--- parser3/src/main/pa_common.C 2003/02/04 15:59:05 1.143.2.15 +++ parser3/src/main/pa_common.C 2004/03/05 13:17:59 1.179 @@ -1,18 +1,26 @@ /** @file Parser: commonly functions. - Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_COMMON_C="$Date: 2003/02/04 15:59:05 $"; +static const char * const IDENT_COMMON_C="$Date: 2004/03/05 13:17:59 $"; #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_globals.h" +#include "pa_vhash.h" +#include "pa_vtable.h" #ifdef CYGWIN #define _GNU_H_WINDOWS32_SOCKETS @@ -29,6 +37,14 @@ int PASCAL closesocket(SOCKET); # endif #endif +#else + +# if defined(WIN32) +# include +# endif + +#endif + // some maybe-undefined constants #ifndef _O_TEXT @@ -48,66 +64,28 @@ int PASCAL closesocket(SOCKET); # endif #endif -// locking constants - -#ifdef HAVE_FLOCK - -static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); } -static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); } -static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); } -static int 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 lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); } -static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); } -static int 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 lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); } -static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); } -static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); } - -#else -#ifdef HAVE_LOCKF +# ifndef INADDR_NONE +# define INADDR_NONE ((ulong) -1) +# endif -#define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1) -static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking -static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); } -static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); } -static int unlock(int fd) { FLOCK(F_TLOCK); } +// defines for globals -#else +#define FILE_STATUS_NAME "status" -#error unable to find file locking func +// globals -#endif -#endif -#endif -#endif +const String file_status_name(FILE_STATUS_NAME); -// defines for statics +// defines #define HTTP_METHOD_NAME "method" #define HTTP_TIMEOUT_NAME "timeout" #define HTTP_HEADERS_NAME "headers" #define HTTP_ANY_STATUS_NAME "any-status" -#define FILE_STATUS_NAME "status" - -// statics - -static StringPtr http_method_name(new String(HTTP_METHOD_NAME)); -static StringPtr http_timeout_name(new String(HTTP_TIMEOUT_NAME)); -static StringPtr http_headers_name(new String(HTTP_HEADERS_NAME)); -static StringPtr http_any_status_name(new String(HTTP_ANY_STATUS_NAME)); -static StringPtr file_status_name(new String(FILE_STATUS_NAME)); +#define HTTP_CHARSET_NAME "charset" +#define HTTP_TABLES_NAME "tables" +#define HTTP_USER "user" +#define HTTP_PASSWORD "password" // defines @@ -115,16 +93,13 @@ static StringPtr file_status_name(new St // functions -void fix_line_breaks(char* buf, size_t& size) { - if(size==0) - return; - +void fix_line_breaks(char *str, size_t& length) { //_asm int 3; - const char* const eob=buf+size; - char* dest=buf; + const char* const eob=str+length; + char* dest=str; // fix DOS: \r\n -> \n // fix Macintosh: \r -> \n - char* bol=buf; + char* bol=str; while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { size_t len=eol-bol; if(dest!=bol) @@ -134,33 +109,41 @@ void fix_line_breaks(char* buf, size_t& if(&eol[1]sin_family=AF_INET; addr->sin_port=htons(port); if(host) { - if(struct hostent *hostIP=gethostbyname(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; @@ -169,25 +152,32 @@ static bool set_addr(struct sockaddr_in return true; } -static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){ +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; int result=0; - ssize_t EOLat=0; + char* EOLat=0; while(true) { - char *buf=(char *)pool.malloc(MAX_STRING); - ssize_t size=recv(sock, buf, MAX_STRING, 0); - if(size<=0) + char buf[MAX_STRING*10]; + ssize_t received_size=recv(sock, buf, sizeof(buf), 0); + if(received_size<=0) break; - response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0); - if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response - StringPtr status_line=response.mid(0, (size_t)EOLat); + 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; - status_line->split(astatus, 0, " ", 1); - StringPtr status_code=astatus.get(1); - result=status_code->as_int(); + 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, + &status_code, "invalid HTTP response status"); } } @@ -195,249 +185,399 @@ static int http_read_response(Pool& pool return result; else throw Exception("http.response", - Exception::undefined_source, - "bad response from host - no status found (size=%lu)", response.size()); + 0, + "bad response from host - no status found (size=%u)", response_size); } /* ********************** request *************************** */ #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) -# define WE_CAN_USE_ALARM +# define PA_USE_ALARM #endif -#ifdef WE_CAN_USE_ALARM +#ifdef PA_USE_ALARM static sigjmp_buf timeout_env; static void timeout_handler(int sig){ siglongjmp(timeout_env, 1); } #endif -static int http_request(Pool& pool, - String& response, - StringPtr origin_string, - const char* host, int port, - const char* request, - int timeout, - bool fail_on_status_ne_200) { +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", - origin_string, + 0, "zero hostname"); //never -#ifdef WE_CAN_USE_ALARM - signal(SIGALRM, timeout_handler); +#ifdef PA_USE_ALARM + signal(SIGALRM, timeout_handler); #endif int sock=-1; - try { - int result; -#ifdef WE_CAN_USE_ALARM - if(sigsetjmp(timeout_env, 1)) - throw Exception("http.timeout", - origin_string, - "timeout occured while retrieving document"); - else { - alarm(timeout); +#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)) + if(!set_addr(&dest, host, port)) throw Exception("http.host", - origin_string, + 0, "can not resolve hostname \"%s\"", host); if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) throw Exception("http.connect", - origin_string, + 0, "can not make socket: %s (%d)", strerror(errno), errno); if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) throw Exception("http.connect", - origin_string, + 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", - origin_string, + 0, "error sending request: %s (%d)", strerror(errno), errno); - result=http_read_response(pool, response, sock, fail_on_status_ne_200); + result=http_read_response(response, response_size, sock, fail_on_status_ne_200); closesocket(sock); -#ifdef WE_CAN_USE_ALARM +#ifdef PA_USE_ALARM alarm(0); - } #endif - return result; - } catch(...) { - if(sock>=0) - closesocket(sock); -#ifdef WE_CAN_USE_ALARM - alarm(0); + return result; + } catch(...) { +#ifdef PA_USE_ALARM + alarm(0); #endif - rethrow; + if(sock>=0) + closesocket(sock); + rethrow; + } +#ifdef PA_USE_ALARM } +#endif } #ifndef DOXYGEN struct Http_pass_header_info { - Pool* pool; - Charset* charset; + 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)<<*key<<": " - << *attributed_meaning_to_string(*info->pool, value, String::UL_HTTP_HEADER, false) - <<"\n"; - - if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT") - info->user_agent_specified=true; -} -static HashStringValuePtr file_read_http(Pool& pool, Charset& charset, StringPtr file_spec, - char*& data, size_t& data_size, - HashStringValue *options=0, bool return_outfields=false) { - HashStringValuePtr result(0); +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<append(*file_spec, String::UL_URI); // tainted pieces -> URI pieces - - CharPtr connect_string_cstr=connect_string->cstr(String::UL_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; + 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(ValuePtr vmethod=options->get(http_method_name)) { + if(Value* vmethod=options->get(HTTP_METHOD_NAME)) { valid_options++; - method=vmethod->as_string(&pool)->cstr(); + method=vmethod->as_string().cstr(); } - if(ValuePtr vtimeout=options->get(http_timeout_name)) { + if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) { valid_options++; timeout=vtimeout->as_int(); } - if(vheaders=options->get(http_headers_name)) { + if((vheaders=options->get(HTTP_HEADERS_NAME))) { valid_options++; } - if(ValuePtr vany_status=options->get(http_any_status_name)) { + 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", - Exception::undefined_source, + 0, "invalid option passed"); - } + } + if(!asked_remote_charset) // defaulting to $request:charset + asked_remote_charset=&charsets.source(); - //making request - String request; - if(method) - request<is_string()) { // allow empty - if(HashStringValue *headers=vheaders->get_hash(connect_string)) { - Http_pass_header_info info={&pool, &charset, &request}; - 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 "\n"; - request<<"\n"; + //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 - String response; - CharPtr request_cstr=request.cstr(String::UL_UNSPECIFIED); - int status_code=http_request(pool, response, - connect_string, host, port, request_cstr, + 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 - int pos=response.pos("\r\n\r\n", 4); - if(pos<1){ + 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, + &connect_string, "bad response from host - no headers found"); - } - StringPtr header_block=response.mid(0, pos); - StringPtr body=response.mid(pos+4, response.size()); + + *headers_end_at=0; + const String header_block(response, headers_end_at-response, true); ArrayString aheaders; - HashStringValuePtr headers(new HashStringValue); - header_block->split(aheaders, 0, "\r\n", 2); + 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 - for(int i=1; ipos(": ", 2); - if(pos<1) + size_t aheaders_count=aheaders.count(); + Charset* real_remote_charset=0; // undetected, yet + for(size_t i=1; icstr().get()); - - headers->put( - line->mid(0, pos)->change_case(pool, charset, String::CC_UPPER), - ValuePtr(new VString(line->mid(pos+2, line->size())))); + &connect_string, + "bad response from host - bad header \"%s\"", line.cstr()); + const String::Body HEADER_NAME= + line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER); + const String& header_value=line.mid(pos+2, line.length()); + if(HEADER_NAME=="CONTENT-TYPE") + real_remote_charset=detect_charset(charsets.source(), header_value); + + // tables + { + Value *valready=(Value *)tables.get(HEADER_NAME); + bool existed=valready!=0; + Table *table; + if(existed) { + // second+ appearence + table=valready->get_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 - data=new(pool) char[data_size=body->size()]; - body->store_to(data, String::UL_AS_IS); - if(return_outfields) { - headers->put(file_status_name, ValuePtr(new VInt(status_code))); - return headers; - } else - return HashStringValuePtr(0); + 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)); + return result; } +#endif + #ifndef DOXYGEN struct File_read_action_info { char **data; size_t *data_size; }; #endif -static void file_read_action(Pool& pool, - struct stat& finfo, - int f, - StringPtr file_spec, const char* fname, bool as_text, - void *context) { +static void file_read_action( + 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=pool.malloc(to_read_size+(as_text?1:0)); + *info.data=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 %lu bytes count not in [0..%lu] valid range", + &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=pool.malloc(1); + *info.data=new(PointerFreeGC) char[1]; *(char*)(*info.data)=0; } else *info.data=0; @@ -445,45 +585,70 @@ static void file_read_action(Pool& pool, return; } } -File_read_result file_read(Pool& pool, Charset& charset, StringPtr file_spec, - char*& data, size_t& data_size, - bool as_text, HashStringValue *params, HashStringValue** out_fields, +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; - if(file_spec->starts_with("http://", 7)) { + File_read_result result={false, 0, 0, 0}; +#ifdef PA_HTTP + if(file_spec.starts_with("http://")) { // fail on read problem - result.fields=file_read_http(pool, charset, file_spec, data, data_size, 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 { - File_read_action_info info={&data, &data_size}; - result.success=file_read_action_under_lock(pool, file_spec, +#endif + if(params && 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 } +#endif if(result.success && as_text) { // UTF-8 signature: EF BB BF - if(data_size>=3) { - char *in=(char *)data; - if((in[0] == '\xEF') && (in[1] == '\xBB') && - (in[2] == '\xBF')) { - data=in+3; data_size-=3;// skip prefix + 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 } } - fix_line_breaks((char *)(data), data_size); - // note: after fixing - ((char*&)(data))[data_size]=0; + fix_line_breaks((char *)(result.str), result.length); } return result; } -bool file_read_action_under_lock(Pool& pool, StringPtr file_spec, +#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", + &file_spec, + "parser is in safe mode: " + "reading files of foreign group and user disabled " + "[recompile parser with --disable-safe-mode configure option], " + "actual filename '%s', " + "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)", + fname, + finfo.st_uid, geteuid(), + finfo.st_gid, getegid()); +} +#endif + +bool file_read_action_under_lock(const String& file_spec, const char* action_name, File_read_action action, void *context, bool as_text, bool fail_on_read_problem) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); + const char* fname=file_spec.cstr(String::L_FILE_SPEC); int f; // first open, next stat: @@ -497,65 +662,60 @@ bool file_read_action_under_lock(Pool& p // later: it seems, even this does not help sometimes if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { try { - if(lock_shared_blocking(f)!=0) + if(pa_lock_shared_blocking(f)!=0) throw Exception("file.lock", - file_spec, + &file_spec, "shared lock failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname.get()); + strerror(errno), errno, fname); struct stat finfo; if(stat(fname, &finfo)!=0) throw Exception("file.missing", // hardly possible: we just opened it OK - file_spec, + &file_spec, "stat failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname.get()); + strerror(errno), errno, fname); #ifdef PA_SAFE_MODE - if(finfo.st_uid/*foreign?*/!=geteuid() - && finfo.st_gid/*foreign?*/!=getegid()) - throw Exception("parser.runtime", - file_spec, - "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'", - fname.get()); + check_safe_mode(finfo, file_spec, fname); #endif - action(pool, finfo, f, file_spec, fname, as_text, context); + action(finfo, f, file_spec, fname, as_text, context); } catch(...) { - unlock(f);close(f); + pa_unlock(f);close(f); if(fail_on_read_problem) rethrow; return false; } - unlock(f);close(f); + pa_unlock(f);close(f); return true; } else { if(fail_on_read_problem) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, - file_spec, + &file_spec, "%s failed: %s (%d), actual filename '%s'", - action_name, strerror(errno), errno, fname.get()); + action_name, strerror(errno), errno, fname); return false; } } -static void create_dir_for_file(StringPtr file_spec) { +static void create_dir_for_file(const String& file_spec) { size_t pos_after=1; - int pos_before; - while((pos_before=file_spec->pos("/", 1, pos_after))>=0) { - mkdir(file_spec->mid(0, pos_before)->cstr(String::UL_FILE_SPEC), 0775); + 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); pos_after=pos_before+1; } } bool file_write_action_under_lock( - StringPtr file_spec, + const String& file_spec, const char* action_name, File_write_action action, void *context, bool as_text, bool do_append, bool do_block, bool fail_on_lock_problem) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); + const char* fname=file_spec.cstr(String::L_FILE_SPEC); int f; if(access(fname, W_OK)!=0) // no create_dir_for_file(file_spec); @@ -564,11 +724,11 @@ bool file_write_action_under_lock( O_CREAT|O_RDWR |(as_text?_O_TEXT:_O_BINARY) |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { - if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) { + if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) { Exception e("file.lock", - file_spec, + &file_spec, "shared lock failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname.get()); + strerror(errno), errno, fname); close(f); if(fail_on_lock_problem) throw e; @@ -582,7 +742,7 @@ bool file_write_action_under_lock( if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower #endif - unlock(f);close(f); + pa_unlock(f);close(f); rethrow; } @@ -590,34 +750,34 @@ bool file_write_action_under_lock( if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov for discovering] #endif - unlock(f);close(f); + pa_unlock(f);close(f); return true; } else throw Exception(errno==EACCES?"file.access":0, - file_spec, + &file_spec, "%s failed: %s (%d), actual filename '%s'", - action_name, strerror(errno), errno, fname.get()); + action_name, strerror(errno), errno, fname); // here should be nothing, see rethrow above } #ifndef DOXYGEN struct File_write_action_info { - const void *data; size_t size; + const char* str; size_t length; }; #endif static void file_write_action(int f, void *context) { File_write_action_info& info=*static_cast(context); - if(info.size) { - int written=write(f, info.data, info.size); + if(info.length) { + int written=write(f, info.str, info.length); if(written<0) throw Exception(0, - Exception::undefined_source, + 0, "write failed: %s (%d)", strerror(errno), errno); } } void file_write( - StringPtr file_spec, - const void *data, size_t size, + const String& file_spec, + const char* data, size_t size, bool as_text, bool do_append) { File_write_action_info info={data, size}; @@ -629,38 +789,38 @@ void file_write( } // throws nothing! [this is required in file_move & file_delete] -static void rmdir(StringPtr file_spec, size_t pos_after) { - int pos_before; - if((pos_before=file_spec->pos("/", 1, pos_after))>=0) +static void rmdir(const String& file_spec, size_t pos_after) { + size_t pos_before; + 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::UL_FILE_SPEC)); + rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC)); } -bool file_delete(StringPtr file_spec, bool fail_on_read_problem) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); +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, + &file_spec, "unlink failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname.get()); + strerror(errno), errno, fname); else return false; rmdir(file_spec, 1); return true; } -void file_move(StringPtr old_spec, StringPtr new_spec) { - CharPtr old_spec_cstr=old_spec->cstr(String::UL_FILE_SPEC); - CharPtr new_spec_cstr=new_spec->cstr(String::UL_FILE_SPEC); +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); create_dir_for_file(new_spec); if(rename(old_spec_cstr, new_spec_cstr)!=0) throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, - old_spec, + &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", - strerror(errno), errno, old_spec_cstr.get(), new_spec_cstr.get()); + strerror(errno), errno, old_spec_cstr, new_spec_cstr); rmdir(old_spec, 1); } @@ -674,13 +834,13 @@ bool entry_exists(const char* fname, str return result; } -bool entry_exists(StringPtr file_spec) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); +bool entry_exists(const String& file_spec) { + const char* fname=file_spec.cstr(String::L_FILE_SPEC); return entry_exists(fname, 0); } -static bool entry_readable(StringPtr file_spec, bool need_dir) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); +static bool entry_readable(const String& file_spec, bool need_dir) { + char* fname=file_spec.cstrm(String::L_FILE_SPEC); if(need_dir) { size_t size=strlen(fname); while(size) { @@ -698,36 +858,36 @@ static bool entry_readable(StringPtr fil } return false; } -bool file_readable(StringPtr file_spec) { +bool file_readable(const String& file_spec) { return entry_readable(file_spec, false); } -bool dir_readable(StringPtr file_spec) { +bool dir_readable(const String& file_spec) { return entry_readable(file_spec, true); } -StringPtr file_readable(StringPtr path, const String& name) { - StringPtr result(new String(*path)); - *result << "/"; - *result << name; - return file_readable(result)?result:StringPtr(0); +const String* file_readable(const String& path, const String& name) { + String& result=*new String(path); + result << "/"; + result << name; + return file_readable(result)?&result:0; } -bool file_executable(StringPtr file_spec) { - return access(file_spec->cstr(String::UL_FILE_SPEC), X_OK)==0; +bool file_executable(const String& file_spec) { + return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0; } -bool file_stat(StringPtr file_spec, +bool file_stat(const String& file_spec, size_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { - CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); - struct stat finfo; + const char* fname=file_spec.cstr(String::L_FILE_SPEC); + struct stat finfo; if(stat(fname, &finfo)!=0) if(fail_on_read_problem) throw Exception("file.missing", - file_spec, + &file_spec, "getting file size failed: %s (%d), real filename '%s'", - strerror(errno), errno, fname.get()); + strerror(errno), errno, fname); else return false; rsize=finfo.st_size; @@ -779,7 +939,7 @@ char* rsplit(char* string, char delim) { } /// @todo less stupid type detection -const char* format(Pool& pool, double value, char* fmt) { +const char* format(double value, char* fmt) { char local_buf[MAX_NUMBER]; size_t size; @@ -794,13 +954,13 @@ const char* format(Pool& pool, double va else size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); - return pool.copy(local_buf, size+1); + return pa_strdup(local_buf, size); } size_t stdout_write(const void *buf, size_t size) { #ifdef WIN32 do{ - int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout); + int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); if(chunk_written<=0) break; size-=chunk_written; @@ -813,8 +973,8 @@ size_t stdout_write(const void *buf, siz #endif } -char* unescape_chars(Pool& pool, const char* cp, int len) { - char* s=pool.malloc(len + 1); +char* unescape_chars(const char* cp, int len) { + char* s=new(PointerFreeGC) char[len + 1]; enum EscapeState { EscapeRest, EscapeFirst, @@ -930,6 +1090,11 @@ 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: @@ -955,7 +1120,7 @@ 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; @@ -970,18 +1135,175 @@ int __snprintf(char* b, size_t s, const return r; } -int pa_sleep(unsigned long secs, unsigned long usecs) { - for (; usecs >= 1000000; ++secs, usecs -= 1000000); +/* 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) +{ + const register 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); +} -#ifdef WIN32 - Sleep(secs * 1000 + usecs / 1000); - return 0; -#else - struct timeval t; - t.tv_sec = secs; - t.tv_usec = usecs; - return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0); -#endif +/** + * 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); } +char* pa_base64(const char *in, size_t len) +{ + /* wont go to more than 2x size (overly conservative) */ + char* result=new(PointerFreeGC) char[len * 2 + 6]; + 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); + return result; +}