--- parser3/src/classes/file.C 2009/01/12 07:46:13 1.181 +++ parser3/src/classes/file.C 2009/05/25 08:11:40 1.192 @@ -1,11 +1,11 @@ /** @file Parser: @b file parser class. - Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_FILE_C="$Date: 2009/01/12 07:46:13 $"; +static const char * const IDENT_FILE_C="$Date: 2009/05/25 08:11:40 $"; #include "pa_config_includes.h" @@ -24,6 +24,7 @@ static const char * const IDENT_FILE_C=" #include "pa_charsets.h" #include "pa_sql_connection.h" #include "pa_md5.h" +#include "pa_vregex.h" // defines @@ -203,14 +204,27 @@ static void _load(Request& r, MethodPara const String& lfile_name=r.absolute(params.as_no_junction(1, FILE_NAME_MUST_NOT_BE_CODE).as_string()); size_t param_index=params.count()-1; - Value* param_value=param_index>=2?¶ms.as_no_junction(param_index, "filename or options must not be code"):0; - HashStringValue* param_hash=param_value?param_value->get_hash():0; - HashStringValue* options=param_hash; + Value* param_value=param_index>1?¶ms.as_no_junction(param_index, "filename or options must not be code"):0; - param_index--; + HashStringValue* options=0; + const char *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.cstr(String::L_FILE_SPEC); + } + } + if(!user_file_name) + user_file_name=lfile_name.cstr(String::L_FILE_SPEC); size_t offset=0; size_t limit=0; + if(options){ options=new HashStringValue(*options); if(Value *voffset=(Value *)options->get(sql_offset_name)){ @@ -221,14 +235,10 @@ static void _load(Request& r, MethodPara } // no check on options count here, see file_read } - File_read_result file=file_read(r.charsets, lfile_name, + File_read_result file=file_load(r, lfile_name, as_text, options, true, 0, offset, limit ); - const char *user_file_name=(param_index>=2)? - params.as_string(param_index, FILE_NAME_MUST_BE_STRING).cstr(String::L_FILE_SPEC) - :lfile_name.cstr(String::L_FILE_SPEC); - Value* vcontent_type=0; if(file.headers){ if(Value* remote_content_type=file.headers->get(HTTP_CONTENT_TYPE_UPPER)) @@ -358,9 +368,8 @@ static void pass_cgi_header_attribute( } static void append_to_argv(Request& r, ArrayString& argv, const String* str){ - if( str->length() ){ + if(!str->is_empty()) argv+=new String(str->cstr_to_string_body(String::L_UNSPECIFIED, 0, &r.charsets), String::L_AS_IS); - } } /// @todo fix `` in perl - they produced flipping consoles and no output to perl @@ -392,7 +401,7 @@ static void _exec_cgi(Request& r, Method 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++) @@ -507,7 +516,7 @@ static void _exec_cgi(Request& r, Method if(file_out->length && is_text_mode(mode_name)){ 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) @@ -563,7 +572,7 @@ 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; @@ -592,7 +601,7 @@ static void _exec_cgi(Request& r, Method 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)); @@ -607,70 +616,41 @@ 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(String::L_UNSPECIFIED); - const char* errptr; - int erroffset; - int options=PCRE_EXTRA | PCRE_DOTALL; - if(r.charsets.source().isUTF8()) - options=options|PCRE_UTF8; - - regexp_code=pcre_compile(pattern, options, - &errptr, &erroffset, - r.charsets.source().pcre_tables); - - if(!regexp_code) - throw Exception(PCRE_EXCEPTION_TYPE, - ®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, false)){ + 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); 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(PCRE_EXCEPTION_TYPE, - regexp, - print_pcre_exec_error_text(exec_result), - 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)); } @@ -705,7 +685,7 @@ static int lastposafter(const String& s, if(beforelast) size=s.length(); size_t at; - while((at=s.pos(String::Body(substr, substr_size), after))!=STRING_NOT_FOUND) { + while((at=s.pos(String::Body(substr), after))!=STRING_NOT_FOUND) { size_t newafter=at+substr_size/*skip substr*/; if(beforelast && newafter==size) break; @@ -760,7 +740,7 @@ static void _dirname(Request& r, MethodP 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) { @@ -847,11 +827,11 @@ 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"); @@ -941,13 +921,13 @@ static void _base64(Request& r, MethodPa } else { // encode: ^f.base64[] const char* encoded=pa_base64_encode(self.value_ptr(), self.value_size()); - r.write_assign_lang(*new String(encoded, 0, true/*tainted. once ?param=base64(something) was needed**/)); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed**/)); } } else { // 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/*tainted. once ?param=base64(something) was needed*/)); + r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed*/)); } }