--- parser3/src/main/pa_common.C 2004/03/05 13:17:59 1.179 +++ parser3/src/main/pa_common.C 2009/09/28 11:39:30 1.252 @@ -1,49 +1,42 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2004 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_COMMON_C="$Date: 2004/03/05 13:17:59 $"; + * BASE64 part + * Authors: Michael Zucchi + * Jeffrey Stedfast + * + * Copyright 2000-2004 Ximian, Inc. (www.ximian.com) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA. + * + */ + +static const char * const IDENT_COMMON_C="$Date: 2009/09/28 11:39:30 $"; #include "pa_common.h" #include "pa_exception.h" #include "pa_hash.h" #include "pa_globals.h" -#include "pa_request_charsets.h" #include "pa_charsets.h" - -#define PA_HTTP - -#ifdef PA_HTTP -#include "pa_vstring.h" -#include "pa_vint.h" -#include "pa_vhash.h" -#include "pa_vtable.h" - -#ifdef CYGWIN -#define _GNU_H_WINDOWS32_SOCKETS -// for PASCAL -#include -// SOCKET -typedef u_int SOCKET; -int PASCAL closesocket(SOCKET); -#else -# if defined(WIN32) -# include -# else -# define closesocket close -# endif -#endif - -#else - -# if defined(WIN32) -# include -# endif - -#endif +#include "pa_http.h" +#include "pa_request_charsets.h" +#include "pcre.h" +#include "pa_request.h" // some maybe-undefined constants @@ -64,10 +57,6 @@ int PASCAL closesocket(SOCKET); # endif #endif -# ifndef INADDR_NONE -# define INADDR_NONE ((ulong) -1) -# endif - // defines for globals #define FILE_STATUS_NAME "status" @@ -76,23 +65,20 @@ int PASCAL closesocket(SOCKET); const String file_status_name(FILE_STATUS_NAME); -// defines - -#define HTTP_METHOD_NAME "method" -#define HTTP_TIMEOUT_NAME "timeout" -#define HTTP_HEADERS_NAME "headers" -#define HTTP_ANY_STATUS_NAME "any-status" -#define HTTP_CHARSET_NAME "charset" -#define HTTP_TABLES_NAME "tables" -#define HTTP_USER "user" -#define HTTP_PASSWORD "password" - -// defines - -#define DEFAULT_USER_AGENT "parser3" - // functions +const char* capitalize(const char* s){ + char* result=pa_strdup(s); + if(result){ + bool upper=true; + for(char* c=result; *c; c++){ + *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c); + upper=strchr("-_ ", *c) != 0; + } + } + return (const char*)result; +} + void fix_line_breaks(char *str, size_t& length) { //_asm int 3; const char* const eob=str+length; @@ -120,518 +106,156 @@ void fix_line_breaks(char *str, size_t& } char* file_read_text(Request_charsets& charsets, - const String& file_spec, - bool fail_on_read_problem, - HashStringValue* params/*, HashStringValue* * out_fields*/) { + const String& file_spec, + bool fail_on_read_problem, + HashStringValue* params, + bool transcode_result) { File_read_result file= - file_read(charsets, file_spec, true, params, fail_on_read_problem); + file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result); return file.success?file.str:0; } -//http request stuff -#ifdef PA_HTTP - -#undef CRLF -#define CRLF "\r\n" - -static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){ - memset(addr, 0, sizeof(*addr)); - addr->sin_family=AF_INET; - addr->sin_port=htons(port); - if(host) { - ulong packed_ip=inet_addr(host); - struct hostent *hostIP = packed_ip==INADDR_NONE? - gethostbyname(host) - : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET); - if(hostIP) - memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); - else - return false; - } else - addr->sin_addr.s_addr=INADDR_ANY; - return true; +char* file_load_text(Request& r, + const String& file_spec, + bool fail_on_read_problem, + HashStringValue* params, + bool transcode_result) { + File_read_result file= + file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result); + return file.success?file.str:0; } -static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){ - response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type - response[(response_size=0)]=0; +/// these options were handled but not checked elsewhere, now check them +int pa_get_valid_file_options_count(HashStringValue& options) { int result=0; - char* EOLat=0; - while(true) { - char buf[MAX_STRING*10]; - ssize_t received_size=recv(sock, buf, sizeof(buf), 0); - if(received_size<=0) - break; - response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/); - memcpy(response+response_size, buf, received_size); - response_size+=received_size; - response[response_size]=0; - - if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response - const String status_line(pa_strdup(response, EOLat-response)); - ArrayString astatus; - size_t pos_after=0; - status_line.split(astatus, pos_after, " "); - const String& status_code=*astatus.get(1); - result=status_code.as_int(); - - if(fail_on_status_ne_200 && result!=200) - throw Exception("http.status", - &status_code, - "invalid HTTP response status"); - } - } - if(result) - return result; - else - throw Exception("http.response", - 0, - "bad response from host - no status found (size=%u)", response_size); -} - -/* ********************** request *************************** */ - -#if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) -# define PA_USE_ALARM -#endif - -#ifdef PA_USE_ALARM -static sigjmp_buf timeout_env; -static void timeout_handler(int sig){ - siglongjmp(timeout_env, 1); -} -#endif - -static int http_request(char*& response, size_t& response_size, - const char* host, int port, - const char* request, - int -#ifdef PA_USE_ALARM - timeout -#endif - , - bool fail_on_status_ne_200) { - if(!host) - throw Exception("http.host", - 0, - "zero hostname"); //never - -#ifdef PA_USE_ALARM - signal(SIGALRM, timeout_handler); -#endif - int sock=-1; -#ifdef PA_USE_ALARM - if(sigsetjmp(timeout_env, 1)) { - // stupid gcc [2.95.4] generated bad code - // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code. - // rewritten simplier [though duplicating closesocket code] - if(sock>=0) - closesocket(sock); - throw Exception("http.timeout", - origin_string, - "timeout occured while retrieving document"); - return 0; // never - } else { - alarm(timeout); -#endif - try { - int result; - struct sockaddr_in dest; - - if(!set_addr(&dest, host, port)) - throw Exception("http.host", - 0, - "can not resolve hostname \"%s\"", host); - - if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) - throw Exception("http.connect", - 0, - "can not make socket: %s (%d)", strerror(errno), errno); - if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) - throw Exception("http.connect", - 0, - "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno); - size_t request_size=strlen(request); - if(send(sock, request, request_size, 0)!=(ssize_t)request_size) - throw Exception("http.connect", - 0, - "error sending request: %s (%d)", strerror(errno), errno); - - result=http_read_response(response, response_size, sock, fail_on_status_ne_200); - closesocket(sock); -#ifdef PA_USE_ALARM - alarm(0); -#endif - return result; - } catch(...) { -#ifdef PA_USE_ALARM - alarm(0); -#endif - if(sock>=0) - closesocket(sock); - rethrow; - } -#ifdef PA_USE_ALARM - } -#endif -} - -#ifndef DOXYGEN -struct Http_pass_header_info { - Request_charsets* charsets; - String* request; - bool user_agent_specified; -}; -#endif -static void http_pass_header(HashStringValue::key_type key, - HashStringValue::value_type value, - Http_pass_header_info *info) { - *info->request <charsets->source(), String::CC_UPPER)=="USER-AGENT") - info->user_agent_specified=true; -} - - -static Charset* detect_charset(Charset& source_charset, const String& content_type_value) { - const String::Body CONTENT_TYPE_VALUE= - content_type_value.change_case(source_charset, String::CC_UPPER); - // content-type: xxx/xxx; source_charset=WE-NEED-THIS - // content-type: xxx/xxx; source_charset="WE-NEED-THIS" - // content-type: xxx/xxx; source_charset="WE-NEED-THIS"; - size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET="); - if(before_charseteq_pos!=STRING_NOT_FOUND) { - size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/; - size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin); - bool quoted=open_quote_pos==charset_begin; - if(quoted) - charset_begin++; // skip opening '"' - size_t charset_end=CONTENT_TYPE_VALUE.length(); - if(quoted) { - size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin); - if(close_quote_pos!=STRING_NOT_FOUND) - charset_end=close_quote_pos; - } else { - size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin); - if(delim_pos!=STRING_NOT_FOUND) - charset_end=delim_pos; - } - const String::Body CHARSET_NAME_BODY= - CONTENT_TYPE_VALUE.mid(charset_begin, charset_end); - - return &charsets.get(CHARSET_NAME_BODY); - } - - return 0; -} - -static const String* basic_authorization_field(const char* user, const char* pass) { - if(!user&& !pass) - return 0; - - String combined; - if(user) - combined<get(HTTP_METHOD_NAME)) { - valid_options++; - method=vmethod->as_string().cstr(); - } - if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { - valid_options++; - timeout=vtimeout->as_int(); - } - if((vheaders=options->get(HTTP_HEADERS_NAME))) { - valid_options++; - } - if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) { - valid_options++; - fail_on_status_ne_200=!vany_status->as_bool(); - } - if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) { - valid_options++; - asked_remote_charset=&::charsets.get(vcharset_name->as_string(). - change_case(charsets.source(), String::CC_UPPER)); - } - if(Value* vuser=options->get(HTTP_USER)) { - valid_options++; - user_cstr=vuser->as_string().cstr(); - } - if(Value* vpassword=options->get(HTTP_PASSWORD)) { - valid_options++; - password_cstr=vpassword->as_string().cstr(); - } - - if(valid_options!=options->count()) - throw Exception("parser.runtime", - 0, - "invalid option passed"); - } - if(!asked_remote_charset) // defaulting to $request:charset - asked_remote_charset=&charsets.source(); - - //preparing request - String& connect_string=*new String; - // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one - connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces - - const char* request_cstr; - { - // influence URLencoding of tainted pieces to String::L_URI lang - Temp_client_charset temp(charsets, *asked_remote_charset); - - const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED); - - const char* current=connect_string_cstr; - if(strncmp(current, "http://", 7)!=0) - throw Exception(0, - &connect_string, - "does not start with http://"); //never - current+=7; - - strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0; - char* host_uri=lsplit(host, '/'); - uri=host_uri?current+(host_uri-1-host):"/"; - char* port_cstr=lsplit(host, ':'); - char* error_pos=0; - port=port_cstr?strtol(port_cstr, &error_pos, 0):80; - - //making request - String request; - if(method) - request<is_string()) { // allow empty - if(HashStringValue *headers=vheaders->get_hash()) { - Http_pass_header_info info={&charsets, &request, false}; - headers->for_each(http_pass_header, &info); - user_agent_specified=info.user_agent_specified; - } else - throw Exception("parser.runtime", - &connect_string, - "headers param must be hash"); - }; - if(!user_agent_specified) // defaulting - request << "user-agent: " DEFAULT_USER_AGENT CRLF; - request << CRLF; - - request_cstr=request.cstr(String::L_UNSPECIFIED); - } - // recode those pieces which are not in String::L_URI lang - // [those violating HTTP standard, but widly used] - request_cstr=Charset::transcode( - String::C(request_cstr, strlen(request_cstr)), - charsets.source(), - *asked_remote_charset); - - //sending request - char* response; - size_t response_size; - int status_code=http_request(response, response_size, - host, port, request_cstr, - timeout, fail_on_status_ne_200); - - //processing results - char* raw_body; size_t raw_body_size; - char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/); - if(headers_end_at) { - raw_body=headers_end_at+4; - raw_body_size=response_size-(raw_body-response); - } else - throw Exception("http.response", - &connect_string, - "bad response from host - no headers found"); - - *headers_end_at=0; - const String header_block(response, headers_end_at-response, true); - - ArrayString aheaders; - result.headers=new HashStringValue; - VHash* vtables=new VHash; - result.headers->put(HTTP_TABLES_NAME, vtables); - HashStringValue& tables=vtables->hash(); - - size_t pos_after=0; - header_block.split(aheaders, pos_after, CRLF); - - //processing headers - size_t aheaders_count=aheaders.count(); - Charset* real_remote_charset=0; // undetected, yet - for(size_t i=1; iget_table(); - } else { - // first appearence - Table::columns_type columns =new ArrayString(1); - *columns+=new String("value"); - table=new Table(columns); - } - // this string becomes next row - ArrayString& row=*new ArrayString(1); - row+=&header_value; - *table+=&row; - // not existed before? add it - if(!existed) - tables.put(HEADER_NAME, new VTable(table)); - } - - result.headers->put(HEADER_NAME, new VString(header_value)); - } - // defaulting to used-asked charset [it's never empty!] - if(!real_remote_charset) - real_remote_charset=asked_remote_charset; - - // output response - String::C real_body=String::C(raw_body, raw_body_size); - if(as_text && raw_body_size) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below - real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source()); - - result.str=const_cast(real_body.str); // hacking a little - result.length=real_body.length; - result.headers->put(file_status_name, new VInt(status_code)); + if(options.get(PA_SQL_LIMIT_NAME)) + result++; + if(options.get(PA_SQL_OFFSET_NAME)) + result++; + if(options.get(PA_COLUMN_SEPARATOR_NAME)) + result++; + if(options.get(PA_COLUMN_ENCLOSER_NAME)) + result++; + if(options.get(PA_CHARSET_NAME)) + result++; return result; } -#endif - #ifndef DOXYGEN struct File_read_action_info { char **data; size_t *data_size; + char* buf; size_t offset; size_t count; }; #endif static void file_read_action( - struct stat& finfo, - int f, - const String& file_spec, const char* /*fname*/, bool as_text, - void *context) { + struct stat& finfo, + int f, + const String& file_spec, const char* /*fname*/, bool as_text, + void *context) { File_read_action_info& info=*static_cast(context); - if(size_t to_read_size=(size_t)finfo.st_size) { - *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)]; + size_t to_read_size=info.count; + if(!to_read_size) + to_read_size=(size_t)finfo.st_size; + assert( !(info.buf && as_text) ); + if(to_read_size) { + if(info.offset) + lseek(f, info.offset, SEEK_SET); + *info.data=info.buf + ? info.buf + : new(PointerFreeGC) char[to_read_size+(as_text?1:0)]; *info.data_size=(size_t)read(f, *info.data, to_read_size); if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) - throw Exception(0, + throw Exception("file.read", &file_spec, "read failed: actually read %u bytes count not in [0..%u] valid range", *info.data_size, to_read_size); } else { // empty file - if(as_text) { - *info.data=new(PointerFreeGC) char[1]; - *(char*)(*info.data)=0; - } else - *info.data=0; + // for both, text and binary: for text we need that terminator, for binary we need nonzero pointer to be able to save such files + *info.data=new(PointerFreeGC) char[1]; + *(char*)(*info.data)=0; *info.data_size=0; return; } } + File_read_result file_read(Request_charsets& charsets, const String& file_spec, - bool as_text, HashStringValue *params, - bool fail_on_read_problem) { + bool as_text, HashStringValue *params, + bool fail_on_read_problem, + char* buf, size_t offset, size_t count, bool transcode_text_result) { File_read_result result={false, 0, 0, 0}; -#ifdef PA_HTTP - if(file_spec.starts_with("http://")) { - // fail on read problem - File_read_http_result http=file_read_http(charsets, file_spec, as_text, params); - result.success=true; - result.str=http.str; - result.length=http.length; - result.headers=http.headers; - } else { -#endif - if(params && params->count()) - throw Exception("parser.runtime", + if(params){ + int valid_options=pa_get_valid_file_options_count(*params); + if(valid_options!=params->count()) + throw Exception(PARSER_RUNTIME, 0, - "invalid option passed"); - - File_read_action_info info={&result.str, &result.length}; - result.success=file_read_action_under_lock(file_spec, - "read", file_read_action, &info, - as_text, fail_on_read_problem); -#ifdef PA_HTTP + INVALID_OPTION_PASSED); } -#endif - if(result.success && as_text) { - // UTF-8 signature: EF BB BF - if(result.length>=3) { - char *in=(char *)result.str; - if(strncmp(in, "\xEF\xBB\xBF", 3)==0) { - result.str=in+3; result.length-=3;// skip prefix + File_read_action_info info={&result.str, &result.length, buf, offset, count}; + + result.success=file_read_action_under_lock(file_spec, + "read", file_read_action, &info, + as_text, fail_on_read_problem); + + if(as_text){ + if(result.success){ + if(result.length>=3 && strncmp(result.str, "\xEF\xBB\xBF", 3)==0){ + // skip UTF-8 signature (BOM code) + result.str+=3; + result.length-=3; + } + + if(result.length && transcode_text_result && params){ // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below + if(Value* vcharset_name=params->get(PA_CHARSET_NAME)){ + Charset asked_charset=::charsets.get(vcharset_name->as_string(). + change_case(charsets.source(), String::CC_UPPER)); + + String::C body=String::C(result.str, result.length); + body=Charset::transcode(body, asked_charset, charsets.source()); + + result.str=const_cast(body.str); // hacking a little + result.length=body.length; + } } } - - fix_line_breaks((char *)(result.str), result.length); + if(result.length) + fix_line_breaks(result.str, result.length); } return result; } +File_read_result file_load(Request& r, const String& file_spec, + bool as_text, HashStringValue *params, + bool fail_on_read_problem, + char* buf, size_t offset, size_t count, bool transcode_text_result) { + + File_read_result result={false, 0, 0, 0}; + if(file_spec.starts_with("http://")) { + if(offset || count) + throw Exception(PARSER_RUNTIME, + 0, + "offset and load options are not supported for HTTP:// file load"); + + // fail on read problem + File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result); + result.success=true; + result.str=http.str; + result.length=http.length; + result.headers=http.headers; + } else + result= + file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result); + + return result; +} + #ifdef PA_SAFE_MODE void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { if(finfo.st_uid/*foreign?*/!=geteuid() && finfo.st_gid/*foreign?*/!=getegid()) - throw Exception("parser.runtime", + throw Exception(PARSER_RUNTIME, &file_spec, "parser is in safe mode: " "reading files of foreign group and user disabled " @@ -648,7 +272,7 @@ bool file_read_action_under_lock(const S const char* action_name, File_read_action action, void *context, bool as_text, bool fail_on_read_problem) { - const char* fname=file_spec.cstr(String::L_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; // first open, next stat: @@ -660,7 +284,7 @@ bool file_read_action_under_lock(const S // they delay update till open, so we would receive "!^test[" string // if would do stat, next open. // later: it seems, even this does not help sometimes - if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { + if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { try { if(pa_lock_shared_blocking(f)!=0) throw Exception("file.lock", @@ -689,7 +313,7 @@ bool file_read_action_under_lock(const S pa_unlock(f);close(f); return true; - } else { + } else { if(fail_on_read_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, &file_spec, @@ -699,23 +323,25 @@ bool file_read_action_under_lock(const S } } -static void create_dir_for_file(const String& file_spec) { +void create_dir_for_file(const String& file_spec) { size_t pos_after=1; size_t pos_before; while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { - mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775); + mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775); pos_after=pos_before+1; } } bool file_write_action_under_lock( const String& file_spec, - const char* action_name, File_write_action action, void *context, + const char* action_name, + File_write_action action, + void *context, bool as_text, bool do_append, bool do_block, bool fail_on_lock_problem) { - const char* fname=file_spec.cstr(String::L_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; if(access(fname, W_OK)!=0) // no create_dir_for_file(file_spec); @@ -762,7 +388,8 @@ bool file_write_action_under_lock( #ifndef DOXYGEN struct File_write_action_info { - const char* str; size_t length; + const char* str; + size_t length; }; #endif static void file_write_action(int f, void *context) { @@ -770,20 +397,34 @@ static void file_write_action(int f, voi if(info.length) { int written=write(f, info.str, info.length); if(written<0) - throw Exception(0, + throw Exception("file.access", 0, "write failed: %s (%d)", strerror(errno), errno); } } void file_write( - const String& file_spec, - const char* data, size_t size, + Request_charsets& charsets, + const String& file_spec, + const char* data, + size_t size, bool as_text, - bool do_append) { + bool do_append, + Charset* asked_charset) { + + if(as_text && asked_charset){ + String::C body=String::C(data, size); + body=Charset::transcode(body, charsets.source(), *asked_charset); + data=body.str; + size=body.length; + }; + File_write_action_info info={data, size}; + file_write_action_under_lock( file_spec, - "write", file_write_action, &info, + "write", + file_write_action, + &info, as_text, do_append); } @@ -794,10 +435,11 @@ static void rmdir(const String& file_spe if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) rmdir(file_spec, pos_before+1); - rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC)); + rmdir(file_spec.mid(0, pos_after-1/* / */).taint_cstr(String::L_FILE_SPEC)); } + bool file_delete(const String& file_spec, bool fail_on_problem) { - const char* fname=file_spec.cstr(String::L_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); if(unlink(fname)!=0) if(fail_on_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, @@ -810,9 +452,10 @@ bool file_delete(const String& file_spec rmdir(file_spec, 1); return true; } + void file_move(const String& old_spec, const String& new_spec) { - const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC); - const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC); + const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC); + const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC); create_dir_for_file(new_spec); @@ -835,12 +478,12 @@ bool entry_exists(const char* fname, str } bool entry_exists(const String& file_spec) { - const char* fname=file_spec.cstr(String::L_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); return entry_exists(fname, 0); } static bool entry_readable(const String& file_spec, bool need_dir) { - char* fname=file_spec.cstrm(String::L_FILE_SPEC); + char* fname=file_spec.taint_cstrm(String::L_FILE_SPEC); if(need_dir) { size_t size=strlen(fname); while(size) { @@ -858,29 +501,33 @@ static bool entry_readable(const String& } return false; } -bool file_readable(const String& file_spec) { + +bool file_exist(const String& file_spec) { return entry_readable(file_spec, false); } -bool dir_readable(const String& file_spec) { + +bool dir_exists(const String& file_spec) { return entry_readable(file_spec, true); } -const String* file_readable(const String& path, const String& name) { + +const String* file_exist(const String& path, const String& name) { String& result=*new String(path); result << "/"; result << name; - return file_readable(result)?&result:0; + return file_exist(result)?&result:0; } + bool file_executable(const String& file_spec) { - return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0; + return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0; } bool file_stat(const String& file_spec, - size_t& rsize, - time_t& ratime, - time_t& rmtime, - time_t& rctime, - bool fail_on_read_problem) { - const char* fname=file_spec.cstr(String::L_FILE_SPEC); + size_t& rsize, + time_t& ratime, + time_t& rmtime, + time_t& rctime, + bool fail_on_read_problem) { + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); struct stat finfo; if(stat(fname, &finfo)!=0) if(fail_on_read_problem) @@ -898,67 +545,146 @@ bool file_stat(const String& file_spec, } char* getrow(char* *row_ref, char delim) { - char* result=*row_ref; - if(result) { + char* result=*row_ref; + if(result) { *row_ref=strchr(result, delim); if(*row_ref) *((*row_ref)++)=0; else if(!*result) return 0; - } - return result; + } + return result; } char* lsplit(char* string, char delim) { - if(string) { + if(string) { char* v=strchr(string, delim); if(v) { *v=0; return v+1; } - } - return 0; + } + return 0; } char* lsplit(char* *string_ref, char delim) { - char* result=*string_ref; + char* result=*string_ref; char* next=lsplit(*string_ref, delim); - *string_ref=next; - return result; + *string_ref=next; + return result; } char* rsplit(char* string, char delim) { - if(string) { + if(string) { char* v=strrchr(string, delim); if(v) { *v=0; return v+1; } - } - return NULL; + } + return NULL; +} + + +// format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx +// flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx +// width, precision: non negative decimal number +enum FormatType { + FormatInvalid, + FormatInt, + FormatUInt, + FormatDouble +}; +FormatType format_type(char* fmt){ + enum FormatState { + Percent, + Flags, + Width, + Precision, + Done + } state=Percent; + + FormatType result=FormatInvalid; + + char* pos=fmt; + while(char c=*(pos++)){ + switch(state){ + case Percent: + if(c=='%'){ + state=Flags; + } else { + return FormatInvalid; // 1st char must be '%' only + } + break; + case Flags: + if(strchr("-+ #0", c)!=0){ + break; + } + // go to the next step + case Width: + if(c=='.'){ + state=Precision; + break; + } + // go to the next step + case Precision: + if(c>='0' && c<='9'){ + if(state == Flags) state=Width; // no more flags + break; + } else if(c=='d' || c=='i'){ + result=FormatInt; + } else if(strchr("feEgG", c)!=0){ + result=FormatDouble; + } else if(strchr("uoxX", c)!=0){ + result=FormatUInt; + } else { + return FormatInvalid; // invalid char + } + state=Done; + break; + case Done: + return FormatInvalid; // no chars allowed after 'type' + } + } + return result; } -/// @todo less stupid type detection + const char* format(double value, char* fmt) { - char local_buf[MAX_NUMBER]; - size_t size; - - if(fmt) - if(strpbrk(fmt, "diouxX")) - if(strpbrk(fmt, "ouxX")) - size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); - else + char local_buf[MAX_NUMBER]; + int size=-1; + + if(fmt && strlen(fmt)){ + switch(format_type(fmt)){ + case FormatDouble: + size=snprintf(local_buf, sizeof(local_buf), fmt, value); + break; + case FormatInt: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value); - else - size=snprintf(local_buf, sizeof(local_buf), fmt, value); - else - size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); - - return pa_strdup(local_buf, size); + break; + case FormatUInt: + size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); + break; + case FormatInvalid: + throw Exception(PARSER_RUNTIME, + 0, + "Incorrect format string '%s' was specified.", fmt); + } + } else + size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); + + if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing + throw Exception(PARSER_RUNTIME, + 0, + "Error occure white executing snprintf with format string '%s'.", fmt); + } + + return pa_strdup(local_buf, (size_t)size); } size_t stdout_write(const void *buf, size_t size) { #ifdef WIN32 + size_t to_write = size; do{ int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); if(chunk_written<=0) @@ -967,47 +693,83 @@ size_t stdout_write(const void *buf, siz buf=((const char*)buf)+chunk_written; } while(size>0); - return size; + return to_write-size; #else return fwrite(buf, 1, size, stdout); #endif } -char* unescape_chars(const char* cp, int len) { - char* s=new(PointerFreeGC) char[len + 1]; - enum EscapeState { - EscapeRest, - EscapeFirst, - EscapeSecond - } escapeState=EscapeRest; - int escapedValue=0; +enum EscapeState { + EscapeRest, + EscapeFirst, + EscapeSecond, + EscapeUnicode +}; + +// @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually) +char* unescape_chars(const char* cp, int len, Charset* charset, bool ignore_plus){ + char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes) + char* dst=s; + EscapeState escapeState=EscapeRest; + uint escapedValue=0; int srcPos=0; - int dstPos=0; - while(srcPos < len) { - int ch=cp[srcPos]; - switch(escapeState) { - case EscapeRest: - if(ch=='%') { - escapeState=EscapeFirst; - } else if(ch=='+') { - s[dstPos++]=' '; - } else { - s[dstPos++]=ch; + short int jsCnt=0; + while(srcPosstore_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?'); + escapeState=EscapeRest; + } + } else { + // not full unicode value + escapeState=EscapeRest; + } + break; } - break; - case EscapeFirst: - escapedValue=hex_value[ch] << 4; - escapeState=EscapeSecond; - break; - case EscapeSecond: - escapedValue +=hex_value[ch]; - s[dstPos++]=escapedValue; - escapeState=EscapeRest; - break; } - srcPos++; + + srcPos++; } - s[dstPos]=0; + + *dst=0; // zero-termination return s; } @@ -1028,55 +790,111 @@ void slashes_to_back_slashes(char* s) { */ #endif -bool StrEqNc(const char* s1, const char* s2, bool strict) { +bool StrStartFromNC(const char* str, const char* substr, bool equal){ while(true) { - if(!(*s1)) { - if(!(*s2)) + if(!(*substr)){ + if(!(*str)) return true; else - return !strict; - } else if(!(*s2)) - return !strict; - if(isalpha(*s1)) { - if(tolower(*s1) !=tolower(*s2)) + return !equal; + } + if(!(*str)) + return false; + if(isalpha((unsigned char)*str)) { + if(tolower((unsigned char)*str)!=tolower((unsigned char)*substr)) return false; - } else if((*s1) !=(*s2)) + } else if((*str) != (*substr)) return false; - s1++; - s2++; + str++; + substr++; + } +} + +size_t strpos(const char *str, const char *substr) { + const char *p = strstr(str, substr); + return (p==0)?STRING_NOT_FOUND:p-str; +} + +// content-type: xxx; charset=WE-NEED-THIS +// content-type: xxx; charset="WE-NEED-THIS" +// content-type: xxx; charset="WE-NEED-THIS"; +Charset* detect_charset(const char* content_type){ + if(content_type){ + char* CONTENT_TYPE=pa_strdup(content_type); + + for(char *p=CONTENT_TYPE; *p; p++) + *p=(char)toupper((unsigned char)*p); + + if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){ + begin+=8; // skip "CHARSET=" + char* end=0; + if(*begin && (*begin=='"' || *begin =='\'')){ + char quote=*begin; + begin++; + end=(char*)strchr(begin, quote); + } + if(!end) + end=(char*)strchr(begin, ';'); + + if(end) + *end=0; // terminator + + return *begin?&charsets.get(begin):0; + } } + return 0; } + static bool isLeap(int year) { - return !( - (year % 4) || ((year % 400) && !(year % 100)) - ); + return !( + (year % 4) || ((year % 400) && !(year % 100)) + ); } int getMonthDays(int year, int month) { - int monthDays[]={ - 31, - isLeap(year) ? 29 : 28, - 31, - 30, - 31, - 30, - 31, - 31, - 30, - 31, - 30, - 31 - }; - return monthDays[month]; -} - -void remove_crlf(char* start, char* end) { - for(char* p=start; p> 6)]; *outptr++ = base64_alphabet [c3 & 0x3f]; /* this is a bit ugly ... */ - if ((++already) >= 19) { + if ((++already) >= BASE64_GROUPS_IN_LINE) { *outptr++ = '\n'; already = 0; } @@ -1236,7 +1057,7 @@ g_mime_utils_base64_encode_step (const u case 2: *saveout++ = *inptr++; case 1: *saveout++ = *inptr++; } - ((char *)save)[0] += inlen; + *(char *)save = *(char *)save+(char)inlen; } /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n", @@ -1295,15 +1116,194 @@ g_mime_utils_base64_encode_close (const return (outptr - out); } -char* pa_base64(const char *in, size_t len) +static unsigned char gmime_base64_rank[256] = { + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63, + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255, + 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255, + 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, + 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255, +}; + +/** + * g_mime_utils_base64_decode_step: + * @in: input stream + * @inlen: max length of data to decode + * @out: output stream + * @state: holds the number of bits that are stored in @save + * @save: leftover bits that have not yet been decoded + * + * Decodes a chunk of base64 encoded data. + * + * Returns the number of bytes decoded (which have been dumped in @out). + **/ +size_t +g_mime_utils_base64_decode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) { - /* wont go to more than 2x size (overly conservative) */ - char* result=new(PointerFreeGC) char[len * 2 + 6]; + const unsigned char *inptr; + unsigned char *outptr; + const unsigned char *inend; + int saved; + unsigned char c; + int i; + + inend = in + inlen; + outptr = out; + + /* convert 4 base64 bytes to 3 normal bytes */ + saved = *save; + i = *state; + inptr = in; + while (inptr < inend) { + c = gmime_base64_rank[*inptr++]; + if (c != 0xff) { + saved = (saved << 6) | c; + i++; + if (i == 4) { + *outptr++ = (unsigned char)(saved >> 16); + *outptr++ = (unsigned char)(saved >> 8); + *outptr++ = (unsigned char)(saved); + i = 0; + } + } + } + + *save = saved; + *state = i; + + /* quick scan back for '=' on the end somewhere */ + /* fortunately we can drop 1 output char for each trailing = (upto 2) */ + i = 2; + while (inptr > in && i) { + inptr--; + if (gmime_base64_rank[*inptr] != 0xff) { + if (*inptr == '=' && outptr > out) + outptr--; + i--; + } + } + + /* if i != 0 then there is a truncation error! */ + return (outptr - out); +} + + +char* pa_base64_encode(const char *in, size_t in_size){ + size_t new_size = ((in_size / 3 + 1) * 4); + new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/; + char* result = new(PointerFreeGC) char[new_size]; int state=0; int save=0; - size_t filled=g_mime_utils_base64_encode_close ((const unsigned char*)in, len, - (unsigned char*)result, &state, &save); - assert(filled <= len * 2 + 6); +#ifndef NDEBUG + size_t filled= +#endif + g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save); + + //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled); + assert(filled <= new_size); return result; } + + +char* pa_base64_encode(const String& file_spec){ + unsigned char* base64=0; + File_base64_action_info info={&base64}; + + file_read_action_under_lock(file_spec, + "pa_base64_encode", file_base64_file_action, &info); + + return (char*)base64; +} + + +static void file_base64_file_action( + struct stat& finfo, + int f, + const String&, const char* /*fname*/, bool, + void *context) { + + if(finfo.st_size) { + File_base64_action_info& info=*static_cast(context); + *info.base64=new(PointerFreeGC) unsigned char[finfo.st_size * 2 + 6]; + unsigned char* base64 = *info.base64; + int state=0; + int save=0; + int nCount; + do { + unsigned char buffer[FILE_BUFFER_SIZE]; + nCount = file_block_read(f, buffer, sizeof(buffer)); + if( nCount ){ + size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save); + base64+=filled; + } + } while(nCount > 0); + g_mime_utils_base64_encode_close (0, 0, base64, &state, &save); + } +} + +void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size) { + /* wont go to more than had (overly conservative) */ + result=new(PointerFreeGC) char[in_size+1/*terminator*/]; + int state=0; + int save=0; + result_size= + g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size, + (unsigned char*)result, &state, &save); + assert(result_size <= in_size); + result[result_size]=0; // for text files +} + + +int file_block_read(const int f, unsigned char* buffer, const size_t size){ + int nCount = read(f, buffer, size); + if (nCount < 0) + throw Exception("file.read", + 0, + "read failed: %s (%d)", strerror(errno), errno); + return nCount; +} + +const unsigned long pa_crc32(const char *in, size_t in_size){ + unsigned long crc32=0xFFFFFFFF; + + InitCrc32Table(); + for(size_t i = 0; i(context); + if(finfo.st_size) { + InitCrc32Table(); + int nCount=0; + do { + unsigned char buffer[FILE_BUFFER_SIZE]; + nCount = file_block_read(f, buffer, sizeof(buffer)); + for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32); + } while(nCount > 0); + } +} +