--- parser3/src/main/pa_common.C 2003/11/06 08:48:59 1.162 +++ parser3/src/main/pa_common.C 2005/11/22 11:12:40 1.210 @@ -1,11 +1,32 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) -*/ -static const char* IDENT_COMMON_C="$Date: 2003/11/06 08:48:59 $"; + * BASE64 part + * Authors: Michael Zucchi + * Jeffrey Stedfast + * + * Copyright 2000-2004 Ximian, Inc. (www.ximian.com) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA. + * + */ + +static const char * const IDENT_COMMON_C="$Date: 2005/11/22 11:12:40 $"; #include "pa_common.h" #include "pa_exception.h" @@ -21,6 +42,7 @@ static const char* IDENT_COMMON_C="$Date #include "pa_vint.h" #include "pa_vhash.h" #include "pa_vtable.h" +#include "pa_socks.h" #ifdef CYGWIN #define _GNU_H_WINDOWS32_SOCKETS @@ -64,61 +86,9 @@ int PASCAL closesocket(SOCKET); # endif #endif -// locking constants -//#define PA_DEBUG_NO_LOCKING - -#ifdef PA_DEBUG_NO_LOCKING - -#ifdef HAVE_FLOCK - -static int pa_lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); } -static int pa_lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); } -static int pa_lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); } -static int pa_unlock(int fd) { return flock(fd, LOCK_UN); } - -#else -#ifdef HAVE__LOCKING - -#define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1) -static int pa_lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); } -static int pa_lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); } -static int pa_lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); } -static int pa_unlock(int fd) { FLOCK(_LK_UNLCK); } - -#else -#ifdef HAVE_FCNTL - -#define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls) -static int pa_lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); } -static int pa_lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); } -static int pa_lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); } -static int pa_unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); } - -#else -#ifdef HAVE_LOCKF - -#define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1) -static int pa_lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking -static int pa_lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); } -static int pa_lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); } -static int pa_unlock(int fd) { FLOCK(F_TLOCK); } - -#else - -#error unable to find file locking func - -#endif -#endif -#endif -#endif - -#else -static int pa_lock_shared_blocking(int fd) { return 0; } -static int pa_lock_exclusive_blocking(int fd) { return 0; } -static int pa_lock_exclusive_nonblocking(int fd) { return 0; } -static int pa_unlock(int fd) { return 0; } - -#endif +# ifndef INADDR_NONE +# define INADDR_NONE ((ulong) -1) +# endif // defines for globals @@ -128,23 +98,18 @@ static int pa_unlock(int fd) { return 0; const String file_status_name(FILE_STATUS_NAME); -// defines for statics +// defines #define HTTP_METHOD_NAME "method" +#define HTTP_FORM_NAME "form" +#define HTTP_BODY_NAME "body" #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" - -// statics - -static const String http_method_name(HTTP_METHOD_NAME); -static const String http_timeout_name(HTTP_TIMEOUT_NAME); -static const String http_headers_name(HTTP_HEADERS_NAME); -static const String http_any_status_name(HTTP_ANY_STATUS_NAME); -static const String http_charset_name(HTTP_CHARSET_NAME); -static const String http_tables_name(HTTP_TABLES_NAME); +#define HTTP_USER "user" +#define HTTP_PASSWORD "password" // defines @@ -187,8 +152,23 @@ char* file_read_text(Request_charsets& c return file.success?file.str:0; } -#ifdef PA_HTTP +/// these options were handled but not checked elsewhere, now check them +static int get_valid_file_options_count(HashStringValue& options) +{ + int result=0; + 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++; + return result; +} + //http request stuff +#ifdef PA_HTTP #undef CRLF #define CRLF "\r\n" @@ -198,44 +178,138 @@ static bool set_addr(struct sockaddr_in addr->sin_family=AF_INET; addr->sin_port=htons(port); if(host) { - if(struct hostent *hostIP=gethostbyname(host)) - memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length); - else - return false; + ulong packed_ip=inet_addr(host); + if(packed_ip!=INADDR_NONE) + memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip)); + else { + struct hostent *hostIP=gethostbyname(host); + 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; } -static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){ +size_t guess_content_length(char* buf) { + char* ptr; + if((ptr=strstr(buf, "Content-Length:"))) // Apache + goto found; + if((ptr=strstr(buf, "content-length:"))) // Parser 3 + goto found; + if((ptr=strstr(buf, "Content-length:"))) // maybe 1 + goto found; + if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2 + goto found; + return 0; +found: + char *error_pos; + size_t result=(size_t)strtol(ptr+15/*strlen("CONTENT-LENGTH:")*/, &error_pos, 0); + + const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/; + if(result>reasonable_initial_max) // sanity check + return reasonable_initial_max; + return 0;//result; +} + +static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) { int result=0; - size_t EOLat=0; - while(true) { - char *buf=new(PointerFreeGC) char[MAX_STRING]; - ssize_t size=recv(sock, buf, MAX_STRING, 0); - if(size<=0) - break; - response.append_strdup(buf, size, String::L_TAINTED); - if(!result && (EOLat=response.pos(CRLF, 2))!=STRING_NOT_FOUND) { // checking status in first response - const String& status_line=response.mid(0, (size_t)EOLat); - 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"); + // fetching some to local buffer, guessing on possible content-length + response_size=0x400*20; // initial size if content-length could not be determined + const size_t preview_size=0x400*20; + char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers + ssize_t received_size=recv(sock, preview_buf, preview_size, 0); + if(received_size==0) + goto done; + if(received_size<0) { + if(int no=pa_socks_errno()) + throw Exception("http.timeout", + 0, + "error receiving response header: %s (%d)", pa_socks_strerr(no), no); + goto done; + } + // detecting response_size + { + preview_buf[received_size]=0; // terminator + if(size_t content_length=guess_content_length(preview_buf)) + response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later + } + + // [gcc is happier this way, see goto above] + { + // allocating initial buf + response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type + char* ptr=response; + size_t todo_size=response_size; + // coping part of already received body + memcpy(ptr, preview_buf, received_size); + ptr+=received_size; + todo_size-=received_size; + + // we use terminator byte for two purposes here: + // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body? + // invariant: all Strings should have zero-terminated buffers + // 2. we use that out-of-size byte to detect if our content-length guess was wrong + // when recv gets more than we expected + // a) we know that the content-length guess was wrong + // b) we have space to put the first byte of extra data + // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0 + while(true) { + received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0); + if(received_size==0) { + response_size-=todo_size; // in case we received less than expected, cut down the reported size + break; + } + if(received_size<0) { + if(int no=pa_socks_errno()) + throw Exception("http.timeout", + 0, + "error receiving response body: %s (%d)", pa_socks_strerr(no), no); + break; + } + // they've touched the terminator? + if((size_t)received_size>todo_size) + { + // that means that our guessed response_size was not big enough + const size_t grow_chunk_size=0x400*0x400; // 1M + response_size+=grow_chunk_size; + size_t ptr_offset=ptr-response; + response=(char*)pa_realloc(response, response_size+1/*terminator*/); + ptr=response+ptr_offset; + todo_size+=grow_chunk_size; + } + // can't do this before realloc: we need 1?1:0); + result=status_code.as_int(); + + if(fail_on_status_ne_200 && result!=200) + throw Exception("http.status", + &status_code, + "invalid HTTP response status"); + } } } +done: if(result) + { + response[response_size]=0; return result; + } else throw Exception("http.response", 0, - "bad response from host - no status found (size=%lu)", response.length()); + "bad response from host - no status found (size=%u)", response_size); } /* ********************** request *************************** */ @@ -246,38 +320,39 @@ static int http_read_response(String& re #ifdef PA_USE_ALARM static sigjmp_buf timeout_env; -static void timeout_handler(int sig){ +static void timeout_handler(int /*sig*/){ siglongjmp(timeout_env, 1); } #endif -static int http_request(String& response, - const char* host, int port, +static int http_request(char*& response, size_t& response_size, + const char* host, short port, const char* request, - int timeout, + int timeout_secs, bool fail_on_status_ne_200) { if(!host) throw Exception("http.host", 0, "zero hostname"); //never + volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning] + int sock=-1; #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] + // rewritten simplier [athough duplicating closesocket code] if(sock>=0) closesocket(sock); throw Exception("http.timeout", - origin_string, + 0, "timeout occured while retrieving document"); return 0; // never } else { - alarm(timeout); + alarm(timeout_secs); #endif try { int result; @@ -288,21 +363,42 @@ static int http_request(String& response 0, "can not resolve hostname \"%s\"", host); - if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) + if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) { + int no=pa_socks_errno(); throw Exception("http.connect", 0, - "can not make socket: %s (%d)", strerror(errno), errno); - if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) + "can not make socket: %s (%d)", pa_socks_strerr(no), no); + } + + // To enable SO_DONTLINGER (that is, disable SO_LINGER) + // l_onoff should be set to zero and setsockopt should be called + linger dont_linger={0,0}; + setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger)); + +#ifdef WIN32 +// SO_*TIMEO can be defined in .h but not implemlemented in protocol, +// failing subsequently with Option not supported by protocol (99) message +// could not suppress that, so leaving this only for win32 + int timeout_ms=timeout_secs*1000; + setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms)); + setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms)); +#endif + + if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) { + int no=pa_socks_errno(); throw Exception("http.connect", 0, - "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno); + "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no); + } size_t request_size=strlen(request); - if(send(sock, request, request_size, 0)!=(ssize_t)request_size) - throw Exception("http.connect", + if(send(sock, request, request_size, 0)!=(ssize_t)request_size) { + int no=pa_socks_errno(); + throw Exception("http.timeout", 0, - "error sending request: %s (%d)", strerror(errno), errno); + "error sending request: %s (%d)", pa_socks_strerr(no), no); + } - result=http_read_response(response, sock, fail_on_status_ne_200); + result=http_read_response(response, response_size, sock, fail_on_status_ne_200); closesocket(sock); #ifdef PA_USE_ALARM alarm(0); @@ -372,48 +468,144 @@ static Charset* detect_charset(Charset& return 0; } +static const String* basic_authorization_field(const char* user, const char* pass) { + if(!user&& !pass) + return 0; + + String combined; + if(user) + combined<key, *row->get(0), info->result); +} +static void form_value2string( + HashStringValue::key_type key, + HashStringValue::value_type value, + String* result) +{ + if(const String* svalue=value->get_string()) + form_string_value2string(key, *svalue, *result); + else if(Table* tvalue=value->get_table()) { + Form_table_value2string_info info(key, *result); + tvalue->for_each(form_table_value2string, &info); + } else + throw Exception(0, + new String(key, String::L_TAINTED), + "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type()); +} +static const char* form2string(HashStringValue& form) { + String string; + form.for_each(form_value2string, &string); + return string.cstr(String::L_UNSPECIFIED); +} #ifndef DOXYGEN struct File_read_http_result { char *str; size_t length; HashStringValue* headers; }; #endif +static void find_headers_end(char* p, + char*& headers_end_at, + char*& raw_body) +{ + raw_body=p; + // \n\n + // \r\n\r\n + while((p=strchr(p, '\n'))) { + headers_end_at=++p; // \n>.< + if(*p=='\r') // \r\n>\r?<\n + p++; + if(*p=='\n') { // \r\n\r>\n?< + raw_body=p+1; + return; + } + } + headers_end_at=0; +} + /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now static File_read_http_result file_read_http(Request_charsets& charsets, const String& file_spec, - HashStringValue *options=0) { + bool as_text, + HashStringValue *options=0) { File_read_http_result result; char host[MAX_STRING]; const char* uri; - int port; - const char* method=0; - int timeout=2; + short port; + const char* method="GET"; bool method_is_get; + HashStringValue* form=0; + const char* body_cstr=0; + int timeout_secs=2; bool fail_on_status_ne_200=true; Value* vheaders=0; Charset *asked_remote_charset=0; + const char* user_cstr=0; + const char* password_cstr=0; if(options) { - int valid_options=0; - if(Value* vmethod=options->get(http_method_name)) { + int valid_options=get_valid_file_options_count(*options); + + if(Value* vmethod=options->get(HTTP_METHOD_NAME)) { valid_options++; method=vmethod->as_string().cstr(); } - if(Value* vtimeout=options->get(http_timeout_name)) { + if(Value* vform=options->get(HTTP_FORM_NAME)) { valid_options++; - timeout=vtimeout->as_int(); - } - if(vheaders=options->get(http_headers_name)) { + form=vform->get_hash(); + } + if(Value* vbody=options->get(HTTP_BODY_NAME)) { valid_options++; - } - if(Value* vany_status=options->get(http_any_status_name)) { + body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED); + } + if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { + valid_options++; + timeout_secs=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)) { + } + 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()) @@ -424,12 +616,18 @@ static File_read_http_result file_read_h if(!asked_remote_charset) // defaulting to $request:charset asked_remote_charset=&charsets.source(); + method_is_get=strcmp(method, "GET")==0; + if(method_is_get && body_cstr) + throw Exception("parser.runtime", + 0, + "you can not use $."HTTP_BODY_NAME" option with method GET"); + //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; + String request_head_and_body; { // influence URLencoding of tainted pieces to String::L_URI lang Temp_client_charset temp(charsets, *asked_remote_charset); @@ -448,21 +646,35 @@ static File_read_http_result file_read_h 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; + port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80; + if(strchr(uri, '?') && form) + throw Exception("parser.runtime", + 0, + "use either uri with ?params or $."HTTP_FORM_NAME" option"); - //making request - String request; - if(method) - request<is_string()) { // allow empty if(HashStringValue *headers=vheaders->get_hash()) { - Http_pass_header_info info={&charsets, &request}; + Http_pass_header_info info={&charsets, &head, false}; headers->for_each(http_pass_header, &info); user_agent_specified=info.user_agent_specified; } else @@ -471,93 +683,115 @@ static File_read_http_result file_read_h "headers param must be hash"); }; if(!user_agent_specified) // defaulting - request << "user-agent: " DEFAULT_USER_AGENT CRLF; - request << CRLF; + head << "user-agent: " DEFAULT_USER_AGENT CRLF; + + if(body_cstr) { + // recode those pieces which are not in String::L_URI lang + // [those violating HTTP standard, but widly used] + body_cstr=Charset::transcode( + String::C(body_cstr, strlen(body_cstr)), + charsets.source(), + *asked_remote_charset); + + head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF; + } + + const char* head_cstr=head.cstr(String::L_UNSPECIFIED); - 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] + head_cstr=Charset::transcode( + String::C(head_cstr, strlen(head_cstr)), + charsets.source(), + *asked_remote_charset); + + // head + end of header + request_head_and_body << head_cstr << CRLF; + // body + if(body_cstr) + request_head_and_body << body_cstr; } - // 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 - String response; - int status_code=http_request(response, - host, port, request_cstr, - timeout, fail_on_status_ne_200); + char* response; + size_t response_size; + int status_code=http_request(response, response_size, + host, port, request_head_and_body.cstr(), + timeout_secs, fail_on_status_ne_200); //processing results - size_t pos=response.pos(CRLF CRLF, 4); - if(pos==STRING_NOT_FOUND || pos<1) { - throw Exception("http.response", - &connect_string, - "bad response from host - no headers found"); - } - const String& header_block=response.mid(0, pos); - const String& raw_body=response.mid(pos+4, response.length()); + char* raw_body; size_t raw_body_size; + char* headers_end_at; + find_headers_end(response, + headers_end_at, + raw_body); + raw_body_size=response_size-(raw_body-response); - 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(); + result.headers->put(HTTP_TABLES_NAME, vtables); 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); + + if(headers_end_at) { + *headers_end_at=0; + const String header_block(String::C(response, headers_end_at-response), true); + + ArrayString aheaders; + HashStringValue& tables=vtables->hash(); + + size_t pos_after=0; + header_block.split(aheaders, pos_after, "\n"); + + //processing headers + size_t aheaders_count=aheaders.count(); + 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)); } - // 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; + result.headers->put(HEADER_NAME, new VString(header_value)); + } + } // output response - String::C real_body=Charset::transcode( - String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()), - *real_remote_charset, - charsets.source()); + 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 + // defaulting to used-asked charset [it's never empty!] + if(!real_remote_charset) + real_remote_charset=asked_remote_charset; + 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; @@ -570,53 +804,70 @@ static File_read_http_result file_read_h #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, + 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, &file_spec, - "read failed: actually read %l bytes count not in [0..%lu] valid range", + "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) { - File_read_result result={false}; + bool fail_on_read_problem, + char* buf, size_t offset, size_t count) { + File_read_result result={false, 0, 0, 0}; #ifdef PA_HTTP 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=file_read_http(charsets, file_spec, params); + 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", - 0, - "invalid option passed"); + if(params) { + int valid_options=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}; + 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); @@ -634,8 +885,6 @@ File_read_result file_read(Request_chars } fix_line_breaks((char *)(result.str), result.length); - // note: after fixing - ((char*&)(result.str))[result.length]=0; } return result; @@ -713,7 +962,7 @@ 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) { @@ -810,10 +1059,10 @@ static void rmdir(const String& file_spe rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC)); } -bool file_delete(const String& file_spec, bool fail_on_read_problem) { +bool file_delete(const String& file_spec, bool fail_on_problem) { const char* fname=file_spec.cstr(String::L_FILE_SPEC); if(unlink(fname)!=0) - if(fail_on_read_problem) + if(fail_on_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, &file_spec, "unlink failed: %s (%d), actual filename '%s'", @@ -973,6 +1222,7 @@ const char* format(double value, char* f 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) @@ -981,7 +1231,7 @@ 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 @@ -994,11 +1244,11 @@ char* unescape_chars(const char* cp, int EscapeFirst, EscapeSecond } escapeState=EscapeRest; - int escapedValue=0; + uchar escapedValue=0; int srcPos=0; int dstPos=0; while(srcPos < len) { - int ch=cp[srcPos]; + uchar ch=(uchar)cp[srcPos]; switch(escapeState) { case EscapeRest: if(ch=='%') { @@ -1010,7 +1260,7 @@ char* unescape_chars(const char* cp, int } break; case EscapeFirst: - escapedValue=hex_value[ch] << 4; + escapedValue=(uchar)(hex_value[ch] << 4); escapeState=EscapeSecond; break; case EscapeSecond: @@ -1051,8 +1301,8 @@ bool StrEqNc(const char* s1, const char* return !strict; } else if(!(*s2)) return !strict; - if(isalpha(*s1)) { - if(tolower(*s1) !=tolower(*s2)) + if(isalpha((unsigned char)*s1)) { + if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2)) return false; } else if((*s1) !=(*s2)) return false; @@ -1104,6 +1354,11 @@ int __vsnprintf(char* b, size_t s, const // note: on win32& maybe somewhere else // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing --s; + + // clients do not check for negative 's', feature: ignore such prints + if((ssize_t)s<0) + return 0; + #if _MSC_VER /* win32: @@ -1129,7 +1384,7 @@ int __vsnprintf(char* b, size_t s, const if(r<0) r=0; - else if(r>s) + else if((size_t)r>s) r=s; #endif b[r]=0; @@ -1143,3 +1398,274 @@ int __snprintf(char* b, size_t s, const va_end(l); return r; } + +/* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */ +/* + * Authors: Michael Zucchi + * Jeffrey Stedfast + * + * Copyright 2000 Helix Code, Inc. (www.helixcode.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 char *base64_alphabet = + "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; + +/** + * g_mime_utils_base64_encode_step: + * @in: input stream + * @inlen: length of the input + * @out: output string + * @state: holds the number of bits that are stored in @save + * @save: leftover bits that have not yet been encoded + * + * Base64 encodes a chunk of data. Performs an 'encode step', only + * encodes blocks of 3 characters to the output at a time, saves + * left-over state in state and save (initialise to 0 on first + * invocation). + * + * Returns the number of bytes encoded. + **/ +static size_t +g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) +{ + register const unsigned char *inptr; + register unsigned char *outptr; + + if (inlen <= 0) + return 0; + + inptr = in; + outptr = out; + + if (inlen + ((unsigned char *)save)[0] > 2) { + const unsigned char *inend = in + inlen - 2; + register int c1 = 0, c2 = 0, c3 = 0; + register int already; + + already = *state; + + switch (((char *)save)[0]) { + case 1: c1 = ((unsigned char *)save)[1]; goto skip1; + case 2: c1 = ((unsigned char *)save)[1]; + c2 = ((unsigned char *)save)[2]; goto skip2; + } + + /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */ + while (inptr < inend) { + c1 = *inptr++; + skip1: + c2 = *inptr++; + skip2: + c3 = *inptr++; + *outptr++ = base64_alphabet [c1 >> 2]; + *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)]; + *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)]; + *outptr++ = base64_alphabet [c3 & 0x3f]; + /* this is a bit ugly ... */ + if ((++already) >= 19) { + *outptr++ = '\n'; + already = 0; + } + } + + ((unsigned char *)save)[0] = 0; + inlen = 2 - (inptr - inend); + *state = already; + } + + //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen)); + + if (inlen > 0) { + register char *saveout; + + /* points to the slot for the next char to save */ + saveout = & (((char *)save)[1]) + ((char *)save)[0]; + + /* inlen can only be 0 1 or 2 */ + switch (inlen) { + case 2: *saveout++ = *inptr++; + case 1: *saveout++ = *inptr++; + } + ((char *)save)[0] += inlen; + } + + /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n", + (int)((char *)save)[0], + (int)((char *)save)[1], + (int)((char *)save)[2]));*/ + + return (outptr - out); +} + +/** + * g_mime_utils_base64_encode_close: + * @in: input stream + * @inlen: length of the input + * @out: output string + * @state: holds the number of bits that are stored in @save + * @save: leftover bits that have not yet been encoded + * + * Base64 encodes the input stream to the output stream. Call this + * when finished encoding data with g_mime_utils_base64_encode_step to + * flush off the last little bit. + * + * Returns the number of bytes encoded. + **/ +static size_t +g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save) +{ + unsigned char *outptr = out; + int c1, c2; + + if (inlen > 0) + outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save); + + c1 = ((unsigned char *)save)[1]; + c2 = ((unsigned char *)save)[2]; + + switch (((unsigned char *)save)[0]) { + case 2: + outptr[2] = base64_alphabet [(c2 & 0x0f) << 2]; + goto skip; + case 1: + outptr[2] = '='; + skip: + outptr[0] = base64_alphabet [c1 >> 2]; + outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)]; + outptr[3] = '='; + outptr += 4; + break; + } + + *outptr++ = 0; + + *save = 0; + *state = 0; + + return (outptr - out); +} + +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) +{ + const register unsigned char *inptr; + register unsigned char *outptr; + const unsigned char *inend; + register guint32 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++ = saved >> 16; + *outptr++ = saved >> 8; + *outptr++ = 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) +{ + /* wont go to more than 2x size (overly conservative) */ + char* result=new(PointerFreeGC) char[in_size * 2 + 6]; + int state=0; + int save=0; +#ifndef NDEBUG + size_t filled= +#endif + g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, + (unsigned char*)result, &state, &save); + assert(filled <= in_size * 2 + 6); + + return result; +} + +void pa_base64_decode(const char *in, size_t in_size, void*& result, size_t& result_size) +{ + /* wont go to more than had (overly conservative) */ + result=static_cast(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); + static_cast(result)[result_size]=0; // for text files +}