Annotation of parser3/src/main/pa_common.C, revision 1.146
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:
1.146 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/02/24 12:26:48 $";
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
1.146 ! paf 196: signal(SIGALRM, timeout_handler);
1.126 paf 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");
1.146 ! paf 209: return 0; // never
1.145 paf 210: } else {
211: alarm(timeout);
212: #endif
1.146 ! paf 213: try {
! 214: int result;
1.126 paf 215: struct sockaddr_in dest;
1.146 ! paf 216:
! 217: if(!set_addr(&dest, host, port))
1.126 paf 218: throw Exception("http.host",
219: origin_string,
1.127 paf 220: "can not resolve hostname \"%s\"", host);
1.126 paf 221:
222: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
223: throw Exception("http.connect",
224: origin_string,
1.127 paf 225: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 226: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
227: throw Exception("http.connect",
228: origin_string,
1.127 paf 229: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 230: size_t request_size=strlen(request);
231: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
232: throw Exception("http.connect",
233: origin_string,
1.127 paf 234: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 235:
1.142 paf 236: result=http_read_response(response, sock, fail_on_status_ne_200);
237: closesocket(sock);
1.145 paf 238: #ifdef PA_USE_ALARM
1.142 paf 239: alarm(0);
1.126 paf 240: #endif
1.146 ! paf 241: } catch(...) {
1.145 paf 242: #ifdef PA_USE_ALARM
1.146 ! paf 243: alarm(0);
1.126 paf 244: #endif
1.146 ! paf 245: if(sock>=0)
! 246: closesocket(sock);
! 247: /*re*/throw;
! 248: }
! 249: return result;
1.126 paf 250: }
251: }
252:
1.127 paf 253: #ifndef DOXYGEN
254: struct Http_pass_header_info {
255: String* request;
256: bool user_agent_specified;
257: };
258: #endif
259: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 260: {
1.127 paf 261: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
262: Pool& pool=i.request->pool();
263:
1.135 paf 264: *(i.request)<<key<<": "
1.136 paf 265: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 266: <<"\n";
267:
1.127 paf 268: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
269: i.user_agent_specified=true;
1.126 paf 270: }
271: static void file_read_http(Pool& pool, const String& file_spec,
272: void*& data, size_t& data_size,
1.127 paf 273: Hash *options=0, Hash** out_fields=0) {
1.126 paf 274: char host[MAX_STRING];
1.129 paf 275: const char* uri;
1.126 paf 276: int port;
277: const char* method="GET";
278: int timeout=2;
1.142 paf 279: bool fail_on_status_ne_200=true;
1.127 paf 280: Value *vheaders=0;
1.126 paf 281:
1.144 paf 282: String& connect_string=*new(pool) String(pool); // must not be local [exception may be reported outside]
1.133 paf 283: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
284: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
285:
286: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
287: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 288: throw Exception(0,
1.133 paf 289: &connect_string,
1.126 paf 290: "does not start with http://"); //never
1.133 paf 291: connect_string_cstr+=7;
1.126 paf 292:
1.133 paf 293: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 294: char* host_uri=lsplit(host, '/');
1.133 paf 295: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 296: char* port_cstr=lsplit(host, ':');
297: char* error_pos=0;
298: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
299:
1.127 paf 300: if(options) {
301: int valid_options=0;
302: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
303: valid_options++;
304: method=vmethod->as_string().cstr();
305: }
306: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
307: valid_options++;
308: timeout=vtimeout->as_int();
309: }
310: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
311: valid_options++;
312: }
1.142 paf 313: if(Value *vany_status=static_cast<Value *>(options->get(*http_any_status_name))) {
314: valid_options++;
315: fail_on_status_ne_200=!vany_status->as_bool();
316: }
317:
1.127 paf 318: if(valid_options!=options->size())
319: throw Exception("parser.runtime",
320: 0,
321: "invalid option passed");
1.133 paf 322: }
1.126 paf 323:
324: //making request
1.127 paf 325: String request(pool);
1.126 paf 326: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 327: bool user_agent_specified=false;
328: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 329: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 330: Http_pass_header_info info={&request};
331: headers->for_each(http_pass_header, &info);
332: user_agent_specified=info.user_agent_specified;
333: } else
334: throw Exception("parser.runtime",
1.133 paf 335: &connect_string,
1.127 paf 336: "headers param must be hash");
337: };
338: if(!user_agent_specified) // defaulting
339: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 340: request<<"\n";
341:
342: //sending request
343: String response(pool);
1.142 paf 344: int status_code=http_request(response,
345: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED),
346: timeout, fail_on_status_ne_200);
1.126 paf 347:
348: //processing results
349: int pos=response.pos("\r\n\r\n", 4);
350: if(pos<1){
351: throw Exception("http.response",
1.133 paf 352: &connect_string,
1.126 paf 353: "bad response from host - no headers found");
354: }
355: String header_block=response.mid(0, pos);
356: String body=response.mid(pos+4, response.size());
357:
358: Array aheaders(pool);
359: Hash& headers=*new(pool) Hash(pool);
360: size_t pos_after_ref=0;
361: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
362:
363: //processing headers
364: for(int i=1;i<aheaders.size();i++) {
365: if(const String *line=aheaders.get_string(i)) {
366: pos=line->pos(": ", 2);
367: if(pos<1)
368: throw Exception("http.response",
1.133 paf 369: &connect_string,
1.126 paf 370: "bad response from host - bad header \"%s\"", line->cstr());
371:
1.128 paf 372: headers.put(
373: line->mid(0, pos).change_case(pool, String::CC_UPPER),
374: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 375: } else
376: throw Exception("http.response",
1.133 paf 377: &connect_string,
1.126 paf 378: "bad response from host - bad headers \"%s\"", header_block.cstr());
379: }
380:
381: // output response
382: data=body.cstr(); data_size=body.size();
1.141 paf 383: if(out_fields) {
1.142 paf 384: headers.put(*file_status_name, new(pool) VInt(pool, status_code));
1.126 paf 385: *out_fields=&headers;
1.141 paf 386: }
1.34 paf 387: }
1.123 paf 388:
389: #ifndef DOXYGEN
390: struct File_read_action_info {
391: void **data; size_t *data_size;
1.126 paf 392: };
1.123 paf 393: #endif
1.126 paf 394: static void file_read_action(Pool& pool,
395: struct stat& finfo,
1.123 paf 396: int f,
1.126 paf 397: const String& file_spec, const char* fname, bool as_text,
1.123 paf 398: void *context) {
1.126 paf 399: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 400: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 401: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
402: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 403:
404: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 405: throw Exception(0,
1.123 paf 406: &file_spec,
407: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 408: *info.data_size, to_read_size);
1.123 paf 409: } else { // empty file
410: if(as_text) {
1.126 paf 411: *info.data=pool.malloc(1);
1.123 paf 412: *(char*)(*info.data)=0;
413: } else
414: *info.data=0;
415: *info.data_size=0;
416: return;
417: }
1.126 paf 418: }
419: bool file_read(Pool& pool, const String& file_spec,
420: void*& data, size_t& data_size,
421: bool as_text, Hash *params, Hash** out_fields,
422: bool fail_on_read_problem) {
423: bool result;
424: if(file_spec.starts_with("http://", 7)) {
425: // fail on read problem
426: file_read_http(pool, file_spec, data, data_size, params, out_fields);
427: result=true;
428: } else {
429: File_read_action_info info={&data, &data_size};
430: result=file_read_action_under_lock(pool, file_spec,
431: "read", file_read_action, &info,
432: as_text, fail_on_read_problem);
433: }
1.123 paf 434:
1.126 paf 435: if(result && as_text) {
1.131 paf 436: // UTF-8 signature: EF BB BF
437: if(data_size>=3) {
438: char *in=(char *)data;
439: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
440: (in[2] == '\xBF')) {
441: data=in+3; data_size-=3;// skip prefix
442: }
443: }
444:
1.126 paf 445: fix_line_breaks((char *)(data), data_size);
1.123 paf 446: // note: after fixing
1.126 paf 447: ((char*&)(data))[data_size]=0;
1.123 paf 448: }
1.126 paf 449:
450: return result;
1.123 paf 451: }
452:
453: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 454: const char* action_name, File_read_action action, void *context,
455: bool as_text,
1.123 paf 456: bool fail_on_read_problem) {
1.126 paf 457: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 458: int f;
459:
460: // first open, next stat:
1.45 paf 461: // directory update of NTFS hard links performed on open.
1.33 paf 462: // ex:
463: // a.html:^test[] and b.html hardlink to a.html
464: // user inserts ! before ^test in a.html
1.126 paf 465: // directory entry of b.html in NTFS not updated at once,
1.35 paf 466: // they delay update till open, so we would receive "!^test[" string
467: // if would do stat, next open.
1.123 paf 468: // later: it seems, even this does not help sometimes
1.98 paf 469: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 470: try {
471: if(lock_shared_blocking(f)!=0)
1.126 paf 472: throw Exception("file.lock",
1.123 paf 473: &file_spec,
474: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 475: strerror(errno), errno, fname);
1.123 paf 476:
1.124 paf 477: struct stat finfo;
478: if(stat(fname, &finfo)!=0)
479: throw Exception("file.missing", // hardly possible: we just opened it OK
480: &file_spec,
481: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 482: strerror(errno), errno, fname);
1.124 paf 483:
1.140 paf 484: #ifdef PA_SAFE_MODE
485: if(finfo.st_uid/*foreign?*/!=geteuid()
486: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 487: throw Exception("parser.runtime",
488: &file_spec,
1.140 paf 489: "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 490: fname);
1.105 paf 491: #endif
1.32 paf 492:
1.126 paf 493: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 494: } catch(...) {
1.126 paf 495: unlock(f);close(f);
1.123 paf 496: if(fail_on_read_problem)
497: /*re*/throw;
498: return false;
499: }
1.87 paf 500:
1.126 paf 501: unlock(f);close(f);
1.72 parser 502: return true;
1.118 paf 503: } else {
504: if(fail_on_read_problem)
1.126 paf 505: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 506: &file_spec,
1.123 paf 507: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 508: action_name, strerror(errno), errno, fname);
1.118 paf 509: return false;
510: }
1.8 paf 511: }
512:
1.63 parser 513: static void create_dir_for_file(const String& file_spec) {
514: size_t pos_after=1;
515: int pos_before;
516: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 517: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 518: pos_after=pos_before+1;
519: }
520: }
521:
1.98 paf 522: bool file_write_action_under_lock(
1.28 paf 523: const String& file_spec,
1.126 paf 524: const char* action_name, File_write_action action, void *context,
525: bool as_text,
526: bool do_append,
527: bool do_block,
1.110 paf 528: bool fail_on_lock_problem) {
1.126 paf 529: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 530: int f;
1.80 paf 531: if(access(fname, W_OK)!=0) // no
1.126 paf 532: create_dir_for_file(file_spec);
1.50 paf 533:
1.80 paf 534: if((f=open(fname,
535: O_CREAT|O_RDWR
536: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 537: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 538: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 539: Exception e("file.lock",
1.110 paf 540: &file_spec,
541: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 542: strerror(errno), errno, fname);
543: close(f);
1.110 paf 544: if(fail_on_lock_problem)
545: throw e;
1.98 paf 546: return false;
547: }
1.96 paf 548:
549: try {
1.126 paf 550: action(f, context);
1.96 paf 551: } catch(...) {
1.138 paf 552: #ifdef HAVE_FTRUNCATE
1.104 paf 553: if(!do_append)
1.125 paf 554: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 555: #endif
1.126 paf 556: unlock(f);close(f);
1.96 paf 557: /*re*/throw;
558: }
1.80 paf 559:
1.138 paf 560: #ifdef HAVE_FTRUNCATE
1.104 paf 561: if(!do_append)
1.125 paf 562: 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 563: #endif
1.126 paf 564: unlock(f);close(f);
1.98 paf 565: return true;
1.80 paf 566: } else
1.126 paf 567: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 568: &file_spec,
1.96 paf 569: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 570: action_name, strerror(errno), errno, fname);
1.96 paf 571: // here should be nothing, see rethrow above
572: }
573:
574: #ifndef DOXYGEN
575: struct File_write_action_info {
576: const void *data; size_t size;
1.126 paf 577: };
1.96 paf 578: #endif
579: static void file_write_action(int f, void *context) {
1.126 paf 580: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 581: if(info.size) {
1.126 paf 582: int written=write(f, info.data, info.size);
1.116 paf 583: if(written<0)
1.126 paf 584: throw Exception(0,
585: 0,
586: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 587: }
1.96 paf 588: }
589: void file_write(
590: const String& file_spec,
591: const void *data, size_t size,
1.126 paf 592: bool as_text,
1.96 paf 593: bool do_append) {
1.126 paf 594: File_write_action_info info={data, size};
1.98 paf 595: file_write_action_under_lock(
1.96 paf 596: file_spec,
1.126 paf 597: "write", file_write_action, &info,
598: as_text,
599: do_append);
1.30 paf 600: }
601:
1.63 parser 602: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 603: static void rmdir(const String& file_spec, size_t pos_after) {
604: int pos_before;
605: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 606: rmdir(file_spec, pos_before+1);
1.50 paf 607:
1.126 paf 608: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 609: }
1.95 paf 610: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 611: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 612: if(unlink(fname)!=0)
1.93 paf 613: if(fail_on_read_problem)
1.126 paf 614: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 615: &file_spec,
616: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 617: strerror(errno), errno, fname);
1.93 paf 618: else
619: return false;
1.50 paf 620:
1.126 paf 621: rmdir(file_spec, 1);
1.93 paf 622: return true;
1.60 parser 623: }
1.95 paf 624: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 625: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
626: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 627:
1.126 paf 628: create_dir_for_file(new_spec);
1.63 parser 629:
1.60 parser 630: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 631: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 632: &old_spec,
633: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 634: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 635:
1.126 paf 636: rmdir(old_spec, 1);
1.31 paf 637: }
638:
1.51 paf 639:
1.126 paf 640: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 641: struct stat lfinfo;
642: bool result=stat(fname, &lfinfo)==0;
643: if(afinfo)
644: *afinfo=lfinfo;
645: return result;
1.119 paf 646: }
647:
648: bool entry_exists(const String& file_spec) {
1.126 paf 649: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
650: return entry_exists(fname, 0);
1.118 paf 651: }
652:
1.51 paf 653: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 654: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 655: if(need_dir) {
1.126 paf 656: size_t size=strlen(fname);
1.120 paf 657: while(size) {
1.126 paf 658: char c=fname[size-1];
1.120 paf 659: if(c=='/' || c=='\\')
660: fname[--size]=0;
661: else
662: break;
663: }
664: }
1.51 paf 665: struct stat finfo;
1.118 paf 666: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 667: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 668: return is_dir==need_dir;
669: }
670: return false;
671: }
1.31 paf 672: bool file_readable(const String& file_spec) {
1.126 paf 673: return entry_readable(file_spec, false);
1.51 paf 674: }
675: bool dir_readable(const String& file_spec) {
1.126 paf 676: return entry_readable(file_spec, true);
1.65 parser 677: }
678: String *file_readable(const String& path, const String& name) {
1.126 paf 679: String *result=new(path.pool()) String(path);
680: *result << "/";
1.65 parser 681: *result << name;
682: return file_readable(*result)?result:0;
1.43 paf 683: }
684: bool file_executable(const String& file_spec) {
1.64 parser 685: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 686: }
687:
1.64 parser 688: bool file_stat(const String& file_spec,
1.58 parser 689: size_t& rsize,
1.126 paf 690: time_t& ratime,
691: time_t& rmtime,
692: time_t& rctime,
1.64 parser 693: bool fail_on_read_problem) {
1.126 paf 694: Pool& pool=file_spec.pool();
695: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 696: struct stat finfo;
697: if(stat(fname, &finfo)!=0)
1.64 parser 698: if(fail_on_read_problem)
1.126 paf 699: throw Exception("file.missing",
1.67 parser 700: &file_spec,
701: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 702: strerror(errno), errno, fname);
1.64 parser 703: else
704: return false;
1.58 parser 705: rsize=finfo.st_size;
706: ratime=finfo.st_atime;
707: rmtime=finfo.st_mtime;
708: rctime=finfo.st_ctime;
1.64 parser 709: return true;
1.18 paf 710: }
711:
1.126 paf 712: char* getrow(char* *row_ref, char delim) {
713: char* result=*row_ref;
1.8 paf 714: if(result) {
1.126 paf 715: *row_ref=strchr(result, delim);
1.8 paf 716: if(*row_ref)
717: *((*row_ref)++)=0;
718: else if(!*result)
719: return 0;
720: }
721: return result;
722: }
723:
1.126 paf 724: char* lsplit(char* string, char delim) {
1.23 paf 725: if(string) {
1.126 paf 726: char* v=strchr(string, delim);
1.8 paf 727: if(v) {
728: *v=0;
729: return v+1;
730: }
731: }
732: return 0;
733: }
734:
1.126 paf 735: char* lsplit(char* *string_ref, char delim) {
736: char* result=*string_ref;
737: char* next=lsplit(*string_ref, delim);
1.8 paf 738: *string_ref=next;
739: return result;
1.9 paf 740: }
741:
1.126 paf 742: char* rsplit(char* string, char delim) {
1.18 paf 743: if(string) {
1.126 paf 744: char* v=strrchr(string, delim);
1.18 paf 745: if(v) {
1.9 paf 746: *v=0;
747: return v+1;
748: }
749: }
750: return NULL;
1.10 paf 751: }
752:
1.37 paf 753: /// @todo less stupid type detection
1.126 paf 754: char* format(Pool& pool, double value, char* fmt) {
755: char local_buf[MAX_NUMBER];
1.108 paf 756: size_t size;
757:
1.10 paf 758: if(fmt)
759: if(strpbrk(fmt, "diouxX"))
760: if(strpbrk(fmt, "ouxX"))
1.126 paf 761: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 762: else
1.126 paf 763: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 764: else
1.126 paf 765: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 766: else
1.126 paf 767: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 768:
1.126 paf 769: char* pool_buf=(char *)pool.malloc(size+1, 4);
770: memcpy(pool_buf, local_buf, size+1);
1.108 paf 771: return pool_buf;
1.12 paf 772: }
773:
1.36 paf 774: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 775: #ifdef WIN32
776: do{
1.126 paf 777: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 778: if(chunk_written<=0)
779: break;
780: size-=chunk_written;
1.36 paf 781: buf=((const char*)buf)+chunk_written;
1.126 paf 782: } while(size>0);
1.12 paf 783:
784: return size;
785: #else
1.126 paf 786: return fwrite(buf, 1, size, stdout);
1.12 paf 787: #endif
1.2 paf 788: }
1.14 paf 789:
1.126 paf 790: char* unescape_chars(Pool& pool, const char* cp, int len) {
791: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 792: enum EscapeState {
1.33 paf 793: EscapeRest,
794: EscapeFirst,
1.14 paf 795: EscapeSecond
796: } escapeState=EscapeRest;
797: int escapedValue=0;
798: int srcPos=0;
799: int dstPos=0;
800: while(srcPos < len) {
1.126 paf 801: int ch=cp[srcPos];
1.14 paf 802: switch(escapeState) {
803: case EscapeRest:
804: if(ch=='%') {
805: escapeState=EscapeFirst;
806: } else if(ch=='+') {
1.126 paf 807: s[dstPos++]=' ';
1.14 paf 808: } else {
809: s[dstPos++]=ch;
810: }
811: break;
812: case EscapeFirst:
813: escapedValue=hex_value[ch] << 4;
814: escapeState=EscapeSecond;
815: break;
816: case EscapeSecond:
1.126 paf 817: escapedValue +=hex_value[ch];
1.14 paf 818: s[dstPos++]=escapedValue;
819: escapeState=EscapeRest;
820: break;
821: }
1.126 paf 822: srcPos++;
1.14 paf 823: }
824: s[dstPos]=0;
825: return s;
1.24 paf 826: }
827:
828: #ifdef WIN32
1.126 paf 829: void back_slashes_to_slashes(char* s) {
1.24 paf 830: if(s)
831: for(; *s; s++)
832: if(*s=='\\')
1.126 paf 833: *s='/';
1.24 paf 834: }
1.42 paf 835: /*
1.126 paf 836: void slashes_to_back_slashes(char* s) {
1.42 paf 837: if(s)
838: for(; *s; s++)
839: if(*s=='/')
1.126 paf 840: *s='\\';
1.42 paf 841: }
842: */
1.24 paf 843: #endif
1.41 paf 844:
1.126 paf 845: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 846: while(true) {
847: if(!(*s1)) {
848: if(!(*s2))
849: return true;
850: else
851: return !strict;
852: } else if(!(*s2))
853: return !strict;
854: if(isalpha(*s1)) {
855: if(tolower(*s1) !=tolower(*s2))
856: return false;
857: } else if((*s1) !=(*s2))
858: return false;
1.126 paf 859: s1++;
860: s2++;
1.41 paf 861: }
1.57 parser 862: }
863:
1.84 paf 864: static bool isLeap(int year) {
1.57 parser 865: return !(
866: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 867: );
1.57 parser 868: }
869:
870: int getMonthDays(int year, int month) {
871: int monthDays[]={
1.126 paf 872: 31,
873: isLeap(year) ? 29 : 28,
874: 31,
875: 30,
876: 31,
877: 30,
878: 31,
879: 31,
880: 30,
881: 31,
882: 30,
1.57 parser 883: 31
1.126 paf 884: };
885: return monthDays[month];
1.41 paf 886: }
1.69 parser 887:
1.126 paf 888: void remove_crlf(char* start, char* end) {
889: for(char* p=start; p<end; p++)
1.69 parser 890: switch(*p) {
1.126 paf 891: case '\n': *p='|'; break;
892: case '\r': *p=' '; break;
1.69 parser 893: }
1.91 paf 894: }
895:
896:
897: /// must be last in this file
898: #undef vsnprintf
1.126 paf 899: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 900: if(!s)
901: return 0;
902:
903: int r;
904: // note: on win32& maybe somewhere else
905: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
906: --s;
907: #if _MSC_VER
908: /*
909: win32:
910: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
911:
912: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
913: */
1.126 paf 914: r=_vsnprintf(b, s, f, l);
1.91 paf 915: if(r<0)
916: r=s;
917: #else
1.126 paf 918: r=vsnprintf(b, s, f, l);
1.91 paf 919: /*
920: solaris:
921: man vsnprintf
922:
923: The snprintf() function returns the number of characters
924: formatted, that is, the number of characters that would have
925: been written to the buffer if it were large enough. If the
926: value of n is 0 on a call to snprintf(), an unspecified
927: value less than 1 is returned.
928: */
929:
930: if(r<0)
931: r=0;
932: else if(r>s)
933: r=s;
934: #endif
935: b[r]=0;
936: return r;
937: }
938:
1.126 paf 939: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 940: va_list l;
1.126 paf 941: va_start(l, f);
942: int r=__vsnprintf(b, s, f, l);
943: va_end(l);
1.91 paf 944: return r;
1.98 paf 945: }
946:
947: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 948: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 949:
950: #ifdef WIN32
1.126 paf 951: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 952: return 0;
953: #else
954: struct timeval t;
955: t.tv_sec = secs;
956: t.tv_usec = usecs;
1.126 paf 957: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 958: #endif
1.135 paf 959: }
960:
961:
962: // attributed meaning
963:
964: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
965: const char month_names[12][4]={
966: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
967: const char days[7][4]={
968: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
969:
970: time_t when=vdate.get_time();
971: struct tm *tms=gmtime(&when);
972: if(!tms)
973: throw Exception(0,
974: 0,
975: "bad time in attribute value (seconds from epoch=%ld)", when);
976: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
977: days[tms->tm_wday],
978: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
979: tms->tm_hour,tms->tm_min,tms->tm_sec);
980: }
981: static void append_attribute_meaning(String& result,
1.136 paf 982: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 983: if(const String *string=value.get_string())
1.136 paf 984: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 985: else
986: if(Value *vdate=value.as(VDATE_TYPE, false)) {
987: char *buf=(char *)result.malloc(MAX_STRING);
988: size_t size=date_attribute(*static_cast<VDate *>(vdate),
989: buf, MAX_STRING);
990:
991: result.APPEND_CLEAN(buf, size, "converted from date", 0);
992: } else
993: throw Exception("parser.runtime",
994: &result,
995: "trying to append here neither string nor date (%s)",
996: value.type());
997: }
998: #ifndef DOXYGEN
999: struct Attributed_meaning_info {
1000: String *header; // header line being constructed
1001: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 1002: bool forced; // do they force that lang?
1.135 paf 1003: };
1004: #endif
1005: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
1006: void *info) {
1007: if(akey==VALUE_NAME)
1008: return;
1009:
1010: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
1011:
1012: // ...; charset=windows1251
1013: *ami.header << "; ";
1.136 paf 1014: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1015: *ami.header << "=";
1.136 paf 1016: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1017: }
1018: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1019: String::Untaint_lang lang, bool forced) {
1.135 paf 1020: String &result=*new(meaning.pool()) String(meaning.pool());
1021: if(Hash *hash=meaning.get_hash(0)) {
1022: // $value(value) $subattribute(subattribute value)
1023: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1024: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1025:
1026: Attributed_meaning_info attributed_meaning_info={&result, lang};
1027: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1028: } else // result value
1.136 paf 1029: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1030:
1031: return result;
1.74 parser 1032: }
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