|
|
| version 1.154, 2003/07/24 11:31:23 | version 1.337, 2025/07/02 23:42:10 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: commonly functions. | Parser: commonly functions. |
| Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2000-2024 Art. Lebedev Studio (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru> |
| */ | */ |
| static const char* IDENT_COMMON_C="$Date$"; | |
| #include "pa_common.h" | #include "pa_common.h" |
| #include "pa_exception.h" | #include "pa_exception.h" |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_globals.h" | #include "pa_globals.h" |
| #include "pa_request_charsets.h" | |
| #include "pa_charsets.h" | #include "pa_charsets.h" |
| #include "pa_http.h" | |
| #include "pa_request_charsets.h" | |
| #include "pa_request.h" | |
| #define PA_HTTP | #include "pa_idna.h" |
| #include "pa_convert_utf.h" | |
| #ifdef PA_HTTP | #ifdef _MSC_VER |
| #include "pa_vstring.h" | |
| #include "pa_vint.h" | |
| #ifdef CYGWIN | |
| #define _GNU_H_WINDOWS32_SOCKETS | |
| // for PASCAL | |
| #include <windows.h> | #include <windows.h> |
| // SOCKET | #include <direct.h> |
| typedef u_int SOCKET; | |
| int PASCAL closesocket(SOCKET); | |
| #else | |
| # if defined(WIN32) | |
| # include <windows.h> | |
| # else | |
| # define closesocket close | |
| # endif | |
| #endif | #endif |
| #ifdef _MSC_VER | |
| #define pa_mkdir(path, mode) _mkdir(path) | |
| #else | #else |
| #define pa_mkdir(path, mode) mkdir(path, mode) | |
| # if defined(WIN32) | |
| # include <windows.h> | |
| # endif | |
| #endif | #endif |
| volatile const char * IDENT_PA_COMMON_C="$Id$" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; | |
| // some maybe-undefined constants | // some maybe-undefined constants |
| #ifndef _O_TEXT | #ifndef _O_TEXT |
| Line 62 int PASCAL closesocket(SOCKET); | Line 49 int PASCAL closesocket(SOCKET); |
| # endif | # endif |
| #endif | #endif |
| // locking constants | |
| //#define PA_DEBUG_NO_LOCKING | |
| #ifdef PA_DEBUG_NO_LOCKING | |
| #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 | |
| #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); } | |
| #else | |
| #error unable to find file locking func | |
| #endif | |
| #endif | |
| #endif | |
| #endif | |
| #else | |
| static int lock_shared_blocking(int fd) { return 0; } | |
| static int lock_exclusive_blocking(int fd) { return 0; } | |
| static int lock_exclusive_nonblocking(int fd) { return 0; } | |
| static int unlock(int fd) { return 0; } | |
| #endif | |
| // defines for globals | // defines for globals |
| #define FILE_STATUS_NAME "status" | #define FILE_STATUS_NAME "status" |
| Line 126 static int unlock(int fd) { return 0; } | Line 57 static int unlock(int fd) { return 0; } |
| const String file_status_name(FILE_STATUS_NAME); | const String file_status_name(FILE_STATUS_NAME); |
| // defines for statics | String sql_bind_name(SQL_BIND_NAME); |
| String sql_limit_name(PA_SQL_LIMIT_NAME); | |
| #define HTTP_METHOD_NAME "method" | String sql_offset_name(PA_SQL_OFFSET_NAME); |
| #define HTTP_TIMEOUT_NAME "timeout" | String sql_default_name(SQL_DEFAULT_NAME); |
| #define HTTP_HEADERS_NAME "headers" | String sql_distinct_name(SQL_DISTINCT_NAME); |
| #define HTTP_ANY_STATUS_NAME "any-status" | String sql_value_type_name(SQL_VALUE_TYPE_NAME); |
| #define HTTP_CHARSET_NAME "charset" | |
| // statics | // forwards |
| static const String http_method_name(HTTP_METHOD_NAME); | const UTF16* pa_utf16_encode(const char* in, Charset& source_charset); |
| 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); | |
| // defines | |
| #define DEFAULT_USER_AGENT "parser3" | |
| // functions | // functions |
| void fix_line_breaks(char *str, size_t& length) { | #ifdef _MSC_VER |
| //_asm int 3; | |
| const char* const eob=str+length; | |
| char* dest=str; | |
| // fix DOS: \r\n -> \n | |
| // fix Macintosh: \r -> \n | |
| char* bol=str; | |
| while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { | |
| size_t len=eol-bol; | |
| if(dest!=bol) | |
| memcpy(dest, bol, len); | |
| dest+=len; | |
| *dest++='\n'; | |
| if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS | int pa_stat(const char *pathname, struct stat *buffer){ |
| bol=eol+2; | const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); |
| length--; | return _wstat64((const wchar_t *)utf16name, buffer); |
| } else // \r, not \n = Macintosh | |
| bol=eol+1; | |
| } | |
| // last piece without \r | |
| if(dest!=bol) | |
| memcpy(dest, bol, eob-bol); | |
| str[length]=0; // terminating | |
| } | } |
| char* file_read_text(Request_charsets& charsets, | int pa_open(const char *pathname, int flags, int mode){ |
| const String& file_spec, | const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); |
| bool fail_on_read_problem, | return _wopen((const wchar_t *)utf16name, flags, mode); |
| HashStringValue* params/*, HashStringValue* * out_fields*/) { | |
| File_read_result file= | |
| file_read(charsets, file_spec, true, params, fail_on_read_problem); | |
| return file.success?file.str:0; | |
| } | |
| #ifdef PA_HTTP | |
| //http request stuff | |
| #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) { | |
| if(struct hostent *hostIP=gethostbyname(host)) | |
| 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){ | FILE *pa_fopen(const char *pathname, const char *mode){ |
| int result=0; | const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); |
| size_t EOLat=0; | const UTF16* utf16mode=pa_utf16_encode(mode, pa_thread_request().charsets.source()); |
| while(true) { | return _wfopen((const wchar_t *)utf16name, (const wchar_t *)utf16mode); |
| 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"); | |
| } | |
| } | |
| if(result) | |
| return result; | |
| else | |
| throw Exception("http.response", | |
| 0, | |
| "bad response from host - no status found (size=%lu)", response.length()); | |
| } | } |
| /* ********************** request *************************** */ | int pa_unlink(const char *pathname){ |
| const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); | |
| #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) | return _wunlink((const wchar_t *)utf16name); |
| # 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(String& response, | #else |
| const char* host, int port, | |
| const char* request, | #define pa_unlink unlink |
| int timeout, | |
| 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, 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 | #endif |
| /// these options were handled but not checked elsewhere, now check them | |
| int pa_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++; | |
| if(options.get(PA_CHARSET_NAME)) | |
| result++; | |
| return result; | |
| } | } |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct Http_pass_header_info { | struct File_read_action_info { |
| Request_charsets* charsets; | char **data; size_t *data_size; |
| String* request; | char* buf; uint64_t offset; size_t limit; |
| bool user_agent_specified; | }; |
| }; | |
| #endif | #endif |
| static void http_pass_header(HashStringValue::key_type key, | |
| HashStringValue::value_type value, | static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) { |
| Http_pass_header_info *info) { | File_read_action_info& info = *static_cast<File_read_action_info *>(context); |
| *info->request <<key<<": " | size_t to_read_size = check_file_size(info.limit && info.limit < (size_t)finfo.st_size ? info.limit : (size_t)finfo.st_size, &file_spec); |
| << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false) | if(to_read_size) { |
| << CRLF; | if(info.offset) |
| pa_lseek(f, info.offset, SEEK_SET); // seek never fails as POSIX allows the file offset to be set beyond the EOF | |
| if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT") | *info.data = info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1); |
| info->user_agent_specified=true; | ssize_t result = read(f, *info.data, to_read_size); |
| if(result<0) | |
| throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno); | |
| *info.data_size = result; | |
| } else { // empty file | |
| // 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 = (char *)pa_malloc_atomic(1); | |
| *(char*)(*info.data) = 0; | |
| *info.data_size = 0; | |
| return; | |
| } | |
| } | } |
| File_read_result file_read_binary(const String& file_spec, bool fail_on_read_problem, char* buf, uint64_t offset, size_t limit) { | |
| File_read_result result = {false, 0, 0, 0}; | |
| File_read_action_info info = {&result.str, &result.length, buf, offset, limit}; | |
| static Charset* detect_charset(Charset& source_charset, const String& content_type_value) { | result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, 0, fail_on_read_problem); |
| const StringBody CONTENT_TYPE_VALUE= | return result; |
| 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 StringBody CHARSET_NAME_BODY= | |
| CONTENT_TYPE_VALUE.mid(charset_begin, charset_end); | |
| return &charsets.get(CHARSET_NAME_BODY); | File_read_result file_read(Request_charsets& charsets, const String& file_spec, |
| bool as_text, HashStringValue *options, | |
| bool fail_on_read_problem, | |
| size_t offset = 0, size_t limit = 0, bool transcode_text_result = true) { | |
| File_read_result result = {false, 0, 0, 0}; | |
| if(options){ | |
| int valid_options = pa_get_valid_file_options_count(*options); | |
| if(valid_options != options->count()) | |
| throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); | |
| } | } |
| return 0; | File_read_action_info info = {&result.str, &result.length, 0, offset, limit}; |
| } | |
| result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); | |
| #ifndef DOXYGEN | if(as_text){ |
| struct File_read_http_result { | if(result.success){ |
| char *str; size_t length; | Charset* asked_charset = 0; |
| HashStringValue* headers; | if(options) |
| }; | if(Value* vcharset_name = options->get(PA_CHARSET_NAME)) |
| #endif | asked_charset = &pa_charsets.get(vcharset_name->as_string()); |
| static File_read_http_result file_read_http(Request_charsets& charsets, | |
| const String& file_spec, | |
| HashStringValue *options=0) { | |
| File_read_http_result result; | |
| char host[MAX_STRING]; | |
| const char* uri; | |
| int port; | |
| const char* method=0; | |
| int timeout=2; | |
| bool fail_on_status_ne_200=true; | |
| Value* vheaders=0; | |
| Charset *asked_remote_charset=0; | |
| if(options) { | |
| int valid_options=0; | |
| if(Value* vmethod=options->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(valid_options!=options->count()) | asked_charset = pa_charsets.checkBOM(result.str, result.length, asked_charset); |
| throw Exception("parser.runtime", | |
| 0, | if(result.length && transcode_text_result && asked_charset){ // length must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below |
| "invalid option passed"); | String::C body = String::C(result.str, result.length); |
| } | body=Charset::transcode(body, *asked_charset, charsets.source()); |
| if(!asked_remote_charset) // defaulting to $request:charset | |
| asked_remote_charset=&charsets.source(); | result.str = const_cast<char*>(body.str); // hacking a little |
| result.length = body.length; | |
| //preparing request | } |
| String& connect_string=*new String; | } |
| // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one | if(result.length) |
| connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces | fix_line_breaks(result.str, result.length); |
| 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<<method; | |
| else | |
| request<<"GET"; | |
| request<< " "<< uri <<" HTTP/1.0" CRLF | |
| "host: "<< host << CRLF; | |
| bool user_agent_specified=false; | |
| if(vheaders && !vheaders->is_string()) { // allow empty | |
| if(HashStringValue *headers=vheaders->get_hash()) { | |
| Http_pass_header_info info={&charsets, &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 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; | |
| int status_code=http_request(response, | |
| host, port, request_cstr, | |
| timeout, 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()); | |
| ArrayString aheaders; | |
| result.headers=new HashStringValue; | |
| 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; i<aheaders_count; i++) { | |
| const String& line=*aheaders.get(i); | |
| pos=line.pos(": ", 2); | |
| if(pos==STRING_NOT_FOUND || pos<1) | |
| throw Exception("http.response", | |
| &connect_string, | |
| "bad response from host - bad header \"%s\"", line.cstr()); | |
| const StringBody 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); | |
| 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=Charset::transcode( | |
| String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()), | |
| *real_remote_charset, | |
| charsets.source()); | |
| result.str=const_cast<char *>(real_body.str); // hacking a little | |
| result.length=real_body.length; | |
| result.headers->put(file_status_name, new VInt(status_code)); | |
| return result; | return result; |
| } | } |
| #endif | File_read_result file_load(Request& r, const String& file_spec, |
| bool as_text, HashStringValue *options, | |
| #ifndef DOXYGEN | bool fail_on_read_problem, |
| struct File_read_action_info { | bool transcode_text_result) { |
| char **data; size_t *data_size; | |
| }; | size_t offset = 0; |
| #endif | size_t limit = 0; |
| static void file_read_action( | |
| struct stat& finfo, | if(options){ |
| int f, | if(Value *voffset = (Value *)options->get(sql_offset_name)) |
| const String& file_spec, const char* fname, bool as_text, | offset = r.process(*voffset).as_int(); |
| void *context) { | if(Value *vlimit = (Value *)options->get(sql_limit_name)) |
| File_read_action_info& info=*static_cast<File_read_action_info *>(context); | limit = r.process(*vlimit).as_int(); |
| if(size_t to_read_size=(size_t)finfo.st_size) { | // no check on options count here |
| *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", | |
| *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; | |
| *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}; | |
| #ifdef PA_HTTP | |
| if(file_spec.starts_with("http://")) { | if(file_spec.starts_with("http://")) { |
| if(offset || limit) | |
| throw Exception(PARSER_RUNTIME, 0, "offset and load options are not supported for HTTP:// file load"); | |
| // fail on read problem | // fail on read problem |
| File_read_http_result http=file_read_http(charsets, file_spec, params); | File_read_http_result http = pa_internal_file_read_http(r, file_spec, as_text, options, transcode_text_result); |
| result.success=true; | |
| result.str=http.str; | |
| result.length=http.length; | |
| result.headers=http.headers; | |
| } else { | |
| #endif | |
| 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) { | File_read_result result = {true, http.str, http.length, http.headers}; |
| // UTF-8 signature: EF BB BF | return result; |
| if(result.length>=3) { | } else |
| char *in=(char *)result.str; | return file_read(r.charsets, file_spec, as_text, options, fail_on_read_problem, offset, limit, transcode_text_result); |
| if(strncmp(in, "\xEB\xBB\xBF", 3)==0) { | } |
| result.str=in+3; result.length-=3;// skip prefix | |
| } | |
| } | |
| fix_line_breaks((char *)(result.str), result.length); | char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem) { |
| // note: after fixing | File_read_result file = file_read(charsets, file_spec, true, 0, fail_on_read_problem); |
| ((char*&)(result.str))[result.length]=0; | return file.success ? file.str : 0; |
| } | } |
| return result; | char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* options, bool transcode_result) { |
| File_read_result file = file_load(r, file_spec, true, options, fail_on_read_problem, transcode_result); | |
| return file.success ? file.str : 0; | |
| } | } |
| #ifdef PA_SAFE_MODE | #ifdef PA_SAFE_MODE |
| void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { | void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) { |
| if(finfo.st_uid/*foreign?*/!=geteuid() | if(finfo.st_uid/*foreign?*/!=geteuid() |
| && finfo.st_gid/*foreign?*/!=getegid()) | && finfo.st_gid/*foreign?*/!=getegid()) |
| throw Exception("parser.runtime", | throw Exception(PARSER_RUNTIME, |
| &file_spec, | &file_spec, |
| "parser is in safe mode: " | "parser is in safe mode: reading files of foreign group and user disabled " |
| "reading files of foreign group and user disabled " | "[recompile parser with --disable-safe-mode configure option], " |
| "[recompile parser with --disable-safe-mode configure option], " | "actual filename '%s', fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)", |
| "actual filename '%s', " | fname, finfo.st_uid, geteuid(), finfo.st_gid, getegid() |
| "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)", | ); |
| fname, | } |
| finfo.st_uid, geteuid(), | #else |
| finfo.st_gid, getegid()); | void check_safe_mode(struct stat, const String&, const char*) { |
| } | } |
| #endif | #endif |
| bool file_read_action_under_lock(const String& file_spec, | |
| const char* action_name, File_read_action action, void *context, | bool file_read_action_under_lock(const String& file_spec, const char* action_name, File_read_action action, void *context, |
| bool as_text, | bool as_text, bool fail_on_read_problem) { |
| 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; | int f; |
| // first open, next stat: | // first open, next stat: |
| Line 636 bool file_read_action_under_lock(const S | Line 261 bool file_read_action_under_lock(const S |
| // they delay update till open, so we would receive "!^test[" string | // they delay update till open, so we would receive "!^test[" string |
| // if would do stat, next open. | // if would do stat, next open. |
| // later: it seems, even this does not help sometimes | // later: it seems, even this does not help sometimes |
| if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { | if((f=pa_open(fname, O_RDONLY | (as_text ? _O_TEXT : _O_BINARY) ))>=0) { |
| try { | try { |
| if(lock_shared_blocking(f)!=0) | int pa_errno=pa_lock_shared_blocking(f); |
| throw Exception("file.lock", | if(pa_errno!=0) |
| &file_spec, | throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(pa_errno), pa_errno, fname); |
| "shared lock failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| struct stat finfo; | struct stat finfo; |
| if(stat(fname, &finfo)!=0) | if(pa_fstat(f, &finfo)!=0) |
| throw Exception("file.missing", // hardly possible: we just opened it OK | 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); |
| "stat failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| #ifdef PA_SAFE_MODE | |
| check_safe_mode(finfo, file_spec, fname); | check_safe_mode(finfo, file_spec, fname); |
| #endif | |
| action(finfo, f, file_spec, fname, as_text, context); | action(finfo, f, file_spec, context); |
| } catch(...) { | } catch(...) { |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| rethrow; | rethrow; |
| return false; | return false; |
| } | } |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| return true; | return true; |
| } else { | } else { |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0, |
| &file_spec, | &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); |
| "%s failed: %s (%d), actual filename '%s'", | |
| action_name, strerror(errno), errno, fname); | |
| return false; | return false; |
| } | } |
| } | } |
| static void create_dir_for_file(const String& file_spec) { | void create_dir_for_file(const String& file_spec) { |
| size_t pos_after=1; | const char *str=file_spec.taint_cstr(String::L_FILE_SPEC); |
| size_t pos_before; | if(str[0]){ |
| while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { | const char *pos=str+1; |
| mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775); | while((pos=strchr(pos,'/')) && pos[1]) { // to avoid trailing /, see #1166 |
| pos_after=pos_before+1; | pa_mkdir(pa_strdup(str,pos-str), 0775); |
| pos++; | |
| } | |
| } | } |
| } | } |
| bool file_write_action_under_lock( | bool file_write_action_under_lock(const String& file_spec, const char* action_name, File_write_action action, void *context, |
| const String& file_spec, | bool as_text, bool do_append, bool do_block, bool fail_on_lock_problem) { |
| const char* action_name, File_write_action action, void *context, | |
| bool as_text, | const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); |
| bool do_append, | |
| bool do_block, | |
| bool fail_on_lock_problem) { | |
| const char* fname=file_spec.cstr(String::L_FILE_SPEC); | |
| int f; | int f; |
| if(access(fname, W_OK)!=0) // no | if(access(fname, W_OK)!=0) // no |
| create_dir_for_file(file_spec); | create_dir_for_file(file_spec); |
| if((f=open(fname, | if((f=pa_open(fname, O_CREAT | O_RDWR | (as_text ? _O_TEXT : _O_BINARY) | (do_append ? O_APPEND : PA_O_TRUNC), 0664))>=0) { |
| O_CREAT|O_RDWR | int pa_errno=do_block ? pa_lock_exclusive_blocking(f) : pa_lock_exclusive_nonblocking(f); |
| |(as_text?_O_TEXT:_O_BINARY) | if(pa_errno!=0) { |
| |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { | Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(pa_errno), pa_errno, fname); |
| if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) { | close(f); |
| Exception e("file.lock", | |
| &file_spec, | |
| "shared lock failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| close(f); | |
| if(fail_on_lock_problem) | if(fail_on_lock_problem) |
| throw e; | throw e; |
| return false; | return false; |
| } | } |
| try { | try { |
| action(f, context); | #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE)) |
| struct stat finfo; | |
| if(pa_fstat(f, &finfo)==0 && finfo.st_mode & 0111) | |
| fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any | |
| #endif | |
| action(f, context); | |
| } catch(...) { | } catch(...) { |
| #ifdef HAVE_FTRUNCATE | #ifdef HAVE_FTRUNCATE |
| if(!do_append) | if(!do_append) |
| ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower | PA_UNUSED int ignore_result=ftruncate(f, lseek(f, 0, SEEK_CUR)); // one cannot use O_TRUNC, read lower |
| #endif | #endif |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| rethrow; | rethrow; |
| } | } |
| #ifdef HAVE_FTRUNCATE | #ifdef HAVE_FTRUNCATE |
| if(!do_append) | if(!do_append) |
| ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering] | PA_UNUSED int ignore_result=ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering] |
| #endif | #endif |
| unlock(f);close(f); | pa_unlock(f);close(f); |
| return true; | return true; |
| } else | } else |
| throw Exception(errno==EACCES?"file.access":0, | throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); |
| &file_spec, | |
| "%s failed: %s (%d), actual filename '%s'", | |
| action_name, strerror(errno), errno, fname); | |
| // here should be nothing, see rethrow above | // here should be nothing, see rethrow above |
| } | } |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct File_write_action_info { | struct File_write_action_info { |
| const char* str; size_t length; | const char* str; |
| size_t length; | |
| }; | }; |
| #endif | #endif |
| static void file_write_action(int f, void *context) { | static void file_write_action(int f, void *context) { |
| File_write_action_info& info=*static_cast<File_write_action_info *>(context); | File_write_action_info& info=*static_cast<File_write_action_info *>(context); |
| if(info.length) { | if(info.length) { |
| int written=write(f, info.str, info.length); | ssize_t written=write(f, info.str, info.length); |
| if(written<0) | if(written<0) |
| throw Exception(0, | throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno); |
| 0, | if((size_t)written!=info.length) |
| "write failed: %s (%d)", strerror(errno), errno); | throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length); |
| } | } |
| } | } |
| void file_write( | void file_write( |
| const String& file_spec, | Request_charsets& charsets, |
| const char* data, size_t size, | const String& file_spec, |
| const char* data, | |
| size_t size, | |
| bool as_text, | 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_info info={data, size}; |
| file_write_action_under_lock( | file_write_action_under_lock( |
| file_spec, | file_spec, |
| "write", file_write_action, &info, | "write", |
| file_write_action, | |
| &info, | |
| as_text, | as_text, |
| do_append); | do_append); |
| } | } |
| // throws nothing! [this is required in file_move & file_delete] | static size_t get_dir(char* fname, size_t helper_length){ |
| static void rmdir(const String& file_spec, size_t pos_after) { | bool dir=false; |
| size_t pos_before; | size_t pos=0; |
| if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) | for(pos=helper_length; pos; pos--){ |
| rmdir(file_spec, pos_before+1); | char c=fname[pos-1]; |
| if(c=='/' || c=='\\'){ | |
| rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC)); | fname[pos-1]=0; |
| } | dir=true; |
| bool file_delete(const String& file_spec, bool fail_on_read_problem) { | } else if(dir) break; |
| const char* fname=file_spec.cstr(String::L_FILE_SPEC); | } |
| if(unlink(fname)!=0) | return pos; |
| if(fail_on_read_problem) | |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | |
| &file_spec, | |
| "unlink failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname); | |
| else | |
| return false; | |
| 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); | |
| 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, | |
| "rename failed: %s (%d), actual filename '%s' to '%s'", | |
| strerror(errno), errno, old_spec_cstr, new_spec_cstr); | |
| rmdir(old_spec, 1); | |
| } | } |
| bool entry_exists(const char* fname, struct stat *afinfo) { | bool entry_exists(const char* fname, struct stat *afinfo) { |
| struct stat lfinfo; | struct stat lfinfo; |
| bool result=stat(fname, &lfinfo)==0; | bool result=pa_stat(fname, &lfinfo)==0; |
| if(afinfo) | if(afinfo) |
| *afinfo=lfinfo; | *afinfo=lfinfo; |
| return result; | return result; |
| } | } |
| bool entry_exists(const String& file_spec) { | bool entry_exists(const String& file_spec) { |
| const char* fname=file_spec.cstr(String::L_FILE_SPEC); | return entry_exists(file_spec.taint_cstr(String::L_FILE_SPEC), 0); |
| return entry_exists(fname, 0); | |
| } | } |
| static bool entry_readable(const String& file_spec, bool need_dir) { | static bool entry_ifdir(char *fname, bool need_dir) { |
| char* fname=file_spec.cstrm(String::L_FILE_SPEC); | if(need_dir){ |
| if(need_dir) { | size_t size=strlen(fname); |
| size_t size=strlen(fname); | |
| while(size) { | while(size) { |
| char c=fname[size-1]; | char c=fname[size-1]; |
| if(c=='/' || c=='\\') | if(c=='/' || c=='\\') |
| fname[--size]=0; | fname[--size]=0; |
| else | else |
| break; | break; |
| } | } |
| } | } |
| struct stat finfo; | struct stat finfo; |
| if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) { | if(entry_exists(fname, &finfo)) { |
| bool is_dir=(finfo.st_mode&S_IFDIR) != 0; | bool is_dir=(finfo.st_mode&S_IFDIR) != 0; |
| return is_dir==need_dir; | return is_dir==need_dir; |
| } | } |
| return false; | return false; |
| } | } |
| bool file_readable(const String& file_spec) { | |
| return entry_readable(file_spec, false); | static bool entry_ifdir(const String& file_spec, bool need_dir) { |
| return entry_ifdir(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir); | |
| } | } |
| bool dir_readable(const String& file_spec) { | |
| return entry_readable(file_spec, true); | // throws nothing! [this is required in file_move & file_delete] |
| static void rmdir(const String& file_spec, size_t pos_after) { | |
| char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC); | |
| size_t length=strlen(dir_spec); | |
| while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){ | |
| #ifdef _MSC_VER | |
| if(!entry_ifdir(dir_spec, true)) | |
| break; | |
| DWORD attrs=GetFileAttributes(dir_spec); | |
| if( | |
| (attrs==INVALID_FILE_ATTRIBUTES) | |
| || !(attrs & FILE_ATTRIBUTE_DIRECTORY) | |
| || (attrs & FILE_ATTRIBUTE_REPARSE_POINT) | |
| ) | |
| break; | |
| #endif | |
| if( rmdir(dir_spec) ) | |
| break; | |
| }; | |
| } | |
| bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) { | |
| const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); | |
| if(pa_unlink(fname)!=0) { | |
| if(fail_on_problem) | |
| throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, | |
| &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); | |
| else | |
| return false; | |
| } | |
| if(!keep_empty_dirs) | |
| rmdir(file_spec, 1); | |
| return true; | |
| } | } |
| const String* file_readable(const String& path, const String& name) { | |
| void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) { | |
| 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); | |
| if(rename(old_spec_cstr, new_spec_cstr)!=0) | |
| throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0, | |
| &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr); | |
| if(!keep_empty_dirs) | |
| rmdir(old_spec, 1); | |
| } | |
| bool file_exist(const String& file_spec) { | |
| return entry_ifdir(file_spec, false); | |
| } | |
| bool dir_exists(const String& file_spec) { | |
| return entry_ifdir(file_spec, true); | |
| } | |
| const String* file_exist(const String& path, const String& name) { | |
| String& result=*new String(path); | String& result=*new String(path); |
| result << "/"; | if(path.last_char() != '/') |
| result << "/"; | |
| result << name; | result << name; |
| return file_readable(result)?&result:0; | return file_exist(result)?&result:0; |
| } | } |
| bool file_executable(const String& file_spec) { | 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, | bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { |
| size_t& rsize, | const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); |
| time_t& ratime, | |
| time_t& rmtime, | |
| time_t& rctime, | |
| bool fail_on_read_problem) { | |
| const char* fname=file_spec.cstr(String::L_FILE_SPEC); | |
| struct stat finfo; | struct stat finfo; |
| if(stat(fname, &finfo)!=0) | if(pa_stat(fname, &finfo)!=0) { |
| if(fail_on_read_problem) | if(fail_on_read_problem) |
| throw Exception("file.missing", | throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname); |
| &file_spec, | |
| "getting file size failed: %s (%d), real filename '%s'", | |
| strerror(errno), errno, fname); | |
| else | else |
| return false; | return false; |
| } | |
| rsize=finfo.st_size; | rsize=finfo.st_size; |
| ratime=finfo.st_atime; | ratime=(time_t)finfo.st_atime; |
| rmtime=finfo.st_mtime; | rmtime=(time_t)finfo.st_mtime; |
| rctime=finfo.st_ctime; | rctime=(time_t)finfo.st_ctime; |
| return true; | |
| } | |
| size_t check_file_size(uint64_t size, const String* file_spec){ | |
| if(size > (uint64_t)pa_file_size_limit) | |
| throw Exception(PARSER_RUNTIME, file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit); | |
| return (size_t)size; | |
| } | |
| /** | |
| String related functions | |
| */ | |
| bool capitalized(const char* s){ | |
| bool upper=true; | |
| for(const char* c=s; *c; c++){ | |
| if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c))) | |
| return false; | |
| upper=strchr("-_ ", *c) != 0; | |
| } | |
| return true; | return true; |
| } | } |
| const char* capitalize(const char* s){ | |
| if(!s || capitalized(s)) | |
| return 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; | |
| } | |
| char *str_lower(const char *s, size_t length){ | |
| char *result=pa_strdup(s, length); | |
| for(char* c=result; *c; c++) | |
| *c=(char)tolower((unsigned char)*c); | |
| return result; | |
| } | |
| char *str_upper(const char *s, size_t length){ | |
| char *result=pa_strdup(s, length); | |
| for(char* c=result; *c; c++) | |
| *c=(char)toupper((unsigned char)*c); | |
| return result; | |
| } | |
| void fix_line_breaks(char *str, size_t& length) { | |
| //_asm int 3; | |
| const char* const eob=str+length; | |
| char* dest=str; | |
| // fix DOS: \r\n -> \n | |
| // fix Macintosh: \r -> \n | |
| char* bol=str; | |
| while(char* eol=(char*)memchr(bol, '\r', eob -bol)) { | |
| size_t len=eol-bol; | |
| if(dest!=bol) | |
| memmove(dest, bol, len); | |
| dest+=len; | |
| *dest++='\n'; | |
| if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS | |
| bol=eol+2; | |
| length--; | |
| } else // \r, not \n = Macintosh | |
| bol=eol+1; | |
| } | |
| // last piece without \r | |
| if(dest!=bol) | |
| memmove(dest, bol, eob-bol); | |
| str[length]=0; // terminating | |
| } | |
| /** | |
| scans for @a delim[default \n] in @a *row_ref, | |
| @return piece of line before it or end of string, if no @a delim found | |
| assigns @a *row_ref to point right after delimiter if there were one | |
| or to zero if no @a delim were found. | |
| */ | |
| char* getrow(char* *row_ref, char delim) { | char* getrow(char* *row_ref, char delim) { |
| char* result=*row_ref; | char* result=*row_ref; |
| if(result) { | if(result) { |
| *row_ref=strchr(result, delim); | *row_ref=strchr(result, delim); |
| if(*row_ref) | if(*row_ref) |
| *((*row_ref)++)=0; | *((*row_ref)++)=0; |
| else if(!*result) | else if(!*result) |
| return 0; | return 0; |
| } | } |
| return result; | return result; |
| } | } |
| char* lsplit(char* string, char delim) { | char* lsplit(char* string, char delim) { |
| if(string) { | if(string) { |
| char* v=strchr(string, delim); | char* v=strchr(string, delim); |
| if(v) { | if(v) { |
| *v=0; | *v=0; |
| return v+1; | return v+1; |
| } | } |
| } | } |
| return 0; | return 0; |
| } | } |
| char* lsplit(char* *string_ref, char delim) { | char* lsplit(char* *string_ref, char delim) { |
| char* result=*string_ref; | char* result=*string_ref; |
| char* next=lsplit(*string_ref, delim); | char* next=lsplit(*string_ref, delim); |
| *string_ref=next; | *string_ref=next; |
| return result; | return result; |
| } | } |
| char* rsplit(char* string, char delim) { | char* rsplit(char* string, char delim) { |
| if(string) { | if(string) { |
| char* v=strrchr(string, delim); | char* v=strrchr(string, delim); |
| if(v) { | if(v) { |
| *v=0; | *v=0; |
| return v+1; | return v+1; |
| } | } |
| } | } |
| return NULL; | return NULL; |
| } | } |
| /// @todo less stupid type detection | void pa_strncpy(char *dst, const char *src, size_t n){ |
| const char* format(double value, char* fmt) { | size_t left = n; |
| char local_buf[MAX_NUMBER]; | |
| size_t size; | if (left != 0 && src) { |
| while (--left != 0) { | |
| if(fmt) | if ((*dst++ = *src++) == '\0') |
| if(strpbrk(fmt, "diouxX")) | return; |
| if(strpbrk(fmt, "ouxX")) | } |
| size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); | } |
| else | if (n != 0) |
| size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value); | *dst = '\0'; |
| else | } |
| size=snprintf(local_buf, sizeof(local_buf), fmt, value); | |
| else | #define STRCAT_STEP(str, len) if(str) { memcpy(ptr, str, len); ptr += len; } |
| size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); | |
| char *pa_strcat(const char *a, const char *b, const char *c) { | |
| return pa_strdup(local_buf, size); | size_t len_a = a ? strlen(a) : 0; |
| size_t len_b = b ? strlen(b) : 0; | |
| size_t len_c = c ? strlen(c) : 0; | |
| char *result=new(PointerFreeGC) char[len_a + len_b + len_c +1/*0*/]; | |
| char *ptr=result; | |
| STRCAT_STEP(a, len_a); | |
| STRCAT_STEP(b, len_b); | |
| STRCAT_STEP(c, len_c); | |
| *ptr='\0'; | |
| return result; | |
| } | |
| const char *pa_filename(const char *path){ | |
| if(!path) | |
| return NULL; | |
| for(const char *c = path + strlen(path) - 1; c >= path; c--) { | |
| if(*c == '/' || *c == '\\') | |
| return c+1; | |
| } | |
| return path; | |
| } | |
| // 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(const char* fmt){ | |
| enum FormatState { | |
| Percent, | |
| Flags, | |
| Width, | |
| Precision, | |
| Done | |
| } state=Percent; | |
| FormatType result=FormatInvalid; | |
| const 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; | |
| } | |
| const char* format_double(double value, const char* fmt) { | |
| 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 FormatUInt: | |
| if(value >= 0){ // on Apple M1 (uint)<negative value> is 0 | |
| size=snprintf(local_buf, sizeof(local_buf), fmt, clip2uint(value)); | |
| break; | |
| } | |
| case FormatInt: | |
| size=snprintf(local_buf, sizeof(local_buf), fmt, clip2int(value)); | |
| break; | |
| case FormatInvalid: | |
| throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt); | |
| } | |
| } else | |
| return pa_itoa(clip2int(value)); | |
| if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing | |
| throw Exception(PARSER_RUNTIME, 0, "Error occurred 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) { | size_t stdout_write(const void *buf, size_t size) { |
| #ifdef WIN32 | #ifdef WIN32 |
| size_t to_write = size; | |
| do{ | do{ |
| int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); | int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); |
| if(chunk_written<=0) | if(chunk_written<=0) |
| break; | break; |
| size-=chunk_written; | size-=chunk_written; |
| buf=((const char*)buf)+chunk_written; | buf=((const char*)buf)+chunk_written; |
| } while(size>0); | } while(size>0); |
| return size; | fflush(stdout); |
| return to_write-size; | |
| #else | #else |
| return fwrite(buf, 1, size, stdout); | size_t result=fwrite(buf, 1, size, stdout); |
| fflush(stdout); | |
| return result; | |
| #endif | #endif |
| } | } |
| char* unescape_chars(const char* cp, int len) { | enum EscapeState { |
| char* s=new(PointerFreeGC) char[len + 1]; | EscapeRest, |
| enum EscapeState { | EscapeFirst, |
| EscapeRest, | EscapeSecond, |
| EscapeFirst, | EscapeUnicode |
| EscapeSecond | }; |
| } escapeState=EscapeRest; | |
| int escapedValue=0; | // @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 js){ | |
| char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes) | |
| XMLByte* dst=(XMLByte *)s; | |
| EscapeState escapeState=EscapeRest; | |
| uint escapedValue=0; | |
| int srcPos=0; | int srcPos=0; |
| int dstPos=0; | short int jsCnt=0; |
| while(srcPos < len) { | while(srcPos<len){ |
| int ch=cp[srcPos]; | uchar c=(uchar)cp[srcPos]; |
| switch(escapeState) { | if(c=='%' || (c=='\\' && js)){ |
| case EscapeRest: | escapeState=EscapeFirst; |
| if(ch=='%') { | } else { |
| escapeState=EscapeFirst; | switch(escapeState) { |
| } else if(ch=='+') { | case EscapeRest: |
| s[dstPos++]=' '; | if(c=='+' && !js){ |
| } else { | *dst++=' '; |
| s[dstPos++]=ch; | } else { |
| *dst++=c; | |
| } | |
| break; | |
| case EscapeFirst: | |
| if(charset && c=='u'){ | |
| // escaped unicode value: %u0430 | |
| jsCnt=0; | |
| escapedValue=0; | |
| escapeState=EscapeUnicode; | |
| } else { | |
| if(isxdigit(c)){ | |
| escapedValue=hex_value[c] << 4; | |
| escapeState=EscapeSecond; | |
| } else { | |
| *dst++=c; | |
| escapeState=EscapeRest; | |
| } | |
| } | |
| break; | |
| case EscapeSecond: | |
| if(isxdigit(c)){ | |
| escapedValue+=hex_value[c]; | |
| *dst++=(char)escapedValue; | |
| } | |
| escapeState=EscapeRest; | |
| break; | |
| case EscapeUnicode: | |
| if(isxdigit(c)){ | |
| escapedValue=(escapedValue << 4) + hex_value[c]; | |
| if(++jsCnt==4){ | |
| // transcode utf8 char to client charset (we can lost some chars here) | |
| charset->store_Char(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; | return s; |
| } | } |
| char *search_stop(char*& current, char cstop_at) { | |
| // sanity check | |
| if(!current) | |
| return 0; | |
| // skip leading WS | |
| while(*current==' ' || *current=='\t') | |
| current++; | |
| if(!*current) | |
| return current=0; | |
| char *result=current; | |
| if(char *pstop_at=strchr(current, cstop_at)) { | |
| *pstop_at=0; | |
| current=pstop_at+1; | |
| } else | |
| current=0; | |
| return result; | |
| } | |
| #ifdef WIN32 | #ifdef WIN32 |
| void back_slashes_to_slashes(char* s) { | void back_slashes_to_slashes(char* s) { |
| if(s) | if(s) |
| Line 994 void back_slashes_to_slashes(char* s) { | Line 906 void back_slashes_to_slashes(char* s) { |
| if(*s=='\\') | if(*s=='\\') |
| *s='/'; | *s='/'; |
| } | } |
| /* | |
| void slashes_to_back_slashes(char* s) { | |
| if(s) | |
| for(; *s; s++) | |
| if(*s=='/') | |
| *s='\\'; | |
| } | |
| */ | |
| #endif | #endif |
| bool StrEqNc(const char* s1, const char* s2, bool strict) { | size_t strpos(const char *str, const char *substr) { |
| while(true) { | const char *p = strstr(str, substr); |
| if(!(*s1)) { | return (p==0)?STRING_NOT_FOUND:p-str; |
| if(!(*s2)) | } |
| return true; | |
| else | size_t remove_crlf(char* start, char* end) { |
| return !strict; | char* from=start; |
| } else if(!(*s2)) | char* to=start; |
| return !strict; | bool skip=false; |
| if(isalpha(*s1)) { | while(from < end){ |
| if(tolower(*s1) !=tolower(*s2)) | switch(*from){ |
| return false; | case '\n': |
| } else if((*s1) !=(*s2)) | case '\r': |
| return false; | case '\t': |
| s1++; | case ' ': |
| s2++; | if(!skip){ |
| *to=' '; | |
| to++; | |
| skip=true; | |
| } | |
| break; | |
| default: | |
| if(from != to) | |
| *to=*from; | |
| to++; | |
| skip=false; | |
| } | |
| from++; | |
| } | |
| return to-start; | |
| } | |
| const char* hex_digits="0123456789ABCDEF"; | |
| const char* hex_string(unsigned char* bytes, size_t size, bool upcase) { | |
| char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/]; | |
| unsigned char *src=bytes; | |
| unsigned char *end=bytes+size; | |
| char *dest=bytes_hex; | |
| const char *hex=upcase? hex_digits : "0123456789abcdef"; | |
| for(; src<end; src++) { | |
| *dest++=hex[*src/0x10]; | |
| *dest++=hex[*src%0x10]; | |
| } | |
| *dest=0; | |
| return bytes_hex; | |
| } | |
| ssize_t file_block_read(const int f, void* buffer, const size_t size){ | |
| ssize_t nCount = read(f, buffer, size); | |
| if (nCount < 0) | |
| throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno); | |
| return nCount; | |
| } | |
| static unsigned long crc32Table[256]; | |
| static void InitCrc32Table() | |
| { | |
| if(crc32Table[1] == 0){ | |
| // This is the official polynomial used by CRC32 in PKZip. | |
| // Often times the polynomial shown reversed as 0x04C11DB7. | |
| static const unsigned long dwPolynomial = 0xEDB88320; | |
| for(int i = 0; i < 256; i++) | |
| { | |
| unsigned long dwCrc = i; | |
| for(int j = 8; j > 0; j--) | |
| { | |
| if(dwCrc & 1) | |
| dwCrc = (dwCrc >> 1) ^ dwPolynomial; | |
| else | |
| dwCrc >>= 1; | |
| } | |
| crc32Table[i] = dwCrc; | |
| } | |
| } | } |
| } | } |
| static bool isLeap(int year) { | inline void CalcCrc32(const unsigned char byte, unsigned long &crc32) |
| return !( | { |
| (year % 4) || ((year % 400) && !(year % 100)) | crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)]; |
| ); | } |
| } | |
| int getMonthDays(int year, int month) { | unsigned long pa_crc32(const char *in, size_t in_size){ |
| int monthDays[]={ | unsigned long crc32=0xFFFFFFFF; |
| 31, | |
| isLeap(year) ? 29 : 28, | InitCrc32Table(); |
| 31, | for(size_t i = 0; i<in_size; i++) |
| 30, | CalcCrc32(in[i], crc32); |
| 31, | |
| 30, | return ~crc32; |
| 31, | } |
| 31, | |
| 30, | static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) { |
| 31, | unsigned long& crc32=*static_cast<unsigned long *>(context); |
| 30, | if(finfo.st_size) { |
| 31 | InitCrc32Table(); |
| }; | int nCount=0; |
| return monthDays[month]; | do { |
| } | unsigned char buffer[FILE_BUFFER_SIZE]; |
| nCount = file_block_read(f, buffer, sizeof(buffer)); | |
| void remove_crlf(char* start, char* end) { | for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32); |
| for(char* p=start; p<end; p++) | } while(nCount > 0); |
| switch(*p) { | } |
| case '\n': *p='|'; break; | } |
| case '\r': *p=' '; break; | |
| unsigned long pa_crc32(const String& file_spec){ | |
| unsigned long crc32=0xFFFFFFFF; | |
| file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32); | |
| return ~crc32; | |
| } | |
| // 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=str_upper(content_type); | |
| 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 ? &pa_charsets.get_direct(begin) : 0; | |
| } | } |
| } | |
| return 0; | |
| } | |
| const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){ | |
| if(!in) | |
| return 0; | |
| String::C sIn(in,strlen(in)); | |
| UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2); | |
| UTF16* utf16_end=utf16; | |
| if(!source_charset.isUTF8()) | |
| sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset); | |
| int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status); | |
| *utf16_end=0; | |
| return utf16; | |
| } | } |
| const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){ | |
| if(!in) | |
| return 0; | |
| const UTF16* utf16_start=in; | |
| const UTF16* utf16_end; | |
| for(utf16_end=in; *utf16_end; utf16_end++); | |
| char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1); | |
| char *result_end = result; | |
| int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status); | |
| *result_end='\0'; | |
| if(asked_charset.isUTF8()) | |
| return result; | |
| return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr(); | |
| } | |
| static bool is_latin(const char *in){ | |
| for(; *in; in++){ | |
| if ((unsigned char)(*in) > 0x7F) | |
| return false; | |
| } | |
| return true; | |
| } | |
| #define MAX_IDNA_LENGTH 256 | |
| const char *pa_idna_encode(const char *in, Charset& source_charset){ | |
| if(!in || is_latin(in)) | |
| return in; | |
| uint32_t utf32[MAX_IDNA_LENGTH]; | |
| uint32_t *utf32_end=utf32; | |
| String::C sIn(in,strlen(in)); | |
| if(!source_charset.isUTF8()) | |
| sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset); | |
| int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status); | |
| *utf32_end=0; | |
| char *result = (char *)pa_malloc(MAX_IDNA_LENGTH); | |
| status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0); | |
| if(status != IDNA_SUCCESS) | |
| throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status)); | |
| return result; | |
| } | |
| const char *pa_idna_decode(const char *in, Charset &asked_charset){ | |
| if(!in || !(*in)) | |
| return in; | |
| uint32_t utf32[MAX_IDNA_LENGTH]; | |
| const uint32_t *utf32_start=utf32; | |
| uint32_t *utf32_end; | |
| int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0); | |
| if(status != IDNA_SUCCESS) | |
| throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status)); | |
| for(utf32_end=utf32; *utf32_end; utf32_end++); | |
| char *result = (char *)pa_malloc(MAX_IDNA_LENGTH); | |
| char *result_end = result; | |
| status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion); | |
| if(status != conversionOK) | |
| throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status); | |
| *result_end='\0'; | |
| if(!asked_charset.isUTF8()) | |
| result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr(); | |
| return result; | |
| } | |
| /// must be last in this file | /// must be last in this file |
| #undef vsnprintf | #undef vsnprintf |
| int __vsnprintf(char* b, size_t s, const char* f, va_list l) { | int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) { |
| if(!s) | if(!s) |
| return 0; | return 0; |
| int r; | int r; |
| // note: on win32& maybe somewhere else | // note: on win32 & maybe somewhere else |
| // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing | // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing |
| // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010 | |
| --s; | --s; |
| #if _MSC_VER | |
| /* | |
| win32: | |
| mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm | |
| if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned | // clients do not check for negative 's', feature: ignore such prints |
| */ | if((ssize_t)s<0) |
| return 0; | |
| #ifdef _MSC_VER | |
| // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned | |
| r=_vsnprintf(b, s, f, l); | r=_vsnprintf(b, s, f, l); |
| if(r<0) | if(r<0) |
| r=s; | r=s; |
| #else | #else |
| r=vsnprintf(b, s, f, l); | r=vsnprintf(b, s, f, l); |
| /* | /* |
| solaris: | solaris: man vsnprintf |
| man vsnprintf | |
| The snprintf() function returns the number of characters | The snprintf() function returns the number of characters |
| formatted, that is, the number of characters that would have | formatted, that is, the number of characters that would have |
| been written to the buffer if it were large enough. If the | been written to the buffer if it were large enough. If the |
| value of n is 0 on a call to snprintf(), an unspecified | value of n is 0 on a call to snprintf(), an unspecified |
| Line 1091 int __vsnprintf(char* b, size_t s, const | Line 1197 int __vsnprintf(char* b, size_t s, const |
| if(r<0) | if(r<0) |
| r=0; | r=0; |
| else if(r>s) | else if((size_t)r>s) |
| r=s; | r=s; |
| #endif | #endif |
| b[r]=0; | b[r]=0; |
| return r; | return r; |
| } | } |
| int __snprintf(char* b, size_t s, const char* f, ...) { | int pa_snprintf(char* b, size_t s, const char* f, ...) { |
| va_list l; | va_list l; |
| va_start(l, f); | va_start(l, f); |
| int r=__vsnprintf(b, s, f, l); | int r=pa_vsnprintf(b, s, f, l); |
| va_end(l); | va_end(l); |
| return r; | return r; |
| } | } |
| int pa_sleep(unsigned long secs, unsigned long usecs) { | |
| for (; usecs >= 1000000; ++secs, usecs -= 1000000); | |
| #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 | |
| } | |