Annotation of parser3/src/main/pa_common.C, revision 1.145
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 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/01/23 08:20:39 $";
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) {
146: char *buf=(char *)response.pool().malloc(MAX_STRING);
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: }
240: #endif
1.142 paf 241: return result;
1.126 paf 242: } catch(...) {
243: if(sock>=0)
244: closesocket(sock);
1.145 ! paf 245: #ifdef PA_USE_ALARM
1.126 paf 246: alarm(0);
247: #endif
248: /*re*/throw;
249: }
250: }
251:
1.127 paf 252: #ifndef DOXYGEN
253: struct Http_pass_header_info {
254: String* request;
255: bool user_agent_specified;
256: };
257: #endif
258: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 259: {
1.127 paf 260: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
261: Pool& pool=i.request->pool();
262:
1.135 paf 263: *(i.request)<<key<<": "
1.136 paf 264: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 265: <<"\n";
266:
1.127 paf 267: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
268: i.user_agent_specified=true;
1.126 paf 269: }
270: static void file_read_http(Pool& pool, const String& file_spec,
271: void*& data, size_t& data_size,
1.127 paf 272: Hash *options=0, Hash** out_fields=0) {
1.126 paf 273: char host[MAX_STRING];
1.129 paf 274: const char* uri;
1.126 paf 275: int port;
276: const char* method="GET";
277: int timeout=2;
1.142 paf 278: bool fail_on_status_ne_200=true;
1.127 paf 279: Value *vheaders=0;
1.126 paf 280:
1.144 paf 281: String& connect_string=*new(pool) String(pool); // must not be local [exception may be reported outside]
1.133 paf 282: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
283: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
284:
285: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
286: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 287: throw Exception(0,
1.133 paf 288: &connect_string,
1.126 paf 289: "does not start with http://"); //never
1.133 paf 290: connect_string_cstr+=7;
1.126 paf 291:
1.133 paf 292: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 293: char* host_uri=lsplit(host, '/');
1.133 paf 294: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 295: char* port_cstr=lsplit(host, ':');
296: char* error_pos=0;
297: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
298:
1.127 paf 299: if(options) {
300: int valid_options=0;
301: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
302: valid_options++;
303: method=vmethod->as_string().cstr();
304: }
305: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
306: valid_options++;
307: timeout=vtimeout->as_int();
308: }
309: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
310: valid_options++;
311: }
1.142 paf 312: if(Value *vany_status=static_cast<Value *>(options->get(*http_any_status_name))) {
313: valid_options++;
314: fail_on_status_ne_200=!vany_status->as_bool();
315: }
316:
1.127 paf 317: if(valid_options!=options->size())
318: throw Exception("parser.runtime",
319: 0,
320: "invalid option passed");
1.133 paf 321: }
1.126 paf 322:
323: //making request
1.127 paf 324: String request(pool);
1.126 paf 325: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 326: bool user_agent_specified=false;
327: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 328: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 329: Http_pass_header_info info={&request};
330: headers->for_each(http_pass_header, &info);
331: user_agent_specified=info.user_agent_specified;
332: } else
333: throw Exception("parser.runtime",
1.133 paf 334: &connect_string,
1.127 paf 335: "headers param must be hash");
336: };
337: if(!user_agent_specified) // defaulting
338: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 339: request<<"\n";
340:
341: //sending request
342: String response(pool);
1.142 paf 343: int status_code=http_request(response,
344: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED),
345: timeout, fail_on_status_ne_200);
1.126 paf 346:
347: //processing results
348: int pos=response.pos("\r\n\r\n", 4);
349: if(pos<1){
350: throw Exception("http.response",
1.133 paf 351: &connect_string,
1.126 paf 352: "bad response from host - no headers found");
353: }
354: String header_block=response.mid(0, pos);
355: String body=response.mid(pos+4, response.size());
356:
357: Array aheaders(pool);
358: Hash& headers=*new(pool) Hash(pool);
359: size_t pos_after_ref=0;
360: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
361:
362: //processing headers
363: for(int i=1;i<aheaders.size();i++) {
364: if(const String *line=aheaders.get_string(i)) {
365: pos=line->pos(": ", 2);
366: if(pos<1)
367: throw Exception("http.response",
1.133 paf 368: &connect_string,
1.126 paf 369: "bad response from host - bad header \"%s\"", line->cstr());
370:
1.128 paf 371: headers.put(
372: line->mid(0, pos).change_case(pool, String::CC_UPPER),
373: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 374: } else
375: throw Exception("http.response",
1.133 paf 376: &connect_string,
1.126 paf 377: "bad response from host - bad headers \"%s\"", header_block.cstr());
378: }
379:
380: // output response
381: data=body.cstr(); data_size=body.size();
1.141 paf 382: if(out_fields) {
1.142 paf 383: headers.put(*file_status_name, new(pool) VInt(pool, status_code));
1.126 paf 384: *out_fields=&headers;
1.141 paf 385: }
1.34 paf 386: }
1.123 paf 387:
388: #ifndef DOXYGEN
389: struct File_read_action_info {
390: void **data; size_t *data_size;
1.126 paf 391: };
1.123 paf 392: #endif
1.126 paf 393: static void file_read_action(Pool& pool,
394: struct stat& finfo,
1.123 paf 395: int f,
1.126 paf 396: const String& file_spec, const char* fname, bool as_text,
1.123 paf 397: void *context) {
1.126 paf 398: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 399: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 400: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
401: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 402:
403: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 404: throw Exception(0,
1.123 paf 405: &file_spec,
406: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 407: *info.data_size, to_read_size);
1.123 paf 408: } else { // empty file
409: if(as_text) {
1.126 paf 410: *info.data=pool.malloc(1);
1.123 paf 411: *(char*)(*info.data)=0;
412: } else
413: *info.data=0;
414: *info.data_size=0;
415: return;
416: }
1.126 paf 417: }
418: bool file_read(Pool& pool, const String& file_spec,
419: void*& data, size_t& data_size,
420: bool as_text, Hash *params, Hash** out_fields,
421: bool fail_on_read_problem) {
422: bool result;
423: if(file_spec.starts_with("http://", 7)) {
424: // fail on read problem
425: file_read_http(pool, file_spec, data, data_size, params, out_fields);
426: result=true;
427: } else {
428: File_read_action_info info={&data, &data_size};
429: result=file_read_action_under_lock(pool, file_spec,
430: "read", file_read_action, &info,
431: as_text, fail_on_read_problem);
432: }
1.123 paf 433:
1.126 paf 434: if(result && as_text) {
1.131 paf 435: // UTF-8 signature: EF BB BF
436: if(data_size>=3) {
437: char *in=(char *)data;
438: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
439: (in[2] == '\xBF')) {
440: data=in+3; data_size-=3;// skip prefix
441: }
442: }
443:
1.126 paf 444: fix_line_breaks((char *)(data), data_size);
1.123 paf 445: // note: after fixing
1.126 paf 446: ((char*&)(data))[data_size]=0;
1.123 paf 447: }
1.126 paf 448:
449: return result;
1.123 paf 450: }
451:
452: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 453: const char* action_name, File_read_action action, void *context,
454: bool as_text,
1.123 paf 455: bool fail_on_read_problem) {
1.126 paf 456: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 457: int f;
458:
459: // first open, next stat:
1.45 paf 460: // directory update of NTFS hard links performed on open.
1.33 paf 461: // ex:
462: // a.html:^test[] and b.html hardlink to a.html
463: // user inserts ! before ^test in a.html
1.126 paf 464: // directory entry of b.html in NTFS not updated at once,
1.35 paf 465: // they delay update till open, so we would receive "!^test[" string
466: // if would do stat, next open.
1.123 paf 467: // later: it seems, even this does not help sometimes
1.98 paf 468: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 469: try {
470: if(lock_shared_blocking(f)!=0)
1.126 paf 471: throw Exception("file.lock",
1.123 paf 472: &file_spec,
473: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 474: strerror(errno), errno, fname);
1.123 paf 475:
1.124 paf 476: struct stat finfo;
477: if(stat(fname, &finfo)!=0)
478: throw Exception("file.missing", // hardly possible: we just opened it OK
479: &file_spec,
480: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 481: strerror(errno), errno, fname);
1.124 paf 482:
1.140 paf 483: #ifdef PA_SAFE_MODE
484: if(finfo.st_uid/*foreign?*/!=geteuid()
485: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 486: throw Exception("parser.runtime",
487: &file_spec,
1.140 paf 488: "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 489: fname);
1.105 paf 490: #endif
1.32 paf 491:
1.126 paf 492: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 493: } catch(...) {
1.126 paf 494: unlock(f);close(f);
1.123 paf 495: if(fail_on_read_problem)
496: /*re*/throw;
497: return false;
498: }
1.87 paf 499:
1.126 paf 500: unlock(f);close(f);
1.72 parser 501: return true;
1.118 paf 502: } else {
503: if(fail_on_read_problem)
1.126 paf 504: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 505: &file_spec,
1.123 paf 506: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 507: action_name, strerror(errno), errno, fname);
1.118 paf 508: return false;
509: }
1.8 paf 510: }
511:
1.63 parser 512: static void create_dir_for_file(const String& file_spec) {
513: size_t pos_after=1;
514: int pos_before;
515: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 516: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 517: pos_after=pos_before+1;
518: }
519: }
520:
1.98 paf 521: bool file_write_action_under_lock(
1.28 paf 522: const String& file_spec,
1.126 paf 523: const char* action_name, File_write_action action, void *context,
524: bool as_text,
525: bool do_append,
526: bool do_block,
1.110 paf 527: bool fail_on_lock_problem) {
1.126 paf 528: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 529: int f;
1.80 paf 530: if(access(fname, W_OK)!=0) // no
1.126 paf 531: create_dir_for_file(file_spec);
1.50 paf 532:
1.80 paf 533: if((f=open(fname,
534: O_CREAT|O_RDWR
535: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 536: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 537: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 538: Exception e("file.lock",
1.110 paf 539: &file_spec,
540: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 541: strerror(errno), errno, fname);
542: close(f);
1.110 paf 543: if(fail_on_lock_problem)
544: throw e;
1.98 paf 545: return false;
546: }
1.96 paf 547:
548: try {
1.126 paf 549: action(f, context);
1.96 paf 550: } catch(...) {
1.138 paf 551: #ifdef HAVE_FTRUNCATE
1.104 paf 552: if(!do_append)
1.125 paf 553: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 554: #endif
1.126 paf 555: unlock(f);close(f);
1.96 paf 556: /*re*/throw;
557: }
1.80 paf 558:
1.138 paf 559: #ifdef HAVE_FTRUNCATE
1.104 paf 560: if(!do_append)
1.125 paf 561: 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 562: #endif
1.126 paf 563: unlock(f);close(f);
1.98 paf 564: return true;
1.80 paf 565: } else
1.126 paf 566: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 567: &file_spec,
1.96 paf 568: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 569: action_name, strerror(errno), errno, fname);
1.96 paf 570: // here should be nothing, see rethrow above
571: }
572:
573: #ifndef DOXYGEN
574: struct File_write_action_info {
575: const void *data; size_t size;
1.126 paf 576: };
1.96 paf 577: #endif
578: static void file_write_action(int f, void *context) {
1.126 paf 579: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 580: if(info.size) {
1.126 paf 581: int written=write(f, info.data, info.size);
1.116 paf 582: if(written<0)
1.126 paf 583: throw Exception(0,
584: 0,
585: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 586: }
1.96 paf 587: }
588: void file_write(
589: const String& file_spec,
590: const void *data, size_t size,
1.126 paf 591: bool as_text,
1.96 paf 592: bool do_append) {
1.126 paf 593: File_write_action_info info={data, size};
1.98 paf 594: file_write_action_under_lock(
1.96 paf 595: file_spec,
1.126 paf 596: "write", file_write_action, &info,
597: as_text,
598: do_append);
1.30 paf 599: }
600:
1.63 parser 601: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 602: static void rmdir(const String& file_spec, size_t pos_after) {
603: int pos_before;
604: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 605: rmdir(file_spec, pos_before+1);
1.50 paf 606:
1.126 paf 607: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 608: }
1.95 paf 609: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 610: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 611: if(unlink(fname)!=0)
1.93 paf 612: if(fail_on_read_problem)
1.126 paf 613: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 614: &file_spec,
615: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 616: strerror(errno), errno, fname);
1.93 paf 617: else
618: return false;
1.50 paf 619:
1.126 paf 620: rmdir(file_spec, 1);
1.93 paf 621: return true;
1.60 parser 622: }
1.95 paf 623: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 624: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
625: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 626:
1.126 paf 627: create_dir_for_file(new_spec);
1.63 parser 628:
1.60 parser 629: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 630: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 631: &old_spec,
632: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 633: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 634:
1.126 paf 635: rmdir(old_spec, 1);
1.31 paf 636: }
637:
1.51 paf 638:
1.126 paf 639: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 640: struct stat lfinfo;
641: bool result=stat(fname, &lfinfo)==0;
642: if(afinfo)
643: *afinfo=lfinfo;
644: return result;
1.119 paf 645: }
646:
647: bool entry_exists(const String& file_spec) {
1.126 paf 648: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
649: return entry_exists(fname, 0);
1.118 paf 650: }
651:
1.51 paf 652: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 653: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 654: if(need_dir) {
1.126 paf 655: size_t size=strlen(fname);
1.120 paf 656: while(size) {
1.126 paf 657: char c=fname[size-1];
1.120 paf 658: if(c=='/' || c=='\\')
659: fname[--size]=0;
660: else
661: break;
662: }
663: }
1.51 paf 664: struct stat finfo;
1.118 paf 665: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 666: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 667: return is_dir==need_dir;
668: }
669: return false;
670: }
1.31 paf 671: bool file_readable(const String& file_spec) {
1.126 paf 672: return entry_readable(file_spec, false);
1.51 paf 673: }
674: bool dir_readable(const String& file_spec) {
1.126 paf 675: return entry_readable(file_spec, true);
1.65 parser 676: }
677: String *file_readable(const String& path, const String& name) {
1.126 paf 678: String *result=new(path.pool()) String(path);
679: *result << "/";
1.65 parser 680: *result << name;
681: return file_readable(*result)?result:0;
1.43 paf 682: }
683: bool file_executable(const String& file_spec) {
1.64 parser 684: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 685: }
686:
1.64 parser 687: bool file_stat(const String& file_spec,
1.58 parser 688: size_t& rsize,
1.126 paf 689: time_t& ratime,
690: time_t& rmtime,
691: time_t& rctime,
1.64 parser 692: bool fail_on_read_problem) {
1.126 paf 693: Pool& pool=file_spec.pool();
694: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 695: struct stat finfo;
696: if(stat(fname, &finfo)!=0)
1.64 parser 697: if(fail_on_read_problem)
1.126 paf 698: throw Exception("file.missing",
1.67 parser 699: &file_spec,
700: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 701: strerror(errno), errno, fname);
1.64 parser 702: else
703: return false;
1.58 parser 704: rsize=finfo.st_size;
705: ratime=finfo.st_atime;
706: rmtime=finfo.st_mtime;
707: rctime=finfo.st_ctime;
1.64 parser 708: return true;
1.18 paf 709: }
710:
1.126 paf 711: char* getrow(char* *row_ref, char delim) {
712: char* result=*row_ref;
1.8 paf 713: if(result) {
1.126 paf 714: *row_ref=strchr(result, delim);
1.8 paf 715: if(*row_ref)
716: *((*row_ref)++)=0;
717: else if(!*result)
718: return 0;
719: }
720: return result;
721: }
722:
1.126 paf 723: char* lsplit(char* string, char delim) {
1.23 paf 724: if(string) {
1.126 paf 725: char* v=strchr(string, delim);
1.8 paf 726: if(v) {
727: *v=0;
728: return v+1;
729: }
730: }
731: return 0;
732: }
733:
1.126 paf 734: char* lsplit(char* *string_ref, char delim) {
735: char* result=*string_ref;
736: char* next=lsplit(*string_ref, delim);
1.8 paf 737: *string_ref=next;
738: return result;
1.9 paf 739: }
740:
1.126 paf 741: char* rsplit(char* string, char delim) {
1.18 paf 742: if(string) {
1.126 paf 743: char* v=strrchr(string, delim);
1.18 paf 744: if(v) {
1.9 paf 745: *v=0;
746: return v+1;
747: }
748: }
749: return NULL;
1.10 paf 750: }
751:
1.37 paf 752: /// @todo less stupid type detection
1.126 paf 753: char* format(Pool& pool, double value, char* fmt) {
754: char local_buf[MAX_NUMBER];
1.108 paf 755: size_t size;
756:
1.10 paf 757: if(fmt)
758: if(strpbrk(fmt, "diouxX"))
759: if(strpbrk(fmt, "ouxX"))
1.126 paf 760: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 761: else
1.126 paf 762: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 763: else
1.126 paf 764: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 765: else
1.126 paf 766: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 767:
1.126 paf 768: char* pool_buf=(char *)pool.malloc(size+1, 4);
769: memcpy(pool_buf, local_buf, size+1);
1.108 paf 770: return pool_buf;
1.12 paf 771: }
772:
1.36 paf 773: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 774: #ifdef WIN32
775: do{
1.126 paf 776: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 777: if(chunk_written<=0)
778: break;
779: size-=chunk_written;
1.36 paf 780: buf=((const char*)buf)+chunk_written;
1.126 paf 781: } while(size>0);
1.12 paf 782:
783: return size;
784: #else
1.126 paf 785: return fwrite(buf, 1, size, stdout);
1.12 paf 786: #endif
1.2 paf 787: }
1.14 paf 788:
1.126 paf 789: char* unescape_chars(Pool& pool, const char* cp, int len) {
790: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 791: enum EscapeState {
1.33 paf 792: EscapeRest,
793: EscapeFirst,
1.14 paf 794: EscapeSecond
795: } escapeState=EscapeRest;
796: int escapedValue=0;
797: int srcPos=0;
798: int dstPos=0;
799: while(srcPos < len) {
1.126 paf 800: int ch=cp[srcPos];
1.14 paf 801: switch(escapeState) {
802: case EscapeRest:
803: if(ch=='%') {
804: escapeState=EscapeFirst;
805: } else if(ch=='+') {
1.126 paf 806: s[dstPos++]=' ';
1.14 paf 807: } else {
808: s[dstPos++]=ch;
809: }
810: break;
811: case EscapeFirst:
812: escapedValue=hex_value[ch] << 4;
813: escapeState=EscapeSecond;
814: break;
815: case EscapeSecond:
1.126 paf 816: escapedValue +=hex_value[ch];
1.14 paf 817: s[dstPos++]=escapedValue;
818: escapeState=EscapeRest;
819: break;
820: }
1.126 paf 821: srcPos++;
1.14 paf 822: }
823: s[dstPos]=0;
824: return s;
1.24 paf 825: }
826:
827: #ifdef WIN32
1.126 paf 828: void back_slashes_to_slashes(char* s) {
1.24 paf 829: if(s)
830: for(; *s; s++)
831: if(*s=='\\')
1.126 paf 832: *s='/';
1.24 paf 833: }
1.42 paf 834: /*
1.126 paf 835: void slashes_to_back_slashes(char* s) {
1.42 paf 836: if(s)
837: for(; *s; s++)
838: if(*s=='/')
1.126 paf 839: *s='\\';
1.42 paf 840: }
841: */
1.24 paf 842: #endif
1.41 paf 843:
1.126 paf 844: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 845: while(true) {
846: if(!(*s1)) {
847: if(!(*s2))
848: return true;
849: else
850: return !strict;
851: } else if(!(*s2))
852: return !strict;
853: if(isalpha(*s1)) {
854: if(tolower(*s1) !=tolower(*s2))
855: return false;
856: } else if((*s1) !=(*s2))
857: return false;
1.126 paf 858: s1++;
859: s2++;
1.41 paf 860: }
1.57 parser 861: }
862:
1.84 paf 863: static bool isLeap(int year) {
1.57 parser 864: return !(
865: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 866: );
1.57 parser 867: }
868:
869: int getMonthDays(int year, int month) {
870: int monthDays[]={
1.126 paf 871: 31,
872: isLeap(year) ? 29 : 28,
873: 31,
874: 30,
875: 31,
876: 30,
877: 31,
878: 31,
879: 30,
880: 31,
881: 30,
1.57 parser 882: 31
1.126 paf 883: };
884: return monthDays[month];
1.41 paf 885: }
1.69 parser 886:
1.126 paf 887: void remove_crlf(char* start, char* end) {
888: for(char* p=start; p<end; p++)
1.69 parser 889: switch(*p) {
1.126 paf 890: case '\n': *p='|'; break;
891: case '\r': *p=' '; break;
1.69 parser 892: }
1.91 paf 893: }
894:
895:
896: /// must be last in this file
897: #undef vsnprintf
1.126 paf 898: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 899: if(!s)
900: return 0;
901:
902: int r;
903: // note: on win32& maybe somewhere else
904: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
905: --s;
906: #if _MSC_VER
907: /*
908: win32:
909: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
910:
911: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
912: */
1.126 paf 913: r=_vsnprintf(b, s, f, l);
1.91 paf 914: if(r<0)
915: r=s;
916: #else
1.126 paf 917: r=vsnprintf(b, s, f, l);
1.91 paf 918: /*
919: solaris:
920: man vsnprintf
921:
922: The snprintf() function returns the number of characters
923: formatted, that is, the number of characters that would have
924: been written to the buffer if it were large enough. If the
925: value of n is 0 on a call to snprintf(), an unspecified
926: value less than 1 is returned.
927: */
928:
929: if(r<0)
930: r=0;
931: else if(r>s)
932: r=s;
933: #endif
934: b[r]=0;
935: return r;
936: }
937:
1.126 paf 938: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 939: va_list l;
1.126 paf 940: va_start(l, f);
941: int r=__vsnprintf(b, s, f, l);
942: va_end(l);
1.91 paf 943: return r;
1.98 paf 944: }
945:
946: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 947: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 948:
949: #ifdef WIN32
1.126 paf 950: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 951: return 0;
952: #else
953: struct timeval t;
954: t.tv_sec = secs;
955: t.tv_usec = usecs;
1.126 paf 956: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 957: #endif
1.135 paf 958: }
959:
960:
961: // attributed meaning
962:
963: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
964: const char month_names[12][4]={
965: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
966: const char days[7][4]={
967: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
968:
969: time_t when=vdate.get_time();
970: struct tm *tms=gmtime(&when);
971: if(!tms)
972: throw Exception(0,
973: 0,
974: "bad time in attribute value (seconds from epoch=%ld)", when);
975: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
976: days[tms->tm_wday],
977: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
978: tms->tm_hour,tms->tm_min,tms->tm_sec);
979: }
980: static void append_attribute_meaning(String& result,
1.136 paf 981: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 982: if(const String *string=value.get_string())
1.136 paf 983: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 984: else
985: if(Value *vdate=value.as(VDATE_TYPE, false)) {
986: char *buf=(char *)result.malloc(MAX_STRING);
987: size_t size=date_attribute(*static_cast<VDate *>(vdate),
988: buf, MAX_STRING);
989:
990: result.APPEND_CLEAN(buf, size, "converted from date", 0);
991: } else
992: throw Exception("parser.runtime",
993: &result,
994: "trying to append here neither string nor date (%s)",
995: value.type());
996: }
997: #ifndef DOXYGEN
998: struct Attributed_meaning_info {
999: String *header; // header line being constructed
1000: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 1001: bool forced; // do they force that lang?
1.135 paf 1002: };
1003: #endif
1004: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
1005: void *info) {
1006: if(akey==VALUE_NAME)
1007: return;
1008:
1009: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
1010:
1011: // ...; charset=windows1251
1012: *ami.header << "; ";
1.136 paf 1013: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1014: *ami.header << "=";
1.136 paf 1015: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1016: }
1017: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1018: String::Untaint_lang lang, bool forced) {
1.135 paf 1019: String &result=*new(meaning.pool()) String(meaning.pool());
1020: if(Hash *hash=meaning.get_hash(0)) {
1021: // $value(value) $subattribute(subattribute value)
1022: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1023: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1024:
1025: Attributed_meaning_info attributed_meaning_info={&result, lang};
1026: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1027: } else // result value
1.136 paf 1028: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1029:
1030: return result;
1.74 parser 1031: }
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