四种json c parser的兼容性比较

四种json c parser的兼容性比较

最近一个移动终端项目的服务器和客户端之间用到了json(http://www.json.org) 来交换数据。json是一种类似于xml的描述性语言，不过相对于xml来说，json这种格式比较简洁，可以降低服务器和客户端之间的流量。它的格式类似于：
    {
      "this": "is",
      "really": "simple",
      "json": "right?"
    }



在json的网站一共有四种C parser，解析json格式我不知道哪种比较实用，花了点时间测试了一下兼容性（我比较关注，性能应该差不多）。现在把结果贴出来供大家参考。

测试的文件从yajl里的cases获取
yajl-0.4.0
    [PASS] /home/lyb/cases/array.json
    [PASS] /home/lyb/cases/codepoints_from_unicode_org.json
    [PASS] /home/lyb/cases/dc_simple_with_comments.json
    [PASS] /home/lyb/cases/deep_arrays.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case_with_comments.json
    [PASS] /home/lyb/cases/doubles.json
    [PASS] /home/lyb/cases/empty_array.json
    [PASS] /home/lyb/cases/escaped_bulgarian.json
    [PASS] /home/lyb/cases/escaped_foobar.json
    [PASS] /home/lyb/cases/isolated_surrogate_marker.json
    [PASS] /home/lyb/cases/nulls_and_bools.json
    [PASS] /home/lyb/cases/simple.json
    [PASS] /home/lyb/cases/simple_with_comments.json
    [PASS] /home/lyb/cases/string_with_escapes.json
    [PASS] /home/lyb/cases/unescaped_bulgarian.json
    total: 54
    pass: 16
    fail: 38



json-c-0.7
    [PASS] /home/lyb/cases/array.json
    [PASS] /home/lyb/cases/codepoints_from_unicode_org.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case_with_comments.json
    [PASS] /home/lyb/cases/doubles.json
    [PASS] /home/lyb/cases/empty_array.json
    [PASS] /home/lyb/cases/escaped_bulgarian.json
    [PASS] /home/lyb/cases/escaped_foobar.json
    [PASS] /home/lyb/cases/integers.json
    [PASS] /home/lyb/cases/invalid_utf8.json
    [PASS] /home/lyb/cases/isolated_surrogate_marker.json
    [PASS] /home/lyb/cases/leading_zero_in_number.json
    [PASS] /home/lyb/cases/missing_integer_after_decimal_point.json
    [PASS] /home/lyb/cases/missing_integer_after_exponent.json
    [PASS] /home/lyb/cases/non_utf8_char_in_string.json
    [PASS] /home/lyb/cases/nulls_and_bools.json
    [PASS] /home/lyb/cases/simple.json
    [PASS] /home/lyb/cases/string_with_invalid_newline.json
    [PASS] /home/lyb/cases/unescaped_bulgarian.json
    total: 54
    pass: 19
    fail: 35



mjson
    [PASS] /home/lyb/cases/array.json
    [PASS] /home/lyb/cases/deep_arrays.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case.json
    [PASS] /home/lyb/cases/doubles.json
    [PASS] /home/lyb/cases/empty_array.json
    [PASS] /home/lyb/cases/escaped_bulgarian.json
    [PASS] /home/lyb/cases/integers.json
    [PASS] /home/lyb/cases/invalid_utf8.json
    [PASS] /home/lyb/cases/non_utf8_char_in_string.json
    [PASS] /home/lyb/cases/nulls_and_bools.json
    [PASS] /home/lyb/cases/simple.json
    [PASS] /home/lyb/cases/unescaped_bulgarian.json
    total: 54
    pass: 12
    fail: 42



json_parser
    [PASS] /home/lyb/cases/array.json
    [PASS] /home/lyb/cases/difficult_json_c_test_case.json
    [PASS] /home/lyb/cases/doubles.json
    [PASS] /home/lyb/cases/empty_array.json
    [PASS] /home/lyb/cases/escaped_bulgarian.json
    [PASS] /home/lyb/cases/integers.json
    [PASS] /home/lyb/cases/invalid_utf8.json
    [PASS] /home/lyb/cases/non_utf8_char_in_string.json
    [PASS] /home/lyb/cases/nulls_and_bools.json
    [PASS] /home/lyb/cases/simple.json
    [PASS] /home/lyb/cases/string_with_escapes.json
    [PASS] /home/lyb/cases/unescaped_bulgarian.json
    total: 54
    pass: 12
    fail: 42



结果分析：

从各个通过的文件数量看， json-c兼容性是最好的

把结果都放到tmp.txt， 执行
    [lyb@localhost json-parser]$  cat tmp.txt | sort | uniq -c | sort -r

          4 [PASS] /home/lyb/cases/unescaped_bulgarian.json
          4 [PASS] /home/lyb/cases/simple.json
          4 [PASS] /home/lyb/cases/nulls_and_bools.json
          4 [PASS] /home/lyb/cases/escaped_bulgarian.json
          4 [PASS] /home/lyb/cases/empty_array.json
          4 [PASS] /home/lyb/cases/doubles.json
          4 [PASS] /home/lyb/cases/difficult_json_c_test_case.json
          4 [PASS] /home/lyb/cases/array.json
          3 [PASS] /home/lyb/cases/non_utf8_char_in_string.json
          3 [PASS] /home/lyb/cases/invalid_utf8.json
          3 [PASS] /home/lyb/cases/integers.json
          2 [PASS] /home/lyb/cases/string_with_escapes.json
          2 [PASS] /home/lyb/cases/isolated_surrogate_marker.json
          2 [PASS] /home/lyb/cases/escaped_foobar.json
          2 [PASS] /home/lyb/cases/difficult_json_c_test_case_with_comments.json
          2 [PASS] /home/lyb/cases/deep_arrays.json
          2 [PASS] /home/lyb/cases/codepoints_from_unicode_org.json
          1 [PASS] /home/lyb/cases/string_with_invalid_newline.json
          1 [PASS] /home/lyb/cases/simple_with_comments.json
          1 [PASS] /home/lyb/cases/missing_integer_after_exponent.json
          1 [PASS] /home/lyb/cases/missing_integer_after_decimal_point.json
          1 [PASS] /home/lyb/cases/leading_zero_in_number.json
          1 [PASS] /home/lyb/cases/dc_simple_with_comments.json



有8个文件都通过了4个parser的测试。
          4 [PASS] /home/lyb/cases/unescaped_bulgarian.json
          4 [PASS] /home/lyb/cases/simple.json
          4 [PASS] /home/lyb/cases/nulls_and_bools.json
          4 [PASS] /home/lyb/cases/escaped_bulgarian.json
          4 [PASS] /home/lyb/cases/empty_array.json
          4 [PASS] /home/lyb/cases/doubles.json
          4 [PASS] /home/lyb/cases/difficult_json_c_test_case.json
          4 [PASS] /home/lyb/cases/array.json



json-c-0.7，mjson， json-parser 都通过了下面3个文件的测试
          3 [PASS] /home/lyb/cases/non_utf8_char_in_string.json
          3 [PASS] /home/lyb/cases/invalid_utf8.json
          3 [PASS] /home/lyb/cases/integers.json


可以看出yajl-0.4.0的非utf-8支持不如其它3个

四个模块的一些特点：
mjson和json-parser只有一两个文件，适合放到单个模块里用
json-c采用autoconf， 适合库形式
yajl采用cmake， 测试做的比较好
mjson和yajl都支持sax事件解析方法

~ END ~