--- parser3/src/main/pa_common.C 2001/04/26 14:55:28 1.52 +++ parser3/src/main/pa_common.C 2003/03/04 14:55:37 1.143.2.21 @@ -1,203 +1,756 @@ /** @file Parser: commonly functions. - Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) - - $Id: pa_common.C,v 1.52 2001/04/26 14:55:28 paf Exp $ + Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -#include "pa_config_includes.h" +static const char* IDENT_COMMON_C="$Date: 2003/03/04 14:55:37 $"; + #include "pa_common.h" -#include "pa_types.h" #include "pa_exception.h" -#include "pa_pool.h" -#include "pa_globals.h" -#include "pa_value.h" #include "pa_hash.h" -#include "pa_string.h" +#include "pa_vstring.h" +#include "pa_vint.h" +#include "pa_globals.h" -#ifndef WIN32 -# ifndef _O_TEXT -# define _O_TEXT 0 +#ifdef CYGWIN +#define _GNU_H_WINDOWS32_SOCKETS +// for PASCAL +#include +// SOCKET +typedef u_int SOCKET; +int PASCAL closesocket(SOCKET); +#else +# if defined(WIN32) +# include +# else +# define closesocket close # endif -# ifndef _O_BINARY -# define _O_BINARY 0 +#endif + +// some maybe-undefined constants + +#ifndef _O_TEXT +# define _O_TEXT 0 +#endif +#ifndef _O_BINARY +# define _O_BINARY 0 +#endif + +#ifdef HAVE_FTRUNCATE +# define PA_O_TRUNC 0 +#else +# ifdef _O_TRUNC +# define PA_O_TRUNC _O_TRUNC +# else +# error you must have either ftruncate function or _O_TRUNC bit declared # endif #endif -#if _MSC_VER +// locking constants -int __vsnprintf(char *b, size_t s, const char *f, va_list l) { - int r=_vsnprintf(b, --s, f, l); - b[s]=0; - return r; +#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 + +#define FILE_STATUS_NAME "status" + +// globals + +StringPtr file_status_name(new String(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 + +static StringPtr http_method_name(new String(HTTP_METHOD_NAME)); +static StringPtr http_timeout_name(new String(HTTP_TIMEOUT_NAME)); +static StringPtr http_headers_name(new String(HTTP_HEADERS_NAME)); +static StringPtr http_any_status_name(new String(HTTP_ANY_STATUS_NAME)); + +// defines + +#define DEFAULT_USER_AGENT "parser3" + +// functions + +void fix_line_breaks(char* buf, size_t& size) { + 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]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; } -int __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); - return r; + +static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){ + int result=0; + ssize_t EOLat=0; + while(true) { + char *buf=new(pool) char[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 -char *file_read_text(Pool& pool, const String& file_spec, bool fail_on_read_problem) { - void *result; - size_t size; - return file_read(pool, file_spec, result, size, true, - fail_on_read_problem)?(char *)result:0; +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 + rethrow; + } +} + +#ifndef DOXYGEN +struct Http_pass_header_info { + Pool* pool; + Charset* charset; + String* request; + bool user_agent_specified; +}; +#endif +static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value, + Http_pass_header_info *info) { + *(info->request)<<*key<<": " + << *attributed_meaning_to_string(*info->pool, value, String::UL_HTTP_HEADER, false) + <<"\n"; + + if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT") + info->user_agent_specified=true; +} +#ifndef DOXYGEN +struct File_read_http_result { + char *data; size_t size; + HashStringValuePtr headers; +}; +#endif +static File_read_http_result file_read_http(Pool& pool, Charset& charset, StringPtr file_spec, + HashStringValue *options=0) { + File_read_http_result 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<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; ipos(": ", 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[1/*as text puts terminating zero later*/+(result.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; } -bool file_read(Pool& pool, const String& file_spec, - void*& data, size_t& read_size, bool as_text, - bool fail_on_read_problem, - size_t offset, size_t limit) { - const char *fname=file_spec.cstr(String::UL_FILE_NAME); + +#ifndef DOXYGEN +struct File_read_action_info { + char **data; size_t *data_size; +}; +#endif +static void file_read_action(Pool& pool, + struct stat& finfo, + int f, + StringPtr file_spec, const char* fname, bool as_text, + void *context) { + File_read_action_info& info=*static_cast(context); + if(size_t to_read_size=(size_t)finfo.st_size) { + *info.data=new(pool) char[to_read_size+(as_text?1:0)]; + *info.data_size=(size_t)read(f, *info.data, to_read_size); + + if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size) + throw Exception(0, + file_spec, + "read failed: actually read %lu bytes count not in [0..%lu] valid range", + *info.data_size, to_read_size); + } else { // empty file + if(as_text) { + *info.data=new(pool) char[1]; + *(char*)(*info.data)=0; + } else + *info.data=0; + *info.data_size=0; + return; + } +} +File_read_result file_read(Pool& pool, Charset& charset, StringPtr file_spec, + bool as_text, HashStringValue *params, + bool fail_on_read_problem) { + File_read_result result; + if(file_spec->starts_with("http://", 7)) { + // fail on read problem + File_read_http_result http=file_read_http(pool, charset, file_spec, params); + result.success=true; + result.data=http.data; + result.size=http.size; + result.headers=http.headers; + } else { + File_read_action_info info={&result.data, &result.size}; + result.success=file_read_action_under_lock(pool, file_spec, + "read", file_read_action, &info, + as_text, fail_on_read_problem); + } + + if(result.success && as_text) { + // UTF-8 signature: EF BB BF + if(result.size>=3) { + char *in=(char *)result.data; + if((in[0] == '\xEF') && (in[1] == '\xBB') && + (in[2] == '\xBF')) { + result.data=in+3; result.size-=3;// skip prefix + } + } + + fix_line_breaks((char *)(result.data), result.size); + // note: after fixing + ((char*&)(result.data))[result.size]=0; + } + + return result; +} + +bool file_read_action_under_lock(Pool& pool, StringPtr file_spec, + const char* action_name, File_read_action action, void *context, + bool as_text, + bool fail_on_read_problem) { + CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); int f; - struct stat finfo; // first open, next stat: // directory update of NTFS hard links performed on open. // ex: // a.html:^test[] and b.html hardlink to a.html // user inserts ! before ^test in a.html - // directory entry of b.html in NTFS not updated at once, + // directory entry of b.html in NTFS not updated at once, // they delay update till open, so we would receive "!^test[" string // if would do stat, next open. - if( - (f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0 && - stat(fname, &finfo)==0) { - /*if(exclusive) - flock(f, LOCK_EX);*/ - size_t max_size=limit?min(offset+limit, finfo.st_size)-offset:finfo.st_size; - if(!max_size) { // eof - data=0; - read_size=0; - return true; - } - data=pool.malloc(max_size+(as_text?1:0)); - if(offset) - lseek(f, offset, SEEK_SET); - read_size=read(f, data, max_size); - /*if(exclusive) - flock(f, LOCK_UN);*/ - close(f); - - if(read_size>=0 && read_size<=max_size) { - if(as_text) - ((char*&)data)[read_size]=0; - } else - PTHROW(0, 0, - &file_spec, - "read failed: actually read %d bytes count not in [0..%lu] valid range", - read_size, (unsigned long)max_size); //never - - return true;//prepare_config(result, remove_empty_lines); - } - if(fail_on_read_problem) - PTHROW(0, 0, - &file_spec, - "read failed: %s (%d)", strerror(errno), errno); - return false; + // later: it seems, even this does not help sometimes + if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) { + try { + if(lock_shared_blocking(f)!=0) + throw Exception("file.lock", + file_spec, + "shared lock failed: %s (%d), actual filename '%s'", + strerror(errno), errno, fname.get()); + + struct stat finfo; + if(stat(fname, &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.get()); + +#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); + } catch(...) { + unlock(f);close(f); + if(fail_on_read_problem) + rethrow; + return false; + } + + unlock(f);close(f); + return true; + } else { + if(fail_on_read_problem) + throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0, + file_spec, + "%s failed: %s (%d), actual filename '%s'", + action_name, strerror(errno), errno, fname.get()); + return false; + } } -void file_write(Pool& pool, - const String& file_spec, - const void *data, size_t size, - bool as_text/*, - bool exclusive*/) { - const char *fname=file_spec.cstr(String::UL_FILE_NAME); +static void create_dir_for_file(StringPtr 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); + pos_after=pos_before+1; + } +} + +bool file_write_action_under_lock( + StringPtr file_spec, + const char* action_name, File_write_action action, void *context, + bool as_text, + bool do_append, + bool do_block, + bool fail_on_lock_problem) { + CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); int f; - if(access(fname, W_OK)!=0) {/*no*/ - 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_NAME), 0775); - pos_after=pos_before+1; + if(access(fname, W_OK)!=0) // no + create_dir_for_file(file_spec); + + if((f=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.get()); + close(f); + if(fail_on_lock_problem) + throw e; + return false; } - if((f=open(fname, O_WRONLY|O_CREAT|_O_BINARY, 0666))>0) - close(f); - } - if(access(fname, R_OK|W_OK)==0) { - int mode=O_RDWR|(as_text?_O_TEXT:_O_BINARY) -#ifdef WIN32 - |O_TRUNC + try { + 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 - ; - if((f=open(fname, mode, 0666))>=0) { - /*if(exclusive) - flock(f, LOCK_EX);*/ - - if(size) write(f, data, size); -#ifndef WIN32 - ftruncate(f, size); -#endif - /*if(exclusive) - flock(f, LOCK_UN);*/ - close(f); - return; + 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); + 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.get()); + // here should be nothing, see rethrow above +} + +#ifndef DOXYGEN +struct File_write_action_info { + const char *data; size_t size; +}; +#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(written<0) + throw Exception(0, + Exception::undefined_source, + "write failed: %s (%d)", strerror(errno), errno); } - PTHROW(0, 0, - &file_spec, - "write failed: %s (%d)", strerror(errno), errno); +} +void file_write( + StringPtr 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); } -static void rmdir(const String& file_spec, size_t pos_after) { +// throws nothing! [this is required in file_move & file_delete] +static void rmdir(StringPtr file_spec, size_t pos_after) { int pos_before; - if((pos_before=file_spec.pos("/", 1, pos_after))>=0) - rmdir(file_spec, pos_before+1); + 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_NAME)); + rmdir(file_spec->mid(0, pos_after-1/* / */)->cstr(String::UL_FILE_SPEC)); } -void file_delete(Pool& pool, const String& file_spec) { - if(unlink(file_spec.cstr(String::UL_FILE_NAME))!=0) - PTHROW(0, 0, - &file_spec, - "unlink failed: %s (%d)", strerror(errno), errno); +bool file_delete(StringPtr file_spec, bool fail_on_read_problem) { + CharPtr 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.get()); + else + return false; - rmdir(file_spec, 1); + 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) { + struct stat lfinfo; + bool result=stat(fname, &lfinfo)==0; + if(afinfo) + *afinfo=lfinfo; + return result; +} -static bool entry_readable(const String& file_spec, bool need_dir) { - const char *fname=file_spec.cstr(String::UL_FILE_NAME); +bool entry_exists(StringPtr file_spec) { + CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); + return entry_exists(fname, 0); +} + +static bool entry_readable(StringPtr file_spec, bool need_dir) { + CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC); + if(need_dir) { + size_t size=strlen(fname); + while(size) { + char c=fname[size-1]; + if(c=='/' || c=='\\') + fname[--size]=0; + else + break; + } + } struct stat finfo; - if(access(fname, R_OK)==0 && stat(fname, &finfo)==0) { - bool is_dir=(bool)(finfo.st_mode&S_IFDIR); + if(access(fname, R_OK)==0 && 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); +bool file_readable(StringPtr file_spec) { + return entry_readable(file_spec, false); } -bool dir_readable(const String& file_spec) { - return entry_readable(file_spec, true); +bool dir_readable(StringPtr file_spec) { + return entry_readable(file_spec, true); } -bool file_executable(const String& file_spec) { - return access(file_spec.cstr(String::UL_FILE_NAME), X_OK)==0; -} - -size_t file_size(const String& file_spec) { - Pool& pool=file_spec.pool(); - const char *fname=file_spec.cstr(String::UL_FILE_NAME); +StringPtr file_readable(String& path, String& name) { + StringPtr result(new String(path)); + *result << "/"; + *result << name; + return file_readable(result)?result:StringPtr(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) - PTHROW(0, 0, - &file_spec, - "write failed: %s (%d)", strerror(errno), errno); - return finfo.st_size; + if(fail_on_read_problem) + throw Exception("file.missing", + file_spec, + "getting file size failed: %s (%d), real filename '%s'", + strerror(errno), errno, fname.get()); + else + return false; + rsize=finfo.st_size; + ratime=finfo.st_atime; + rmtime=finfo.st_mtime; + rctime=finfo.st_ctime; + return true; } -char *getrow(char **row_ref, char delim) { - char *result=*row_ref; +char* getrow(char* *row_ref, char delim) { + char* result=*row_ref; if(result) { - *row_ref=strchr(result, delim); + *row_ref=strchr(result, delim); if(*row_ref) *((*row_ref)++)=0; else if(!*result) @@ -206,9 +759,9 @@ char *getrow(char **row_ref, char delim) return result; } -char *lsplit(char *string, char delim) { +char* lsplit(char* string, char delim) { if(string) { - char *v=strchr(string, delim); + char* v=strchr(string, delim); if(v) { *v=0; return v+1; @@ -217,16 +770,16 @@ char *lsplit(char *string, char delim) { return 0; } -char *lsplit(char **string_ref, char delim) { - char *result=*string_ref; - char *next=lsplit(*string_ref, delim); +char* lsplit(char* *string_ref, char delim) { + char* result=*string_ref; + char* next=lsplit(*string_ref, delim); *string_ref=next; return result; } -char *rsplit(char *string, char delim) { +char* rsplit(char* string, char delim) { if(string) { - char *v=strrchr(string, delim); + char* v=strrchr(string, delim); if(v) { *v=0; return v+1; @@ -236,40 +789,42 @@ char *rsplit(char *string, char delim) { } /// @todo less stupid type detection -char *format(Pool& pool, double value, char *fmt) { - char *result=(char *)pool.malloc(MAX_NUMBER); +const char* format(Pool& pool, double value, char* fmt) { + char local_buf[MAX_NUMBER]; + size_t size; + if(fmt) if(strpbrk(fmt, "diouxX")) if(strpbrk(fmt, "ouxX")) - snprintf(result, MAX_NUMBER, fmt, (uint)value ); + size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value); else - snprintf(result, MAX_NUMBER, fmt, (int)value ); + size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value); else - snprintf(result, MAX_NUMBER, fmt, value); + size=snprintf(local_buf, sizeof(local_buf), fmt, value); else - snprintf(result, MAX_NUMBER, "%d", (int)value); + size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value); - return result; + return pool.copy(local_buf, size+1); } size_t stdout_write(const void *buf, size_t size) { #ifdef WIN32 do{ - int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout); + int chunk_written=fwrite(buf, 1, min(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; #else - return fwrite(buf, 1, size, stdout); + return fwrite(buf, 1, size, stdout); #endif } -const char *unescape_chars(Pool& pool, const char *cp, int len) { - char *s=(char *)pool.malloc(len + 1); +char* unescape_chars(Pool& pool, const char* cp, int len) { + char* s=new(pool) char[len + 1]; enum EscapeState { EscapeRest, EscapeFirst, @@ -279,13 +834,13 @@ const char *unescape_chars(Pool& pool, c int srcPos=0; int dstPos=0; while(srcPos < len) { - int ch=cp[srcPos]; + int ch=cp[srcPos]; switch(escapeState) { case EscapeRest: if(ch=='%') { escapeState=EscapeFirst; } else if(ch=='+') { - s[dstPos++]=' '; + s[dstPos++]=' '; } else { s[dstPos++]=ch; } @@ -295,72 +850,35 @@ const char *unescape_chars(Pool& pool, c escapeState=EscapeSecond; break; case EscapeSecond: - escapedValue +=hex_value[ch]; + escapedValue +=hex_value[ch]; s[dstPos++]=escapedValue; escapeState=EscapeRest; break; } - srcPos++; + srcPos++; } s[dstPos]=0; return s; } -/// used by attributed_meaning_to_string / append_attribute_subattribute -struct Attributed_meaning_info { - String *header; // header line being constructed - String::Untaint_lang lang; // language in which to append to that line -}; -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, true); - *ami.header << "="; - ami.header->append(static_cast(avalue)->as_string(), ami.lang, true); -} -const String& attributed_meaning_to_string(Value& meaning, - String::Untaint_lang lang) { - String &result=*new(meaning.pool()) String(meaning.pool()); - if(Hash *hash=meaning.get_hash()) { - // $value(value) $subattribute(subattribute value) - if(Value *value=static_cast(hash->get(*value_name))) - result.append(value->as_string(), lang, true); - - Attributed_meaning_info attributed_meaning_info={ - &result, - lang - }; - hash->for_each(append_attribute_subattribute, &attributed_meaning_info); - } else // result value - result.append(meaning.as_string(), lang, true); - - return result; -} - #ifdef WIN32 -void back_slashes_to_slashes(char *s) { +void back_slashes_to_slashes(char* s) { if(s) for(; *s; s++) if(*s=='\\') - *s='/'; + *s='/'; } /* -void slashes_to_back_slashes(char *s) { +void slashes_to_back_slashes(char* s) { if(s) for(; *s; s++) if(*s=='/') - *s='\\'; + *s='\\'; } */ #endif -bool StrEqNc(const char *s1, const char *s2, bool strict) { +bool StrEqNc(const char* s1, const char* s2, bool strict) { while(true) { if(!(*s1)) { if(!(*s2)) @@ -374,44 +892,106 @@ bool StrEqNc(const char *s1, const char return false; } else if((*s1) !=(*s2)) return false; - s1++; - s2++; + s1++; + s2++; } } -char *unquote(char*& current, char stop_at) { - char *result=current; - char *dest=current; - // skip leading WS - while(*current==' ' || *current=='\t') - current++; - if(!*current) - return current=0; - - bool quoted=*current=='"'; - if(quoted) - current++; - - for(; *current; ) { - if(quoted) - switch(*current) { - case '\\': // "...\c - if(current[1]) - current++; - break; - case '"': - current++; - goto break2; - } - else - if(*current==stop_at) - break; +static bool isLeap(int year) { + return !( + (year % 4) || ((year % 400) && !(year % 100)) + ); +} + +int getMonthDays(int year, int month) { + int monthDays[]={ + 31, + isLeap(year) ? 29 : 28, + 31, + 30, + 31, + 30, + 31, + 31, + 30, + 31, + 30, + 31 + }; + return monthDays[month]; +} + +void remove_crlf(char* start, char* end) { + for(char* p=start; ps) + r=s; +#endif + b[r]=0; + return r; } + +int __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); + 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 +} + +