--- parser3/src/classes/file.C 2008/02/14 09:10:23 1.166 +++ parser3/src/classes/file.C 2012/03/03 00:21:47 1.217 @@ -1,16 +1,14 @@ /** @file Parser: @b file parser class. - Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2012 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_FILE_C="$Date: 2008/02/14 09:10:23 $"; +static const char * const IDENT_FILE_C="$Date: 2012/03/03 00:21:47 $"; #include "pa_config_includes.h" -#include "pcre.h" - #include "classes.h" #include "pa_vmethod_frame.h" @@ -26,11 +24,11 @@ static const char * const IDENT_FILE_C=" #include "pa_charsets.h" #include "pa_sql_connection.h" #include "pa_md5.h" +#include "pa_vregex.h" +#include "pa_version.h" // defines -#define TEXT_MODE_NAME "text" -#define BINARY_MODE_NAME "binary" #define STDIN_EXEC_PARAM_NAME "stdin" #define CHARSET_EXEC_PARAM_NAME "charset" @@ -45,15 +43,9 @@ extern String sql_offset_name; class MFile: public Methoded { public: // VStateless_class - - Value* create_new_value(Pool&, HashStringValue&) { return new VFile(); } - -public: // Methoded - bool used_directly() { return true; } - + Value* create_new_value(Pool&) { return new VFile(); } public: MFile(); - }; // global variable @@ -114,34 +106,31 @@ static const String::Body cdate_name("cd // methods -static bool is_valid_mode (const String& mode) { - return (mode==TEXT_MODE_NAME || mode==BINARY_MODE_NAME); -} - -static bool is_text_mode(const String& mode) { - if(mode==TEXT_MODE_NAME) - return true; - if(mode==BINARY_MODE_NAME) - return false; - throw Exception(PARSER_RUNTIME, - &mode, - "is invalid mode, must be either '"TEXT_MODE_NAME"' or '"BINARY_MODE_NAME"'"); -} - static void _save(Request& r, MethodParams& params) { - Value& vmode_name=params.as_no_junction(0, MODE_MUST_NOT_BE_CODE); + bool is_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); Value& vfile_name=params.as_no_junction(1, FILE_NAME_MUST_NOT_BE_CODE); + Charset* asked_charset=0; + if(params.count()>2) + if(HashStringValue* options=params.as_hash(2)){ + int valid_options=0; + if(Value* vcharset_name=options->get(PA_CHARSET_NAME)){ + asked_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + // save - GET_SELF(r, VFile).save(r.absolute(vfile_name.as_string()), - is_text_mode(vmode_name.as_string())); + GET_SELF(r, VFile).save(r.charsets, r.absolute(vfile_name.as_string()), is_text, asked_charset); } static void _delete(Request& r, MethodParams& params) { - Value& vfile_name=params.as_no_junction(0, FILE_NAME_MUST_NOT_BE_CODE); + const String& file_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); // unlink - file_delete(r.absolute(vfile_name.as_string())); + file_delete(r.absolute(file_name)); } static void _move(Request& r, MethodParams& params) { @@ -155,10 +144,10 @@ static void _move(Request& r, MethodPara } static void copy_process_source( - struct stat& , - int from_file, - const String& , const char* /*fname*/, bool, - void *context) { + struct stat& , + int from_file, + const String& , const char* /*fname*/, bool, + void *context) { int& to_file=*static_cast(context); int nCount=0; @@ -167,7 +156,7 @@ static void copy_process_source( nCount = file_block_read(from_file, buffer, sizeof(buffer)); int written=write(to_file, buffer, nCount); if( written < 0 ) - throw Exception(0, + throw Exception("file.access", 0, "write failed: %s (%d)", strerror(errno), errno); @@ -194,81 +183,151 @@ static void _copy(Request& r, MethodPara } static void _load_pass_param( - HashStringValue::key_type key, - HashStringValue::value_type value, - HashStringValue *dest) { + HashStringValue::key_type key, + HashStringValue::value_type value, + HashStringValue *dest) { dest->put(key, value); } + static void _load(Request& r, MethodParams& params) { - Value& vmode_name=params.as_no_junction(0, MODE_MUST_NOT_BE_CODE); - const String& lfile_name=r.absolute(params.as_no_junction(1, FILE_NAME_MUST_NOT_BE_CODE).as_string()); - Value* third_param=params.count()>2?¶ms.as_no_junction(2, "filename or options must not be code") - :0; - HashStringValue* third_param_hash=third_param?third_param->get_hash():0; - size_t alt_filename_param_index=2; - if(third_param_hash) - alt_filename_param_index++; + bool as_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); + const String& lfile_name=r.absolute(params.as_string(1, FILE_NAME_MUST_NOT_BE_CODE)); + + size_t param_index=params.count()-1; + Value* param_value=param_index>1?¶ms.as_no_junction(param_index, "file name or options must not be code"):0; + + HashStringValue* options=0; + const String* user_file_name=0; + + if(param_value){ + options=param_value->get_hash(); + if(options || param_index>2) + param_index--; + if(param_index>1){ + const String& luser_file_name=params.as_string(param_index, FILE_NAME_MUST_BE_STRING); + if(!luser_file_name.is_empty()) + user_file_name=&luser_file_name; + } + } + if(!user_file_name) + user_file_name=&lfile_name; - HashStringValue* options=third_param_hash; size_t offset=0; size_t limit=0; - if(options) { + + if(options){ options=new HashStringValue(*options); - if(Value *voffset=(Value *)options->get(sql_offset_name)) { + if(Value *voffset=(Value *)options->get(sql_offset_name)){ offset=r.process_to_value(*voffset).as_int(); } - if(Value *vlimit=(Value *)options->get(sql_limit_name)) { + if(Value *vlimit=(Value *)options->get(sql_limit_name)){ limit=r.process_to_value(*vlimit).as_int(); } // no check on options count here, see file_read } - File_read_result file=file_read(r.charsets, lfile_name, - is_text_mode(vmode_name.as_string()), - options, true, 0, offset, limit + File_read_result file=file_load(r, lfile_name, + as_text, options, true, 0, offset, limit ); - const char *user_file_name=params.count()>alt_filename_param_index? - params.as_string(alt_filename_param_index, FILE_NAME_MUST_BE_STRING).cstr() - :lfile_name.cstr(String::L_FILE_SPEC); - Value* vcontent_type=0; - if(file.headers) - { - if(Value* remote_content_type=file.headers->get("CONTENT-TYPE")) + if(file.headers){ + if(Value* remote_content_type=file.headers->get(HTTP_CONTENT_TYPE_UPPER)) vcontent_type=new VString(*new String(remote_content_type->as_string().cstr())); } - if(!vcontent_type) - vcontent_type=new VString(r.mime_type_of(user_file_name)); VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, file.str, file.length, user_file_name, vcontent_type); - if(file.headers) + self.set(true/*tainted*/, file.str, file.length, user_file_name, vcontent_type, &r); + + self.set_mode(as_text); + + if(file.headers){ file.headers->for_each(_load_pass_param, &self.fields()); + } else { + size_t size; + time_t atime, mtime, ctime; + + file_stat(lfile_name, size, atime, mtime, ctime); + + HashStringValue& ff=self.fields(); + ff.put(adate_name, new VDate(atime)); + ff.put(mdate_name, new VDate(mtime)); + ff.put(cdate_name, new VDate(ctime)); + } } static void _create(Request& r, MethodParams& params) { - Value& vmode_name=params.as_no_junction(0, MODE_MUST_NOT_BE_CODE); - if(!is_text_mode(vmode_name.as_string())) - throw Exception(PARSER_RUNTIME, - 0, - "only text mode is currently supported"); - - const char* user_file_name_cstr=r.absolute( - params.as_no_junction(1, FILE_NAME_MUST_NOT_BE_CODE).as_string()).cstr(String::L_FILE_SPEC); + const String* mode=0; + const String* file_name=0; + bool is_text=true; + + // new format: ^file::create[string-or-file-content[;$.mode[text|binary] $.name[...] $.content-type[...] $.charset[...] ]] + size_t content_index=0; + size_t options_index=1; + bool extended_options=true; + + if(params.count()>=3){ + // old format: ^file::create[text|binary;file-name;string-or-file-content[;options]] + mode=¶ms.as_string(0, MODE_MUST_NOT_BE_CODE); + is_text=VFile::is_text_mode(*mode); + file_name=¶ms.as_string(1, FILE_NAME_MUST_NOT_BE_CODE); + content_index=2; + options_index=3; + extended_options=false; + } + + VString* vcontent_type=0; + Charset* asked_charset=0; + if(params.count()>options_index) + if(HashStringValue* options=params.as_hash(options_index)) { + int valid_options=0; + if(extended_options) { + if(Value* vmode=options->get(MODE_NAME)) { + mode=&vmode->as_string(); + is_text=VFile::is_text_mode(*mode); + valid_options++; + } + if(Value* vfile_name=options->get(NAME_NAME)) { + file_name=&vfile_name->as_string(); + valid_options++; + } + } + if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) { + asked_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + valid_options++; + } + if(Value* value=options->get(CONTENT_TYPE_NAME)) { + vcontent_type=new VString(value->as_string()); + valid_options++; + } + if(valid_options != options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } - const String& content=params.as_string(2, "content must be string"); - const char* content_cstr=content.cstr(String::L_UNSPECIFIED); // explode content, honor tainting changes + Value& vcontent=params.as_no_junction(content_index, "content must be string or file"); - VString* vcontent_type=new VString(r.mime_type_of(user_file_name_cstr)); - VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, content_cstr, strlen(content_cstr), user_file_name_cstr, vcontent_type); + + if(const String* content_str=vcontent.get_string()){ + String::Body body=content_str->cstr_to_string_body_untaint(String::L_AS_IS); // explode content, honor tainting changes + if(asked_charset && is_text) + body=Charset::transcode(body, r.charsets.source(), *asked_charset); + self.set(true/*tainted*/, body.cstr(), body.length()); + self.set_mode(is_text); + } else { + if(asked_charset) + throw Exception(PARSER_RUNTIME, 0, "charset option can not be used with file-content"); + self.set(*vcontent.as_vfile(String::L_AS_IS)); + if(mode) + self.set_mode(is_text); + } + + self.set_name(file_name); + + self.set_content_type(vcontent_type, file_name, &r); } static void _stat(Request& r, MethodParams& params) { - Value& vfile_name=params.as_no_junction(0, FILE_NAME_MUST_NOT_BE_CODE); - - const String& lfile_name=vfile_name.as_string(); + const String& lfile_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); size_t size; time_t atime, mtime, ctime; @@ -277,12 +336,12 @@ static void _stat(Request& r, MethodPara atime, mtime, ctime); VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, 0/*no bytes*/, size); + + self.set(true/*tainted*/, 0/*no bytes*/, size, &lfile_name, 0, &r); HashStringValue& ff=self.fields(); ff.put(adate_name, new VDate(atime)); ff.put(mdate_name, new VDate(mtime)); ff.put(cdate_name, new VDate(ctime)); - ff.put(content_type_name, new VString(r.mime_type_of(lfile_name.cstr(String::L_FILE_SPEC)))); } static bool is_safe_env_key(const char* key) { @@ -291,6 +350,7 @@ static bool is_safe_env_key(const char* if(!(c>='A' && c<='Z' || c>='0' && c<='9' || c=='_' || c=='-')) return false; } +#ifdef PA_SAFE_MODE if(strncasecmp(key, "HTTP_", 5)==0) return true; if(strncasecmp(key, "CGI_", 4)==0) @@ -300,6 +360,9 @@ static bool is_safe_env_key(const char* return true; } return false; +#else + return true; +#endif } #ifndef DOXYGEN struct Append_env_pair_info { @@ -309,9 +372,9 @@ struct Append_env_pair_info { }; #endif static void append_env_pair( - HashStringValue::key_type akey, - HashStringValue::value_type avalue, - Append_env_pair_info *info) { + HashStringValue::key_type akey, + HashStringValue::value_type avalue, + Append_env_pair_info *info) { if(akey==STDIN_EXEC_PARAM_NAME) { info->vstdin=avalue; } else if(akey==CHARSET_EXEC_PARAM_NAME) { @@ -321,7 +384,7 @@ static void append_env_pair( throw Exception(PARSER_RUNTIME, new String(akey, String::L_TAINTED), "not safe environment variable"); - info->env->put(akey, avalue->as_string().cstr_to_string_body(String::L_UNSPECIFIED, 0, info->charsets)); + info->env->put(akey, avalue->as_string().cstr_to_string_body_untaint(String::L_AS_IS, 0, info->charsets)); } } #ifndef DOXYGEN @@ -332,60 +395,45 @@ struct Pass_cgi_header_attribute_info { }; #endif static void pass_cgi_header_attribute( - ArrayString::element_type astring, - Pass_cgi_header_attribute_info* info) { + ArrayString::element_type astring, + Pass_cgi_header_attribute_info* info) { size_t colon_pos=astring->pos(':'); if(colon_pos!=STRING_NOT_FOUND) { const String& key=astring->mid(0, colon_pos).change_case( *info->charset, String::CC_UPPER); Value* value=new VString(astring->mid(colon_pos+1, astring->length()).trim()); info->fields->put(key, value); - if(key=="CONTENT-TYPE") + if(key==HTTP_CONTENT_TYPE_UPPER) info->content_type=value; } } static void append_to_argv(Request& r, ArrayString& argv, const String* str){ - if( str->length() ){ - argv+=new String(str->cstr_to_string_body(String::L_UNSPECIFIED, 0, &r.charsets), String::L_AS_IS); - } -} - -inline size_t strpos(const char *s1, const char *s2) { - const char *p = strstr(s1, s2); - return (p==0)?STRING_NOT_FOUND:p-s1; + if(!str->is_empty()) + argv+=new String(str->cstr_to_string_body_untaint(String::L_AS_IS, 0, &r.charsets), String::L_AS_IS); } /// @todo fix `` in perl - they produced flipping consoles and no output to perl -static void _exec_cgi(Request& r, MethodParams& params, - bool cgi) { - - Value& first_param=params.as_no_junction(0, FIRST_ARG_MUST_NOT_BE_CODE); - - bool is_mode_specified=is_valid_mode(first_param.as_string()); - const String& mode_name=(is_mode_specified) ? first_param.as_string() : *new String(TEXT_MODE_NAME); - - size_t param_index=1; - if(!is_mode_specified){ - --param_index; +static void _exec_cgi(Request& r, MethodParams& params, bool cgi) { + bool is_text=true; + size_t param_index=0; + const String& mode=params.as_string(0, FIRST_ARG_MUST_NOT_BE_CODE); + if(VFile::is_valid_mode(mode)) { + is_text=VFile::is_text_mode(mode); + param_index++; } if(param_index>=params.count()) - throw Exception(PARSER_RUNTIME, - 0, - "file name must be specified"); + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); - - Value& vfile_name=params.as_no_junction(param_index++, FILE_NAME_MUST_NOT_BE_CODE); - - const String& script_name=r.absolute(vfile_name.as_string()); + const String& script_name=r.absolute(params.as_string(param_index++, FILE_NAME_MUST_NOT_BE_CODE)); HashStringString env; #define ECSTR(name, value_cstr) \ if(value_cstr) \ env.put( \ String::Body(#name), \ - String::Body(value_cstr, 0)); \ + String::Body(*value_cstr?value_cstr:0)); \ // passing SAPI::environment if(const char *const *pairs=SAPI::environment(r.sapi_info)) { while(const char* pair=*pairs++) @@ -398,6 +446,7 @@ static void _exec_cgi(Request& r, Method // const ECSTR(GATEWAY_INTERFACE, "CGI/1.1"); + ECSTR("PARSER_VERSION", PARSER_VERSION); // from Request.info ECSTR(DOCUMENT_ROOT, r.request_info.document_root); ECSTR(PATH_TRANSLATED, r.request_info.path_translated); @@ -405,15 +454,11 @@ static void _exec_cgi(Request& r, Method ECSTR(QUERY_STRING, r.request_info.query_string); ECSTR(REQUEST_URI, r.request_info.uri); ECSTR(CONTENT_TYPE, r.request_info.content_type); - char content_length_cstr[MAX_NUMBER]; - snprintf(content_length_cstr, MAX_NUMBER, "%u", r.request_info.content_length); - //String content_length(content_length_cstr); - ECSTR(CONTENT_LENGTH, content_length_cstr); + ECSTR(CONTENT_LENGTH, format(r.request_info.content_length, "%u")); // SCRIPT_* env.put(String::Body("SCRIPT_NAME"), script_name); //env.put(String::Body("SCRIPT_FILENAME"), ??&script_name); - bool stdin_specified=false; // environment & stdin from param String *in=new String(); Charset *charset=0; // default script works raw_in 'source' charset = no transcoding needed @@ -435,11 +480,11 @@ static void _exec_cgi(Request& r, Method } // $.stdin if(info.vstdin) { - stdin_specified=true; if(const String* sstdin=info.vstdin->get_string()) { - in->append(*sstdin, String::L_CLEAN, true); + // untaint stdin + in = new String(sstdin->cstr_to_string_body_untaint(String::L_AS_IS), String::L_AS_IS); } else - if(VFile* vfile=static_cast(info.vstdin->as("file", false))) + if(VFile* vfile=static_cast(info.vstdin->as("file"))) in->append_know_length((const char* )vfile->value_ptr(), vfile->value_size(), String::L_TAINTED); else throw Exception(PARSER_RUNTIME, @@ -452,9 +497,8 @@ static void _exec_cgi(Request& r, Method // argv from params ArrayString argv; if(param_index < params.count()) { - // influence tainting - // main target -- URLencoding of tainted pieces to String::L_URI lang - Temp_client_charset temp(r.charsets, charset? *charset: r.charsets.source()); + // influence tainting + Temp_client_charset temp(r.charsets, charset? *charset: r.charsets.source()); for(size_t i=param_index; ilength && is_text_mode(mode_name)){ + if(file_out->length && is_text){ fix_line_breaks(file_out->str, file_out->length); // treat output as string - String *real_out = new String(file_out->str, file_out->length); + String *real_out = new String(file_out->str); // transcode out if necessary if(charset) @@ -534,13 +578,13 @@ static void _exec_cgi(Request& r, Method break; default: // 00 unix_header_break=false; // calm down, compiler - throw Exception(0, + throw Exception("file.execute", 0, "output does not contain CGI header; " "exit status=%d; stdoutsize=%u; stdout: \"%s\"; stderrsize=%u; stderr: \"%s\"", execution.status, - (size_t)file_out->length, (file_out->length) ? (file_out->str) : "", - (size_t)real_err->length(), real_err->cstr()); + file_out->length, (file_out->length) ? (file_out->str) : "", + real_err->length(), real_err->cstr()); break; //never reached } @@ -556,7 +600,7 @@ static void _exec_cgi(Request& r, Method } file_out->str[header_break_pos] = 0; - String *header=new String(file_out->str, header_break_pos); + String *header=new String(file_out->str); unsigned long headersize = header_break_pos+eol_marker_size*2; file_out->str += headersize; file_out->length -= headersize; @@ -565,7 +609,7 @@ static void _exec_cgi(Request& r, Method self.set(false/*not tainted*/, file_out->str, file_out->length); // $fields << header - if(header && eol_marker) { + if(header) { ArrayString rows; size_t pos_after=0; header->split(rows, pos_after, eol_marker); @@ -581,11 +625,13 @@ static void _exec_cgi(Request& r, Method self.set(false/*not tainted*/, file_out->str, file_out->length); } + self.set_mode(is_text); + // $status self.fields().put(file_status_name, new VInt(execution.status)); // $stderr - if(real_err->length()) + if(!real_err->is_empty()) self.fields().put( String::Body("stderr"), new VString(*real_err)); @@ -600,66 +646,41 @@ static void _cgi(Request& r, MethodParam static void _list(Request& r, MethodParams& params) { Value& relative_path=params.as_no_junction(0, "path must not be code"); - const String* regexp; - pcre *regexp_code; - const int ovecsize=(1/*match*/)*3; - int ovector[ovecsize]; - if(params.count()>1) { - regexp=¶ms.as_no_junction(1, "regexp must not be code").as_string(); - - const char* pattern=regexp->cstr(); - const char* errptr; - int erroffset; - regexp_code=pcre_compile(pattern, PCRE_EXTRA | PCRE_DOTALL, - &errptr, &erroffset, - r.charsets.source().pcre_tables); - - if(!regexp_code) - throw Exception(0, - ®exp->mid(erroffset, regexp->length()), - "regular expression syntax error - %s", errptr); - } else { - regexp=0; // not used, just to calm down compiler - regexp_code=0; + VRegex* vregex=0; + VRegexCleaner vrcleaner; + if(params.count()>1){ + Value& regexp=params.as_no_junction(1, "regexp must not be code"); + if(regexp.is_defined()){ + if(Value* value=regexp.as(VREGEX_TYPE)){ + vregex=static_cast(value); + } else { + vregex=new VRegex(r.charsets.source(), ®exp.as_string(), 0/*options*/); + vregex->study(); + vrcleaner.vregex=vregex; + } + } } - - const char* absolute_path_cstr=r.absolute(relative_path.as_string()).cstr(String::L_FILE_SPEC); + const char* absolute_path_cstr=r.absolute(relative_path.as_string()).taint_cstr(String::L_FILE_SPEC); Table::columns_type columns(new ArrayString); *columns+=new String("name"); Table& table=*new Table(columns); + const int ovector_size=(1/*match*/)*3; + int ovector[ovector_size]; + LOAD_DIR(absolute_path_cstr, const char* file_name_cstr=ffblk.ff_name; size_t file_name_size=strlen(file_name_cstr); - bool suits=true; - if(regexp_code) { - int exec_result=pcre_exec(regexp_code, 0, - ffblk.ff_name, file_name_size, 0, - 0, ovector, ovecsize); - - if(exec_result==PCRE_ERROR_NOMATCH) - suits=false; - else if(exec_result<0) { - (*pcre_free)(regexp_code); - throw Exception(0, - regexp, - "regular expression execute (%d)", - exec_result); - } - } - if(suits) { + if(!vregex || vregex->exec(ffblk.ff_name, file_name_size, ovector, ovector_size)>=0) { Table::element_type row(new ArrayString); - *row+=new String(pa_strdup(file_name_cstr, file_name_size), file_name_size, true); + *row+=new String(pa_strdup(file_name_cstr, file_name_size), String::L_TAINTED); table+=row; } ); - if(regexp_code) - pcre_free(regexp_code); - // write out result r.write_no_lang(*new VTable(&table)); } @@ -689,23 +710,10 @@ static void _lock(Request& r, MethodPara &info); } -static int lastposafter(const String& s, size_t after, const char* substr, size_t substr_size, bool beforelast=false) { - size_t size=0; // just to calm down compiler - if(beforelast) - size=s.length(); - size_t at; - while((at=s.pos(String::Body(substr, substr_size), after))!=STRING_NOT_FOUND) { - size_t newafter=at+substr_size/*skip substr*/; - if(beforelast && newafter==size) - break; - after=newafter; - } - - return after; -} - static void _find(Request& r, MethodParams& params) { - const String& file_name=params.as_no_junction(0, FILE_NAME_MUST_NOT_BE_CODE).as_string(); + const String& file_name=params.as_string(0, FILE_NAME_MUST_NOT_BE_CODE); + Value* not_found_code=(params.count()==2)?¶ms.as_junction(1, "not-found param must be code"):0; + const String* file_spec; if(file_name.first_char()=='/') file_spec=&file_name; @@ -735,40 +743,38 @@ static void _find(Request& r, MethodPara } // no way, not found - if(params.count()==2) { - Value& not_found_code=params.as_junction(1, "not-found param must be code"); - r.write_pass_lang(r.process(not_found_code)); - } + if(not_found_code) + r.write_pass_lang(r.process(*not_found_code)); } static void _dirname(Request& r, MethodParams& params) { const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); - // /a/some.tar.gz > /a + // /a/some.tar.gz > /a // /a/b/ > /a int afterslash=lastposafter(file_spec, 0, "/", 1, true); if(afterslash>0) r.write_assign_lang(file_spec.mid(0, afterslash==1?1:afterslash-1)); else - r.write_assign_lang(String(".", 1)); + r.write_assign_lang(String(".")); } static void _basename(Request& r, MethodParams& params) { const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); - // /a/some.tar.gz > some.tar.gz + // /a/some.tar.gz > some.tar.gz int afterslash=lastposafter(file_spec, 0, "/", 1); r.write_assign_lang(file_spec.mid(afterslash, file_spec.length())); } static void _justname(Request& r, MethodParams& params) { const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); - // /a/some.tar.gz > some.tar + // /a/some.tar.gz > some.tar int afterslash=lastposafter(file_spec, 0, "/", 1); int afterdot=lastposafter(file_spec, afterslash, ".", 1); r.write_assign_lang(file_spec.mid(afterslash, afterdot!=afterslash?afterdot-1:file_spec.length())); } static void _justext(Request& r, MethodParams& params) { const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); - // /a/some.tar.gz > gz + // /a/some.tar.gz > gz int afterdot=lastposafter(file_spec, 0, ".", 1); if(afterdot>0) r.write_assign_lang(file_spec.mid(afterdot, file_spec.length())); @@ -836,14 +842,14 @@ public: break; case 1: if(!user_file_name) // user not specified? - user_file_name=new String(str, length, true); + user_file_name=new String(str, String::L_TAINTED); break; case 2: if(!user_content_type) // user not specified? - user_content_type=new String(str, length, true); + user_content_type=new String(str, String::L_TAINTED); break; default: - error=SQL_Error(PARSER_RUNTIME, "result must not contain more then one row, three rows"); + error=SQL_Error(PARSER_RUNTIME, "result must not contain more then one row, three columns"); return true; } return false; @@ -859,13 +865,15 @@ static void _sql(Request& r, MethodParam Temp_lang temp_lang(r, String::L_SQL); const String& statement_string=r.process_to_string(statement); - const char* statement_cstr= - statement_string.cstr(String::L_UNSPECIFIED, r.connection()); + const char* statement_cstr=statement_string.untaint_cstr(r.flang, r.connection()); + File_sql_event_handlers handlers(statement_string, statement_cstr); + ulong limit=SQL_NO_LIMIT; + ulong offset=0; + if(params.count()>1) - if(HashStringValue* options= - params.as_no_junction(1, PARAM_MUST_NOT_BE_CODE).get_hash()) { + if(HashStringValue* options=params.as_hash(1)){ int valid_options=0; if(Value* vfilename=options->get(NAME_NAME)) { valid_options++; @@ -875,17 +883,23 @@ static void _sql(Request& r, MethodParam valid_options++; handlers.user_content_type=&vcontent_type->as_string(); } + if(Value* vlimit=options->get(sql_limit_name)) { + valid_options++; + limit=(ulong)r.process_to_value(*vlimit).as_double(); + } + if(Value* voffset=options->get(sql_offset_name)) { + valid_options++; + offset=(ulong)r.process_to_value(*voffset).as_double(); + } if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } r.connection()->query( statement_cstr, 0, 0, - 0, 0, + offset, limit, handlers, statement_string); @@ -894,39 +908,79 @@ static void _sql(Request& r, MethodParam 0, "produced no result"); - const char* user_file_name_cstr=handlers.user_file_name? handlers.user_file_name->cstr(): 0; - - VString* vcontent_type=handlers.user_content_type? - new VString(*handlers.user_content_type) - : user_file_name_cstr? - new VString(r.mime_type_of(user_file_name_cstr)) - : 0; VFile& self=GET_SELF(r, VFile); - self.set(true/*tainted*/, handlers.value.str, handlers.value.length, user_file_name_cstr, vcontent_type); + + self.set(true/*tainted*/, handlers.value.str, handlers.value.length, handlers.user_file_name + , handlers.user_content_type ? new VString(*handlers.user_content_type) : 0 + , &r); + self.set_mode(false/*binary*/); } static void _base64(Request& r, MethodParams& params) { - bool dynamic = !(&r.get_self() == file_class); - if ( dynamic ){ - VFile& self=GET_SELF(r, VFile); - if(params.count()) { - // decode - const char* cstr=params.as_string(0, PARAMETER_MUST_BE_STRING).cstr(); - char* decoded_cstr=0; - size_t decoded_size=0; - pa_base64_decode(cstr, strlen(cstr), decoded_cstr, decoded_size); - if(decoded_cstr && decoded_size) - self.set(true/*tainted*/, decoded_cstr, decoded_size); - } else { - // encode - const char* encoded=pa_base64_encode(self.value_ptr(), self.value_size()); - r.write_assign_lang(*new String(encoded, 0, true/*once ?param=base64(something) was needed*/)); - } + bool dynamic=!(&r.get_self() == file_class); + if(dynamic) { + VFile& self=GET_SELF(r, VFile); + if(params.count()) { + // decode: + // ^file::base64[encoded] // backward + // ^file::base64[mode;user-file-name;encoded[;$.content-type[...] $.strict(true|false)]] + bool is_text=false; + bool strict=false; + VString* vcontent_type=0; + const String* user_file_name=0; + size_t param_index=0; + + if(params.count() > 1) { + if(params.count() < 3) + throw Exception(PARSER_RUNTIME, + 0, + "constructor can not have less then 3 parameters (has %d parameters)", + params.count()); // actually it accepts 1 parameter (backward) + + is_text=VFile::is_text_mode(params.as_string(0, MODE_MUST_NOT_BE_CODE)); + user_file_name=¶ms.as_string(1, FILE_NAME_MUST_BE_STRING); + + if(params.count() == 4) + if(HashStringValue* options=params.as_hash(3)) { + int valid_options=0; + if(Value* value=options->get(CONTENT_TYPE_NAME)) { + vcontent_type=new VString(value->as_string()); + valid_options++; + } + if(Value* vstrict=options->get(BASE64_STRICT_OPTION_NAME)) { + strict=r.process_to_value(*vstrict).as_bool(); + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + + param_index=2; + } + + const char* encoded=params.as_string(param_index, PARAMETER_MUST_BE_STRING).cstr(); + + char* decoded=0; + size_t length=0; + pa_base64_decode(encoded, strlen(encoded), decoded, length, strict); + + if(length && is_text) + fix_line_breaks(decoded, length); + + self.set(true/*tainted*/, decoded, length, user_file_name, vcontent_type, &r); + + if(params.count() > 1) + self.set_mode(is_text); + } else { + // encode: ^f.base64[] + const char* encoded=pa_base64_encode(self.value_ptr(), self.value_size()); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed**/)); + } } else { - // encode + // encode: ^file:base64[filespec] const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); const char* encoded=pa_base64_encode(r.absolute(file_spec)); - r.write_assign_lang(*new String(encoded, 0, true/*once ?param=base64(something) was needed*/)); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed*/)); } } @@ -938,9 +992,7 @@ static void _crc32(Request& r, MethodPar const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); crc32=pa_crc32(r.absolute(file_spec)); } else { - throw Exception(PARSER_RUNTIME, - 0, - "file name must be defined"); + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); } } else { // ^file.crc32[] @@ -952,10 +1004,10 @@ static void _crc32(Request& r, MethodPar static void file_md5_file_action( - struct stat& finfo, - int f, - const String& , const char* /*fname*/, bool, - void *context) + struct stat& finfo, + int f, + const String& , const char* /*fname*/, bool, + void *context) { PA_MD5_CTX& md5context=*static_cast(context); if(finfo.st_size) { @@ -1000,9 +1052,7 @@ static void _md5(Request& r, MethodParam const String& file_spec=params.as_string(0, FILE_NAME_MUST_BE_STRING); md5=pa_md5(r.absolute(file_spec)); } else { - throw Exception(PARSER_RUNTIME, - 0, - "file name must be defined"); + throw Exception(PARSER_RUNTIME, 0, FILE_NAME_MUST_BE_SPECIFIED); } } else { // ^file.md5[] @@ -1016,12 +1066,13 @@ static void _md5(Request& r, MethodParam // constructor MFile::MFile(): Methoded("file") { - // ^file::create[text;user-name;string] - // ^file::create[binary;user-name;SOMEDAY SOMETHING] - add_native_method("create", Method::CT_DYNAMIC, _create, 3, 3); + // ^file::create[text|binary;file-name;string-or-file[;options hash]] + // ^file::create[string-or-file[;options hash]] + add_native_method("create", Method::CT_DYNAMIC, _create, 1, 4); // ^file.save[mode;file-name] - add_native_method("save", Method::CT_DYNAMIC, _save, 2, 2); + // ^file.save[mode;file-name;$.charset[...]] + add_native_method("save", Method::CT_DYNAMIC, _save, 2, 3); // ^file:delete[file-name] add_native_method("delete", Method::CT_STATIC, _delete, 1, 1); @@ -1031,7 +1082,9 @@ MFile::MFile(): Methoded("file") { // ^file::load[mode;disk-name] // ^file::load[mode;disk-name;user-name] - add_native_method("load", Method::CT_DYNAMIC, _load, 2, 3); + // ^file::load[mode;disk-name;user-name;options hash] + // ^file::load[mode;disk-name;options hash] + add_native_method("load", Method::CT_DYNAMIC, _load, 2, 4); // ^file::stat[disk-name] add_native_method("stat", Method::CT_DYNAMIC, _stat, 1, 1); @@ -1057,28 +1110,33 @@ MFile::MFile(): Methoded("file") { // ^file:find[file-name]{when-not-found} add_native_method("find", Method::CT_STATIC, _find, 1, 2); - // ^file:dirname[/a/some.tar.gz]=/a + // ^file:dirname[/a/some.tar.gz]=/a // ^file:dirname[/a/b/]=/a add_native_method("dirname", Method::CT_STATIC, _dirname, 1, 1); - // ^file:basename[/a/some.tar.gz]=some.tar.gz - add_native_method("basename", Method::CT_STATIC, _basename, 1, 1); - // ^file:justname[/a/some.tar.gz]=some.tar + // ^file:basename[/a/some.tar.gz]=some.tar.gz + add_native_method("basename", Method::CT_STATIC, _basename, 1, 1); + // ^file:justname[/a/some.tar.gz]=some.tar add_native_method("justname", Method::CT_STATIC, _justname, 1, 1); - // ^file:justext[/a/some.tar.gz]=gz + // ^file:justext[/a/some.tar.gz]=gz add_native_method("justext", Method::CT_STATIC, _justext, 1, 1); - // /some/page.html: ^file:fullpath[a.gif] => /some/a.gif + // /some/page.html: ^file:fullpath[a.gif] => /some/a.gif add_native_method("fullpath", Method::CT_STATIC, _fullpath, 1, 1); - // ^file.sql-string[] + // ^file.sql-string[] add_native_method("sql-string", Method::CT_DYNAMIC, _sql_string, 0, 0); - // ^file::sql[[alt_name]]{} + // ^file::sql{} + // ^file::sql{}[options hash] add_native_method("sql", Method::CT_DYNAMIC, _sql, 1, 2); - // ^file::base64[string] << decode - // ^file.base64[] << encode - // ^file:base64[file-name] << encode - add_native_method("base64", Method::CT_ANY, _base64, 0, 1); + // encode: + // ^file.base64[] + // ^file:base64[file-name] + // decode: + // ^file::base64[encoded] // backward + // ^file::base64[mode;user-file-name;encoded] + // ^file::base64[mode;user-file-name;encoded;$.content-type[...]] + add_native_method("base64", Method::CT_ANY, _base64, 0, 4); // ^file.crc32[] // ^file:crc32[file-name]