|
|
| version 1.143.2.19, 2003/02/14 17:38:29 | version 1.332, 2024/09/28 14:37:53 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: commonly functions. | Parser: commonly functions. |
| Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2000-2023 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_vstring.h" | |
| #include "pa_vint.h" | |
| #include "pa_globals.h" | #include "pa_globals.h" |
| #include "pa_charsets.h" | |
| #include "pa_http.h" | |
| #include "pa_request_charsets.h" | |
| #include "pa_request.h" | |
| #include "pa_idna.h" | |
| #include "pa_convert_utf.h" | |
| #ifdef CYGWIN | #ifdef _MSC_VER |
| #define _GNU_H_WINDOWS32_SOCKETS | |
| // for PASCAL | |
| #include <windows.h> | #include <windows.h> |
| // SOCKET | #include <direct.h> |
| typedef u_int SOCKET; | #endif |
| int PASCAL closesocket(SOCKET); | |
| #ifdef _MSC_VER | |
| #define pa_mkdir(path, mode) _mkdir(path) | |
| #else | #else |
| # if defined(WIN32) | #define pa_mkdir(path, mode) mkdir(path, mode) |
| # include <windows.h> | |
| # else | |
| # define closesocket close | |
| # 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 48 int PASCAL closesocket(SOCKET); | Line 49 int PASCAL closesocket(SOCKET); |
| # endif | # endif |
| #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 | |
| #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 | |
| // defines for globals | // defines for globals |
| #define FILE_STATUS_NAME "status" | #define FILE_STATUS_NAME "status" |
| // globals | // globals |
| StringPtr file_status_name(new String(FILE_STATUS_NAME)); | const String file_status_name(FILE_STATUS_NAME); |
| // defines for statics | |
| #define HTTP_METHOD_NAME "method" | |
| #define HTTP_TIMEOUT_NAME "timeout" | |
| #define HTTP_HEADERS_NAME "headers" | |
| #define HTTP_ANY_STATUS_NAME "any-status" | |
| // statics | String sql_bind_name(SQL_BIND_NAME); |
| String sql_limit_name(PA_SQL_LIMIT_NAME); | |
| String sql_offset_name(PA_SQL_OFFSET_NAME); | |
| String sql_default_name(SQL_DEFAULT_NAME); | |
| String sql_distinct_name(SQL_DISTINCT_NAME); | |
| String sql_value_type_name(SQL_VALUE_TYPE_NAME); | |
| static StringPtr http_method_name(new String(HTTP_METHOD_NAME)); | // forwards |
| 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)); | |
| // defines | const UTF16* pa_utf16_encode(const char* in, Charset& source_charset); |
| #define DEFAULT_USER_AGENT "parser3" | |
| // functions | // functions |
| void fix_line_breaks(char* buf, size_t& size) { | #ifdef _MSC_VER |
| if(size==0) | |
| return; | |
| //_asm int 3; | |
| const char* const eob=buf+size; | |
| char* dest=buf; | |
| // fix DOS: \r\n -> \n | |
| // fix Macintosh: \r -> \n | |
| char* bol=buf; | |
| 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()); |
| size--; | return _wstat64((const wchar_t *)utf16name, buffer); |
| } else // \r, not \n = Macintosh | |
| bol=eol+1; | |
| } | |
| // last piece without \r, including terminating 0 | |
| if(dest!=bol) | |
| memcpy(dest, bol, eob-bol); | |
| } | } |
| char* file_read_text(Pool& pool, Charset& charset, | int pa_open(const char *pathname, int flags, int mode){ |
| StringPtr 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/*, HashStringValuePtr* out_fields*/) { | |
| File_read_result file= | |
| file_read(pool, charset, file_spec, true, params, fail_on_read_problem); | |
| return file.success?file.data:0; | |
| } | |
| //http request stuff | |
| /* ************************ http stuff *********************** */ | |
| 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(Pool& pool, 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()); |
| ssize_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=(char *)pool.malloc(MAX_STRING); | |
| ssize_t size=recv(sock, buf, MAX_STRING, 0); | |
| if(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); | |
| ArrayString astatus; | |
| status_line->split(astatus, 0, " ", 1); | |
| StringPtr 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", | |
| Exception::undefined_source, | |
| "bad response from host - no status found (size=%lu)", response.size()); | |
| } | |
| /* ********************** request *************************** */ | |
| #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) | |
| # define WE_CAN_USE_ALARM | |
| #endif | |
| #ifdef WE_CAN_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) { | |
| if(!host) | |
| throw Exception("http.host", | |
| origin_string, | |
| "zero hostname"); //never | |
| #ifdef WE_CAN_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); | |
| #endif | |
| struct sockaddr_in dest; | |
| if(!set_addr(&dest, host, port)) | |
| throw Exception("http.host", | |
| origin_string, | |
| "can not resolve hostname \"%s\"", host); | |
| if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) | |
| throw Exception("http.connect", | |
| origin_string, | |
| "can not make socket: %s (%d)", strerror(errno), errno); | |
| if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) | |
| throw Exception("http.connect", | |
| origin_string, | |
| "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, | |
| "error sending request: %s (%d)", strerror(errno), errno); | |
| result=http_read_response(pool, response, sock, fail_on_status_ne_200); | |
| closesocket(sock); | |
| #ifdef WE_CAN_USE_ALARM | |
| alarm(0); | |
| } | |
| #endif | |
| return result; | |
| } catch(...) { | |
| if(sock>=0) | |
| closesocket(sock); | |
| #ifdef WE_CAN_USE_ALARM | |
| alarm(0); | |
| #endif | #endif |
| rethrow; | |
| } | |
| } | |
| #ifndef DOXYGEN | /// these options were handled but not checked elsewhere, now check them |
| struct Http_pass_header_info { | int pa_get_valid_file_options_count(HashStringValue& options) { |
| Pool* pool; | int result=0; |
| Charset* charset; | if(options.get(PA_SQL_LIMIT_NAME)) |
| String* request; | result++; |
| bool user_agent_specified; | if(options.get(PA_SQL_OFFSET_NAME)) |
| }; | result++; |
| #endif | if(options.get(PA_COLUMN_SEPARATOR_NAME)) |
| static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value, | result++; |
| Http_pass_header_info *info) { | if(options.get(PA_COLUMN_ENCLOSER_NAME)) |
| *(info->request)<<*key<<": " | result++; |
| << *attributed_meaning_to_string(*info->pool, value, String::UL_HTTP_HEADER, false) | if(options.get(PA_CHARSET_NAME)) |
| <<"\n"; | result++; |
| if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT") | |
| info->user_agent_specified=true; | |
| } | |
| #ifndef DOXYGEN | |
| struct File_read_http_info { | |
| char **data; size_t *data_size; | |
| HashStringValuePtr headers; | |
| }; | |
| #endif | |
| static File_read_http_info file_read_http(Pool& pool, Charset& charset, StringPtr file_spec, | |
| HashStringValue *options=0) { | |
| File_read_http_info result; | |
| char host[MAX_STRING]; | |
| const char* uri; | |
| int port; | |
| CharPtr method(0); | |
| int timeout=2; | |
| bool fail_on_status_ne_200=true; | |
| ValuePtr vheaders(0); | |
| StringPtr connect_string(new String()); | |
| // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one | |
| connect_string->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; | |
| if(options) { | |
| int valid_options=0; | |
| if(ValuePtr vmethod=options->get(http_method_name)) { | |
| valid_options++; | |
| method=vmethod->as_string(&pool)->cstr(); | |
| } | |
| if(ValuePtr vtimeout=options->get(http_timeout_name)) { | |
| valid_options++; | |
| timeout=vtimeout->as_int(); | |
| } | |
| if(vheaders=options->get(http_headers_name)) { | |
| valid_options++; | |
| } | |
| if(ValuePtr vany_status=options->get(http_any_status_name)) { | |
| valid_options++; | |
| fail_on_status_ne_200=!vany_status->as_bool(); | |
| } | |
| if(valid_options!=options->count()) | |
| throw Exception("parser.runtime", | |
| Exception::undefined_source, | |
| "invalid option passed"); | |
| } | |
| //making request | |
| String request; | |
| if(method) | |
| request<<method; | |
| else | |
| request<<"GET"; | |
| request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; | |
| bool user_agent_specified=false; | |
| if(vheaders && !vheaders->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"; | |
| //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, | |
| timeout, fail_on_status_ne_200); | |
| //processing results | |
| int pos=response.pos("\r\n\r\n", 4); | |
| if(pos<1){ | |
| throw Exception("http.response", | |
| connect_string, | |
| "bad response from host - no headers found"); | |
| } | |
| StringPtr header_block=response.mid(0, pos); | |
| StringPtr body=response.mid(pos+4, response.size()); | |
| ArrayString aheaders; | |
| result.headers=HashStringValuePtr(new HashStringValue); | |
| header_block->split(aheaders, 0, "\r\n", 2); | |
| //processing headers | |
| for(int i=1; i<aheaders.count(); i++) { | |
| StringPtr line=aheaders.get(i); | |
| pos=line->pos(": ", 2); | |
| if(pos<1) | |
| throw Exception("http.response", | |
| connect_string, | |
| "bad response from host - bad header \"%s\"", line->cstr().get()); | |
| result.headers->put( | |
| line->mid(0, pos)->change_case(pool, charset, String::CC_UPPER), | |
| ValuePtr(new VString(line->mid(pos+2, line->size())))); | |
| } | |
| // output response | |
| *result.data=new(pool) char[*result.data_size=body->size()]; | |
| body->store_to(*result.data, String::UL_AS_IS); | |
| result.headers->put(file_status_name, ValuePtr(new VInt(status_code))); | |
| return result; | return result; |
| } | } |
| #ifndef DOXYGEN | #ifndef DOXYGEN |
| struct File_read_action_info { | struct File_read_action_info { |
| char **data; size_t *data_size; | char **data; size_t *data_size; |
| char* buf; uint64_t offset; size_t limit; | |
| }; | }; |
| #endif | #endif |
| static void file_read_action(Pool& pool, | |
| struct stat& finfo, | static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) { |
| int f, | File_read_action_info& info = *static_cast<File_read_action_info *>(context); |
| StringPtr file_spec, const char* fname, bool as_text, | 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); |
| void *context) { | if(to_read_size) { |
| File_read_action_info& info=*static_cast<File_read_action_info *>(context); | if(info.offset) |
| if(size_t to_read_size=(size_t)finfo.st_size) { | pa_lseek(f, info.offset, SEEK_SET); // seek never fails as POSIX allows the file offset to be set beyond the EOF |
| *info.data=pool.malloc(to_read_size+(as_text?1:0)); | *info.data = info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1); |
| *info.data_size=(size_t)read(f, *info.data, to_read_size); | ssize_t result = read(f, *info.data, to_read_size); |
| if(result<0) | |
| if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) | throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno); |
| throw Exception(0, | *info.data_size = result; |
| file_spec, | |
| "read failed: actually read %lu bytes count not in [0..%lu] valid range", | |
| *info.data_size, to_read_size); | |
| } else { // empty file | } else { // empty file |
| if(as_text) { | // 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=pool.malloc(1); | *info.data = (char *)pa_malloc_atomic(1); |
| *(char*)(*info.data)=0; | *(char*)(*info.data) = 0; |
| } else | *info.data_size = 0; |
| *info.data=0; | |
| *info.data_size=0; | |
| return; | return; |
| } | } |
| } | } |
| File_read_result file_read(Pool& pool, Charset& charset, StringPtr file_spec, | |
| bool as_text, HashStringValue *params, | File_read_result file_read_binary(const String& file_spec, bool fail_on_read_problem, char* buf, uint64_t offset, size_t limit) { |
| bool fail_on_read_problem) { | File_read_result result = {false, 0, 0, 0}; |
| File_read_result result; | File_read_action_info info = {&result.str, &result.length, buf, offset, limit}; |
| if(file_spec->starts_with("http://", 7)) { | |
| // fail on read problem | result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, 0, fail_on_read_problem); |
| result.headers=file_read_http(pool, charset, file_spec, params).headers; | return result; |
| result.success=true; | } |
| } else { | |
| File_read_action_info info={&result.data, &result.size}; | File_read_result file_read(Request_charsets& charsets, const String& file_spec, |
| result.success=file_read_action_under_lock(pool, file_spec, | bool as_text, HashStringValue *options, |
| "read", file_read_action, &info, | bool fail_on_read_problem, |
| as_text, 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); | |
| } | } |
| if(result.success && as_text) { | File_read_action_info info = {&result.str, &result.length, 0, offset, limit}; |
| // UTF-8 signature: EF BB BF | |
| if(result.size>=3) { | result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); |
| char *in=(char *)result.data; | |
| if((in[0] == '\xEF') && (in[1] == '\xBB') && | if(as_text){ |
| (in[2] == '\xBF')) { | if(result.success){ |
| result.data=in+3; result.size-=3;// skip prefix | Charset* asked_charset = 0; |
| if(options) | |
| if(Value* vcharset_name = options->get(PA_CHARSET_NAME)) | |
| asked_charset = &pa_charsets.get(vcharset_name->as_string()); | |
| asked_charset = pa_charsets.checkBOM(result.str, result.length, asked_charset); | |
| 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 | |
| String::C body = String::C(result.str, result.length); | |
| body=Charset::transcode(body, *asked_charset, charsets.source()); | |
| result.str = const_cast<char*>(body.str); // hacking a little | |
| result.length = body.length; | |
| } | } |
| } | } |
| if(result.length) | |
| fix_line_breaks((char *)(result.data), result.size); | fix_line_breaks(result.str, result.length); |
| // note: after fixing | |
| ((char*&)(result.data))[result.size]=0; | |
| } | } |
| return result; | return result; |
| } | } |
| bool file_read_action_under_lock(Pool& pool, StringPtr file_spec, | File_read_result file_load(Request& r, const String& file_spec, |
| bool as_text, HashStringValue *options, | |
| bool fail_on_read_problem, | |
| bool transcode_text_result) { | |
| size_t offset = 0; | |
| size_t limit = 0; | |
| if(options){ | |
| if(Value *voffset = (Value *)options->get(sql_offset_name)) | |
| offset = r.process(*voffset).as_int(); | |
| if(Value *vlimit = (Value *)options->get(sql_limit_name)) | |
| limit = r.process(*vlimit).as_int(); | |
| // no check on options count here | |
| } | |
| 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 | |
| File_read_http_result http = pa_internal_file_read_http(r, file_spec, as_text, options, transcode_text_result); | |
| File_read_result result = {true, http.str, http.length, http.headers}; | |
| return result; | |
| } else | |
| return file_read(r.charsets, file_spec, as_text, options, fail_on_read_problem, offset, limit, transcode_text_result); | |
| } | |
| char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem) { | |
| File_read_result file = file_read(charsets, file_spec, true, 0, fail_on_read_problem); | |
| return file.success ? file.str : 0; | |
| } | |
| 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 | |
| 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() | |
| ); | |
| } | |
| #else | |
| void check_safe_mode(struct stat, const String&, const char*) { | |
| } | |
| #endif | |
| bool file_read_action_under_lock(const String& file_spec, | |
| const char* action_name, File_read_action action, void *context, | 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) { |
| CharPtr fname=file_spec->cstr(String::UL_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 502 bool file_read_action_under_lock(Pool& p | Line 254 bool file_read_action_under_lock(Pool& p |
| // 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) | if(pa_lock_shared_blocking(f)!=0) |
| throw Exception("file.lock", | throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); |
| file_spec, | |
| "shared lock failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname.get()); | |
| 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.get()); | check_safe_mode(finfo, file_spec, 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()); | |
| #endif | |
| action(pool, 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.get()); | |
| return false; | return false; |
| } | } |
| } | } |
| static void create_dir_for_file(StringPtr 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); |
| int pos_before; | if(str[0]){ |
| while((pos_before=file_spec->pos("/", 1, pos_after))>=0) { | const char *pos=str+1; |
| mkdir(file_spec->mid(0, pos_before)->cstr(String::UL_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( |
| StringPtr file_spec, | const String& file_spec, |
| const char* action_name, File_write_action action, void *context, | const char* action_name, |
| File_write_action action, | |
| void *context, | |
| bool as_text, | bool as_text, |
| bool do_append, | bool do_append, |
| bool do_block, | bool do_block, |
| bool fail_on_lock_problem) { | bool fail_on_lock_problem) { |
| CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); | const char* fname=file_spec.taint_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 | O_CREAT|O_RDWR |
| |(as_text?_O_TEXT:_O_BINARY) | |(as_text?_O_TEXT:_O_BINARY) |
| |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) { | |(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", | Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); |
| file_spec, | |
| "shared lock failed: %s (%d), actual filename '%s'", | |
| strerror(errno), errno, fname.get()); | |
| close(f); | close(f); |
| if(fail_on_lock_problem) | if(fail_on_lock_problem) |
| throw e; | throw e; |
| Line 583 bool file_write_action_under_lock( | Line 322 bool file_write_action_under_lock( |
| } | } |
| 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 | 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; |
| } | } |
| Line 597 bool file_write_action_under_lock( | Line 341 bool file_write_action_under_lock( |
| 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] | 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.get()); | |
| // 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 *data; size_t size; | 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.size) { | if(info.length) { |
| int written=write(f, info.data, info.size); | 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); |
| Exception::undefined_source, | 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( |
| StringPtr 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(StringPtr file_spec, size_t pos_after) { | bool dir=false; |
| int pos_before; | size_t pos=0; |
| if((pos_before=file_spec->pos("/", 1, pos_after))>=0) | 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::UL_FILE_SPEC)); | fname[pos-1]=0; |
| } | dir=true; |
| bool file_delete(StringPtr file_spec, bool fail_on_read_problem) { | } else if(dir) break; |
| CharPtr fname=file_spec->cstr(String::UL_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.get()); | |
| 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); | |
| 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.get(), new_spec_cstr.get()); | |
| 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(StringPtr file_spec) { | bool entry_exists(const String& file_spec) { |
| CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); | return entry_exists(file_spec.taint_cstr(String::L_FILE_SPEC), 0); |
| return entry_exists(fname, 0); | |
| } | } |
| static bool entry_readable(StringPtr file_spec, bool need_dir) { | static bool entry_ifdir(char *fname, bool need_dir) { |
| CharPtr fname=file_spec->cstr(String::UL_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(StringPtr 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); | |
| } | |
| // 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(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; | |
| } | |
| 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); | |
| } | } |
| bool dir_readable(StringPtr file_spec) { | |
| return entry_readable(file_spec, true); | const String* file_exist(const String& path, const String& name) { |
| String& result=*new String(path); | |
| if(path.last_char() != '/') | |
| result << "/"; | |
| result << name; | |
| return file_exist(result)?&result:0; | |
| } | |
| bool file_executable(const String& file_spec) { | |
| return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0; | |
| } | } |
| StringPtr file_readable(String& path, String& name) { | |
| StringPtr result(new String(path)); | bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { |
| *result << "/"; | const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); |
| *result << name; | struct stat finfo; |
| return file_readable(result)?result:StringPtr(0); | if(pa_stat(fname, &finfo)!=0) { |
| } | |
| bool file_executable(StringPtr file_spec) { | |
| return access(file_spec->cstr(String::UL_FILE_SPEC), X_OK)==0; | |
| } | |
| bool file_stat(StringPtr 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; | |
| if(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.get()); | |
| 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; | 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; | |
| } | |
| 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(Pool& pool, 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 pool.copy(local_buf, size+1); | 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 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(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(Pool& pool, const char* cp, int len) { | enum EscapeState { |
| char* s=pool.malloc(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) | |
| char* dst=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((XMLByte*&)dst, (XMLCh)escapedValue, '?'); | |
| escapeState=EscapeRest; | |
| } | |
| } else { | |
| // not full unicode value | |
| escapeState=EscapeRest; | |
| } | |
| break; | |
| } | } |
| break; | |
| case EscapeFirst: | |
| escapedValue=hex_value[ch] << 4; | |
| escapeState=EscapeSecond; | |
| break; | |
| case EscapeSecond: | |
| escapedValue +=hex_value[ch]; | |
| s[dstPos++]=escapedValue; | |
| escapeState=EscapeRest; | |
| break; | |
| } | } |
| srcPos++; | |
| srcPos++; | |
| } | } |
| s[dstPos]=0; | |
| *dst=0; // zero-termination | |
| return s; | 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 865 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; | |
| } | |
| } | |
| } | |
| inline void CalcCrc32(const unsigned char byte, unsigned long &crc32) | |
| { | |
| crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)]; | |
| } | |
| unsigned long pa_crc32(const char *in, size_t in_size){ | |
| unsigned long crc32=0xFFFFFFFF; | |
| InitCrc32Table(); | |
| for(size_t i = 0; i<in_size; i++) | |
| CalcCrc32(in[i], crc32); | |
| return ~crc32; | |
| } | |
| static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) { | |
| unsigned long& crc32=*static_cast<unsigned long *>(context); | |
| if(finfo.st_size) { | |
| InitCrc32Table(); | |
| int nCount=0; | |
| do { | |
| unsigned char buffer[FILE_BUFFER_SIZE]; | |
| nCount = file_block_read(f, buffer, sizeof(buffer)); | |
| for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32); | |
| } while(nCount > 0); | |
| } | } |
| } | } |
| static bool isLeap(int year) { | unsigned long pa_crc32(const String& file_spec){ |
| return !( | unsigned long crc32=0xFFFFFFFF; |
| (year % 4) || ((year % 400) && !(year % 100)) | file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32); |
| ); | return ~crc32; |
| } | } |
| int getMonthDays(int year, int month) { | // content-type: xxx; charset=WE-NEED-THIS |
| int monthDays[]={ | // content-type: xxx; charset="WE-NEED-THIS" |
| 31, | // content-type: xxx; charset="WE-NEED-THIS"; |
| isLeap(year) ? 29 : 28, | Charset* detect_charset(const char* content_type){ |
| 31, | if(content_type){ |
| 30, | char* CONTENT_TYPE=str_upper(content_type); |
| 31, | |
| 30, | if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){ |
| 31, | begin+=8; // skip "CHARSET=" |
| 31, | char* end=0; |
| 30, | if(*begin && (*begin=='"' || *begin =='\'')){ |
| 31, | char quote=*begin; |
| 30, | begin++; |
| 31 | end=(char*)strchr(begin, quote); |
| }; | } |
| return monthDays[month]; | if(!end) |
| } | end=(char*)strchr(begin, ';'); |
| void remove_crlf(char* start, char* end) { | if(end) |
| for(char* p=start; p<end; p++) | *end=0; // terminator |
| switch(*p) { | |
| case '\n': *p='|'; break; | return *begin ? &pa_charsets.get_direct(begin) : 0; |
| case '\r': *p=' '; break; | |
| } | } |
| } | |
| 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 962 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 | |
| } | |