--- parser3/src/classes/json.C 2012/06/21 14:22:09 1.24 +++ parser3/src/classes/json.C 2016/11/01 23:10:40 1.48 @@ -1,7 +1,7 @@ /** @file Parser: @b json parser class. - Copyright (c) 2000-2012 Art. Lebedev Studio (http://www.artlebedev.com) + Copyright (c) 2000-2015 Art. Lebedev Studio (http://www.artlebedev.com) */ #include "classes.h" @@ -12,13 +12,13 @@ #include "pa_charset.h" #include "pa_charsets.h" -#include "JSON_parser.h" +#include "pa_json.h" #ifdef XML #include "pa_vxdoc.h" #endif -volatile const char * IDENT_JSON_C="$Id: json.C,v 1.24 2012/06/21 14:22:09 moko Exp $"; +volatile const char * IDENT_JSON_C="$Id: json.C,v 1.48 2016/11/01 23:10:40 moko Exp $"; // class @@ -29,7 +29,7 @@ public: // global variable -DECLARE_CLASS_VAR(json, new MJson, 0); +DECLARE_CLASS_VAR(json, new MJson); // methods struct Json { @@ -47,10 +47,12 @@ struct Json { String::Language taint; bool handle_double; + bool handle_int; enum Distinct { D_EXCEPTION, D_FIRST, D_LAST, D_ALL } distinct; Json(Charset* acharset): stack(), key_stack(), key(NULL), result(NULL), hook_object(NULL), hook_array(NULL), - request(NULL), charset(acharset), taint(String::L_TAINTED), handle_double(true), distinct(D_EXCEPTION){} + request(NULL), charset(acharset), taint(String::L_TAINTED), handle_double(true), handle_int(true), + distinct(D_EXCEPTION){} bool set_distinct(const String &value){ if (value == "first") distinct = D_FIRST; @@ -91,11 +93,11 @@ static void set_json_value(Json *json, V } } -String* json_string(Json *json, const JSON_value* value){ +String* json_string(Json *json, const char *value, uint32_t length){ String::C result = json->charset !=NULL ? - Charset::transcode(String::C(value->vu.str.value, value->vu.str.length), UTF8_charset, *json->charset) : - String::C(pa_strdup(value->vu.str.value, value->vu.str.length), value->vu.str.length); - return new String(result.str, json->taint, result.length); + Charset::transcode(String::C(value, length), pa_UTF8_charset, *json->charset) : + String::C(pa_strdup(value, length), length); + return new String(result, json->taint); } static Value *json_hook(Request &r, Junction *hook, String* key, Value* value){ @@ -105,13 +107,13 @@ static Value *json_hook(Request &r, Junc frame.store_params(params, 2); r.execute_method(frame); - return &frame.result().as_value(); + return &frame.result(); } -static int json_callback(Json *json, int type, const JSON_value* value) +static int json_callback(Json *json, int type, const char *value, uint32_t length) { switch(type) { - case JSON_T_OBJECT_BEGIN:{ + case JSON_OBJECT_BEGIN:{ VHash *v = new VHash(); if (json->hook_object){ json->key_stack.push(json->key); @@ -122,7 +124,7 @@ static int json_callback(Json *json, int json->stack.push(v); break; } - case JSON_T_OBJECT_END:{ + case JSON_OBJECT_END:{ if (json->hook_object){ String* key = json->key_stack.pop(); json->result = json_hook(*json->request, json->hook_object, key, json->stack.pop()); @@ -136,7 +138,7 @@ static int json_callback(Json *json, int } break; } - case JSON_T_ARRAY_BEGIN:{ + case JSON_ARRAY_BEGIN:{ VHash *v = new VHash(); if (json->hook_array){ json->key_stack.push(json->key); @@ -147,7 +149,7 @@ static int json_callback(Json *json, int json->stack.push(v); break; } - case JSON_T_ARRAY_END: + case JSON_ARRAY_END: // libjson supports array at top level, we too if (json->hook_array){ String* key = json->key_stack.pop(); @@ -161,75 +163,149 @@ static int json_callback(Json *json, int json->result = json->stack.pop(); } break; - case JSON_T_KEY: - json->key = json_string(json, value); + case JSON_KEY: + json->key = json_string(json, value, length); break; - case JSON_T_INTEGER: - set_json_value(json, new VDouble((double)value->vu.integer_value)); + case JSON_INT: + if (json->handle_int){ + set_json_value(json, new VDouble( json_string(json, value, length)->as_double() )); + } else { + // JSON_STRING + set_json_value(json, new VString(*json_string(json, value, length))); + } break; - case JSON_T_FLOAT: + case JSON_FLOAT: if (json->handle_double){ - set_json_value(json, new VDouble( json_string(json, value)->as_double() )); + set_json_value(json, new VDouble( json_string(json, value, length)->as_double() )); break; - } // else is JSON_T_STRING - case JSON_T_STRING: - set_json_value(json, new VString(*json_string(json, value))); + } // else is JSON_STRING + case JSON_STRING: + set_json_value(json, new VString(*json_string(json, value, length))); break; - case JSON_T_NULL: + case JSON_NULL: set_json_value(json, VVoid::get()); break; - case JSON_T_TRUE: + case JSON_TRUE: set_json_value(json, &VBool::get(true)); break; - case JSON_T_FALSE: + case JSON_FALSE: set_json_value(json, &VBool::get(false)); - break; + break; } - return 1; + return 0; } static const char* json_error_message(int error_code){ static const char* error_messages[] = { NULL, - "invalid char", - "invalid keyword", - "invalid escape sequence", - "invalid unicode sequence", - "invalid number", - "nesting depth reached", - "unbalanced collection", - "expected key", - "expected colon", - "out of memory" + "out of memory", + "bad character", + "stack empty", + "pop unexpected mode", + "nesting limit", + "data limit", + "comment not allowed by config", + "unexpected character", + "missing unicode low surrogate", + "unexpected unicode low surrogate", + "error comma out of structure", + "error in a callback" }; return error_messages[error_code]; } extern String::Language get_untaint_lang(const String& lang_name); +#define SOURCE_MAX_LEN 60 + +void json_exception_with_source(Request& r, const char* msg, const char* json, int offset){ + int i; + + int line=0; + int start=0; + int end=strlen(json); + + if(offset>end) + offset=end; + + for(i = 0; i < offset; i++){ + if(json[i]=='\n'){ + line++; + } + } + + if(offset > SOURCE_MAX_LEN/2) + start = offset - SOURCE_MAX_LEN/2; + + for(i = offset-1; i>=start; i--){ + if(json[i]=='\n'){ + start=i+1; + break; + } + } + + if(start+SOURCE_MAX_LEN < end) + end=start+SOURCE_MAX_LEN; + + for(i = offset+1; i0){ + String s_source(pa_strdup(source,source_offset)); + source_offset=s_source.length(r.charsets.source()); + } + } + + throw Exception("json.parse", 0, "%s at line %d\n%s\n%*s", msg, line+1, source, source_offset+1, "^"); +} + static void _parse(Request& r, MethodParams& params) { const String& json_string=params.as_string(0, "json must be string"); Json json(r.charsets.source().isUTF8() ? NULL : &(r.charsets.source())); - JSON_config config; - init_JSON_config(&config); - - config.depth = 19; - config.callback = (JSON_parser_callback)&json_callback; - config.allow_comments = 1; - config.handle_floats_manually = 1; - config.callback_ctx = &json; + json_config config = { + 0, // buffer_initial_size + 128, // max_nesting + 0, // max_data + 1, // allow_c_comments + 1, // allow_yaml_comments + pa_malloc, + pa_realloc, + pa_free + }; if(params.count() == 2) if(HashStringValue* options=params.as_hash(1)) { int valid_options=0; if(Value* value=options->get("depth")) { - config.depth=r.process_to_value(*value).as_int(); + config.max_nesting=r.process(*value).as_int(); valid_options++; } if(Value* value=options->get("double")) { - json.handle_double=r.process_to_value(*value).as_bool(); + json.handle_double=r.process(*value).as_bool(); + valid_options++; + } + if(Value* value=options->get("int")) { + json.handle_int=r.process(*value).as_bool(); valid_options++; } if(Value* value=options->get("distinct")) { @@ -260,26 +336,34 @@ static void _parse(Request& r, MethodPar throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } - const String::Body json_body = json_string.cstr_to_string_body_untaint(String::L_JSON, 0, &(r.charsets)); - const char *json_cstr = json.charset != NULL ? Charset::transcode(json_body, *json.charset, UTF8_charset).cstr() : json_body.cstr(); + const String::Body json_body = json_string.cstr_to_string_body_untaint(String::L_JSON, r.connection(false), &r.charsets); + const char *json_cstr = json.charset != NULL ? Charset::transcode(json_body, *json.charset, pa_UTF8_charset).cstr() : json_body.cstr(); - struct JSON_parser_struct* jc = new_JSON_parser(&config); + json_parser parser; + if(int result = json_parser_init(&parser, &config, (json_parser_callback)&json_callback, &json)) + throw Exception("json.parse", 0, "%s", json_error_message(result)); - for (const char *c=json_cstr; *c; c++){ - if (!JSON_parser_char(jc, *((const unsigned char *)c))) { - throw Exception("json.parse", 0, "%s at byte %d", json_error_message(JSON_parser_get_last_error(jc)), c-json_cstr); - } - } + if(!*json_cstr) + throw Exception("json.parse", 0, "empty string is not valid json"); - if (!JSON_parser_done(jc)) { - throw Exception("json.parse", 0, "%s at the end", json_error_message(JSON_parser_get_last_error(jc))); - } - - delete_JSON_parser(jc); + const char *first_quote=strchr(json_cstr,'"'); + if(first_quote && first_quote>json_cstr && *(--first_quote) == '\\') + json_exception_with_source(r, "illegal quote escape, json may be tainted", json_cstr, first_quote-json_cstr); + + uint32_t processed; + if(int result = json_parser_string(&parser, json_cstr, strlen(json_cstr), &processed)) + json_exception_with_source(r, json_error_message(result), json_cstr, processed); + + if (!json_parser_is_done(&parser)) + json_exception_with_source(r, "unexpected end of json data", json_cstr, processed); - if (json.result) r.write_no_lang(*json.result); + json_parser_free(&parser); + + if (json.result) r.write(*json.result); } +const uint ANTI_ENDLESS_JSON_STRING_RECOURSION=128; + char *get_indent(uint level){ static char* cache[ANTI_ENDLESS_JSON_STRING_RECOURSION]={}; if (!cache[level]){ @@ -291,21 +375,34 @@ char *get_indent(uint level){ return cache[level]; } +class Json_string_recoursion { + Json_options& foptions; +public: + Json_string_recoursion(Json_options& aoptions) : foptions(aoptions) { + if(++foptions.json_string_recoursion==ANTI_ENDLESS_JSON_STRING_RECOURSION) + throw Exception(PARSER_RUNTIME, 0, "call canceled - endless json recursion detected"); + } + ~Json_string_recoursion() { + if(foptions.json_string_recoursion) + foptions.json_string_recoursion--; + } +}; + const String& value_json_string(String::Body key, Value& v, Json_options& options); -const String* Json_options::hash_json_string(HashStringValue &hash) { - if(!hash.count()) +const String* Json_options::hash_json_string(HashStringValue *hash) { + if(!hash || !hash->count()) return new String("{}", String::L_AS_IS); - uint level = r->json_string_recoursion_go_down(); + Json_string_recoursion go_down(*this); String& result = *new String("{\n", String::L_AS_IS); if (indent){ String *delim=NULL; - indent=get_indent(level); - for(HashStringValue::Iterator i(hash); i; i.next() ){ + indent=get_indent(json_string_recoursion); + for(HashStringValue::Iterator i(*hash); i; i.next() ){ if (delim){ result << *delim; } else { @@ -314,12 +411,12 @@ const String* Json_options::hash_json_st } result << String(i.key(), String::L_JSON) << "\":" << value_json_string(i.key(), *i.value(), *this); } - result << "\n" << (indent=get_indent(level-1)) << "}"; + result << "\n" << (indent=get_indent(json_string_recoursion-1)) << "}"; } else { bool need_delim=false; - for(HashStringValue::Iterator i(hash); i; i.next() ){ + for(HashStringValue::Iterator i(*hash); i; i.next() ){ result << (need_delim ? ",\n\"" : "\""); result << String(i.key(), String::L_JSON) << "\":" << value_json_string(i.key(), *i.value(), *this); need_delim=true; @@ -328,25 +425,27 @@ const String* Json_options::hash_json_st } - r->json_string_recoursion_go_up(); return &result; } static bool based_on(HashStringValue::key_type key, HashStringValue::value_type /*value*/, Value* v) { return v->is(key.cstr()); } - + const String& value_json_string(String::Body key, Value& v, Json_options& options) { if(options.methods) { Value* method=options.methods->get(v.type()); if(!method){ method=options.methods->first_that(based_on, &v); - options.methods->put(key, method ? method : VVoid::get()); + options.methods->put(v.type(), method ? method : VVoid::get()); } if(method && !method->is_void()) { Junction* junction=method->get_junction(); VMethodFrame frame(*junction->method, options.r->method_frame, junction->self); + HashStringValue* params_hash=options.params && options.indent ? options.params->get_hash() : NULL; + Temp_hash_value indent(params_hash, "indent", new VString(*new String(options.indent, String::L_AS_IS))); + Value *params[]={new VString(*new String(key, String::L_JSON)), &v, options.params ? options.params : VVoid::get()}; frame.store_params(params, 3); @@ -365,7 +464,7 @@ static void _string(Request& r, MethodPa if(params.count() == 2) if(HashStringValue* options=params.as_hash(1)) { - json.params=params.get(1); + json.params=¶ms[1]; HashStringValue* methods=new HashStringValue(); int valid_options=0; HashStringValue* vvalue; @@ -373,15 +472,18 @@ static void _string(Request& r, MethodPa String::Body key=i.key(); Value* value=i.value(); if(key == "skip-unknown"){ - json.skip_unknown=r.process_to_value(*value).as_bool(); + json.skip_unknown=r.process(*value).as_bool(); valid_options++; } else if(key == "date" && value->is_string()){ const String& svalue=value->as_string(); if(!json.set_date_format(svalue)) - throw Exception(PARSER_RUNTIME, &svalue, "must be 'sql-string', 'gmt-string' or 'unix-timestamp'"); + throw Exception(PARSER_RUNTIME, &svalue, "must be 'sql-string', 'gmt-string', 'iso-string' or 'unix-timestamp'"); valid_options++; } else if(key == "indent"){ - json.indent=r.process_to_value(*value).as_bool() ? "":NULL; + if(value->is_string()){ + json.indent=value->as_string().cstr(); + json.json_string_recoursion=strlen(json.indent); + } else json.indent=r.process(*value).as_bool() ? "" : NULL; valid_options++; } else if(key == "table" && value->is_string()){ const String& svalue=value->as_string(); @@ -393,6 +495,11 @@ static void _string(Request& r, MethodPa if(!json.set_file_format(svalue)) throw Exception(PARSER_RUNTIME, &svalue, "must be 'base64', 'text' or 'stat'"); valid_options++; + } else if(key == "void" && value->is_string()){ + const String& svalue=value->as_string(); + if(!json.set_void_format(svalue)) + throw Exception(PARSER_RUNTIME, &svalue, "must be 'string' or 'null'"); + valid_options++; #ifdef XML } else if(key == "xdoc" && (vvalue = value->get_hash())){ json.xdoc_options=new XDocOutputOptions(); @@ -411,11 +518,13 @@ static void _string(Request& r, MethodPa throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); // special handling for $._default - if(VHash* vhash=static_cast(params[1].as(VHASH_TYPE))) + if(VHashBase* vhash=static_cast(params[1].as(VHASH_TYPE))) if(Value* value=vhash->get_default()) { - Junction* junction=value->get_junction(); - if(!junction || !junction->method || !junction->method->params_names || junction->method->params_names->count() != 3) - throw Exception(PARSER_RUNTIME, 0, "$.%s must be parser method with 3 parameters", HASH_DEFAULT_ELEMENT_NAME); + if(!value->is_string()){ + Junction* junction=value->get_junction(); + if(!junction || !junction->method || !junction->method->params_names || junction->method->params_names->count() != 3) + throw Exception(PARSER_RUNTIME, 0, "$._default must be string or parser method with 3 parameters"); + } json.default_method=value; } @@ -423,9 +532,9 @@ static void _string(Request& r, MethodPa json.methods=methods; } - const String& result_string=value_json_string(String::Body(), params[0], json); - String::Body result_body=result_string.cstr_to_string_body_untaint(String::L_JSON, 0, &r.charsets); - r.write_pass_lang(*new String(result_body, String::L_AS_IS)); + const String& result_string=value_json_string(String::Body(), r.process(params[0]), json); + String::Body result_body=result_string.cstr_to_string_body_untaint(String::L_JSON, r.connection(false), &r.charsets); + r.write(*new String(result_body, String::L_AS_IS)); } // constructor