--- parser3/src/main/pa_common.C 2003/05/30 10:45:37 1.151 +++ parser3/src/main/pa_common.C 2020/12/08 21:30:46 1.321 @@ -1,25 +1,36 @@ /** @file Parser: commonly functions. - Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2000-2017 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_COMMON_C="$Date: 2003/05/30 10:45:37 $"; - #include "pa_common.h" #include "pa_exception.h" -#include "pa_globals.h" #include "pa_hash.h" -#include "pa_vstring.h" -#include "pa_vdate.h" +#include "pa_globals.h" +#include "pa_charsets.h" +#include "pa_http.h" +#include "pa_request_charsets.h" +#include "pcre.h" +#include "pa_request.h" + +#include "pa_idna.h" +#include "pa_convert_utf.h" + +#ifdef _MSC_VER +#include +#include +#endif -#ifdef WIN32 -# include +#ifdef _MSC_VER +#define pa_mkdir(path, mode) _mkdir(path) #else -# define closesocket close +#define pa_mkdir(path, mode) mkdir(path, mode) #endif +volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.321 2020/12/08 21:30:46 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H; + // some maybe-undefined constants #ifndef _O_TEXT @@ -39,441 +50,200 @@ static const char* IDENT_COMMON_C="$Date # endif #endif -// locking constants +// defines for globals -#ifdef HAVE_FLOCK +#define FILE_STATUS_NAME "status" -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); } +// globals -#else -#ifdef HAVE__LOCKING +const String file_status_name(FILE_STATUS_NAME); -#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); } +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); -#else -#ifdef HAVE_FCNTL +// forwards -#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); } +const UTF16* pa_utf16_encode(const char* in, Charset& source_charset); -#else -#ifdef HAVE_LOCKF +// functions -#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); } +#ifdef _MSC_VER -#else +int pa_stat(const char *pathname, struct stat *buffer){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + return _wstat64((const wchar_t *)utf16name, buffer); +} -#error unable to find file locking func +int pa_open(const char *pathname, int flags, int mode){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + return _wopen((const wchar_t *)utf16name, flags, mode); +} + +FILE *pa_fopen(const char *pathname, const char *mode){ + const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source()); + const UTF16* utf16mode=pa_utf16_encode(mode, pa_thread_request().charsets.source()); + return _wfopen((const wchar_t *)utf16name, (const wchar_t *)utf16mode); +} -#endif -#endif -#endif #endif -#define DEFAULT_USER_AGENT "parser3" +/// 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 +struct File_read_action_info { + char **data; size_t *data_size; + char* buf; uint64_t offset; size_t limit; +}; +#endif -void fix_line_breaks(char* buf, size_t& size) { - if(size==0) +static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) { + File_read_action_info& info = *static_cast(context); + size_t to_read_size = check_file_size(info.limit && info.limit < finfo.st_size ? info.limit : finfo.st_size, &file_spec); + if(to_read_size) { + 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 + *info.data = info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1); + 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; - - //_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]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; +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}; + + result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, 0, fail_on_read_problem); + return result; } -static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){ - const String* status_code=0; - ssize_t EOLat=0; - while(true) { - char *buf=(char *)response.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(!status_code && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response - const String& status_line=response.mid(0, (size_t)EOLat); - Array astatus(response.pool()); - size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1); - status_code=astatus.get_string(1); - - if(fail_on_status_ne_200 && *status_code!="200") - throw Exception("http.status", - status_code, - "invalid HTTP response status"); - } +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); } - if(status_code) - return status_code->as_int(); - else - throw Exception("http.response", - 0, - "bad response from host - no status found (size=%lu)", response.size()); -} -/* ********************** request *************************** */ + File_read_action_info info = {&result.str, &result.length, 0, offset, limit}; -#if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP) -# define PA_USE_ALARM -#endif + result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem); -#ifdef PA_USE_ALARM -static sigjmp_buf timeout_env; -static void timeout_handler(int sig){ - siglongjmp(timeout_env, 1); -} -#endif + if(as_text){ + if(result.success){ + Charset* asked_charset = 0; + if(options) + if(Value* vcharset_name = options->get(PA_CHARSET_NAME)) + asked_charset = &pa_charsets.get(vcharset_name->as_string()); -static int http_request(String& response, - const String *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 + asked_charset = pa_charsets.checkBOM(result.str, result.length, asked_charset); -#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(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()); - 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(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); - /*re*/throw; + result.str = const_cast(body.str); // hacking a little + result.length = body.length; + } } -#ifdef PA_USE_ALARM + if(result.length) + fix_line_breaks(result.str, result.length); } -#endif -} -#ifndef DOXYGEN -struct Http_pass_header_info { - String* request; - bool user_agent_specified; -}; -#endif -static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info) -{ - Http_pass_header_info& i=*static_cast(info); - Pool& pool=i.request->pool(); - - *(i.request)<(value), String::UL_HTTP_HEADER, false) - <<"\n"; - - if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT") - i.user_agent_specified=true; + return result; } -static void file_read_http(Pool& pool, const String& file_spec, - void*& data, size_t& data_size, - Hash *options=0, Hash** out_fields=0) { - char host[MAX_STRING]; - const char* uri; - int port; - const char* method="GET"; - int timeout=2; - bool fail_on_status_ne_200=true; - Value *vheaders=0; - - String& connect_string=*new(pool) String(pool); // must not be local [exception may be reported outside] - // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one - connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces - - char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED); - if(strncmp(connect_string_cstr, "http://", 7)!=0) - throw Exception(0, - &connect_string, - "does not start with http://"); //never - connect_string_cstr+=7; - - strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0; - char* host_uri=lsplit(host, '/'); - uri=host_uri?connect_string_cstr+(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(Value *vmethod=static_cast(options->get(*http_method_name))) { - valid_options++; - method=vmethod->as_string().cstr(); - } - if(Value *vtimeout=static_cast(options->get(*http_timeout_name))) { - valid_options++; - timeout=vtimeout->as_int(); - } - if(vheaders=static_cast(options->get(*http_headers_name))) { - valid_options++; - } - if(Value *vany_status=static_cast(options->get(*http_any_status_name))) { - valid_options++; - fail_on_status_ne_200=!vany_status->as_bool(); - } - if(valid_options!=options->size()) - throw Exception("parser.runtime", - 0, - "invalid option passed"); - } - - //making request - String request(pool); - request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n"; - bool user_agent_specified=false; - if(vheaders && !vheaders->is_string()) { // allow empty - if(Hash *headers=vheaders->get_hash(&connect_string)) { - Http_pass_header_info info={&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(pool); - int status_code=http_request(response, - &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), - 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"); +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 } - String header_block=response.mid(0, pos); - String body=response.mid(pos+4, response.size()); - - Array aheaders(pool); - Hash& headers=*new(pool) Hash(pool); - size_t pos_after_ref=0; - header_block.split(aheaders, &pos_after_ref, "\r\n", 2); - - //processing headers - for(int i=1;ipos(": ", 2); - if(pos<1) - throw Exception("http.response", - &connect_string, - "bad response from host - bad header \"%s\"", line->cstr()); - - headers.put( - line->mid(0, pos).change_case(pool, String::CC_UPPER), - new(pool) VString(line->mid(pos+2, line->size()))); - } else - throw Exception("http.response", - &connect_string, - "bad response from host - bad headers \"%s\"", header_block.cstr()); - } - - // output response - data=body.cstr(); data_size=body.size(); - if(out_fields) { - headers.put(*file_status_name, new(pool) VInt(pool, status_code)); - *out_fields=&headers; - } -} -#ifndef DOXYGEN -struct File_read_action_info { - void **data; size_t *data_size; -}; -#endif -static void file_read_action(Pool& pool, - struct stat& finfo, - int f, - const String& file_spec, const char* fname, bool as_text, - void *context) { - File_read_action_info& info=*static_cast(context); - if(size_t to_read_size=(size_t)finfo.st_size) { - *info.data=pool.malloc(to_read_size+(as_text?1:0), 3); - *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=pool.malloc(1); - *(char*)(*info.data)=0; - } else - *info.data=0; - *info.data_size=0; - return; - } -} -bool file_read(Pool& pool, const String& file_spec, - void*& data, size_t& data_size, - bool as_text, Hash *params, Hash** out_fields, - bool fail_on_read_problem) { - bool result; - if(file_spec.starts_with("http://", 7)) { + 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(pool, file_spec, data, data_size, params, out_fields); - result=true; - } else { - File_read_action_info info={&data, &data_size}; - result=file_read_action_under_lock(pool, file_spec, - "read", file_read_action, &info, - as_text, fail_on_read_problem); - } + File_read_http_result http = pa_internal_file_read_http(r, file_spec, as_text, options, transcode_text_result); - if(result && as_text) { - // UTF-8 signature: EF BB BF - if(data_size>=3) { - char *in=(char *)data; - if((in[0] == '\xEF') && (in[1] == '\xBB') && - (in[2] == '\xBF')) { - data=in+3; data_size-=3;// skip prefix - } - } + 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); +} - fix_line_breaks((char *)(data), data_size); - // note: after fixing - ((char*&)(data))[data_size]=0; - } +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; +} - 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) { - 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 " + 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()); + "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(Pool& pool, const String& file_spec, + + +bool file_read_action_under_lock(const String& file_spec, const char* action_name, File_read_action action, void *context, bool as_text, bool fail_on_read_problem) { - const char* fname=file_spec.cstr(String::UL_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; // first open, next stat: @@ -485,75 +255,65 @@ bool file_read_action_under_lock(Pool& p // they delay update till open, so we would receive "!^test[" string // if would do stat, next open. // later: it seems, even this does not help sometimes - if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { + if((f=pa_open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { try { - if(lock_shared_blocking(f)!=0) - throw Exception("file.lock", - &file_spec, - "shared lock failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname); + if(pa_lock_shared_blocking(f)!=0) + throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); 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 - &file_spec, - "stat failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname); + &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname); -#ifdef PA_SAFE_MODE check_safe_mode(finfo, file_spec, fname); -#endif - action(pool, finfo, f, file_spec, fname, as_text, context); + action(finfo, f, file_spec, context); } catch(...) { - unlock(f);close(f); + pa_unlock(f);close(f); if(fail_on_read_problem) - /*re*/throw; - return false; + rethrow; + return false; } - unlock(f);close(f); + pa_unlock(f);close(f); return true; - } else { + } else { if(fail_on_read_problem) - throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, - &file_spec, - "%s failed: %s (%d), actual filename '%s'", - action_name, strerror(errno), errno, fname); + throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0, + &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); 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; - int pos_before; - while((pos_before=file_spec.pos("/", 1, pos_after))>=0) { - mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775); + size_t pos_before; + while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) { + pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775); pos_after=pos_before+1; } } bool file_write_action_under_lock( const String& file_spec, - const char* action_name, File_write_action action, void *context, + const char* action_name, + File_write_action action, + void *context, bool as_text, bool do_append, bool do_block, bool fail_on_lock_problem) { - const char* fname=file_spec.cstr(String::UL_FILE_SPEC); + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); int f; if(access(fname, W_OK)!=0) // no create_dir_for_file(file_spec); - 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) { - if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) { - Exception e("file.lock", - &file_spec, - "shared lock failed: %s (%d), actual filename '%s'", - strerror(errno), errno, fname); + if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) { + 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) throw e; @@ -561,284 +321,568 @@ bool file_write_action_under_lock( } 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(...) { #ifdef HAVE_FTRUNCATE if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower #endif - unlock(f);close(f); - /*re*/throw; + pa_unlock(f);close(f); + rethrow; } #ifdef HAVE_FTRUNCATE if(!do_append) ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov for discovering] #endif - unlock(f);close(f); + pa_unlock(f);close(f); return true; } else - throw Exception(errno==EACCES?"file.access":0, - &file_spec, - "%s failed: %s (%d), actual filename '%s'", - action_name, strerror(errno), errno, fname); + throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname); // here should be nothing, see rethrow above } #ifndef DOXYGEN struct File_write_action_info { - const void *data; size_t size; + const char* str; + size_t length; }; #endif + static void file_write_action(int f, void *context) { File_write_action_info& info=*static_cast(context); - if(info.size) { - int written=write(f, info.data, info.size); + if(info.length) { + ssize_t written=write(f, info.str, info.length); if(written<0) - throw Exception(0, - 0, - "write failed: %s (%d)", strerror(errno), errno); + throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno); + if((size_t)written!=info.length) + throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length); } } + void file_write( - const String& file_spec, - const void *data, size_t size, + Request_charsets& charsets, + const String& file_spec, + const char* data, + size_t size, bool as_text, - bool do_append) { - File_write_action_info info={data, size}; - file_write_action_under_lock( - file_spec, - "write", file_write_action, &info, - as_text, - do_append); -} - -// throws nothing! [this is required in file_move & file_delete] -static void rmdir(const String& file_spec, size_t pos_after) { - int pos_before; - if((pos_before=file_spec.pos("/", 1, pos_after))>=0) - rmdir(file_spec, pos_before+1); - - rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC)); -} -bool file_delete(const String& file_spec, bool fail_on_read_problem) { - const char* fname=file_spec.cstr(String::UL_FILE_SPEC); - if(unlink(fname)!=0) - 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; + bool do_append, + Charset* asked_charset) { - 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::UL_FILE_SPEC); - const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC); - - create_dir_for_file(new_spec); + 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; + }; - 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); + File_write_action_info info={data, size}; - rmdir(old_spec, 1); + file_write_action_under_lock( + file_spec, + "write", + file_write_action, + &info, + as_text, + do_append); +} + +static size_t get_dir(char* fname, size_t helper_length){ + bool dir=false; + size_t pos=0; + for(pos=helper_length; pos; pos--){ + char c=fname[pos-1]; + if(c=='/' || c=='\\'){ + fname[pos-1]=0; + dir=true; + } else if(dir) break; + } + return pos; } - bool entry_exists(const char* fname, struct stat *afinfo) { struct stat lfinfo; - bool result=stat(fname, &lfinfo)==0; + bool result=pa_stat(fname, &lfinfo)==0; if(afinfo) *afinfo=lfinfo; return result; } bool entry_exists(const String& file_spec) { - const char* fname=file_spec.cstr(String::UL_FILE_SPEC); - return entry_exists(fname, 0); + return entry_exists(file_spec.taint_cstr(String::L_FILE_SPEC), 0); } -static bool entry_readable(const String& file_spec, bool need_dir) { - char* fname=file_spec.cstr(String::UL_FILE_SPEC); - if(need_dir) { - size_t size=strlen(fname); +static bool entry_ifdir(char *fname, bool need_dir) { + if(need_dir){ + size_t size=strlen(fname); while(size) { - char c=fname[size-1]; + char c=fname[size-1]; if(c=='/' || c=='\\') fname[--size]=0; else break; } } + 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; return is_dir==need_dir; } 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); +} + +// 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_readable(const String& file_spec) { - return entry_readable(file_spec, true); + +bool dir_exists(const String& file_spec) { + return entry_ifdir(file_spec, true); } -String *file_readable(const String& path, const String& name) { - String *result=new(path.pool()) String(path); - *result << "/"; - *result << name; - return file_readable(*result)?result:0; + +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.cstr(String::UL_FILE_SPEC), X_OK)==0; + return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0; } -bool file_stat(const String& file_spec, - size_t& rsize, - time_t& ratime, - time_t& rmtime, - time_t& rctime, - bool fail_on_read_problem) { - Pool& pool=file_spec.pool(); - const char* fname=file_spec.cstr(String::UL_FILE_SPEC); - struct stat finfo; - if(stat(fname, &finfo)!=0) +bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) { + const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC); + struct stat finfo; + if(pa_stat(fname, &finfo)!=0) { if(fail_on_read_problem) - throw Exception("file.missing", - &file_spec, - "getting file size failed: %s (%d), real filename '%s'", - strerror(errno), errno, fname); + throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname); else return false; + } rsize=finfo.st_size; - ratime=finfo.st_atime; - rmtime=finfo.st_mtime; - rctime=finfo.st_ctime; + ratime=(time_t)finfo.st_atime; + rmtime=(time_t)finfo.st_mtime; + 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; } +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 helper_length){ + char *result=pa_strdup(s, helper_length); + for(char* c=result; *c; c++) + *c=(char)tolower((unsigned char)*c); + return result; +} + +char *str_upper(const char *s, size_t helper_length){ + char *result=pa_strdup(s, helper_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]= 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 FormatInt: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value); - else - size=snprintf(local_buf, sizeof(local_buf), fmt, value); - else - size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); - - char* pool_buf=(char *)pool.malloc(size+1, 4); - memcpy(pool_buf, local_buf, size+1); - return pool_buf; + break; + case FormatUInt: + size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); + break; + case FormatInvalid: + throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt); + } + } else + size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); + + if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing + throw Exception(PARSER_RUNTIME, 0, "Error 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) { #ifdef WIN32 + size_t to_write = size; do{ - int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout); + int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout); if(chunk_written<=0) break; size-=chunk_written; buf=((const char*)buf)+chunk_written; - } while(size>0); + } while(size>0); - return size; + fflush(stdout); + return to_write-size; #else - return fwrite(buf, 1, size, stdout); + size_t result=fwrite(buf, 1, size, stdout); + fflush(stdout); + return result; #endif } -char* unescape_chars(Pool& pool, const char* cp, int len) { - char* s=(char *)pool.malloc(len + 1, 5); - enum EscapeState { - EscapeRest, - EscapeFirst, - EscapeSecond - } escapeState=EscapeRest; - int escapedValue=0; +enum EscapeState { + EscapeRest, + EscapeFirst, + EscapeSecond, + EscapeUnicode +}; + +// @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually) +char* unescape_chars(const char* cp, int len, Charset* charset, bool 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 dstPos=0; - while(srcPos < len) { - int ch=cp[srcPos]; - switch(escapeState) { - case EscapeRest: - if(ch=='%') { - escapeState=EscapeFirst; - } else if(ch=='+') { - s[dstPos++]=' '; - } else { - s[dstPos++]=ch; + short int jsCnt=0; + while(srcPosstore_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?'); + escapeState=EscapeRest; + } + } else { + // not full unicode value + escapeState=EscapeRest; + } + break; } - break; - case EscapeFirst: - escapedValue=hex_value[ch] << 4; - escapeState=EscapeSecond; - break; - case EscapeSecond: - escapedValue +=hex_value[ch]; - s[dstPos++]=escapedValue; - escapeState=EscapeRest; - break; } - srcPos++; + + srcPos++; } - s[dstPos]=0; + + *dst=0; // zero-termination return s; } +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 void back_slashes_to_slashes(char* s) { if(s) @@ -846,95 +890,289 @@ void back_slashes_to_slashes(char* s) { if(*s=='\\') *s='/'; } -/* -void slashes_to_back_slashes(char* s) { - if(s) - for(; *s; s++) - if(*s=='/') - *s='\\'; -} -*/ #endif -bool StrEqNc(const char* s1, const char* s2, bool strict) { - while(true) { - if(!(*s1)) { - if(!(*s2)) - return true; - else - return !strict; - } else if(!(*s2)) - return !strict; - if(isalpha(*s1)) { - if(tolower(*s1) !=tolower(*s2)) - return false; - } else if((*s1) !=(*s2)) - return false; - s1++; - s2++; +size_t strpos(const char *str, const char *substr) { + const char *p = strstr(str, substr); + return (p==0)?STRING_NOT_FOUND:p-str; +} + +int remove_crlf(char* start, char* end) { + char* from=start; + char* to=start; + bool skip=false; + while(from < end){ + switch(*from){ + case '\n': + case '\r': + case '\t': + case ' ': + 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 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(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); + } +} + +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 #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) return 0; 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 + // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010 --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 –1 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); if(r<0) r=s; #else r=vsnprintf(b, s, f, l); /* - solaris: - man vsnprintf + solaris: 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 been written to the buffer if it were large enough. If the value of n is 0 on a call to snprintf(), an unspecified @@ -943,105 +1181,18 @@ int __vsnprintf(char* b, size_t s, const if(r<0) r=0; - else if(r>s) + else if((size_t)r>s) r=s; #endif b[r]=0; 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_start(l, f); - int r=__vsnprintf(b, s, f, l); - va_end(l); + va_start(l, f); + int r=pa_vsnprintf(b, s, f, l); + va_end(l); 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 -} - - -// attributed meaning - -/// http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.3 -static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) { - const char month_names[12][4]={ - "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"}; - const char days[7][4]={ - "Sun","Mon","Tue","Wed","Thu","Fri","Sat"}; - - time_t when=vdate.get_time(); - struct tm *tms=gmtime(&when); - if(!tms) - throw Exception(0, - 0, - "bad time in attribute value (seconds from epoch=%ld)", when); - return snprintf(buf, MAX_STRING, "%s, %.2d %s %.4d %.2d:%.2d:%.2d GMT", - days[tms->tm_wday], - tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900, - tms->tm_hour,tms->tm_min,tms->tm_sec); -} -static void append_attribute_meaning(String& result, - Value& value, String::Untaint_lang lang, bool forced) { - if(const String *string=value.get_string()) - result.append(string->join_chains(result.pool(), 0), lang, forced); - else - if(Value *vdate=value.as(VDATE_TYPE, false)) { - char *buf=(char *)result.malloc(MAX_STRING); - size_t size=date_attribute(*static_cast(vdate), - buf, MAX_STRING); - - result.APPEND_CLEAN(buf, size, "converted from date", 0); - } else - throw Exception("parser.runtime", - &result, - "trying to append here neither string nor date (%s)", - value.type()); -} -#ifndef DOXYGEN -struct Attributed_meaning_info { - String *header; // header line being constructed - String::Untaint_lang lang; // language in which to append to that line - bool forced; // do they force that lang? -}; -#endif -static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue, - void *info) { - if(akey==VALUE_NAME) - return; - - Attributed_meaning_info& ami=*static_cast(info); - - // ...; charset=windows1251 - *ami.header << "; "; - ami.header->append(akey, ami.lang, ami.forced); - *ami.header << "="; - append_attribute_meaning(*ami.header, *static_cast(avalue), ami.lang, ami.forced); -} -const String& attributed_meaning_to_string(Value& meaning, - String::Untaint_lang lang, bool forced) { - String &result=*new(meaning.pool()) String(meaning.pool()); - if(Hash *hash=meaning.get_hash(0)) { - // $value(value) $subattribute(subattribute value) - if(Value *value=static_cast(hash->get(*value_name))) - append_attribute_meaning(result, *value, lang, forced); - - Attributed_meaning_info attributed_meaning_info={&result, lang}; - hash->for_each(append_attribute_subattribute, &attributed_meaning_info); - } else // result value - append_attribute_meaning(result, meaning, lang, forced); - - return result; -}