data) {
if (!data->complete) {
luaL_error(L, "decode not complete");
}
lua_rawgeti(L, LUA_REGISTRYINDEX, data->result_ref);
lua_rawgeti(L, -1, 1);
lua_remove(L, -2);
return 1;
}
static int sjson_decoder_result(lua_State *L) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, 1, "sjson.decoder");
return sjson_decoder_result_int(L, data);
}
static void sjson_free_working_data(lua_State *L, JSN_DATA *data) {
jsonsl_t jsn = data->jsn;
int i;
for (i = 0; i < jsn->levels_max; i++) {
luaL_unref(L, LUA_REGISTRYINDEX, jsn->stack[i].lua_object_ref);
jsn->stack[i].lua_object_ref = LUA_NOREF;
}
luaL_unref(L, LUA_REGISTRYINDEX, data->metatable);
data->metatable = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->pos_ref);
data->pos_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->buffer_ref);
data->buffer_ref = LUA_NOREF;
}
static int sjson_decoder_write_int(lua_State *L, int udata_pos, int string_pos) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, udata_pos, "sjson.decoder");
size_t len;
const char *str = luaL_checklstring(L, string_pos, &len);
if (data->error) {
luaL_error(L, "JSON parse error: previous call");
}
if (!data->complete) {
data->L = L;
// Merge into any existing buffer and deal with discard
if (data->buffer_ref != LUA_NOREF) {
luaL_Buffer b;
luaL_buffinit(L, &b);
lua_rawgeti(L, LUA_REGISTRYINDEX, data->buffer_ref);
size_t prev_len;
const char *prev_buffer = luaL_checklstring(L, -1, &prev_len);
lua_pop(L, 1); // But string still referenced so it cannot move
int discard = data->min_needed - data->min_available;
prev_buffer += discard;
prev_len -= discard;
if (prev_len > 0) {
luaL_addlstring(&b, prev_buffer, prev_len);
}
data->min_available += discard;
luaL_unref(L, LUA_REGISTRYINDEX, data->buffer_ref);
data->buffer_ref = LUA_NOREF;
lua_pushvalue(L, string_pos);
luaL_addvalue(&b);
luaL_pushresult(&b);
} else {
lua_pushvalue(L, string_pos);
}
size_t blen;
data->buffer = luaL_checklstring(L, -1, &blen);
data->buffer_len = blen;
data->buffer_ref = lua_ref(L, 1);
jsonsl_feed(data->jsn, str, len);
if (data->error) {
luaL_error(L, "JSON parse error: %s", data->error);
}
}
if (data->complete) {
// We no longer need the buffer
sjson_free_working_data(L, data);
return sjson_decoder_result_int(L, data);
}
return 0;
}
static int sjson_decoder_write(lua_State *L) {
return sjson_decoder_write_int(L, 1, 2);
}
static int sjson_decode(lua_State *L) {
int push_count = sjson_decoder_int(L, 2);
if (push_count != 1) {
luaL_error(L, "Internal error in sjson.deocder");
}
luaL_checkudata(L, -1, "sjson.decoder");
push_count = sjson_decoder_write_int(L, -1, 1);
if (push_count != 1) {
luaL_error(L, "Incomplete JSON object passed to sjson.decode");
}
// Now we have two items on the stack -- the udata and the result
lua_remove(L, -2);
return 1;
}
static int sjson_decoder_destructor(lua_State *L) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, 1, "sjson.decoder");
sjson_free_working_data(L, data);
data->jsn = NULL;
luaL_unref(L, LUA_REGISTRYINDEX, data->result_ref);
data->result_ref = LUA_NOREF;
DBG_PRINTF("Destructor called\n");
return 0;
}
//
//--------------------------------- ENCODER BELOW
//
//
//
//#undef DBG_PRINTF
//#define DBG_PRINTF printf
typedef struct {
int lua_object_ref;
// for arrays
// 0 -> [
// 1 -> first element
// 2 -> ,
// 3 -> second element
// 4 -> ]
// for objects
// 0 -> { firstkey :
// 1 -> first value
// 2 -> , secondkey :
// 3 -> second value
// 4 -> }
short offset;
// -1 for objects
// 0 -> n maximum integer key = n
short size;
int lua_key_ref;
} ENC_DATA_STATE;
typedef struct {
ENC_DATA_STATE *stack;
int nlevels;
int level;
int current_str_ref;
int null_ref;
int offset;
} ENC_DATA;
static int sjson_encoder_get_table_size(lua_State *L, int argno) {
// Returns -1 for object, otherwise the maximum integer key value found.
lua_pushvalue(L, argno);
// stack now contains: -1 => table
lua_pushnil(L);
// stack now contains: -1 => nil; -2 => table
//
int maxkey = 0;
while (lua_next(L, -2)) {
lua_pop(L, 1);
// stack now contains: -1 => key; -2 => table
if (lua_type(L, -1) == LUA_TNUMBER) {
int val = lua_tointeger(L, -1);
if (val > maxkey) {
maxkey = val;
} else if (val <= 0) {
maxkey = -1;
lua_pop(L, 1);
break;
}
} else {
maxkey = -1;
lua_pop(L, 1);
break;
}
}
lua_pop(L, 1);
return maxkey;
}
static void enc_pop_stack(lua_State *L, ENC_DATA *data) {
if (data->level < 0) {
luaL_error(L, "encoder stack underflow");
}
ENC_DATA_STATE *state = &data->stack[data->level];
lua_unref(L, state->lua_object_ref);
state->lua_object_ref = LUA_NOREF;
lua_unref(L, state->lua_key_ref);
state->lua_key_ref = LUA_REFNIL;
data->level--;
}
static void enc_push_stack(lua_State *L, ENC_DATA *data, int argno) {
if (++data->level >= data->nlevels) {
luaL_error(L, "encoder stack overflow");
}
lua_pushvalue(L, argno);
ENC_DATA_STATE *state = &data->stack[data->level];
state->lua_object_ref = lua_ref(L, 1);
state->size = sjson_encoder_get_table_size(L, argno);
state->offset = 0; // We haven't started on this one yet
}
static int sjson_encoder(lua_State *L) {
int nlevels = DEFAULT_DEPTH;
int argno = 1;
// Validate first arg is a table
luaL_checktype(L, argno++, LUA_TTABLE);
if (lua_type(L, argno) == LUA_TTABLE) {
lua_getfield(L, argno, "depth");
nlevels = lua_tointeger(L, argno);
if (nlevels == 0) {
nlevels = DEFAULT_DEPTH;
}
if (nlevels < 4) {
nlevels = 4;
}
if (nlevels > 1000) {
nlevels = 1000;
}
lua_pop(L, 1);
}
ENC_DATA *data = (ENC_DATA *) lua_newuserdata(L, sizeof(ENC_DATA) + nlevels * sizeof(ENC_DATA_STATE));
// Associate its metatable
luaL_getmetatable(L, "sjson.encoder");
lua_setmetatable(L, -2);
data->nlevels = nlevels;
data->level = -1;
data->stack = (ENC_DATA_STATE *) (data + 1);
data->current_str_ref = LUA_NOREF;
int i;
for (i = 0; i < nlevels; i++) {
data->stack[i].lua_object_ref = LUA_NOREF;
data->stack[i].lua_key_ref = LUA_REFNIL;
}
enc_push_stack(L, data, 1);
data->null_ref = LUA_REFNIL;
if (lua_type(L, argno) == LUA_TTABLE) {
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
lua_getfield(L, argno, "null");
data->null_ref = lua_ref(L, 1);
}
return 1;
}
static void encode_lua_object(lua_State *L, ENC_DATA *data, int argno, const char *prefix, const char *suffix) {
luaL_Buffer b;
luaL_buffinit(L, &b);
luaL_addstring(&b, prefix);
int type = lua_type(L, argno);
if (type == LUA_TSTRING) {
// Check to see if it is the NULL value
if (data->null_ref != LUA_REFNIL) {
lua_rawgeti(L, LUA_REGISTRYINDEX, data->null_ref);
if (lua_equal(L, -1, -2)) {
type = LUA_TNIL;
}
lua_pop(L, 1);
}
}
switch (type) {
default:
luaL_error(L, "Cannot encode type %d", type);
break;
case LUA_TLIGHTUSERDATA:
case LUA_TNIL:
luaL_addstring(&b, "null");
break;
case LUA_TBOOLEAN:
luaL_addstring(&b, lua_toboolean(L, argno) ? "true" : "false");
break;
case LUA_TNUMBER:
{
lua_pushvalue(L, argno);
size_t len;
const char *str = lua_tolstring(L, -1, &len);
char value[len + 1];
strcpy(value, str);
lua_pop(L, 1);
luaL_addstring(&b, value);
break;
}
case LUA_TSTRING:
{
luaL_addchar(&b, '"');
size_t len;
const char *str = lua_tolstring(L, argno, &len);
while (len > 0) {
if ((*str & 0xff) < 0x20) {
char value[8];
value[0] = '\\';
char *d = value + 1;
switch(*str) {
case '\f':
*d++ = 'f';
break;
case '\n':
*d++ = 'n';
break;
case '\t':
*d++ = 't';
break;
case '\r':
*d++ = 'r';
break;
case '\b':
*d++ = 'b';
break;
default:
*d++ = 'u';
*d++ = '0';
*d++ = '0';
*d++ = "0123456789abcdef"[(*str >> 4) & 0xf];
*d++ = "0123456789abcdef"[(*str ) & 0xf];
break;
}
*d = '\0';
luaL_addstring(&b, value);
} else {
luaL_addchar(&b, *str);
}
str++;
len--;
}
luaL_addchar(&b, '"');
break;
}
}
luaL_addstring(&b, suffix);
luaL_pushresult(&b);
}
static int sjson_encoder_next_value_is_table(lua_State *L) {
int count = 10;
while ((lua_type(L, -1) == LUA_TFUNCTION
#ifdef LUA_TLIGHTFUNCTION
|| lua_type(L, -1) == LUA_TLIGHTFUNCTION
#endif
) && count-- > 0) {
// call it and use the return value
lua_call(L, 0, 1); // Expecting replacement value
}
return (lua_type(L, -1) == LUA_TTABLE);
}
static void sjson_encoder_make_next_chunk(lua_State *L, ENC_DATA *data) {
if (data->level < 0) {
return;
}
luaL_Buffer b;
luaL_buffinit(L, &b);
// Ending condition
while (data->level >= 0 && !b.lvl) {
ENC_DATA_STATE *state = &data->stack[data->level];
int finished = 0;
if (state->size >= 0) {
if (state->offset == 0) {
// start of object or whatever
luaL_addchar(&b, '[');
}
if (state->offset == state->size << 1) {
luaL_addchar(&b, ']');
finished = 1;
} else if ((state->offset & 1) == 0) {
if (state->offset > 0) {
luaL_addchar(&b, ',');
}
} else {
// output the value
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_rawgeti(L, -1, (state->offset >> 1) + 1);
if (sjson_encoder_next_value_is_table(L)) {
enc_push_stack(L, data, -1);
lua_pop(L, 2);
state->offset++;
continue;
}
encode_lua_object(L, data, -1, "", "");
lua_remove(L, -2);
lua_remove(L, -2);
luaL_addvalue(&b);
}
state->offset++;
} else {
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_object_ref);
// stack now contains: -1 => table
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_key_ref);
// stack now contains: -1 => nil or key; -2 => table
if (lua_next(L, -2)) {
// save the key
if (state->offset & 1) {
lua_unref(L, state->lua_key_ref);
state->lua_key_ref = LUA_NOREF;
// Duplicate the key
lua_pushvalue(L, -2);
state->lua_key_ref = lua_ref(L, 1);
}
if ((state->offset & 1) == 0) {
// copy the key so that lua_tostring does not modify the original
lua_pushvalue(L, -2);
// stack now contains: -1 => key; -2 => value; -3 => key; -4 => table
// key
lua_tostring(L, -1);
encode_lua_object(L, data, -1, state->offset ? "," : "{", ":");
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
} else {
if (sjson_encoder_next_value_is_table(L)) {
enc_push_stack(L, data, -1);
lua_pop(L, 3);
state->offset++;
continue;
}
encode_lua_object(L, data, -1, "", "");
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
}
luaL_addvalue(&b);
} else {
lua_pop(L, 1);
// We have got to the end
luaL_addchar(&b, '}');
finished = 1;
}
state->offset++;
}
if (finished) {
enc_pop_stack(L, data);
}
}
luaL_pushresult(&b);
data->current_str_ref = lua_ref(L, 1);
data->offset = 0;
}
static int sjson_encoder_read_int(lua_State *L, ENC_DATA *data, int readsize) {
luaL_Buffer b;
luaL_buffinit(L, &b);
size_t len;
do {
// Fill the buffer with (up to) readsize characters
if (data->current_str_ref != LUA_NOREF) {
// this is not allowed
lua_rawgeti(L, LUA_REGISTRYINDEX, data->current_str_ref);
const char *str = lua_tolstring(L, -1, &len);
lua_pop(L, 1); // Note that we still have the string referenced so it can't go away
int amnt = len - data->offset;;
if (amnt > readsize) {
amnt = readsize;
}
luaL_addlstring(&b, str + data->offset, amnt);
data->offset += amnt;
readsize -= amnt;
if (data->offset == len) {
lua_unref(L, data->current_str_ref);
data->current_str_ref = LUA_NOREF;
}
}
if (readsize > 0) {
// Make the next chunk
sjson_encoder_make_next_chunk(L, data);
}
} while (readsize > 0 && data->current_str_ref != LUA_NOREF);
luaL_pushresult(&b);
lua_tolstring(L, -1, &len);
if (len == 0) {
// we have got to the end
lua_pop(L, 1);
return 0;
}
return 1;
}
static int sjson_encoder_read(lua_State *L) {
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, 1, "sjson.encoder");
int readsize = 1024;
if (lua_type(L, 2) == LUA_TNUMBER) {
readsize = lua_tointeger(L, 2);
if (readsize < 1) {
readsize = 1;
}
}
return sjson_encoder_read_int(L, data, readsize);
}
static int sjson_encode(lua_State *L) {
sjson_encoder(L);
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, -1, "sjson.encoder");
int rc = sjson_encoder_read_int(L, data, 1000000);
lua_remove(L, -(rc + 1));
return rc;
}
static int sjson_encoder_destructor(lua_State *L) {
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, 1, "sjson.encoder");
int i;
for (i = 0; i < data->nlevels; i++) {
luaL_unref(L, LUA_REGISTRYINDEX, data->stack[i].lua_object_ref);
luaL_unref(L, LUA_REGISTRYINDEX, data->stack[i].lua_key_ref);
}
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
luaL_unref(L, LUA_REGISTRYINDEX, data->current_str_ref);
DBG_PRINTF("Destructor called\n");
return 0;
}
#ifdef LOCAL_LUA
static const luaL_Reg sjson_encoder_map[] = {
{ "read", sjson_encoder_read },
{ "__gc", sjson_encoder_destructor },
{ NULL, NULL }
};
static const luaL_Reg sjson_decoder_map[] = {
{ "write", sjson_decoder_write },
{ "result", sjson_decoder_result },
{ "__gc", sjson_decoder_destructor },
{ NULL, NULL }
};
static const luaL_Reg sjsonlib[] = {
{ "decode", sjson_decode },
{ "decoder", sjson_decoder },
{ "encode", sjson_encode },
{ "encoder", sjson_encoder },
{NULL, NULL}
};
#else
static const LUA_REG_TYPE sjson_encoder_map[] = {
{ LSTRKEY( "read" ), LFUNCVAL( sjson_encoder_read ) },
{ LSTRKEY( "__gc" ), LFUNCVAL( sjson_encoder_destructor ) },
{ LSTRKEY( "__index" ), LROVAL( sjson_encoder_map ) },
{ LNILKEY, LNILVAL }
};
static const LUA_REG_TYPE sjson_decoder_map[] = {
{ LSTRKEY( "write" ), LFUNCVAL( sjson_decoder_write ) },
{ LSTRKEY( "result" ), LFUNCVAL( sjson_decoder_result ) },
{ LSTRKEY( "__gc" ), LFUNCVAL( sjson_decoder_destructor ) },
{ LSTRKEY( "__index" ), LROVAL( sjson_decoder_map ) },
{ LNILKEY, LNILVAL }
};
static const LUA_REG_TYPE sjson_map[] = {
{ LSTRKEY( "encode" ), LFUNCVAL( sjson_encode ) },
{ LSTRKEY( "decode" ), LFUNCVAL( sjson_decode ) },
{ LSTRKEY( "encoder" ), LFUNCVAL( sjson_encoder ) },
{ LSTRKEY( "decoder" ), LFUNCVAL( sjson_decoder ) },
{ LSTRKEY( "NULL" ), LUDATA( 0 ) },
{ LNILKEY, LNILVAL }
};
#endif
LUALIB_API int luaopen_sjson (lua_State *L) {
#ifdef LOCAL_LUA
luaL_register(L, LUA_SJSONLIBNAME, sjsonlib);
lua_getglobal(L, LUA_SJSONLIBNAME);
lua_pushstring(L, "NULL");
lua_pushlightuserdata(L, 0);
lua_settable(L, -3);
lua_pop(L, 1);
luaL_newmetatable(L, "sjson.encoder");
luaL_register(L, NULL, sjson_encoder_map);
lua_setfield(L, -1, "__index");
luaL_newmetatable(L, "sjson.decoder");
luaL_register(L, NULL, sjson_decoder_map);
lua_setfield(L, -1, "__index");
#else
luaL_rometatable(L, "sjson.decoder", (void *)sjson_decoder_map);
luaL_rometatable(L, "sjson.encoder", (void *)sjson_encoder_map);
#endif
return 1;
}
#ifndef LOCAL_LUA
NODEMCU_MODULE(SJSON, "sjson", sjson_map, luaopen_sjson);
#endif
/* Copyright (C) 2012-2015 Mark Nunberg.
*
* See included LICENSE file for license details.
*/
#include "jsonsl.h"
#include <assert.h>
#include <limits.h>
#include <ctype.h>
#ifdef JSONSL_USE_METRICS
#define XMETRICS \
X(STRINGY_INSIGNIFICANT) \
X(STRINGY_SLOWPATH) \
X(ALLOWED_WHITESPACE) \
X(QUOTE_FASTPATH) \
X(SPECIAL_FASTPATH) \
X(SPECIAL_WSPOP) \
X(SPECIAL_SLOWPATH) \
X(GENERIC) \
X(STRUCTURAL_TOKEN) \
X(SPECIAL_SWITCHFIRST) \
X(STRINGY_CATCH) \
X(NUMBER_FASTPATH) \
X(ESCAPES) \
X(TOTAL) \
struct jsonsl_metrics_st {
#define X(m) \
unsigned long metric_##m;
XMETRICS
#undef X
};
static struct jsonsl_metrics_st GlobalMetrics = { 0 };
static unsigned long GenericCounter[0x100] = { 0 };
static unsigned long StringyCatchCounter[0x100] = { 0 };
#define INCR_METRIC(m) \
GlobalMetrics.metric_##m++;
#define INCR_GENERIC(c) \
INCR_METRIC(GENERIC); \
GenericCounter[c]++; \
#define INCR_STRINGY_CATCH(c) \
INCR_METRIC(STRINGY_CATCH); \
StringyCatchCounter[c]++;
JSONSL_API
void jsonsl_dump_global_metrics(void)
{
int ii;
printf("JSONSL Metrics:\n");
#define X(m) \
printf("\t%-30s %20lu (%0.2f%%)\n", #m, GlobalMetrics.metric_##m, \
(float)((float)(GlobalMetrics.metric_##m/(float)GlobalMetrics.metric_TOTAL)) * 100);
XMETRICS
#undef X
printf("Generic Characters:\n");
for (ii = 0; ii < 0xff; ii++) {
if (GenericCounter[ii]) {
printf("\t[ %c ] %lu\n", ii, GenericCounter[ii]);
}
}
printf("Weird string loop\n");
for (ii = 0; ii < 0xff; ii++) {
if (StringyCatchCounter[ii]) {
printf("\t[ %c ] %lu\n", ii, StringyCatchCounter[ii]);
}
}
}
#else
#define INCR_METRIC(m)
#define INCR_GENERIC(c)
#define INCR_STRINGY_CATCH(c)
JSONSL_API
void jsonsl_dump_global_metrics(void) { }
#endif /* JSONSL_USE_METRICS */
#define CASE_DIGITS \
case '1': \
case '2': \
case '3': \
case '4': \
case '5': \
case '6': \
case '7': \
case '8': \
case '9': \
case '0':
static unsigned extract_special(unsigned);
static int is_special_end(unsigned);
static int is_allowed_whitespace(unsigned);
static int is_allowed_escape(unsigned);
static int is_simple_char(unsigned);
static char get_escape_equiv(unsigned);
JSONSL_API
size_t jsonsl_get_size(int nlevels)
{
return sizeof (struct jsonsl_st) + ( (nlevels-1) * sizeof (struct jsonsl_state_st)) ;
}
JSONSL_API
jsonsl_t jsonsl_init(jsonsl_t jsn, int nlevels)
{
unsigned int ii;
memset(jsn, 0, jsonsl_get_size(nlevels));
jsn->levels_max = nlevels;
jsn->max_callback_level = -1;
jsonsl_reset(jsn);
for (ii = 0; ii < jsn->levels_max; ii++) {
jsn->stack[ii].level = ii;
}
return jsn;
}
JSONSL_API
jsonsl_t jsonsl_new(int nlevels)
{
struct jsonsl_st *jsn = (struct jsonsl_st *)
calloc(1, jsonsl_get_size(nlevels));
if (jsn) {
jsonsl_init(jsn, nlevels);
}
return jsn;
}
JSONSL_API
void jsonsl_reset(jsonsl_t jsn)
{
jsn->tok_last = 0;
jsn->can_insert = 1;
jsn->pos = 0;
jsn->level = 0;
jsn->stopfl = 0;
jsn->in_escape = 0;
jsn->expecting = 0;
}
JSONSL_API
void jsonsl_destroy(jsonsl_t jsn)
{
if (jsn) {
free(jsn);
}
}
#define FASTPARSE_EXHAUSTED 1
#define FASTPARSE_BREAK 0
/*
* This function is meant to accelerate string parsing, reducing the main loop's
* check if we are indeed a string.
*
* @param jsn the parser
* @param[in,out] bytes_p A pointer to the current buffer (i.e. current position)
* @param[in,out] nbytes_p A pointer to the current size of the buffer
* @return true if all bytes have been exhausted (and thus the main loop can
* return), false if a special character was examined which requires greater
* examination.
*/
static int
jsonsl__str_fastparse(jsonsl_t jsn,
const jsonsl_uchar_t **bytes_p, size_t *nbytes_p)
{
const jsonsl_uchar_t *bytes = *bytes_p;
const jsonsl_uchar_t *end;
for (end = bytes + *nbytes_p; bytes != end; bytes++) {
if (
#ifdef JSONSL_USE_WCHAR
*bytes >= 0x100 ||
#endif /* JSONSL_USE_WCHAR */
(is_simple_char(*bytes))) {
INCR_METRIC(TOTAL);
INCR_METRIC(STRINGY_INSIGNIFICANT);
} else {
/* Once we're done here, re-calculate the position variables */
jsn->pos += (bytes - *bytes_p);
*nbytes_p -= (bytes - *bytes_p);
*bytes_p = bytes;
return FASTPARSE_BREAK;
}
}
/* Once we're done here, re-calculate the position variables */
jsn->pos += (bytes - *bytes_p);
return FASTPARSE_EXHAUSTED;
}
/* Functions exactly like str_fastparse, except it also accepts a 'state'
* argument, since the number's value is updated in the state. */
static int
jsonsl__num_fastparse(jsonsl_t jsn,
const jsonsl_uchar_t **bytes_p, size_t *nbytes_p,
struct jsonsl_state_st *state)
{
int exhausted = 1;
size_t nbytes = *nbytes_p;
const jsonsl_uchar_t *bytes = *bytes_p;
for (; nbytes; nbytes--, bytes++) {
jsonsl_uchar_t c = *bytes;
if (isdigit(c)) {
INCR_METRIC(TOTAL);
INCR_METRIC(NUMBER_FASTPATH);
state->nelem = (state->nelem * 10) + (c - 0x30);
} else {
exhausted = 0;
break;
}
}
jsn->pos += (*nbytes_p - nbytes);
if (exhausted) {
return FASTPARSE_EXHAUSTED;
}
*nbytes_p = nbytes;
*bytes_p = bytes;
return FASTPARSE_BREAK;
}
JSONSL_API
void
jsonsl_feed(jsonsl_t jsn, const jsonsl_char_t *bytes, size_t nbytes)
{
#define INVOKE_ERROR(eb) \
if (jsn->error_callback(jsn, JSONSL_ERROR_##eb, state, (char*)c)) { \
goto GT_AGAIN; \
} \
return;
#define STACK_PUSH \
if (jsn->level >= (levels_max-1)) { \
jsn->error_callback(jsn, JSONSL_ERROR_LEVELS_EXCEEDED, state, (char*)c); \
return; \
} \
state = jsn->stack + (++jsn->level); \
state->ignore_callback = jsn->stack[jsn->level-1].ignore_callback; \
state->pos_begin = jsn->pos;
#define STACK_POP_NOPOS \
state->pos_cur = jsn->pos; \
state = jsn->stack + (--jsn->level);
#define STACK_POP \
STACK_POP_NOPOS; \
state->pos_cur = jsn->pos;
#define CALLBACK_AND_POP_NOPOS(T) \
state->pos_cur = jsn->pos; \
DO_CALLBACK(T, POP); \
state->nescapes = 0; \
state = jsn->stack + (--jsn->level);
#define CALLBACK_AND_POP(T) \
CALLBACK_AND_POP_NOPOS(T); \
state->pos_cur = jsn->pos;
#define SPECIAL_POP \
CALLBACK_AND_POP(SPECIAL); \
jsn->expecting = 0; \
jsn->tok_last = 0; \
#define CUR_CHAR (*(jsonsl_uchar_t*)c)
#define DO_CALLBACK(T, action) \
if (jsn->call_##T && \
jsn->max_callback_level > state->level && \
state->ignore_callback == 0) { \
\
if (jsn->action_callback_##action) { \
jsn->action_callback_##action(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
} else if (jsn->action_callback) { \
jsn->action_callback(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
} \
if (jsn->stopfl) { return; } \
}
/**
* Verifies that we are able to insert the (non-string) item into a hash.
*/
#define ENSURE_HVAL \
if (state->nelem % 2 == 0 && state->type == JSONSL_T_OBJECT) { \
INVOKE_ERROR(HKEY_EXPECTED); \
}
#define VERIFY_SPECIAL(lit) \
if (CUR_CHAR != (lit)[jsn->pos - state->pos_begin]) { \
INVOKE_ERROR(SPECIAL_EXPECTED); \
}
#define STATE_SPECIAL_LENGTH \
(state)->nescapes
#define IS_NORMAL_NUMBER \
((state)->special_flags == JSONSL_SPECIALf_UNSIGNED || \
(state)->special_flags == JSONSL_SPECIALf_SIGNED)
#define STATE_NUM_LAST jsn->tok_last
#define CONTINUE_NEXT_CHAR() continue
const jsonsl_uchar_t *c = (jsonsl_uchar_t*)bytes;
size_t levels_max = jsn->levels_max;
struct jsonsl_state_st *state = jsn->stack + jsn->level;
jsn->base = bytes;
for (; nbytes; nbytes--, jsn->pos++, c++) {
unsigned state_type;
INCR_METRIC(TOTAL);
GT_AGAIN:
state_type = state->type;
/* Most common type is typically a string: */
if (state_type & JSONSL_Tf_STRINGY) {
/* Special escape handling for some stuff */
if (jsn->in_escape) {
jsn->in_escape = 0;
if (!is_allowed_escape(CUR_CHAR)) {
INVOKE_ERROR(ESCAPE_INVALID);
} else if (CUR_CHAR == 'u') {
DO_CALLBACK(UESCAPE, UESCAPE);
if (jsn->return_UESCAPE) {
return;
}
}
CONTINUE_NEXT_CHAR();
}
if (jsonsl__str_fastparse(jsn, &c, &nbytes) ==
FASTPARSE_EXHAUSTED) {
/* No need to readjust variables as we've exhausted the iterator */
return;
} else {
if (CUR_CHAR == '"') {
goto GT_QUOTE;
} else if (CUR_CHAR == '\\') {
goto GT_ESCAPE;
} else {
INVOKE_ERROR(WEIRD_WHITESPACE);
}
}
INCR_METRIC(STRINGY_SLOWPATH);
} else if (state_type == JSONSL_T_SPECIAL) {
/* Fast track for signed/unsigned */
if (IS_NORMAL_NUMBER) {
if (jsonsl__num_fastparse(jsn, &c, &nbytes, state) ==
FASTPARSE_EXHAUSTED) {
return;
} else {
goto GT_SPECIAL_NUMERIC;
}
} else if (state->special_flags == JSONSL_SPECIALf_DASH) {
if (!isdigit(CUR_CHAR)) {
INVOKE_ERROR(INVALID_NUMBER);
}
if (CUR_CHAR == '0') {
state->special_flags = JSONSL_SPECIALf_ZERO|JSONSL_SPECIALf_SIGNED;
} else if (isdigit(CUR_CHAR)) {
state->special_flags = JSONSL_SPECIALf_SIGNED;
state->nelem = CUR_CHAR - 0x30;
} else {
INVOKE_ERROR(INVALID_NUMBER);
}
CONTINUE_NEXT_CHAR();
} else if (state->special_flags == JSONSL_SPECIALf_ZERO) {
if (isdigit(CUR_CHAR)) {
/* Following a zero! */
INVOKE_ERROR(INVALID_NUMBER);
}
/* Unset the 'zero' flag: */
if (state->special_flags & JSONSL_SPECIALf_SIGNED) {
state->special_flags = JSONSL_SPECIALf_SIGNED;
} else {
state->special_flags = JSONSL_SPECIALf_UNSIGNED;
}
goto GT_SPECIAL_NUMERIC;
}
if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
GT_SPECIAL_NUMERIC:
switch (CUR_CHAR) {
CASE_DIGITS
STATE_NUM_LAST = '1';
CONTINUE_NEXT_CHAR();
case '.':
if (state->special_flags & JSONSL_SPECIALf_FLOAT) {
INVOKE_ERROR(INVALID_NUMBER);
}
state->special_flags |= JSONSL_SPECIALf_FLOAT;
STATE_NUM_LAST = '.';
CONTINUE_NEXT_CHAR();
case 'e':
case 'E':
if (state->special_flags & JSONSL_SPECIALf_EXPONENT) {
INVOKE_ERROR(INVALID_NUMBER);
}
state->special_flags |= JSONSL_SPECIALf_EXPONENT;
STATE_NUM_LAST = 'e';
CONTINUE_NEXT_CHAR();
case '-':
case '+':
if (STATE_NUM_LAST != 'e') {
INVOKE_ERROR(INVALID_NUMBER);
}
STATE_NUM_LAST = '-';
CONTINUE_NEXT_CHAR();
default:
if (is_special_end(CUR_CHAR)) {
goto GT_SPECIAL_POP;
}
INVOKE_ERROR(INVALID_NUMBER);
break;
}
}
/* else if (!NUMERIC) */
if (!is_special_end(CUR_CHAR)) {
STATE_SPECIAL_LENGTH++;
/* Verify TRUE, FALSE, NULL */
if (state->special_flags == JSONSL_SPECIALf_TRUE) {
VERIFY_SPECIAL("true");
} else if (state->special_flags == JSONSL_SPECIALf_FALSE) {
VERIFY_SPECIAL("false");
} else if (state->special_flags == JSONSL_SPECIALf_NULL) {
VERIFY_SPECIAL("null");
}
INCR_METRIC(SPECIAL_FASTPATH);
CONTINUE_NEXT_CHAR();
}
GT_SPECIAL_POP:
jsn->can_insert = 0;
if (IS_NORMAL_NUMBER) {
/* Nothing */
} else if (state->special_flags == JSONSL_SPECIALf_ZERO ||
state->special_flags == (JSONSL_SPECIALf_ZERO|JSONSL_SPECIALf_SIGNED)) {
/* 0 is unsigned! */
state->special_flags = JSONSL_SPECIALf_UNSIGNED;
} else if (state->special_flags == JSONSL_SPECIALf_DASH) {
/* Still in dash! */
INVOKE_ERROR(INVALID_NUMBER);
} else if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
/* Check that we're not at the end of a token */
if (STATE_NUM_LAST != '1') {
INVOKE_ERROR(INVALID_NUMBER);
}
} else if (state->special_flags == JSONSL_SPECIALf_TRUE) {
if (STATE_SPECIAL_LENGTH != 4) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
state->nelem = 1;
} else if (state->special_flags == JSONSL_SPECIALf_FALSE) {
if (STATE_SPECIAL_LENGTH != 5) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
} else if (state->special_flags == JSONSL_SPECIALf_NULL) {
if (STATE_SPECIAL_LENGTH != 4) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
}
SPECIAL_POP;
jsn->expecting = ',';
if (is_allowed_whitespace(CUR_CHAR)) {
CONTINUE_NEXT_CHAR();
}
/**
* This works because we have a non-whitespace token
* which is not a special token. If this is a structural
* character then it will be gracefully handled by the
* switch statement. Otherwise it will default to the 'special'
* state again,
*/
goto GT_STRUCTURAL_TOKEN;
} else if (is_allowed_whitespace(CUR_CHAR)) {
INCR_METRIC(ALLOWED_WHITESPACE);
/* So we're not special. Harmless insignificant whitespace
* passthrough
*/
CONTINUE_NEXT_CHAR();
} else if (extract_special(CUR_CHAR)) {
/* not a string, whitespace, or structural token. must be special */
goto GT_SPECIAL_BEGIN;
}
INCR_GENERIC(CUR_CHAR);
if (CUR_CHAR == '"') {
GT_QUOTE:
jsn->can_insert = 0;
switch (state_type) {
/* the end of a string or hash key */
case JSONSL_T_STRING:
CALLBACK_AND_POP(STRING);
CONTINUE_NEXT_CHAR();
case JSONSL_T_HKEY:
CALLBACK_AND_POP(HKEY);
CONTINUE_NEXT_CHAR();
case JSONSL_T_OBJECT:
state->nelem++;
if ( (state->nelem-1) % 2 ) {
/* Odd, this must be a hash value */
if (jsn->tok_last != ':') {
INVOKE_ERROR(MISSING_TOKEN);
}
jsn->expecting = ','; /* Can't figure out what to expect next */
jsn->tok_last = 0;
STACK_PUSH;
state->type = JSONSL_T_STRING;
DO_CALLBACK(STRING, PUSH);
} else {
/* hash key */
if (jsn->expecting != '"') {
INVOKE_ERROR(STRAY_TOKEN);
}
jsn->tok_last = 0;
jsn->expecting = ':';
STACK_PUSH;
state->type = JSONSL_T_HKEY;
DO_CALLBACK(HKEY, PUSH);
}
CONTINUE_NEXT_CHAR();
case JSONSL_T_LIST:
state->nelem++;
STACK_PUSH;
state->type = JSONSL_T_STRING;
jsn->expecting = ',';
jsn->tok_last = 0;
DO_CALLBACK(STRING, PUSH);
CONTINUE_NEXT_CHAR();
case JSONSL_T_SPECIAL:
INVOKE_ERROR(STRAY_TOKEN);
break;
default:
INVOKE_ERROR(STRING_OUTSIDE_CONTAINER);
break;
} /* switch(state->type) */
} else if (CUR_CHAR == '\\') {
GT_ESCAPE:
INCR_METRIC(ESCAPES);
/* Escape */
if ( (state->type & JSONSL_Tf_STRINGY) == 0 ) {
INVOKE_ERROR(ESCAPE_OUTSIDE_STRING);
}
state->nescapes++;
jsn->in_escape = 1;
CONTINUE_NEXT_CHAR();
} /* " or \ */
GT_STRUCTURAL_TOKEN:
switch (CUR_CHAR) {
case ':':
INCR_METRIC(STRUCTURAL_TOKEN);
if (jsn->expecting != CUR_CHAR) {
INVOKE_ERROR(STRAY_TOKEN);
}
jsn->tok_last = ':';
jsn->can_insert = 1;
jsn->expecting = '"';
CONTINUE_NEXT_CHAR();
case ',':
INCR_METRIC(STRUCTURAL_TOKEN);
/**
* The comma is one of the more generic tokens.
* In the context of an OBJECT, the can_insert flag
* should never be set, and no other action is
* necessary.
*/
if (jsn->expecting != CUR_CHAR) {
/* make this branch execute only when we haven't manually
* just placed the ',' in the expecting register.
*/
INVOKE_ERROR(STRAY_TOKEN);
}
if (state->type == JSONSL_T_OBJECT) {
/* end of hash value, expect a string as a hash key */
jsn->expecting = '"';
} else {
jsn->can_insert = 1;
}
jsn->tok_last = ',';
jsn->expecting = '"';
CONTINUE_NEXT_CHAR();
/* new list or object */
/* hashes are more common */
case '{':
case '[':
INCR_METRIC(STRUCTURAL_TOKEN);
if (!jsn->can_insert) {
INVOKE_ERROR(CANT_INSERT);
}
ENSURE_HVAL;
state->nelem++;
STACK_PUSH;
/* because the constants match the opening delimiters, we can do this: */
state->type = CUR_CHAR;
state->nelem = 0;
jsn->can_insert = 1;
if (CUR_CHAR == '{') {
/* If we're a hash, we expect a key first, which is quouted */
jsn->expecting = '"';
}
if (CUR_CHAR == JSONSL_T_OBJECT) {
DO_CALLBACK(OBJECT, PUSH);
} else {
DO_CALLBACK(LIST, PUSH);
}
jsn->tok_last = 0;
CONTINUE_NEXT_CHAR();
/* closing of list or object */
case '}':
case ']':
INCR_METRIC(STRUCTURAL_TOKEN);
if (jsn->tok_last == ',' && jsn->options.allow_trailing_comma == 0) {
INVOKE_ERROR(TRAILING_COMMA);
}
jsn->can_insert = 0;
jsn->level--;
jsn->expecting = ',';
jsn->tok_last = 0;
if (CUR_CHAR == ']') {
if (state->type != '[') {
INVOKE_ERROR(BRACKET_MISMATCH);
}
DO_CALLBACK(LIST, POP);
} else {
if (state->type != '{') {
INVOKE_ERROR(BRACKET_MISMATCH);
} else if (state->nelem && state->nelem % 2 != 0) {
INVOKE_ERROR(VALUE_EXPECTED);
}
DO_CALLBACK(OBJECT, POP);
}
state = jsn->stack + jsn->level;
state->pos_cur = jsn->pos;
CONTINUE_NEXT_CHAR();
default:
GT_SPECIAL_BEGIN:
/**
* Not a string, not a structural token, and not benign whitespace.
* Technically we should iterate over the character always, but since
* we are not doing full numerical/value decoding anyway (but only hinting),
* we only check upon entry.
*/
if (state->type != JSONSL_T_SPECIAL) {
int special_flags = extract_special(CUR_CHAR);
if (!special_flags) {
/**
* Try to do some heuristics here anyway to figure out what kind of
* error this is. The 'special' case is a fallback scenario anyway.
*/
if (CUR_CHAR == '\0') {
INVOKE_ERROR(FOUND_NULL_BYTE);
} else if (CUR_CHAR < 0x20) {
INVOKE_ERROR(WEIRD_WHITESPACE);
} else {
INVOKE_ERROR(SPECIAL_EXPECTED);
}
}
ENSURE_HVAL;
state->nelem++;
if (!jsn->can_insert) {
INVOKE_ERROR(CANT_INSERT);
}
STACK_PUSH;
state->type = JSONSL_T_SPECIAL;
state->special_flags = special_flags;
STATE_SPECIAL_LENGTH = 1;
if (special_flags == JSONSL_SPECIALf_UNSIGNED) {
state->nelem = CUR_CHAR - 0x30;
STATE_NUM_LAST = '1';
} else {
STATE_NUM_LAST = '-';
state->nelem = 0;
}
DO_CALLBACK(SPECIAL, PUSH);
}
CONTINUE_NEXT_CHAR();
}
}
}
JSONSL_API
const char* jsonsl_strerror(jsonsl_error_t err)
{
if (err == JSONSL_ERROR_SUCCESS) {
return "SUCCESS";
}
#define X(t) \
if (err == JSONSL_ERROR_##t) \
return #t;
JSONSL_XERR;
#undef X
return "<UNKNOWN_ERROR>";
}
JSONSL_API
const char *jsonsl_strtype(jsonsl_type_t type)
{
#define X(o,c) \
if (type == JSONSL_T_##o) \
return #o;
JSONSL_XTYPE
#undef X
return "UNKNOWN TYPE";
}
/*
*
* JPR/JSONPointer functions
*
*
*/
#ifndef JSONSL_NO_JPR
static
jsonsl_jpr_type_t
populate_component(char *in,
struct jsonsl_jpr_component_st *component,
char **next,
jsonsl_error_t *errp)
{
unsigned long pctval;
char *c = NULL, *outp = NULL, *end = NULL;
size_t input_len;
jsonsl_jpr_type_t ret = JSONSL_PATH_NONE;
if (*next == NULL || *(*next) == '\0') {
return JSONSL_PATH_NONE;
}
/* Replace the next / with a NULL */
*next = strstr(in, "/");
if (*next != NULL) {
*(*next) = '\0'; /* drop the forward slash */
input_len = *next - in;
end = *next;
*next += 1; /* next character after the '/' */
} else {
input_len = strlen(in);
end = in + input_len + 1;
}
component->pstr = in;
/* Check for special components of interest */
if (*in == JSONSL_PATH_WILDCARD_CHAR && input_len == 1) {
/* Lone wildcard */
ret = JSONSL_PATH_WILDCARD;
goto GT_RET;
} else if (isdigit(*in)) {
/* ASCII Numeric */
char *endptr;
component->idx = strtoul(in, &endptr, 10);
if (endptr && *endptr == '\0') {
ret = JSONSL_PATH_NUMERIC;
goto GT_RET;
}
}
/* Default, it's a string */
ret = JSONSL_PATH_STRING;
for (c = outp = in; c < end; c++, outp++) {
char origc;
if (*c != '%') {
goto GT_ASSIGN;
}
/*
* c = { [+0] = '%', [+1] = 'b', [+2] = 'e', [+3] = '\0' }
*/
/* Need %XX */
if (c+2 >= end) {
*errp = JSONSL_ERROR_PERCENT_BADHEX;
return JSONSL_PATH_INVALID;
}
if (! (isxdigit(*(c+1)) && isxdigit(*(c+2))) ) {
*errp = JSONSL_ERROR_PERCENT_BADHEX;
return JSONSL_PATH_INVALID;
}
/* Temporarily null-terminate the characters */
origc = *(c+3);
*(c+3) = '\0';
pctval = strtoul(c+1, NULL, 16);
*(c+3) = origc;
*outp = (char) pctval;
c += 2;
continue;
GT_ASSIGN:
*outp = *c;
}
/* Null-terminate the string */
for (; outp < c; outp++) {
*outp = '\0';
}
GT_RET:
component->ptype = ret;
if (ret != JSONSL_PATH_WILDCARD) {
component->len = strlen(component->pstr);
}
return ret;
}
JSONSL_API
jsonsl_jpr_t
jsonsl_jpr_new(const char *path, jsonsl_error_t *errp)
{
char *my_copy = NULL;
int count, curidx;
struct jsonsl_jpr_st *ret = NULL;
struct jsonsl_jpr_component_st *components = NULL;
size_t origlen;
jsonsl_error_t errstacked;
#define JPR_BAIL(err) *errp = err; goto GT_ERROR;
if (errp == NULL) {
errp = &errstacked;
}
if (path == NULL || *path != '/') {
JPR_BAIL(JSONSL_ERROR_JPR_NOROOT);
return NULL;
}
count = 1;
path++;
{
const char *c = path;
for (; *c; c++) {
if (*c == '/') {
count++;
if (*(c+1) == '/') {
JPR_BAIL(JSONSL_ERROR_JPR_DUPSLASH);
}
}
}
}
if(*path) {
count++;
}
components = (struct jsonsl_jpr_component_st *)
malloc(sizeof(*components) * count);
if (!components) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
my_copy = (char *)malloc(strlen(path) + 1);
if (!my_copy) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
strcpy(my_copy, path);
components[0].ptype = JSONSL_PATH_ROOT;
if (*my_copy) {
char *cur = my_copy;
int pathret = JSONSL_PATH_STRING;
curidx = 1;
while (pathret > 0 && curidx < count) {
pathret = populate_component(cur, components + curidx, &cur, errp);
if (pathret > 0) {
curidx++;
} else {
break;
}
}
if (pathret == JSONSL_PATH_INVALID) {
JPR_BAIL(JSONSL_ERROR_JPR_BADPATH);
}
} else {
curidx = 1;
}
path--; /*revert path to leading '/' */
origlen = strlen(path) + 1;
ret = (struct jsonsl_jpr_st *)malloc(sizeof(*ret));
if (!ret) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
ret->orig = (char *)malloc(origlen);
if (!ret->orig) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
ret->components = components;
ret->ncomponents = curidx;
ret->basestr = my_copy;
ret->norig = origlen-1;
strcpy(ret->orig, path);
return ret;
GT_ERROR:
free(my_copy);
free(components);
if (ret) {
free(ret->orig);
}
free(ret);
return NULL;
#undef JPR_BAIL
}
void jsonsl_jpr_destroy(jsonsl_jpr_t jpr)
{
free(jpr->components);
free(jpr->basestr);
free(jpr->orig);
free(jpr);
}
/**
* Call when there is a possibility of a match, either as a final match or
* as a path within a match
* @param jpr The JPR path
* @param component Component corresponding to the current element
* @param prlevel The level of the *parent*
* @param chtype The type of the child
* @return Match status
*/
static jsonsl_jpr_match_t
jsonsl__match_continue(jsonsl_jpr_t jpr,
const struct jsonsl_jpr_component_st *component,
unsigned prlevel, unsigned chtype)
{
const struct jsonsl_jpr_component_st *next_comp = component + 1;
if (prlevel == jpr->ncomponents - 1) {
/* This is the match. Check the expected type of the match against
* the child */
if (jpr->match_type == 0 || jpr->match_type == chtype) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
}
if (chtype == JSONSL_T_LIST) {
if (next_comp->ptype == JSONSL_PATH_NUMERIC) {
return JSONSL_MATCH_POSSIBLE;
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
} else if (chtype == JSONSL_T_OBJECT) {
if (next_comp->ptype == JSONSL_PATH_NUMERIC) {
return JSONSL_MATCH_TYPE_MISMATCH;
} else {
return JSONSL_MATCH_POSSIBLE;
}
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
}
JSONSL_API
jsonsl_jpr_match_t
jsonsl_path_match(jsonsl_jpr_t jpr,
const struct jsonsl_state_st *parent,
const struct jsonsl_state_st *child,
const char *key, size_t nkey)
{
const struct jsonsl_jpr_component_st *comp;
if (!parent) {
/* No parent. Return immediately since it's always a match */
return jsonsl__match_continue(jpr, jpr->components, 0, child->type);
}
comp = jpr->components + parent->level;
/* note that we don't need to verify the type of the match, this is
* always done through the previous call to jsonsl__match_continue.
* If we are in a POSSIBLE tree then we can be certain the types (at
* least at this level) are correct */
if (parent->type == JSONSL_T_OBJECT) {
if (comp->len != nkey || strncmp(key, comp->pstr, nkey) != 0) {
return JSONSL_MATCH_NOMATCH;
}
} else {
if (comp->idx != parent->nelem - 1) {
return JSONSL_MATCH_NOMATCH;
}
}
return jsonsl__match_continue(jpr, comp, parent->level, child->type);
}
JSONSL_API
jsonsl_jpr_match_t
jsonsl_jpr_match(jsonsl_jpr_t jpr,
unsigned int parent_type,
unsigned int parent_level,
const char *key,
size_t nkey)
{
/* find our current component. This is the child level */
int cmpret;
struct jsonsl_jpr_component_st *p_component;
p_component = jpr->components + parent_level;
if (parent_level >= jpr->ncomponents) {
return JSONSL_MATCH_NOMATCH;
}
/* Lone query for 'root' element. Always matches */
if (parent_level == 0) {
if (jpr->ncomponents == 1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
/* Wildcard, always matches */
if (p_component->ptype == JSONSL_PATH_WILDCARD) {
if (parent_level == jpr->ncomponents-1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
/* Check numeric array index. This gets its special block so we can avoid
* string comparisons */
if (p_component->ptype == JSONSL_PATH_NUMERIC) {
if (parent_type == JSONSL_T_LIST) {
if (p_component->idx != nkey) {
/* Wrong index */
return JSONSL_MATCH_NOMATCH;
} else {
if (parent_level == jpr->ncomponents-1) {
/* This is the last element of the path */
return JSONSL_MATCH_COMPLETE;
} else {
/* Intermediate element */
return JSONSL_MATCH_POSSIBLE;
}
}
} else if (p_component->is_arridx) {
/* Numeric and an array index (set explicitly by user). But not
* a list for a parent */
return JSONSL_MATCH_TYPE_MISMATCH;
}
} else if (parent_type == JSONSL_T_LIST) {
return JSONSL_MATCH_TYPE_MISMATCH;
}
/* Check lengths */
if (p_component->len != nkey) {
return JSONSL_MATCH_NOMATCH;
}
/* Check string comparison */
cmpret = strncmp(p_component->pstr, key, nkey);
if (cmpret == 0) {
if (parent_level == jpr->ncomponents-1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
return JSONSL_MATCH_NOMATCH;
}
JSONSL_API
void jsonsl_jpr_match_state_init(jsonsl_t jsn,
jsonsl_jpr_t *jprs,
size_t njprs)
{
size_t ii, *firstjmp;
if (njprs == 0) {
return;
}
jsn->jprs = (jsonsl_jpr_t *)malloc(sizeof(jsonsl_jpr_t) * njprs);
jsn->jpr_count = njprs;
jsn->jpr_root = (size_t*)calloc(1, sizeof(size_t) * njprs * jsn->levels_max);
memcpy(jsn->jprs, jprs, sizeof(jsonsl_jpr_t) * njprs);
/* Set the initial jump table values */
firstjmp = jsn->jpr_root;
for (ii = 0; ii < njprs; ii++) {
firstjmp[ii] = ii+1;
}
}
JSONSL_API
void jsonsl_jpr_match_state_cleanup(jsonsl_t jsn)
{
if (jsn->jpr_count == 0) {
return;
}
free(jsn->jpr_root);
free(jsn->jprs);
jsn->jprs = NULL;
jsn->jpr_root = NULL;
jsn->jpr_count = 0;
}
/**
* This function should be called exactly once on each element...
* This should also be called in recursive order, since we rely
* on the parent having been initalized for a match.
*
* Since the parent is checked for a match as well, we maintain a 'serial' counter.
* Whenever we traverse an element, we expect the serial to be the same as a global
* integer. If they do not match, we re-initialize the context, and set the serial.
*
* This ensures a type of consistency without having a proactive reset by the
* main lexer itself.
*
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_match_state(jsonsl_t jsn,
struct jsonsl_state_st *state,
const char *key,
size_t nkey,
jsonsl_jpr_match_t *out)
{
struct jsonsl_state_st *parent_state;
jsonsl_jpr_t ret = NULL;
/* Jump and JPR tables for our own state and the parent state */
size_t *jmptable, *pjmptable;
size_t jmp_cur, ii, ourjmpidx;
if (!jsn->jpr_root) {
*out = JSONSL_MATCH_NOMATCH;
return NULL;
}
pjmptable = jsn->jpr_root + (jsn->jpr_count * (state->level-1));
jmptable = pjmptable + jsn->jpr_count;
/* If the parent cannot match, then invalidate it */
if (*pjmptable == 0) {
*jmptable = 0;
*out = JSONSL_MATCH_NOMATCH;
return NULL;
}
parent_state = jsn->stack + state->level - 1;
if (parent_state->type == JSONSL_T_LIST) {
nkey = (size_t) parent_state->nelem;
}
*jmptable = 0;
ourjmpidx = 0;
memset(jmptable, 0, sizeof(int) * jsn->jpr_count);
for (ii = 0; ii < jsn->jpr_count; ii++) {
jmp_cur = pjmptable[ii];
if (jmp_cur) {
jsonsl_jpr_t jpr = jsn->jprs[jmp_cur-1];
*out = jsonsl_jpr_match(jpr,
parent_state->type,
parent_state->level,
key, nkey);
if (*out == JSONSL_MATCH_COMPLETE) {
ret = jpr;
*jmptable = 0;
return ret;
} else if (*out == JSONSL_MATCH_POSSIBLE) {
jmptable[ourjmpidx] = ii+1;
ourjmpidx++;
}
} else {
break;
}
}
if (!*jmptable) {
*out = JSONSL_MATCH_NOMATCH;
}
return NULL;
}
JSONSL_API
const char *jsonsl_strmatchtype(jsonsl_jpr_match_t match)
{
#define X(T,v) \
if ( match == JSONSL_MATCH_##T ) \
return #T;
JSONSL_XMATCH
#undef X
return "<UNKNOWN>";
}
#endif /* JSONSL_WITH_JPR */
static char *
jsonsl__writeutf8(uint32_t pt, char *out)
{
#define ADD_OUTPUT(c) *out = (char)(c); out++;
if (pt < 0x80) {
ADD_OUTPUT(pt);
} else if (pt < 0x800) {
ADD_OUTPUT((pt >> 6) | 0xC0);
ADD_OUTPUT((pt & 0x3F) | 0x80);
} else if (pt < 0x10000) {
ADD_OUTPUT((pt >> 12) | 0xE0);
ADD_OUTPUT(((pt >> 6) & 0x3F) | 0x80);
ADD_OUTPUT((pt & 0x3F) | 0x80);
} else {
ADD_OUTPUT((pt >> 18) | 0xF0);
ADD_OUTPUT(((pt >> 12) & 0x3F) | 0x80);
ADD_OUTPUT(((pt >> 6) & 0x3F) | 0x80);
ADD_OUTPUT((pt & 0x3F) | 0x80);
}
return out;
#undef ADD_OUTPUT
}
/* Thanks snej (https://github.com/mnunberg/jsonsl/issues/9) */
static int
jsonsl__digit2int(char ch) {
int d = ch - '0';
if ((unsigned) d < 10) {
return d;
}
d = ch - 'a';
if ((unsigned) d < 6) {
return d + 10;
}
d = ch - 'A';
if ((unsigned) d < 6) {
return d + 10;
}
return -1;
}
/* Assume 's' is at least 4 bytes long */
static int
jsonsl__get_uescape_16(const char *s)
{
int ret = 0;
int cur;
#define GET_DIGIT(off) \
cur = jsonsl__digit2int(s[off]); \
if (cur == -1) { return -1; } \
ret |= (cur << (12 - (off * 4)));
GET_DIGIT(0);
GET_DIGIT(1);
GET_DIGIT(2);
GET_DIGIT(3);
#undef GET_DIGIT
return ret;
}
/**
* Utility function to convert escape sequences
*/
JSONSL_API
size_t jsonsl_util_unescape_ex(const char *in,
char *out,
size_t len,
const int toEscape[128],
unsigned *oflags,
jsonsl_error_t *err,
const char **errat)
{
const unsigned char *c = (const unsigned char*)in;
char *begin_p = out;
unsigned oflags_s;
uint16_t last_codepoint = 0;
if (!oflags) {
oflags = &oflags_s;
}
*oflags = 0;
#define UNESCAPE_BAIL(e,offset) \
*err = JSONSL_ERROR_##e; \
if (errat) { \
*errat = (const char*)(c+ (ptrdiff_t)(offset)); \
} \
return 0;
for (; len; len--, c++, out++) {
int uescval;
if (*c != '\\') {
/* Not an escape, so we don't care about this */
goto GT_ASSIGN;
}
if (len < 2) {
UNESCAPE_BAIL(ESCAPE_INVALID, 0);
}
if (!is_allowed_escape(c[1])) {
UNESCAPE_BAIL(ESCAPE_INVALID, 1)
}
if ((toEscape && toEscape[(unsigned char)c[1] & 0x7f] == 0 &&
c[1] != '\\' && c[1] != '"')) {
/* if we don't want to unescape this string, write the escape sequence to the output */
*out++ = *c++;
if (--len == 0)
break;
goto GT_ASSIGN;
}
if (c[1] != 'u') {
/* simple skip-and-replace using pre-defined maps.
* TODO: should the maps actually reflect the desired
* replacement character in toEscape?
*/
char esctmp = get_escape_equiv(c[1]);
if (esctmp) {
/* Check if there is a corresponding replacement */
*out = esctmp;
} else {
/* Just gobble up the 'reverse-solidus' */
*out = c[1];
}
len--;
c++;
/* do not assign, just continue */
continue;
}
/* next == 'u' */
if (len < 6) {
/* Need at least six characters.. */
UNESCAPE_BAIL(UESCAPE_TOOSHORT, 2);
}
uescval = jsonsl__get_uescape_16((const char *)c + 2);
if (uescval == -1) {
UNESCAPE_BAIL(PERCENT_BADHEX, -1);
} else if (uescval == 0) {
UNESCAPE_BAIL(INVALID_CODEPOINT, 2);
}
if (last_codepoint) {
uint16_t w1 = last_codepoint, w2 = (uint16_t)uescval;
uint32_t cp;
if (uescval < 0xDC00 || uescval > 0xDFFF) {
UNESCAPE_BAIL(INVALID_CODEPOINT, -1);
}
cp = (w1 & 0x3FF) << 10;
cp |= (w2 & 0x3FF);
cp += 0x10000;
out = jsonsl__writeutf8(cp, out) - 1;
last_codepoint = 0;
} else if (uescval < 0xD800 || uescval > 0xDFFF) {
*oflags |= JSONSL_SPECIALf_NONASCII;
out = jsonsl__writeutf8(uescval, out) - 1;
} else if (uescval > 0xD7FF && uescval < 0xDC00) {
*oflags |= JSONSL_SPECIALf_NONASCII;
last_codepoint = (uint16_t)uescval;
out--;
} else {
UNESCAPE_BAIL(INVALID_CODEPOINT, 2);
}
/* Post uescape cleanup */
len -= 5; /* Gobble up 5 chars after 'u' */
c += 5;
continue;
/* Only reached by previous branches */
GT_ASSIGN:
*out = *c;
}
if (last_codepoint) {
*err = JSONSL_ERROR_INVALID_CODEPOINT;
return 0;
}
*err = JSONSL_ERROR_SUCCESS;
return out - begin_p;
}
/**
* Character Table definitions.
* These were all generated via srcutil/genchartables.pl
*/
/**
* This table contains the beginnings of non-string
* allowable (bareword) values.
*/
static const unsigned short Special_Table[0x80] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2c */
/* 0x2d */ JSONSL_SPECIALf_DASH /* <-> */, /* 0x2d */
/* 0x2e */ 0,0, /* 0x2f */
/* 0x30 */ JSONSL_SPECIALf_ZERO /* <0> */, /* 0x30 */
/* 0x31 */ JSONSL_SPECIALf_UNSIGNED /* <1> */, /* 0x31 */
/* 0x32 */ JSONSL_SPECIALf_UNSIGNED /* <2> */, /* 0x32 */
/* 0x33 */ JSONSL_SPECIALf_UNSIGNED /* <3> */, /* 0x33 */
/* 0x34 */ JSONSL_SPECIALf_UNSIGNED /* <4> */, /* 0x34 */
/* 0x35 */ JSONSL_SPECIALf_UNSIGNED /* <5> */, /* 0x35 */
/* 0x36 */ JSONSL_SPECIALf_UNSIGNED /* <6> */, /* 0x36 */
/* 0x37 */ JSONSL_SPECIALf_UNSIGNED /* <7> */, /* 0x37 */
/* 0x38 */ JSONSL_SPECIALf_UNSIGNED /* <8> */, /* 0x38 */
/* 0x39 */ JSONSL_SPECIALf_UNSIGNED /* <9> */, /* 0x39 */
/* 0x3a */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x59 */
/* 0x5a */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x65 */
/* 0x66 */ JSONSL_SPECIALf_FALSE /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ JSONSL_SPECIALf_NULL /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0,0,0, /* 0x73 */
/* 0x74 */ JSONSL_SPECIALf_TRUE /* <t> */ /* 0x74 */
};
// Bit tables are order such that the MSB is bit 0.
//
/**
* Contains characters which signal the termination of any of the 'special' bareword
* values.
*/
static const char Special_Endings[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
/* 0x09 */ 1 /* <TAB> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 0,0, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 1 /* <SP> */, /* 0x20 */
/* 0x21 */ 0, /* 0x21 */
/* 0x22 */ 1 /* " */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0, /* 0x2b */
/* 0x2c */ 1 /* , */, /* 0x2c */
/* 0x2d */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x39 */
/* 0x3a */ 1 /* : */, /* 0x3a */
/* 0x3b */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5a */
/* 0x5b */ 1 /* [ */, /* 0x5b */
/* 0x5c */ 1 /* \ */, /* 0x5c */
/* 0x5d */ 1 /* ] */, /* 0x5d */
/* 0x5e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x7a */
/* 0x7b */ 1 /* { */, /* 0x7b */
/* 0x7c */ 0, /* 0x7c */
/* 0x7d */ 1 /* } */, /* 0x7d */
/* 0x7e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x9d */
/* 0x9e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xbd */
/* 0xbe */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xdd */
/* 0xde */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xfd */
/* 0xfe */ 0 /* 0xfe */
};
static const uint32_t Special_Endings_bits[0x80 / 32] = {
0b00000000110010000000000000000000,
0b10100000000010000000000000100000,
0b00000000000000000000000000011100,
0b00000000000000000000000000010100
};
/**
* This table contains entries for the allowed whitespace as per RFC 4627
*/
static const char Allowed_Whitespace[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
/* 0x09 */ 1 /* <TAB> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 0,0, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 1 /* <SP> */, /* 0x20 */
/* 0x21 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x40 */
/* 0x41 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x60 */
/* 0x61 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x80 */
/* 0x81 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xa0 */
/* 0xa1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xc0 */
/* 0xc1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xe0 */
/* 0xe1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 /* 0xfe */
};
static const uint32_t Allowed_Whitespace_bits = 0b00000000011001000000000000000000;
static const char String_No_Passthrough[0x100] = {
/* 0x00 */ 1 /* <NUL> */, /* 0x00 */
/* 0x01 */ 1 /* <SOH> */, /* 0x01 */
/* 0x02 */ 1 /* <STX> */, /* 0x02 */
/* 0x03 */ 1 /* <ETX> */, /* 0x03 */
/* 0x04 */ 1 /* <EOT> */, /* 0x04 */
/* 0x05 */ 1 /* <ENQ> */, /* 0x05 */
/* 0x06 */ 1 /* <ACK> */, /* 0x06 */
/* 0x07 */ 1 /* <BEL> */, /* 0x07 */
/* 0x08 */ 1 /* <BS> */, /* 0x08 */
/* 0x09 */ 1 /* <HT> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 1 /* <VT> */, /* 0x0b */
/* 0x0c */ 1 /* <FF> */, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 1 /* <SO> */, /* 0x0e */
/* 0x0f */ 1 /* <SI> */, /* 0x0f */
/* 0x10 */ 1 /* <DLE> */, /* 0x10 */
/* 0x11 */ 1 /* <DC1> */, /* 0x11 */
/* 0x12 */ 1 /* <DC2> */, /* 0x12 */
/* 0x13 */ 1 /* <DC3> */, /* 0x13 */
/* 0x14 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x21 */
/* 0x22 */ 1 /* <"> */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x42 */
/* 0x43 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5b */
/* 0x5c */ 1 /* <\> */, /* 0x5c */
/* 0x5d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x7c */
/* 0x7d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x9c */
/* 0x9d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xbc */
/* 0xbd */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xdc */
/* 0xdd */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xfc */
/* 0xfd */ 0,0, /* 0xfe */
};
/**
* Allowable two-character 'common' escapes:
*/
static const char Allowed_Escapes[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0, /* 0x21 */
/* 0x22 */ 1 /* <"> */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2e */
/* 0x2f */ 1 /* </> */, /* 0x2f */
/* 0x30 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x4f */
/* 0x50 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5b */
/* 0x5c */ 1 /* <\> */, /* 0x5c */
/* 0x5d */ 0,0,0,0,0, /* 0x61 */
/* 0x62 */ 1 /* <b> */, /* 0x62 */
/* 0x63 */ 0,0,0, /* 0x65 */
/* 0x66 */ 1 /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ 1 /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0, /* 0x71 */
/* 0x72 */ 1 /* <r> */, /* 0x72 */
/* 0x73 */ 0, /* 0x73 */
/* 0x74 */ 1 /* <t> */, /* 0x74 */
/* 0x75 */ 1 /* <u> */, /* 0x75 */
/* 0x76 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x95 */
/* 0x96 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb5 */
/* 0xb6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd5 */
/* 0xd6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf5 */
/* 0xf6 */ 0,0,0,0,0,0,0,0,0, /* 0xfe */
};
static const uint32_t Allowed_Escapes_bits[0x80 / 32] = {
0b00000000000000000000000000000000,
0b00100000000000010000000000000000,
0b00000000000000000000000000001000,
0b00100010000000100010110000000000
};
/**
* This table contains the _values_ for a given (single) escaped character.
*/
static unsigned char Escape_Equivs[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x3f */
/* 0x40 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5f */
/* 0x60 */ 0,0, /* 0x61 */
/* 0x62 */ 8 /* <b> */, /* 0x62 */
/* 0x63 */ 0,0,0, /* 0x65 */
/* 0x66 */ 12 /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ 10 /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0, /* 0x71 */
/* 0x72 */ 13 /* <r> */, /* 0x72 */
/* 0x73 */ 0, /* 0x73 */
/* 0x74 */ 9 /* <t> */, /* 0x74 */
/* 0x75 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x94 */
/* 0x95 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb4 */
/* 0xb5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd4 */
/* 0xd5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf4 */
/* 0xf5 */ 0,0,0,0,0,0,0,0,0,0 /* 0xfe */
};
/* Definitions of above-declared static functions */
static char get_escape_equiv(unsigned c) {
switch(c) {
case 'b':
return '\b';
case 'n':
return '\n';
case 'r':
return '\r';
case 't':
return '\t';
case 'f':
return '\f';
}
return 0;
}
static unsigned extract_special(unsigned c) {
return (c < 0x80) ? Special_Table[c & 0xff] : 0;
}
static int is_special_end(unsigned c) {
return (c < 0x80) && (Special_Endings_bits[c >> 5] & (1 << (31 - (c & 31))));
}
static int is_allowed_whitespace(unsigned c) {
return c == ' ' || (c < 0x20 && (Allowed_Whitespace_bits & (1 << (31 - c))));
}
static int is_allowed_escape(unsigned c) {
return (c < 0x80) && (Allowed_Escapes_bits[c >> 5] & (1 << (31 - (c & 31))));
}
static int is_simple_char(unsigned c) {
return !(c < 0x14 || c == '"' || c == '\\');
}
/* Clean up all our macros! */
#undef INCR_METRIC
#undef INCR_GENERIC
#undef INCR_STRINGY_CATCH
#undef CASE_DIGITS
#undef INVOKE_ERROR
#undef STACK_PUSH
#undef STACK_POP_NOPOS
#undef STACK_POP
#undef CALLBACK_AND_POP_NOPOS
#undef CALLBACK_AND_POP
#undef SPECIAL_POP
#undef CUR_CHAR
#undef DO_CALLBACK
#undef ENSURE_HVAL
#undef VERIFY_SPECIAL
#undef STATE_SPECIAL_LENGTH
#undef IS_NORMAL_NUMBER
#undef STATE_NUM_LAST
#undef FASTPARSE_EXHAUSTED
#undef FASTPARSE_BREAK
/**
* JSON Simple/Stacked/Stateful Lexer.
* - Does not buffer data
* - Maintains state
* - Callback oriented
* - Lightweight and fast. One source file and one header file
*
* Copyright (C) 2012-2015 Mark Nunberg
* See included LICENSE file for license details.
*/
#ifndef JSONSL_H_
#define JSONSL_H_
#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <sys/types.h>
#include <wchar.h>
#ifdef __cplusplus
extern "C" {
#endif /* __cplusplus */
#ifdef JSONSL_USE_WCHAR
typedef jsonsl_char_t wchar_t;
typedef jsonsl_uchar_t unsigned wchar_t;
#else
typedef char jsonsl_char_t;
typedef unsigned char jsonsl_uchar_t;
#endif /* JSONSL_USE_WCHAR */
/* Stolen from http-parser.h, and possibly others */
#if defined(_WIN32) && !defined(__MINGW32__) && (!defined(_MSC_VER) || _MSC_VER<1600)
typedef __int8 int8_t;
typedef unsigned __int8 uint8_t;
typedef __int16 int16_t;
typedef unsigned __int16 uint16_t;
typedef __int32 int32_t;
typedef unsigned __int32 uint32_t;
typedef __int64 int64_t;
typedef unsigned __int64 uint64_t;
#if !defined(_MSC_VER) || _MSC_VER<1400
typedef unsigned int size_t;
typedef int ssize_t;
#endif
#else
#include <stdint.h>
#endif
#if (!defined(JSONSL_STATE_GENERIC)) && (!defined(JSONSL_STATE_USER_FIELDS))
#define JSONSL_STATE_GENERIC
#endif /* !defined JSONSL_STATE_GENERIC */
#ifdef JSONSL_STATE_GENERIC
#define JSONSL_STATE_USER_FIELDS
#endif /* JSONSL_STATE_GENERIC */
/* Additional fields for component object */
#ifndef JSONSL_JPR_COMPONENT_USER_FIELDS
#define JSONSL_JPR_COMPONENT_USER_FIELDS
#endif
#ifndef JSONSL_API
/**
* We require a /DJSONSL_DLL so that users already using this as a static
* or embedded library don't get confused
*/
#if defined(_WIN32) && defined(JSONSL_DLL)
#define JSONSL_API __declspec(dllexport)
#else
#define JSONSL_API
#endif /* _WIN32 */
#endif /* !JSONSL_API */
#ifndef JSONSL_INLINE
#if defined(_MSC_VER)
#define JSONSL_INLINE __inline
#elif defined(__GNUC__)
#define JSONSL_INLINE __inline__
#else
#define JSONSL_INLINE inline
#endif /* _MSC_VER or __GNUC__ */
#endif /* JSONSL_INLINE */
#define JSONSL_MAX_LEVELS 512
struct jsonsl_st;
typedef struct jsonsl_st *jsonsl_t;
typedef struct jsonsl_jpr_st* jsonsl_jpr_t;
/**
* This flag is true when AND'd against a type whose value
* must be in "quoutes" i.e. T_HKEY and T_STRING
*/
#define JSONSL_Tf_STRINGY 0xffff00
/**
* Constant representing the special JSON types.
* The values are special and aid in speed (the OBJECT and LIST
* values are the char literals of their openings).
*
* Their actual value is a character which attempts to resemble
* some mnemonic reference to the actual type.
*
* If new types are added, they must fit into the ASCII printable
* range (so they should be AND'd with 0x7f and yield something
* meaningful)
*/
#define JSONSL_XTYPE \
X(STRING, '"'|JSONSL_Tf_STRINGY) \
X(HKEY, '#'|JSONSL_Tf_STRINGY) \
X(OBJECT, '{') \
X(LIST, '[') \
X(SPECIAL, '^') \
X(UESCAPE, 'u')
typedef enum {
#define X(o, c) \
JSONSL_T_##o = c,
JSONSL_XTYPE
JSONSL_T_UNKNOWN = '?',
/* Abstract 'root' object */
JSONSL_T_ROOT = 0
#undef X
} jsonsl_type_t;
/**
* Subtypes for T_SPECIAL. We define them as flags
* because more than one type can be applied to a
* given object.
*/
#define JSONSL_XSPECIAL \
X(NONE, 0) \
X(SIGNED, 1<<0) \
X(UNSIGNED, 1<<1) \
X(TRUE, 1<<2) \
X(FALSE, 1<<3) \
X(NULL, 1<<4) \
X(FLOAT, 1<<5) \
X(EXPONENT, 1<<6) \
X(NONASCII, 1<<7)
typedef enum {
#define X(o,b) \
JSONSL_SPECIALf_##o = b,
JSONSL_XSPECIAL
#undef X
/* Handy flags for checking */
JSONSL_SPECIALf_UNKNOWN = 1 << 8,
/** @private Private */
JSONSL_SPECIALf_ZERO = 1 << 9 | JSONSL_SPECIALf_UNSIGNED,
/** @private */
JSONSL_SPECIALf_DASH = 1 << 10,
/** Type is numeric */
JSONSL_SPECIALf_NUMERIC = (JSONSL_SPECIALf_SIGNED| JSONSL_SPECIALf_UNSIGNED),
/** Type is a boolean */
JSONSL_SPECIALf_BOOLEAN = (JSONSL_SPECIALf_TRUE|JSONSL_SPECIALf_FALSE),
/** Type is an "extended", not integral type (but numeric) */
JSONSL_SPECIALf_NUMNOINT = (JSONSL_SPECIALf_FLOAT|JSONSL_SPECIALf_EXPONENT)
} jsonsl_special_t;
/**
* These are the various types of stack (or other) events
* which will trigger a callback.
* Like the type constants, this are also mnemonic
*/
#define JSONSL_XACTION \
X(PUSH, '+') \
X(POP, '-') \
X(UESCAPE, 'U') \
X(ERROR, '!')
typedef enum {
#define X(a,c) \
JSONSL_ACTION_##a = c,
JSONSL_XACTION
JSONSL_ACTION_UNKNOWN = '?'
#undef X
} jsonsl_action_t;
/**
* Various errors which may be thrown while parsing JSON
*/
#define JSONSL_XERR \
/* Trailing garbage characters */ \
X(GARBAGE_TRAILING) \
/* We were expecting a 'special' (numeric, true, false, null) */ \
X(SPECIAL_EXPECTED) \
/* The 'special' value was incomplete */ \
X(SPECIAL_INCOMPLETE) \
/* Found a stray token */ \
X(STRAY_TOKEN) \
/* We were expecting a token before this one */ \
X(MISSING_TOKEN) \
/* Cannot insert because the container is not ready */ \
X(CANT_INSERT) \
/* Found a '\' outside a string */ \
X(ESCAPE_OUTSIDE_STRING) \
/* Found a ':' outside of a hash */ \
X(KEY_OUTSIDE_OBJECT) \
/* found a string outside of a container */ \
X(STRING_OUTSIDE_CONTAINER) \
/* Found a null byte in middle of string */ \
X(FOUND_NULL_BYTE) \
/* Current level exceeds limit specified in constructor */ \
X(LEVELS_EXCEEDED) \
/* Got a } as a result of an opening [ or vice versa */ \
X(BRACKET_MISMATCH) \
/* We expected a key, but got something else instead */ \
X(HKEY_EXPECTED) \
/* We got an illegal control character (bad whitespace or something) */ \
X(WEIRD_WHITESPACE) \
/* Found a \u-escape, but there were less than 4 following hex digits */ \
X(UESCAPE_TOOSHORT) \
/* Invalid two-character escape */ \
X(ESCAPE_INVALID) \
/* Trailing comma */ \
X(TRAILING_COMMA) \
/* An invalid number was passed in a numeric field */ \
X(INVALID_NUMBER) \
/* Value is missing for object */ \
X(VALUE_EXPECTED) \
/* The following are for JPR Stuff */ \
\
/* Found a literal '%' but it was only followed by a single valid hex digit */ \
X(PERCENT_BADHEX) \
/* jsonpointer URI is malformed '/' */ \
X(JPR_BADPATH) \
/* Duplicate slash */ \
X(JPR_DUPSLASH) \
/* No leading root */ \
X(JPR_NOROOT) \
/* Allocation failure */ \
X(ENOMEM) \
/* Invalid unicode codepoint detected (in case of escapes) */ \
X(INVALID_CODEPOINT)
typedef enum {
JSONSL_ERROR_SUCCESS = 0,
#define X(e) \
JSONSL_ERROR_##e,
JSONSL_XERR
#undef X
JSONSL_ERROR_GENERIC
} jsonsl_error_t;
/**
* A state is a single level of the stack.
* Non-private data (i.e. the 'data' field, see the STATE_GENERIC section)
* will remain in tact until the item is popped.
*
* As a result, it means a parent state object may be accessed from a child
* object, (the parents fields will all be valid). This allows a user to create
* an ad-hoc hierarchy on top of the JSON one.
*
*/
struct jsonsl_state_st {
/**
* The JSON object type
*/
unsigned int type;
/**
* The position (in terms of number of bytes since the first call to
* jsonsl_feed()) at which the state was first pushed. This includes
* opening tokens, if applicable.
*
* @note For strings (i.e. type & JSONSL_Tf_STRINGY is nonzero) this will
* be the position of the first quote.
*
* @see jsonsl_st::pos which contains the _current_ position and can be
* used during a POP callback to get the length of the element.
*/
size_t pos_begin;
/**FIXME: This is redundant as the same information can be derived from
* jsonsl_st::pos at pop-time */
size_t pos_cur;
/** If this element is special, then its extended type is here */
unsigned short special_flags;
/**
* Level of recursion into nesting. This is mainly a convenience
* variable, as this can technically be deduced from the lexer's
* level parameter (though the logic is not that simple)
*/
unsigned short level;
/**
* how many elements in the object/list.
* For objects (hashes), an element is either
* a key or a value. Thus for one complete pair,
* nelem will be 2.
*
* For special types, this will hold the sum of the digits.
* This only holds true for values which are simple signed/unsigned
* numbers. Otherwise a special flag is set, and extra handling is not
* performed.
*/
uint32_t nelem;
/*TODO: merge this and special_flags into a union */
/**
* Useful for an opening nest, this will prevent a callback from being
* invoked on this item or any of its children
*/
int ignore_callback : 1;
/**
* Counter which is incremented each time an escape ('\') is encountered.
* This is used internally for non-string types and should only be
* inspected by the user if the state actually represents a string
* type.
*/
unsigned int nescapes : 31;
/**
* Put anything you want here. if JSONSL_STATE_USER_FIELDS is here, then
* the macro expansion happens here.
*
* You can use these fields to store hierarchical or 'tagging' information
* for specific objects.
*
* See the documentation above for the lifetime of the state object (i.e.
* if the private data points to allocated memory, it should be freed
* when the object is popped, as the state object will be re-used)
*/
#ifndef JSONSL_STATE_GENERIC
JSONSL_STATE_USER_FIELDS
#else
/**
* Otherwise, this is a simple void * pointer for anything you want
*/
void *data;
#endif /* JSONSL_STATE_USER_FIELDS */
};
/**Gets the number of elements in the list.
* @param st The state. Must be of type JSONSL_T_LIST
* @return number of elements in the list
*/
#define JSONSL_LIST_SIZE(st) ((st)->nelem)
/**Gets the number of key-value pairs in an object
* @param st The state. Must be of type JSONSL_T_OBJECT
* @return the number of key-value pairs in the object
*/
#define JSONSL_OBJECT_SIZE(st) ((st)->nelem / 2)
/**Gets the numeric value.
* @param st The state. Must be of type JSONSL_T_SPECIAL and
* special_flags must have the JSONSL_SPECIALf_NUMERIC flag
* set.
* @return the numeric value of the state.
*/
#define JSONSL_NUMERIC_VALUE(st) ((st)->nelem)
/*
* So now we need some special structure for keeping the
* JPR info in sync. Preferrably all in a single block
* of memory (there's no need for separate allocations.
* So we will define a 'table' with the following layout
*
* Level nPosbl JPR1_last JPR2_last JPR3_last
*
* 0 1 NOMATCH POSSIBLE POSSIBLE
* 1 0 NOMATCH NOMATCH COMPLETE
* [ table ends here because no further path is possible]
*
* Where the JPR..n corresponds to the number of JPRs
* requested, and nPosble is a quick flag to determine
*
* the number of possibilities. In the future this might
* be made into a proper 'jump' table,
*
* Since we always mark JPRs from the higher levels descending
* into the lower ones, a prospective child match would first
* look at the parent table to check the possibilities, and then
* see which ones were possible..
*
* Thus, the size of this blob would be (and these are all ints here)
* nLevels * nJPR * 2.
*
* the 'Width' of the table would be nJPR*2, and the 'height' would be
* nlevels
*/
/**
* This is called when a stack change ocurs.
*
* @param jsn The lexer
* @param action The type of action, this can be PUSH or POP
* @param state A pointer to the stack currently affected by the action
* @param at A pointer to the position of the input buffer which triggered
* this action.
*/
typedef void (*jsonsl_stack_callback)(
jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st* state,
const jsonsl_char_t *at);
/**
* This is called when an error is encountered.
* Sometimes it's possible to 'erase' characters (by replacing them
* with whitespace). If you think you have corrected the error, you
* can return a true value, in which case the parser will backtrack
* and try again.
*
* @param jsn The lexer
* @param error The error which was thrown
* @param state the current state
* @param a pointer to the position of the input buffer which triggered
* the error. Note that this is not const, this is because you have the
* possibility of modifying the character in an attempt to correct the
* error
*
* @return zero to bail, nonzero to try again (this only makes sense if
* the input buffer has been modified by this callback)
*/
typedef int (*jsonsl_error_callback)(
jsonsl_t jsn,
jsonsl_error_t error,
struct jsonsl_state_st* state,
jsonsl_char_t *at);
struct jsonsl_st {
/** Public, read-only */
/** This is the current level of the stack */
unsigned int level;
/** Flag set to indicate we should stop processing */
unsigned int stopfl;
/**
* This is the current position, relative to the beginning
* of the stream.
*/
size_t pos;
/** This is the 'bytes' variable passed to feed() */
const jsonsl_char_t *base;
/** Callback invoked for PUSH actions */
jsonsl_stack_callback action_callback_PUSH;
/** Callback invoked for POP actions */
jsonsl_stack_callback action_callback_POP;
/** Default callback for any action, if neither PUSH or POP callbacks are defined */
jsonsl_stack_callback action_callback;
/**
* Do not invoke callbacks for objects deeper than this level.
* NOTE: This field establishes the lower bound for ignored callbacks,
* and is thus misnamed. `min_ignore_level` would actually make more
* sense, but we don't want to break API.
*/
unsigned int max_callback_level;
/** The error callback. Invoked when an error happens. Should not be NULL */
jsonsl_error_callback error_callback;
/* these are boolean flags you can modify. You will be called
* about notification for each of these types if the corresponding
* variable is true.
*/
/**
* @name Callback Booleans.
* These determine whether a callback is to be invoked for certain types of objects
* @{*/
/** Boolean flag to enable or disable the invokcation for events on this type*/
int call_SPECIAL;
int call_OBJECT;
int call_LIST;
int call_STRING;
int call_HKEY;
/*@}*/
/**
* @name u-Escape handling
* Special handling for the \\u-f00d type sequences. These are meant
* to be translated back into the corresponding octet(s).
* A special callback (if set) is invoked with *at=='u'. An application
* may wish to temporarily suspend parsing and handle the 'u-' sequence
* internally (or not).
*/
/*@{*/
/** Callback to be invoked for a u-escape */
jsonsl_stack_callback action_callback_UESCAPE;
/** Boolean flag, whether to invoke the callback */
int call_UESCAPE;
/** Boolean flag, whether we should return after encountering a u-escape:
* the callback is invoked and then we return if this is true
*/
int return_UESCAPE;
/*@}*/
struct {
int allow_trailing_comma;
} options;
/** Put anything here */
void *data;
/*@{*/
/** Private */
int in_escape;
char expecting;
char tok_last;
int can_insert;
unsigned int levels_max;
#ifndef JSONSL_NO_JPR
size_t jpr_count;
jsonsl_jpr_t *jprs;
/* Root pointer for JPR matching information */
size_t *jpr_root;
#endif /* JSONSL_NO_JPR */
/*@}*/
/**
* This is the stack. Its upper bound is levels_max, or the
* nlevels argument passed to jsonsl_new. If you modify this structure,
* make sure that this member is last.
*/
struct jsonsl_state_st stack[1];
};
/**
* Creates a new lexer object, with capacity for recursion up to nlevels
*
* @param nlevels maximum recursion depth
*/
JSONSL_API
jsonsl_t jsonsl_new(int nlevels);
JSONSL_API
jsonsl_t jsonsl_init(jsonsl_t jsn, int nlevels);
JSONSL_API
size_t jsonsl_get_size(int nlevels);
/**
* Feeds data into the lexer.
*
* @param jsn the lexer object
* @param bytes new data to be fed
* @param nbytes size of new data
*/
JSONSL_API
void jsonsl_feed(jsonsl_t jsn, const jsonsl_char_t *bytes, size_t nbytes);
/**
* Resets the internal parser state. This does not free the parser
* but does clean it internally, so that the next time feed() is called,
* it will be treated as a new stream
*
* @param jsn the lexer
*/
JSONSL_API
void jsonsl_reset(jsonsl_t jsn);
/**
* Frees the lexer, cleaning any allocated memory taken
*
* @param jsn the lexer
*/
JSONSL_API
void jsonsl_destroy(jsonsl_t jsn);
/**
* Gets the 'parent' element, given the current one
*
* @param jsn the lexer
* @param cur the current nest, which should be a struct jsonsl_nest_st
*/
static JSONSL_INLINE
struct jsonsl_state_st *jsonsl_last_state(const jsonsl_t jsn,
const struct jsonsl_state_st *state)
{
/* Don't complain about overriding array bounds */
if (state->level > 1) {
return jsn->stack + state->level - 1;
} else {
return NULL;
}
}
/**
* Gets the state of the last fully consumed child of this parent. This is
* only valid in the parent's POP callback.
*
* @param the lexer
* @return A pointer to the child.
*/
static JSONSL_INLINE
struct jsonsl_state_st *jsonsl_last_child(const jsonsl_t jsn,
const struct jsonsl_state_st *parent)
{
return jsn->stack + (parent->level + 1);
}
/**Call to instruct the parser to stop parsing and return. This is valid
* only from within a callback */
static JSONSL_INLINE
void jsonsl_stop(jsonsl_t jsn)
{
jsn->stopfl = 1;
}
/**
* This enables receiving callbacks on all events. Doesn't do
* anything special but helps avoid some boilerplate.
* This does not touch the UESCAPE callbacks or flags.
*/
static JSONSL_INLINE
void jsonsl_enable_all_callbacks(jsonsl_t jsn)
{
jsn->call_HKEY = 1;
jsn->call_STRING = 1;
jsn->call_OBJECT = 1;
jsn->call_SPECIAL = 1;
jsn->call_LIST = 1;
}
/**
* A macro which returns true if the current state object can
* have children. This means a list type or an object type.
*/
#define JSONSL_STATE_IS_CONTAINER(state) \
(state->type == JSONSL_T_OBJECT || state->type == JSONSL_T_LIST)
/**
* These two functions, dump a string representation
* of the error or type, respectively. They will never
* return NULL
*/
JSONSL_API
const char* jsonsl_strerror(jsonsl_error_t err);
JSONSL_API
const char* jsonsl_strtype(jsonsl_type_t jt);
/**
* Dumps global metrics to the screen. This is a noop unless
* jsonsl was compiled with JSONSL_USE_METRICS
*/
JSONSL_API
void jsonsl_dump_global_metrics(void);
/* This macro just here for editors to do code folding */
#ifndef JSONSL_NO_JPR
/**
* @name JSON Pointer API
*
* JSONPointer API. This isn't really related to the lexer (at least not yet)
* JSONPointer provides an extremely simple specification for providing
* locations within JSON objects. We will extend it a bit and allow for
* providing 'wildcard' characters by which to be able to 'query' the stream.
*
* See http://tools.ietf.org/html/draft-pbryan-zyp-json-pointer-00
*
* Currently I'm implementing the 'single query' API which can only use a single
* query component. In the future I will integrate my yet-to-be-published
* Boyer-Moore-esque prefix searching implementation, in order to allow
* multiple paths to be merged into one for quick and efficient searching.
*
*
* JPR (as we'll refer to it within the source) can be used by splitting
* the components into mutliple sections, and incrementally 'track' each
* component. When JSONSL delivers a 'pop' callback for a string, or a 'push'
* callback for an object, we will check to see whether the index matching
* the component corresponding to the current level contains a match
* for our path.
*
* In order to do this properly, a structure must be maintained within the
* parent indicating whether its children are possible matches. This flag
* will be 'inherited' by call children which may conform to the match
* specification, and discarded by all which do not (thereby eliminating
* their children from inheriting it).
*
* A successful match is a complete one. One can provide multiple paths with
* multiple levels of matches e.g.
* /foo/bar/baz/^/blah
*
* @{
*/
/** The wildcard character */
#ifndef JSONSL_PATH_WILDCARD_CHAR
#define JSONSL_PATH_WILDCARD_CHAR '^'
#endif /* WILDCARD_CHAR */
#define JSONSL_XMATCH \
X(COMPLETE,1) \
X(POSSIBLE,0) \
X(NOMATCH,-1) \
X(TYPE_MISMATCH, -2)
typedef enum {
#define X(T,v) \
JSONSL_MATCH_##T = v,
JSONSL_XMATCH
#undef X
JSONSL_MATCH_UNKNOWN
} jsonsl_jpr_match_t;
typedef enum {
JSONSL_PATH_STRING = 1,
JSONSL_PATH_WILDCARD,
JSONSL_PATH_NUMERIC,
JSONSL_PATH_ROOT,
/* Special */
JSONSL_PATH_INVALID = -1,
JSONSL_PATH_NONE = 0
} jsonsl_jpr_type_t;
struct jsonsl_jpr_component_st {
/** The string the component points to */
char *pstr;
/** if this is a numeric type, the number is 'cached' here */
unsigned long idx;
/** The length of the string */
size_t len;
/** The type of component (NUMERIC or STRING) */
jsonsl_jpr_type_t ptype;
/** Set this to true to enforce type checking between dict keys and array
* indices. jsonsl_jpr_match() will return TYPE_MISMATCH if it detects
* that an array index is actually a child of a dictionary. */
short is_arridx;
/* Extra fields (for more advanced searches. Default is empty) */
JSONSL_JPR_COMPONENT_USER_FIELDS
};
struct jsonsl_jpr_st {
/** Path components */
struct jsonsl_jpr_component_st *components;
size_t ncomponents;
/**Type of the match to be expected. If nonzero, will be compared against
* the actual type */
unsigned match_type;
/** Base of allocated string for components */
char *basestr;
/** The original match string. Useful for returning to the user */
char *orig;
size_t norig;
};
/**
* Create a new JPR object.
*
* @param path the JSONPointer path specification.
* @param errp a pointer to a jsonsl_error_t. If this function returns NULL,
* then more details will be in this variable.
*
* @return a new jsonsl_jpr_t object, or NULL on error.
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_new(const char *path, jsonsl_error_t *errp);
/**
* Destroy a JPR object
*/
JSONSL_API
void jsonsl_jpr_destroy(jsonsl_jpr_t jpr);
/**
* Match a JSON object against a type and specific level
*
* @param jpr the JPR object
* @param parent_type the type of the parent (should be T_LIST or T_OBJECT)
* @param parent_level the level of the parent
* @param key the 'key' of the child. If the parent is an array, this should be
* empty.
* @param nkey - the length of the key. If the parent is an array (T_LIST), then
* this should be the current index.
*
* NOTE: The key of the child means any kind of associative data related to the
* element. Thus: <<< { "foo" : [ >>,
* the opening array's key is "foo".
*
* @return a status constant. This indicates whether a match was excluded, possible,
* or successful.
*/
JSONSL_API
jsonsl_jpr_match_t jsonsl_jpr_match(jsonsl_jpr_t jpr,
unsigned int parent_type,
unsigned int parent_level,
const char *key, size_t nkey);
/**
* Alternate matching algorithm. This matching algorithm does not use
* JSONPointer but relies on a more structured searching mechanism. It
* assumes that there is a clear distinction between array indices and
* object keys. In this case, the jsonsl_path_component_st::ptype should
* be set to @ref JSONSL_PATH_NUMERIC for an array index (the
* jsonsl_path_comonent_st::is_arridx field will be removed in a future
* version).
*
* @param jpr The path
* @param parent The parent structure. Can be NULL if this is the root object
* @param child The child structure. Should not be NULL
* @param key Object key, if an object
* @param nkey Length of object key
* @return Status constant if successful
*
* @note
* For successful matching, both the key and the path itself should be normalized
* to contain 'proper' utf8 sequences rather than utf16 '\uXXXX' escapes. This
* should currently be done in the application. Another version of this function
* may use a temporary buffer in such circumstances (allocated by the application).
*
* Since this function also checks the state of the child, it should only
* be called on PUSH callbacks, and not POP callbacks
*/
JSONSL_API
jsonsl_jpr_match_t
jsonsl_path_match(jsonsl_jpr_t jpr,
const struct jsonsl_state_st *parent,
const struct jsonsl_state_st *child,
const char *key, size_t nkey);
/**
* Associate a set of JPR objects with a lexer instance.
* This should be called before the lexer has been fed any data (and
* behavior is undefined if you don't adhere to this).
*
* After using this function, you may subsequently call match_state() on
* given states (presumably from within the callbacks).
*
* Note that currently the first JPR is the quickest and comes
* pre-allocated with the state structure. Further JPR objects
* are chained.
*
* @param jsn The lexer
* @param jprs An array of jsonsl_jpr_t objects
* @param njprs How many elements in the jprs array.
*/
JSONSL_API
void jsonsl_jpr_match_state_init(jsonsl_t jsn,
jsonsl_jpr_t *jprs,
size_t njprs);
/**
* This follows the same semantics as the normal match,
* except we infer parent and type information from the relevant state objects.
* The match status (for all possible JPR objects) is set in the *out parameter.
*
* If a match has succeeded, then its JPR object will be returned. In all other
* instances, NULL is returned;
*
* @param jpr The jsonsl_jpr_t handle
* @param state The jsonsl_state_st which is a candidate
* @param key The hash key (if applicable, can be NULL if parent is list)
* @param nkey Length of hash key (if applicable, can be zero if parent is list)
* @param out A pointer to a jsonsl_jpr_match_t. This will be populated with
* the match result
*
* @return If a match was completed in full, then the JPR object containing
* the matching path will be returned. Otherwise, the return is NULL (note, this
* does not mean matching has failed, it can still be part of the match: check
* the out parameter).
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_match_state(jsonsl_t jsn,
struct jsonsl_state_st *state,
const char *key,
size_t nkey,
jsonsl_jpr_match_t *out);
/**
* Cleanup any memory allocated and any states set by
* match_state_init() and match_state()
* @param jsn The lexer
*/
JSONSL_API
void jsonsl_jpr_match_state_cleanup(jsonsl_t jsn);
/**
* Return a string representation of the match result returned by match()
*/
JSONSL_API
const char *jsonsl_strmatchtype(jsonsl_jpr_match_t match);
/* @}*/
/**
* Utility function to convert escape sequences into their original form.
*
* The decoders I've sampled do not seem to specify a standard behavior of what
* to escape/unescape.
*
* RFC 4627 Mandates only that the quoute, backslash, and ASCII control
* characters (0x00-0x1f) be escaped. It is often common for applications
* to escape a '/' - however this may also be desired behavior. the JSON
* spec is not clear on this, and therefore jsonsl leaves it up to you.
*
* Additionally, sometimes you may wish to _normalize_ JSON. This is specifically
* true when dealing with 'u-escapes' which can be expressed perfectly fine
* as utf8. One use case for normalization is JPR string comparison, in which
* case two effectively equivalent strings may not match because one is using
* u-escapes and the other proper utf8. To normalize u-escapes only, pass in
* an empty `toEscape` table, enabling only the `u` index.
*
* @param in The input string.
* @param out An allocated output (should be the same size as in)
* @param len the size of the buffer
* @param toEscape - A sparse array of characters to unescape. Characters
* which are not present in this array, e.g. toEscape['c'] == 0 will be
* ignored and passed to the output in their original form.
* @param oflags If not null, and a \uXXXX escape expands to a non-ascii byte,
* then this variable will have the SPECIALf_NONASCII flag on.
*
* @param err A pointer to an error variable. If an error ocurrs, it will be
* set in this variable
* @param errat If not null and an error occurs, this will be set to point
* to the position within the string at which the offending character was
* encountered.
*
* @return The effective size of the output buffer.
*
* @note
* This function now encodes the UTF8 equivalents of utf16 escapes (i.e.
* 'u-escapes'). Previously this would encode the escapes as utf16 literals,
* which while still correct in some sense was confusing for many (especially
* considering that the inputs were variations of char).
*
* @note
* The output buffer will never be larger than the input buffer, since
* standard escape sequences (i.e. '\t') occupy two bytes in the source
* but only one byte (when unescaped) in the output. Likewise u-escapes
* (i.e. \uXXXX) will occupy six bytes in the source, but at the most
* two bytes when escaped.
*/
JSONSL_API
size_t jsonsl_util_unescape_ex(const char *in,
char *out,
size_t len,
const int toEscape[128],
unsigned *oflags,
jsonsl_error_t *err,
const char **errat);
/**
* Convenience macro to avoid passing too many parameters
*/
#define jsonsl_util_unescape(in, out, len, toEscape, err) \
jsonsl_util_unescape_ex(in, out, len, toEscape, NULL, err, NULL)
#endif /* JSONSL_NO_JPR */
#ifdef __cplusplus
}
#endif /* __cplusplus */
#endif /* JSONSL_H_ */
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment