Annotation of parser3/src/main/pa_common.C, revision 1.145.2.2
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.143 paf 4: Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)
1.101 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.111 paf 6: */
1.16 paf 7:
1.145.2.2! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/03/17 14:48:05 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.14 paf 12: #include "pa_globals.h"
1.126 paf 13: #include "pa_hash.h"
14: #include "pa_vstring.h"
1.135 paf 15: #include "pa_vdate.h"
1.1 paf 16:
1.98 paf 17: #ifdef WIN32
18: # include <windows.h>
1.126 paf 19: #else
20: # define closesocket close
1.98 paf 21: #endif
22:
1.93 paf 23: // some maybe-undefined constants
24:
1.82 paf 25: #ifndef _O_TEXT
26: # define _O_TEXT 0
27: #endif
28: #ifndef _O_BINARY
29: # define _O_BINARY 0
1.47 paf 30: #endif
1.80 paf 31:
1.138 paf 32: #ifdef HAVE_FTRUNCATE
33: # define PA_O_TRUNC 0
34: #else
35: # ifdef _O_TRUNC
36: # define PA_O_TRUNC _O_TRUNC
37: # else
38: # error you must have either ftruncate function or _O_TRUNC bit declared
39: # endif
40: #endif
41:
1.93 paf 42: // locking constants
43:
1.99 paf 44: #ifdef HAVE_FLOCK
45:
46: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
47: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
48: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
49: static int unlock(int fd) { return flock(fd, LOCK_UN); }
50:
1.98 paf 51: #else
1.99 paf 52: #ifdef HAVE__LOCKING
1.98 paf 53:
1.126 paf 54: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 55: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
56: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
57: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
58: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 59:
1.99 paf 60: #else
61: #ifdef HAVE_FCNTL
1.93 paf 62:
1.126 paf 63: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 64: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
65: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
66: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
67: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 68:
69: #else
70: #ifdef HAVE_LOCKF
1.99 paf 71:
1.126 paf 72: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 73: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
74: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
75: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
76: static int unlock(int fd) { FLOCK(F_TLOCK); }
77:
1.93 paf 78: #else
1.99 paf 79:
80: #error unable to find file locking func
81:
82: #endif
1.93 paf 83: #endif
84: #endif
85: #endif
86:
1.127 paf 87: #define DEFAULT_USER_AGENT "parser3"
88:
89:
1.126 paf 90: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 91: if(size==0)
92: return;
93:
1.87 paf 94: //_asm int 3;
1.126 paf 95: const char* const eob=buf+size;
96: char* dest=buf;
1.72 parser 97: // fix DOS: \r\n -> \n
98: // fix Macintosh: \r -> \n
1.126 paf 99: char* bol=buf;
1.137 paf 100: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 101: size_t len=eol-bol;
102: if(dest!=bol)
1.126 paf 103: memcpy(dest, bol, len);
1.72 parser 104: dest+=len;
1.126 paf 105: *dest++='\n';
1.72 parser 106:
1.126 paf 107: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 108: bol=eol+2;
1.126 paf 109: size--;
110: } else // \r, not \n = Macintosh
1.72 parser 111: bol=eol+1;
112: }
113: // last piece without \r, including terminating 0
114: if(dest!=bol)
1.126 paf 115: memcpy(dest, bol, eob-bol);
1.72 parser 116: }
1.18 paf 117:
1.126 paf 118: char* file_read_text(Pool& pool, const String& file_spec,
119: bool fail_on_read_problem,
120: Hash *params, Hash** out_fields) {
1.72 parser 121: void *result; size_t size;
1.126 paf 122: return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;
123: }
124:
125: //http request stuff
126: /* ************************ http stuff *********************** */
127:
128: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
129: memset(addr, 0, sizeof(*addr));
130: addr->sin_family=AF_INET;
131: addr->sin_port=htons(port);
132: if(host) {
133: if(struct hostent *hostIP=gethostbyname(host))
134: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
135: else
136: return false;
137: } else
138: addr->sin_addr.s_addr=INADDR_ANY;
139: return true;
140: }
141:
1.142 paf 142: static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
143: const String* status_code=0;
1.130 paf 144: ssize_t EOLat=0;
1.126 paf 145: while(true) {
1.145.2.1 paf 146: char *buf=(char *)response.pool().malloc_atomic(MAX_STRING);
1.126 paf 147: ssize_t size=recv(sock, buf, MAX_STRING, 0);
148: if(size<=0)
149: break;
1.130 paf 150: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
1.142 paf 151: if(!status_code && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
152: const String& status_line=response.mid(0, (size_t)EOLat);
153: Array astatus(response.pool());
154: size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
155: status_code=astatus.get_string(1);
156:
157: if(fail_on_status_ne_200 && *status_code!="200")
158: throw Exception("http.status",
159: status_code,
160: "invalid HTTP response status");
161: }
162: }
163: if(status_code)
164: return status_code->as_int();
165: else
166: throw Exception("http.response",
167: 0,
168: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 169: }
170:
171: /* ********************** request *************************** */
172:
173: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 174: # define PA_USE_ALARM
1.126 paf 175: #endif
176:
1.145 paf 177: #ifdef PA_USE_ALARM
1.126 paf 178: static sigjmp_buf timeout_env;
179: static void timeout_handler(int sig){
180: siglongjmp(timeout_env, 1);
181: }
182: #endif
183:
1.142 paf 184: static int http_request(String& response,
1.126 paf 185: const String *origin_string,
186: const char* host, int port,
187: const char* request,
1.142 paf 188: int timeout,
189: bool fail_on_status_ne_200){
1.126 paf 190: if(!host)
191: throw Exception("http.host",
192: origin_string,
193: "zero hostname"); //never
194:
1.145 paf 195: #ifdef PA_USE_ALARM
1.126 paf 196: signal(SIGALRM, timeout_handler);
197: #endif
198: int sock=-1;
1.145 paf 199: #ifdef PA_USE_ALARM
200: if(sigsetjmp(timeout_env, 1)) {
201: // stupid gcc [2.95.4] generated bad code
202: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
203: // rewritten simplier [though duplicating closesocket code]
204: if(sock>=0)
205: closesocket(sock);
206: throw Exception("http.timeout",
207: origin_string,
208: "timeout occured while retrieving document");
209: } else {
210: alarm(timeout);
211: #endif
1.126 paf 212: try {
1.142 paf 213: int result;
1.126 paf 214: struct sockaddr_in dest;
215:
216: if(!set_addr(&dest, host, port))
217: throw Exception("http.host",
218: origin_string,
1.127 paf 219: "can not resolve hostname \"%s\"", host);
1.126 paf 220:
221: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
222: throw Exception("http.connect",
223: origin_string,
1.127 paf 224: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 225: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
226: throw Exception("http.connect",
227: origin_string,
1.127 paf 228: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 229: size_t request_size=strlen(request);
230: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
231: throw Exception("http.connect",
232: origin_string,
1.127 paf 233: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 234:
1.142 paf 235: result=http_read_response(response, sock, fail_on_status_ne_200);
236: closesocket(sock);
1.145 paf 237: #ifdef PA_USE_ALARM
1.142 paf 238: alarm(0);
1.126 paf 239: #endif
1.142 paf 240: return result;
1.126 paf 241: } catch(...) {
242: if(sock>=0)
243: closesocket(sock);
1.145 paf 244: #ifdef PA_USE_ALARM
1.126 paf 245: alarm(0);
246: #endif
247: /*re*/throw;
248: }
249: }
250:
1.127 paf 251: #ifndef DOXYGEN
252: struct Http_pass_header_info {
253: String* request;
254: bool user_agent_specified;
255: };
256: #endif
257: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 258: {
1.127 paf 259: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
260: Pool& pool=i.request->pool();
261:
1.135 paf 262: *(i.request)<<key<<": "
1.136 paf 263: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 264: <<"\n";
265:
1.127 paf 266: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
267: i.user_agent_specified=true;
1.126 paf 268: }
269: static void file_read_http(Pool& pool, const String& file_spec,
270: void*& data, size_t& data_size,
1.127 paf 271: Hash *options=0, Hash** out_fields=0) {
1.126 paf 272: char host[MAX_STRING];
1.129 paf 273: const char* uri;
1.126 paf 274: int port;
275: const char* method="GET";
276: int timeout=2;
1.142 paf 277: bool fail_on_status_ne_200=true;
1.127 paf 278: Value *vheaders=0;
1.126 paf 279:
1.144 paf 280: String& connect_string=*new(pool) String(pool); // must not be local [exception may be reported outside]
1.133 paf 281: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
282: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
283:
284: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
285: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 286: throw Exception(0,
1.133 paf 287: &connect_string,
1.126 paf 288: "does not start with http://"); //never
1.133 paf 289: connect_string_cstr+=7;
1.126 paf 290:
1.133 paf 291: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 292: char* host_uri=lsplit(host, '/');
1.133 paf 293: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 294: char* port_cstr=lsplit(host, ':');
295: char* error_pos=0;
296: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
297:
1.127 paf 298: if(options) {
299: int valid_options=0;
300: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
301: valid_options++;
302: method=vmethod->as_string().cstr();
303: }
304: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
305: valid_options++;
306: timeout=vtimeout->as_int();
307: }
308: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
309: valid_options++;
310: }
1.142 paf 311: if(Value *vany_status=static_cast<Value *>(options->get(*http_any_status_name))) {
312: valid_options++;
313: fail_on_status_ne_200=!vany_status->as_bool();
314: }
315:
1.127 paf 316: if(valid_options!=options->size())
317: throw Exception("parser.runtime",
318: 0,
319: "invalid option passed");
1.133 paf 320: }
1.126 paf 321:
322: //making request
1.127 paf 323: String request(pool);
1.126 paf 324: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 325: bool user_agent_specified=false;
326: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 327: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 328: Http_pass_header_info info={&request};
329: headers->for_each(http_pass_header, &info);
330: user_agent_specified=info.user_agent_specified;
331: } else
332: throw Exception("parser.runtime",
1.133 paf 333: &connect_string,
1.127 paf 334: "headers param must be hash");
335: };
336: if(!user_agent_specified) // defaulting
337: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 338: request<<"\n";
339:
340: //sending request
341: String response(pool);
1.142 paf 342: int status_code=http_request(response,
343: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED),
344: timeout, fail_on_status_ne_200);
1.126 paf 345:
346: //processing results
347: int pos=response.pos("\r\n\r\n", 4);
348: if(pos<1){
349: throw Exception("http.response",
1.133 paf 350: &connect_string,
1.126 paf 351: "bad response from host - no headers found");
352: }
353: String header_block=response.mid(0, pos);
354: String body=response.mid(pos+4, response.size());
355:
356: Array aheaders(pool);
357: Hash& headers=*new(pool) Hash(pool);
358: size_t pos_after_ref=0;
359: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
360:
361: //processing headers
362: for(int i=1;i<aheaders.size();i++) {
363: if(const String *line=aheaders.get_string(i)) {
364: pos=line->pos(": ", 2);
365: if(pos<1)
366: throw Exception("http.response",
1.133 paf 367: &connect_string,
1.126 paf 368: "bad response from host - bad header \"%s\"", line->cstr());
369:
1.128 paf 370: headers.put(
371: line->mid(0, pos).change_case(pool, String::CC_UPPER),
372: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 373: } else
374: throw Exception("http.response",
1.133 paf 375: &connect_string,
1.126 paf 376: "bad response from host - bad headers \"%s\"", header_block.cstr());
377: }
378:
379: // output response
380: data=body.cstr(); data_size=body.size();
1.141 paf 381: if(out_fields) {
1.142 paf 382: headers.put(*file_status_name, new(pool) VInt(pool, status_code));
1.126 paf 383: *out_fields=&headers;
1.141 paf 384: }
1.34 paf 385: }
1.123 paf 386:
387: #ifndef DOXYGEN
388: struct File_read_action_info {
389: void **data; size_t *data_size;
1.126 paf 390: };
1.123 paf 391: #endif
1.126 paf 392: static void file_read_action(Pool& pool,
393: struct stat& finfo,
1.123 paf 394: int f,
1.126 paf 395: const String& file_spec, const char* fname, bool as_text,
1.123 paf 396: void *context) {
1.126 paf 397: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 398: if(size_t to_read_size=(size_t)finfo.st_size) {
1.145.2.1 paf 399: *info.data=pool.malloc_atomic(to_read_size+(as_text?1:0), 3);
1.126 paf 400: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 401:
402: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 403: throw Exception(0,
1.123 paf 404: &file_spec,
405: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 406: *info.data_size, to_read_size);
1.123 paf 407: } else { // empty file
408: if(as_text) {
1.145.2.1 paf 409: *info.data=pool.malloc_atomic(1);
1.123 paf 410: *(char*)(*info.data)=0;
411: } else
412: *info.data=0;
413: *info.data_size=0;
414: return;
415: }
1.126 paf 416: }
417: bool file_read(Pool& pool, const String& file_spec,
418: void*& data, size_t& data_size,
419: bool as_text, Hash *params, Hash** out_fields,
420: bool fail_on_read_problem) {
421: bool result;
422: if(file_spec.starts_with("http://", 7)) {
423: // fail on read problem
424: file_read_http(pool, file_spec, data, data_size, params, out_fields);
425: result=true;
426: } else {
427: File_read_action_info info={&data, &data_size};
428: result=file_read_action_under_lock(pool, file_spec,
429: "read", file_read_action, &info,
430: as_text, fail_on_read_problem);
431: }
1.123 paf 432:
1.126 paf 433: if(result && as_text) {
1.131 paf 434: // UTF-8 signature: EF BB BF
435: if(data_size>=3) {
436: char *in=(char *)data;
437: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
438: (in[2] == '\xBF')) {
439: data=in+3; data_size-=3;// skip prefix
440: }
441: }
442:
1.126 paf 443: fix_line_breaks((char *)(data), data_size);
1.123 paf 444: // note: after fixing
1.126 paf 445: ((char*&)(data))[data_size]=0;
1.123 paf 446: }
1.126 paf 447:
448: return result;
1.123 paf 449: }
450:
451: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 452: const char* action_name, File_read_action action, void *context,
453: bool as_text,
1.123 paf 454: bool fail_on_read_problem) {
1.126 paf 455: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 456: int f;
457:
458: // first open, next stat:
1.45 paf 459: // directory update of NTFS hard links performed on open.
1.33 paf 460: // ex:
461: // a.html:^test[] and b.html hardlink to a.html
462: // user inserts ! before ^test in a.html
1.126 paf 463: // directory entry of b.html in NTFS not updated at once,
1.35 paf 464: // they delay update till open, so we would receive "!^test[" string
465: // if would do stat, next open.
1.123 paf 466: // later: it seems, even this does not help sometimes
1.98 paf 467: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 468: try {
469: if(lock_shared_blocking(f)!=0)
1.126 paf 470: throw Exception("file.lock",
1.123 paf 471: &file_spec,
472: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 473: strerror(errno), errno, fname);
1.123 paf 474:
1.124 paf 475: struct stat finfo;
476: if(stat(fname, &finfo)!=0)
477: throw Exception("file.missing", // hardly possible: we just opened it OK
478: &file_spec,
479: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 480: strerror(errno), errno, fname);
1.124 paf 481:
1.140 paf 482: #ifdef PA_SAFE_MODE
483: if(finfo.st_uid/*foreign?*/!=geteuid()
484: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 485: throw Exception("parser.runtime",
486: &file_spec,
1.140 paf 487: "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
1.126 paf 488: fname);
1.105 paf 489: #endif
1.32 paf 490:
1.126 paf 491: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 492: } catch(...) {
1.126 paf 493: unlock(f);close(f);
1.123 paf 494: if(fail_on_read_problem)
495: /*re*/throw;
496: return false;
497: }
1.87 paf 498:
1.126 paf 499: unlock(f);close(f);
1.72 parser 500: return true;
1.118 paf 501: } else {
502: if(fail_on_read_problem)
1.126 paf 503: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 504: &file_spec,
1.123 paf 505: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 506: action_name, strerror(errno), errno, fname);
1.118 paf 507: return false;
508: }
1.8 paf 509: }
510:
1.63 parser 511: static void create_dir_for_file(const String& file_spec) {
512: size_t pos_after=1;
513: int pos_before;
514: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 515: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 516: pos_after=pos_before+1;
517: }
518: }
519:
1.98 paf 520: bool file_write_action_under_lock(
1.28 paf 521: const String& file_spec,
1.126 paf 522: const char* action_name, File_write_action action, void *context,
523: bool as_text,
524: bool do_append,
525: bool do_block,
1.110 paf 526: bool fail_on_lock_problem) {
1.126 paf 527: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 528: int f;
1.80 paf 529: if(access(fname, W_OK)!=0) // no
1.126 paf 530: create_dir_for_file(file_spec);
1.50 paf 531:
1.80 paf 532: if((f=open(fname,
533: O_CREAT|O_RDWR
534: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 535: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 536: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 537: Exception e("file.lock",
1.110 paf 538: &file_spec,
539: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 540: strerror(errno), errno, fname);
541: close(f);
1.110 paf 542: if(fail_on_lock_problem)
543: throw e;
1.98 paf 544: return false;
545: }
1.96 paf 546:
547: try {
1.126 paf 548: action(f, context);
1.96 paf 549: } catch(...) {
1.138 paf 550: #ifdef HAVE_FTRUNCATE
1.104 paf 551: if(!do_append)
1.125 paf 552: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 553: #endif
1.126 paf 554: unlock(f);close(f);
1.96 paf 555: /*re*/throw;
556: }
1.80 paf 557:
1.138 paf 558: #ifdef HAVE_FTRUNCATE
1.104 paf 559: if(!do_append)
1.125 paf 560: ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
1.138 paf 561: #endif
1.126 paf 562: unlock(f);close(f);
1.98 paf 563: return true;
1.80 paf 564: } else
1.126 paf 565: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 566: &file_spec,
1.96 paf 567: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 568: action_name, strerror(errno), errno, fname);
1.96 paf 569: // here should be nothing, see rethrow above
570: }
571:
572: #ifndef DOXYGEN
573: struct File_write_action_info {
574: const void *data; size_t size;
1.126 paf 575: };
1.96 paf 576: #endif
577: static void file_write_action(int f, void *context) {
1.126 paf 578: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 579: if(info.size) {
1.126 paf 580: int written=write(f, info.data, info.size);
1.116 paf 581: if(written<0)
1.126 paf 582: throw Exception(0,
583: 0,
584: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 585: }
1.96 paf 586: }
587: void file_write(
588: const String& file_spec,
589: const void *data, size_t size,
1.126 paf 590: bool as_text,
1.96 paf 591: bool do_append) {
1.126 paf 592: File_write_action_info info={data, size};
1.98 paf 593: file_write_action_under_lock(
1.96 paf 594: file_spec,
1.126 paf 595: "write", file_write_action, &info,
596: as_text,
597: do_append);
1.30 paf 598: }
599:
1.63 parser 600: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 601: static void rmdir(const String& file_spec, size_t pos_after) {
602: int pos_before;
603: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 604: rmdir(file_spec, pos_before+1);
1.50 paf 605:
1.126 paf 606: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 607: }
1.95 paf 608: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 609: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 610: if(unlink(fname)!=0)
1.93 paf 611: if(fail_on_read_problem)
1.126 paf 612: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 613: &file_spec,
614: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 615: strerror(errno), errno, fname);
1.93 paf 616: else
617: return false;
1.50 paf 618:
1.126 paf 619: rmdir(file_spec, 1);
1.93 paf 620: return true;
1.60 parser 621: }
1.95 paf 622: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 623: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
624: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 625:
1.126 paf 626: create_dir_for_file(new_spec);
1.63 parser 627:
1.60 parser 628: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 629: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 630: &old_spec,
631: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 632: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 633:
1.126 paf 634: rmdir(old_spec, 1);
1.31 paf 635: }
636:
1.51 paf 637:
1.126 paf 638: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 639: struct stat lfinfo;
640: bool result=stat(fname, &lfinfo)==0;
641: if(afinfo)
642: *afinfo=lfinfo;
643: return result;
1.119 paf 644: }
645:
646: bool entry_exists(const String& file_spec) {
1.126 paf 647: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
648: return entry_exists(fname, 0);
1.118 paf 649: }
650:
1.51 paf 651: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 652: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 653: if(need_dir) {
1.126 paf 654: size_t size=strlen(fname);
1.120 paf 655: while(size) {
1.126 paf 656: char c=fname[size-1];
1.120 paf 657: if(c=='/' || c=='\\')
658: fname[--size]=0;
659: else
660: break;
661: }
662: }
1.51 paf 663: struct stat finfo;
1.118 paf 664: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 665: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 666: return is_dir==need_dir;
667: }
668: return false;
669: }
1.31 paf 670: bool file_readable(const String& file_spec) {
1.126 paf 671: return entry_readable(file_spec, false);
1.51 paf 672: }
673: bool dir_readable(const String& file_spec) {
1.126 paf 674: return entry_readable(file_spec, true);
1.65 parser 675: }
676: String *file_readable(const String& path, const String& name) {
1.126 paf 677: String *result=new(path.pool()) String(path);
678: *result << "/";
1.65 parser 679: *result << name;
680: return file_readable(*result)?result:0;
1.43 paf 681: }
682: bool file_executable(const String& file_spec) {
1.64 parser 683: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 684: }
685:
1.64 parser 686: bool file_stat(const String& file_spec,
1.58 parser 687: size_t& rsize,
1.126 paf 688: time_t& ratime,
689: time_t& rmtime,
690: time_t& rctime,
1.64 parser 691: bool fail_on_read_problem) {
1.126 paf 692: Pool& pool=file_spec.pool();
693: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 694: struct stat finfo;
695: if(stat(fname, &finfo)!=0)
1.64 parser 696: if(fail_on_read_problem)
1.126 paf 697: throw Exception("file.missing",
1.67 parser 698: &file_spec,
699: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 700: strerror(errno), errno, fname);
1.64 parser 701: else
702: return false;
1.58 parser 703: rsize=finfo.st_size;
704: ratime=finfo.st_atime;
705: rmtime=finfo.st_mtime;
706: rctime=finfo.st_ctime;
1.64 parser 707: return true;
1.18 paf 708: }
709:
1.126 paf 710: char* getrow(char* *row_ref, char delim) {
711: char* result=*row_ref;
1.8 paf 712: if(result) {
1.126 paf 713: *row_ref=strchr(result, delim);
1.8 paf 714: if(*row_ref)
715: *((*row_ref)++)=0;
716: else if(!*result)
717: return 0;
718: }
719: return result;
720: }
721:
1.126 paf 722: char* lsplit(char* string, char delim) {
1.23 paf 723: if(string) {
1.126 paf 724: char* v=strchr(string, delim);
1.8 paf 725: if(v) {
726: *v=0;
727: return v+1;
728: }
729: }
730: return 0;
731: }
732:
1.126 paf 733: char* lsplit(char* *string_ref, char delim) {
734: char* result=*string_ref;
735: char* next=lsplit(*string_ref, delim);
1.8 paf 736: *string_ref=next;
737: return result;
1.9 paf 738: }
739:
1.126 paf 740: char* rsplit(char* string, char delim) {
1.18 paf 741: if(string) {
1.126 paf 742: char* v=strrchr(string, delim);
1.18 paf 743: if(v) {
1.9 paf 744: *v=0;
745: return v+1;
746: }
747: }
748: return NULL;
1.10 paf 749: }
750:
1.37 paf 751: /// @todo less stupid type detection
1.126 paf 752: char* format(Pool& pool, double value, char* fmt) {
753: char local_buf[MAX_NUMBER];
1.108 paf 754: size_t size;
755:
1.10 paf 756: if(fmt)
757: if(strpbrk(fmt, "diouxX"))
758: if(strpbrk(fmt, "ouxX"))
1.126 paf 759: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 760: else
1.126 paf 761: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 762: else
1.126 paf 763: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 764: else
1.126 paf 765: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 766:
1.145.2.1 paf 767: char* pool_buf=(char *)pool.malloc_atomic(size+1, 4);
1.126 paf 768: memcpy(pool_buf, local_buf, size+1);
1.108 paf 769: return pool_buf;
1.12 paf 770: }
771:
1.36 paf 772: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 773: #ifdef WIN32
774: do{
1.126 paf 775: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 776: if(chunk_written<=0)
777: break;
778: size-=chunk_written;
1.36 paf 779: buf=((const char*)buf)+chunk_written;
1.126 paf 780: } while(size>0);
1.12 paf 781:
782: return size;
783: #else
1.126 paf 784: return fwrite(buf, 1, size, stdout);
1.12 paf 785: #endif
1.2 paf 786: }
1.14 paf 787:
1.126 paf 788: char* unescape_chars(Pool& pool, const char* cp, int len) {
1.145.2.1 paf 789: char* s=(char *)pool.malloc_atomic(len + 1, 5);
1.14 paf 790: enum EscapeState {
1.33 paf 791: EscapeRest,
792: EscapeFirst,
1.14 paf 793: EscapeSecond
794: } escapeState=EscapeRest;
795: int escapedValue=0;
796: int srcPos=0;
797: int dstPos=0;
798: while(srcPos < len) {
1.126 paf 799: int ch=cp[srcPos];
1.14 paf 800: switch(escapeState) {
801: case EscapeRest:
802: if(ch=='%') {
803: escapeState=EscapeFirst;
804: } else if(ch=='+') {
1.126 paf 805: s[dstPos++]=' ';
1.14 paf 806: } else {
807: s[dstPos++]=ch;
808: }
809: break;
810: case EscapeFirst:
811: escapedValue=hex_value[ch] << 4;
812: escapeState=EscapeSecond;
813: break;
814: case EscapeSecond:
1.126 paf 815: escapedValue +=hex_value[ch];
1.14 paf 816: s[dstPos++]=escapedValue;
817: escapeState=EscapeRest;
818: break;
819: }
1.126 paf 820: srcPos++;
1.14 paf 821: }
822: s[dstPos]=0;
823: return s;
1.24 paf 824: }
825:
826: #ifdef WIN32
1.126 paf 827: void back_slashes_to_slashes(char* s) {
1.24 paf 828: if(s)
829: for(; *s; s++)
830: if(*s=='\\')
1.126 paf 831: *s='/';
1.24 paf 832: }
1.42 paf 833: /*
1.126 paf 834: void slashes_to_back_slashes(char* s) {
1.42 paf 835: if(s)
836: for(; *s; s++)
837: if(*s=='/')
1.126 paf 838: *s='\\';
1.42 paf 839: }
840: */
1.24 paf 841: #endif
1.41 paf 842:
1.126 paf 843: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 844: while(true) {
845: if(!(*s1)) {
846: if(!(*s2))
847: return true;
848: else
849: return !strict;
850: } else if(!(*s2))
851: return !strict;
852: if(isalpha(*s1)) {
853: if(tolower(*s1) !=tolower(*s2))
854: return false;
855: } else if((*s1) !=(*s2))
856: return false;
1.126 paf 857: s1++;
858: s2++;
1.41 paf 859: }
1.57 parser 860: }
861:
1.84 paf 862: static bool isLeap(int year) {
1.57 parser 863: return !(
864: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 865: );
1.57 parser 866: }
867:
868: int getMonthDays(int year, int month) {
869: int monthDays[]={
1.126 paf 870: 31,
871: isLeap(year) ? 29 : 28,
872: 31,
873: 30,
874: 31,
875: 30,
876: 31,
877: 31,
878: 30,
879: 31,
880: 30,
1.57 parser 881: 31
1.126 paf 882: };
883: return monthDays[month];
1.41 paf 884: }
1.69 parser 885:
1.126 paf 886: void remove_crlf(char* start, char* end) {
887: for(char* p=start; p<end; p++)
1.69 parser 888: switch(*p) {
1.126 paf 889: case '\n': *p='|'; break;
890: case '\r': *p=' '; break;
1.69 parser 891: }
1.91 paf 892: }
893:
894:
895: /// must be last in this file
896: #undef vsnprintf
1.126 paf 897: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 898: if(!s)
899: return 0;
900:
901: int r;
902: // note: on win32& maybe somewhere else
903: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
904: --s;
905: #if _MSC_VER
906: /*
907: win32:
908: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
909:
910: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
911: */
1.126 paf 912: r=_vsnprintf(b, s, f, l);
1.91 paf 913: if(r<0)
914: r=s;
915: #else
1.126 paf 916: r=vsnprintf(b, s, f, l);
1.91 paf 917: /*
918: solaris:
919: man vsnprintf
920:
921: The snprintf() function returns the number of characters
922: formatted, that is, the number of characters that would have
923: been written to the buffer if it were large enough. If the
924: value of n is 0 on a call to snprintf(), an unspecified
925: value less than 1 is returned.
926: */
927:
928: if(r<0)
929: r=0;
930: else if(r>s)
931: r=s;
932: #endif
933: b[r]=0;
934: return r;
935: }
936:
1.126 paf 937: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 938: va_list l;
1.126 paf 939: va_start(l, f);
940: int r=__vsnprintf(b, s, f, l);
941: va_end(l);
1.91 paf 942: return r;
1.98 paf 943: }
944:
945: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 946: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 947:
948: #ifdef WIN32
1.126 paf 949: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 950: return 0;
951: #else
952: struct timeval t;
953: t.tv_sec = secs;
954: t.tv_usec = usecs;
1.126 paf 955: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 956: #endif
1.135 paf 957: }
958:
959:
960: // attributed meaning
961:
962: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
963: const char month_names[12][4]={
964: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
965: const char days[7][4]={
966: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
967:
968: time_t when=vdate.get_time();
969: struct tm *tms=gmtime(&when);
970: if(!tms)
971: throw Exception(0,
972: 0,
973: "bad time in attribute value (seconds from epoch=%ld)", when);
974: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
975: days[tms->tm_wday],
976: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
977: tms->tm_hour,tms->tm_min,tms->tm_sec);
978: }
979: static void append_attribute_meaning(String& result,
1.136 paf 980: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 981: if(const String *string=value.get_string())
1.136 paf 982: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 983: else
984: if(Value *vdate=value.as(VDATE_TYPE, false)) {
1.145.2.1 paf 985: char *buf=(char *)result.malloc_atomic(MAX_STRING);
1.135 paf 986: size_t size=date_attribute(*static_cast<VDate *>(vdate),
987: buf, MAX_STRING);
988:
989: result.APPEND_CLEAN(buf, size, "converted from date", 0);
990: } else
991: throw Exception("parser.runtime",
992: &result,
993: "trying to append here neither string nor date (%s)",
994: value.type());
995: }
996: #ifndef DOXYGEN
997: struct Attributed_meaning_info {
998: String *header; // header line being constructed
999: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 1000: bool forced; // do they force that lang?
1.135 paf 1001: };
1002: #endif
1003: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
1004: void *info) {
1005: if(akey==VALUE_NAME)
1006: return;
1007:
1008: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
1009:
1010: // ...; charset=windows1251
1011: *ami.header << "; ";
1.136 paf 1012: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1013: *ami.header << "=";
1.136 paf 1014: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1015: }
1016: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1017: String::Untaint_lang lang, bool forced) {
1.135 paf 1018: String &result=*new(meaning.pool()) String(meaning.pool());
1019: if(Hash *hash=meaning.get_hash(0)) {
1020: // $value(value) $subattribute(subattribute value)
1021: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1022: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1023:
1024: Attributed_meaning_info attributed_meaning_info={&result, lang};
1025: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1026: } else // result value
1.136 paf 1027: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1028:
1029: return result;
1.74 parser 1030: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_common.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main