Annotation of parser3/src/main/pa_common.C, revision 1.135
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.102 paf 4: Copyright(c) 2001, 2002 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:
1.135 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2002/11/29 07:48:38 $";
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:
1.93 paf 32: // locking constants
33:
1.99 paf 34: #ifdef HAVE_FLOCK
35:
36: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
37: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
38: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
39: static int unlock(int fd) { return flock(fd, LOCK_UN); }
40:
1.98 paf 41: #else
1.99 paf 42: #ifdef HAVE__LOCKING
1.98 paf 43:
1.126 paf 44: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 45: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
46: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
47: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
48: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 49:
1.99 paf 50: #else
51: #ifdef HAVE_FCNTL
1.93 paf 52:
1.126 paf 53: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 54: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
55: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
56: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
57: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 58:
59: #else
60: #ifdef HAVE_LOCKF
1.99 paf 61:
1.126 paf 62: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 63: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
64: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
65: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
66: static int unlock(int fd) { FLOCK(F_TLOCK); }
67:
1.93 paf 68: #else
1.99 paf 69:
70: #error unable to find file locking func
71:
72: #endif
1.93 paf 73: #endif
74: #endif
75: #endif
76:
1.127 paf 77: #define DEFAULT_USER_AGENT "parser3"
78:
79:
1.126 paf 80:
81: static char* strnchr(char* buf, size_t size, char c) {
1.103 paf 82: // sanity check
83: if(!buf)
84: return 0;
85:
1.126 paf 86: for(; size-->0; buf++) {
1.86 paf 87: if(*buf==c)
88: return buf;
89: }
90:
91: return 0;
92: }
93:
1.126 paf 94: void fix_line_breaks(char* buf, size_t& size) {
1.87 paf 95: //_asm int 3;
1.126 paf 96: const char* const eob=buf+size;
97: char* dest=buf;
1.72 parser 98: // fix DOS: \r\n -> \n
99: // fix Macintosh: \r -> \n
1.126 paf 100: char* bol=buf;
101: while(char* eol=strnchr(bol, eob -bol, '\r')) {
1.72 parser 102: size_t len=eol-bol;
103: if(dest!=bol)
1.126 paf 104: memcpy(dest, bol, len);
1.72 parser 105: dest+=len;
1.126 paf 106: *dest++='\n';
1.72 parser 107:
1.126 paf 108: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 109: bol=eol+2;
1.126 paf 110: size--;
111: } else // \r, not \n = Macintosh
1.72 parser 112: bol=eol+1;
113: }
114: // last piece without \r, including terminating 0
115: if(dest!=bol)
1.126 paf 116: memcpy(dest, bol, eob-bol);
1.72 parser 117: }
1.18 paf 118:
1.126 paf 119: char* file_read_text(Pool& pool, const String& file_spec,
120: bool fail_on_read_problem,
121: Hash *params, Hash** out_fields) {
1.72 parser 122: void *result; size_t size;
1.126 paf 123: return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;
124: }
125:
126: //http request stuff
127: /* ************************ http stuff *********************** */
128:
129: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
130: memset(addr, 0, sizeof(*addr));
131: addr->sin_family=AF_INET;
132: addr->sin_port=htons(port);
133: if(host) {
134: if(struct hostent *hostIP=gethostbyname(host))
135: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
136: else
137: return false;
138: } else
139: addr->sin_addr.s_addr=INADDR_ANY;
140: return true;
141: }
142:
143: static void http_read_response(String& response, int sock){
1.130 paf 144: bool check_status=true;
145: ssize_t EOLat=0;
1.126 paf 146: while(true) {
147: char *buf=(char *)response.pool().malloc(MAX_STRING);
148: ssize_t size=recv(sock, buf, MAX_STRING, 0);
149: if(size<=0)
150: break;
1.130 paf 151: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
152: if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
153: check_status=false;
154:
155: const String& status_line=response.mid(0, (size_t)EOLat);
156: Array astatus(response.pool());
157: size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
158: const String* status_code=astatus.get_string(1);
159:
160: if(!status_code || *status_code!="200")
161: throw Exception("http.status",
162: status_code,
163: "invalid HTTP response status");
164: }
1.126 paf 165: }
1.130 paf 166: if(check_status)
167: throw Exception("http.response",
168: 0,
169: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 170: }
171:
172: /* ********************** request *************************** */
173:
174: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
175: # define WE_CAN_USE_ALARM
176: #endif
177:
178: #ifdef WE_CAN_USE_ALARM
179: static sigjmp_buf timeout_env;
180: static void timeout_handler(int sig){
181: siglongjmp(timeout_env, 1);
182: }
183: #endif
184:
185: static void http_request(String& response,
186: const String *origin_string,
187: const char* host, int port,
188: const char* request,
189: int timeout){
190: if(!host)
191: throw Exception("http.host",
192: origin_string,
193: "zero hostname"); //never
194:
195: #ifdef WE_CAN_USE_ALARM
196: signal(SIGALRM, timeout_handler);
197: #endif
198: int sock=-1;
199: try {
200: #ifdef WE_CAN_USE_ALARM
201: if(sigsetjmp(timeout_env, 1))
202: throw Exception("http.timeout",
203: origin_string,
204: "timeout occured while retrieving document");
205: else {
206: alarm(timeout);
207: #endif
208: struct sockaddr_in dest;
209:
210: if(!set_addr(&dest, host, port))
211: throw Exception("http.host",
212: origin_string,
1.127 paf 213: "can not resolve hostname \"%s\"", host);
1.126 paf 214:
215: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
216: throw Exception("http.connect",
217: origin_string,
1.127 paf 218: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 219: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
220: throw Exception("http.connect",
221: origin_string,
1.127 paf 222: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 223: size_t request_size=strlen(request);
224: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
225: throw Exception("http.connect",
226: origin_string,
1.127 paf 227: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 228:
229: http_read_response(response, sock);
230: if(sock>=0)
231: closesocket(sock);
232: #ifdef WE_CAN_USE_ALARM
233: }
234: #endif
235: } catch(...) {
236: if(sock>=0)
237: closesocket(sock);
238: #ifdef WE_CAN_USE_ALARM
239: alarm(0);
240: #endif
241: /*re*/throw;
242: }
243: }
244:
1.127 paf 245: #ifndef DOXYGEN
246: struct Http_pass_header_info {
247: String* request;
248: bool user_agent_specified;
249: };
250: #endif
251: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 252: {
1.127 paf 253: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
254: Pool& pool=i.request->pool();
255:
1.135 ! paf 256: *(i.request)<<key<<": "
! 257: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_URI)
! 258: <<"\n";
! 259:
1.127 paf 260: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
261: i.user_agent_specified=true;
1.126 paf 262: }
263: static void file_read_http(Pool& pool, const String& file_spec,
264: void*& data, size_t& data_size,
1.127 paf 265: Hash *options=0, Hash** out_fields=0) {
1.126 paf 266: char host[MAX_STRING];
1.129 paf 267: const char* uri;
1.126 paf 268: int port;
269: const char* method="GET";
270: int timeout=2;
1.127 paf 271: Value *vheaders=0;
1.126 paf 272:
1.133 paf 273: String connect_string(pool);
274: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
275: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
276:
277: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
278: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 279: throw Exception(0,
1.133 paf 280: &connect_string,
1.126 paf 281: "does not start with http://"); //never
1.133 paf 282: connect_string_cstr+=7;
1.126 paf 283:
1.133 paf 284: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 285: char* host_uri=lsplit(host, '/');
1.133 paf 286: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 287: char* port_cstr=lsplit(host, ':');
288: char* error_pos=0;
289: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
290:
1.127 paf 291: if(options) {
292: int valid_options=0;
293: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
294: valid_options++;
295: method=vmethod->as_string().cstr();
296: }
297: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
298: valid_options++;
299: timeout=vtimeout->as_int();
300: }
301: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
302: valid_options++;
303: }
304: if(valid_options!=options->size())
305: throw Exception("parser.runtime",
306: 0,
307: "invalid option passed");
1.133 paf 308: }
1.126 paf 309:
310: //making request
1.127 paf 311: String request(pool);
1.126 paf 312: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 313: bool user_agent_specified=false;
314: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 315: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 316: Http_pass_header_info info={&request};
317: headers->for_each(http_pass_header, &info);
318: user_agent_specified=info.user_agent_specified;
319: } else
320: throw Exception("parser.runtime",
1.133 paf 321: &connect_string,
1.127 paf 322: "headers param must be hash");
323: };
324: if(!user_agent_specified) // defaulting
325: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 326: request<<"\n";
327:
328: //sending request
329: String response(pool);
330: http_request(response,
1.133 paf 331: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout);
1.126 paf 332:
333: //processing results
334: int pos=response.pos("\r\n\r\n", 4);
335: if(pos<1){
336: throw Exception("http.response",
1.133 paf 337: &connect_string,
1.126 paf 338: "bad response from host - no headers found");
339: }
340: String header_block=response.mid(0, pos);
341: String body=response.mid(pos+4, response.size());
342:
343: Array aheaders(pool);
344: Hash& headers=*new(pool) Hash(pool);
345: size_t pos_after_ref=0;
346: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
347:
348: //processing status code
349: const String *status_line=aheaders.get_string(0);
350: if(!status_line){
351: throw Exception("http.response",
1.133 paf 352: &connect_string,
1.126 paf 353: "bad response from host - no status line ");
354: }
355: //processing headers
356: for(int i=1;i<aheaders.size();i++) {
357: if(const String *line=aheaders.get_string(i)) {
358: pos=line->pos(": ", 2);
359: if(pos<1)
360: throw Exception("http.response",
1.133 paf 361: &connect_string,
1.126 paf 362: "bad response from host - bad header \"%s\"", line->cstr());
363:
1.128 paf 364: headers.put(
365: line->mid(0, pos).change_case(pool, String::CC_UPPER),
366: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 367: } else
368: throw Exception("http.response",
1.133 paf 369: &connect_string,
1.126 paf 370: "bad response from host - bad headers \"%s\"", header_block.cstr());
371: }
372:
373: // output response
374: data=body.cstr(); data_size=body.size();
1.130 paf 375: if(out_fields)
1.126 paf 376: *out_fields=&headers;
1.34 paf 377: }
1.123 paf 378:
379: #ifndef DOXYGEN
380: struct File_read_action_info {
381: void **data; size_t *data_size;
1.126 paf 382: };
1.123 paf 383: #endif
1.126 paf 384: static void file_read_action(Pool& pool,
385: struct stat& finfo,
1.123 paf 386: int f,
1.126 paf 387: const String& file_spec, const char* fname, bool as_text,
1.123 paf 388: void *context) {
1.126 paf 389: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 390: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 391: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
392: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 393:
394: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 395: throw Exception(0,
1.123 paf 396: &file_spec,
397: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 398: *info.data_size, to_read_size);
1.123 paf 399: } else { // empty file
400: if(as_text) {
1.126 paf 401: *info.data=pool.malloc(1);
1.123 paf 402: *(char*)(*info.data)=0;
403: } else
404: *info.data=0;
405: *info.data_size=0;
406: return;
407: }
1.126 paf 408: }
409: bool file_read(Pool& pool, const String& file_spec,
410: void*& data, size_t& data_size,
411: bool as_text, Hash *params, Hash** out_fields,
412: bool fail_on_read_problem) {
413: bool result;
414: if(file_spec.starts_with("http://", 7)) {
415: // fail on read problem
416: file_read_http(pool, file_spec, data, data_size, params, out_fields);
417: result=true;
418: } else {
419: File_read_action_info info={&data, &data_size};
420: result=file_read_action_under_lock(pool, file_spec,
421: "read", file_read_action, &info,
422: as_text, fail_on_read_problem);
423: }
1.123 paf 424:
1.126 paf 425: if(result && as_text) {
1.131 paf 426: // UTF-8 signature: EF BB BF
427: if(data_size>=3) {
428: char *in=(char *)data;
429: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
430: (in[2] == '\xBF')) {
431: data=in+3; data_size-=3;// skip prefix
432: }
433: }
434:
1.126 paf 435: fix_line_breaks((char *)(data), data_size);
1.123 paf 436: // note: after fixing
1.126 paf 437: ((char*&)(data))[data_size]=0;
1.123 paf 438: }
1.126 paf 439:
440: return result;
1.123 paf 441: }
442:
443: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 444: const char* action_name, File_read_action action, void *context,
445: bool as_text,
1.123 paf 446: bool fail_on_read_problem) {
1.126 paf 447: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 448: int f;
449:
450: // first open, next stat:
1.45 paf 451: // directory update of NTFS hard links performed on open.
1.33 paf 452: // ex:
453: // a.html:^test[] and b.html hardlink to a.html
454: // user inserts ! before ^test in a.html
1.126 paf 455: // directory entry of b.html in NTFS not updated at once,
1.35 paf 456: // they delay update till open, so we would receive "!^test[" string
457: // if would do stat, next open.
1.123 paf 458: // later: it seems, even this does not help sometimes
1.98 paf 459: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 460: try {
461: if(lock_shared_blocking(f)!=0)
1.126 paf 462: throw Exception("file.lock",
1.123 paf 463: &file_spec,
464: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 465: strerror(errno), errno, fname);
1.123 paf 466:
1.124 paf 467: struct stat finfo;
468: if(stat(fname, &finfo)!=0)
469: throw Exception("file.missing", // hardly possible: we just opened it OK
470: &file_spec,
471: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 472: strerror(errno), errno, fname);
1.124 paf 473:
1.105 paf 474: #ifdef NO_FOREIGN_GROUP_FILES
1.132 paf 475: if(finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 476: throw Exception("parser.runtime",
477: &file_spec,
1.123 paf 478: "parser reading files of foreign group disabled [recompile parser without --disable-foreign-group-files configure option], actual filename '%s'",
1.126 paf 479: fname);
1.105 paf 480: #endif
1.32 paf 481:
1.126 paf 482: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 483: } catch(...) {
1.126 paf 484: unlock(f);close(f);
1.123 paf 485: if(fail_on_read_problem)
486: /*re*/throw;
487: return false;
488: }
1.87 paf 489:
1.126 paf 490: unlock(f);close(f);
1.72 parser 491: return true;
1.118 paf 492: } else {
493: if(fail_on_read_problem)
1.126 paf 494: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 495: &file_spec,
1.123 paf 496: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 497: action_name, strerror(errno), errno, fname);
1.118 paf 498: return false;
499: }
1.8 paf 500: }
501:
1.63 parser 502: static void create_dir_for_file(const String& file_spec) {
503: size_t pos_after=1;
504: int pos_before;
505: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 506: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 507: pos_after=pos_before+1;
508: }
509: }
510:
1.98 paf 511: bool file_write_action_under_lock(
1.28 paf 512: const String& file_spec,
1.126 paf 513: const char* action_name, File_write_action action, void *context,
514: bool as_text,
515: bool do_append,
516: bool do_block,
1.110 paf 517: bool fail_on_lock_problem) {
1.126 paf 518: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 519: int f;
1.80 paf 520: if(access(fname, W_OK)!=0) // no
1.126 paf 521: create_dir_for_file(file_spec);
1.50 paf 522:
1.80 paf 523: if((f=open(fname,
524: O_CREAT|O_RDWR
525: |(as_text?_O_TEXT:_O_BINARY)
1.122 paf 526: |(do_append?O_APPEND:0), 0664))>=0) {
1.99 paf 527: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 528: Exception e("file.lock",
1.110 paf 529: &file_spec,
530: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 531: strerror(errno), errno, fname);
532: close(f);
1.110 paf 533: if(fail_on_lock_problem)
534: throw e;
1.98 paf 535: return false;
536: }
1.96 paf 537:
538: try {
1.126 paf 539: action(f, context);
1.96 paf 540: } catch(...) {
1.104 paf 541: if(!do_append)
1.125 paf 542: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.126 paf 543: unlock(f);close(f);
1.96 paf 544: /*re*/throw;
545: }
1.80 paf 546:
1.104 paf 547: if(!do_append)
1.125 paf 548: ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
1.126 paf 549: unlock(f);close(f);
1.98 paf 550: return true;
1.80 paf 551: } else
1.126 paf 552: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 553: &file_spec,
1.96 paf 554: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 555: action_name, strerror(errno), errno, fname);
1.96 paf 556: // here should be nothing, see rethrow above
557: }
558:
559: #ifndef DOXYGEN
560: struct File_write_action_info {
561: const void *data; size_t size;
1.126 paf 562: };
1.96 paf 563: #endif
564: static void file_write_action(int f, void *context) {
1.126 paf 565: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 566: if(info.size) {
1.126 paf 567: int written=write(f, info.data, info.size);
1.116 paf 568: if(written<0)
1.126 paf 569: throw Exception(0,
570: 0,
571: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 572: }
1.96 paf 573: }
574: void file_write(
575: const String& file_spec,
576: const void *data, size_t size,
1.126 paf 577: bool as_text,
1.96 paf 578: bool do_append) {
1.126 paf 579: File_write_action_info info={data, size};
1.98 paf 580: file_write_action_under_lock(
1.96 paf 581: file_spec,
1.126 paf 582: "write", file_write_action, &info,
583: as_text,
584: do_append);
1.30 paf 585: }
586:
1.63 parser 587: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 588: static void rmdir(const String& file_spec, size_t pos_after) {
589: int pos_before;
590: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 591: rmdir(file_spec, pos_before+1);
1.50 paf 592:
1.126 paf 593: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 594: }
1.95 paf 595: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 596: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 597: if(unlink(fname)!=0)
1.93 paf 598: if(fail_on_read_problem)
1.126 paf 599: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 600: &file_spec,
601: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 602: strerror(errno), errno, fname);
1.93 paf 603: else
604: return false;
1.50 paf 605:
1.126 paf 606: rmdir(file_spec, 1);
1.93 paf 607: return true;
1.60 parser 608: }
1.95 paf 609: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 610: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
611: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 612:
1.126 paf 613: create_dir_for_file(new_spec);
1.63 parser 614:
1.60 parser 615: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 616: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 617: &old_spec,
618: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 619: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 620:
1.126 paf 621: rmdir(old_spec, 1);
1.31 paf 622: }
623:
1.51 paf 624:
1.126 paf 625: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 626: struct stat lfinfo;
627: bool result=stat(fname, &lfinfo)==0;
628: if(afinfo)
629: *afinfo=lfinfo;
630: return result;
1.119 paf 631: }
632:
633: bool entry_exists(const String& file_spec) {
1.126 paf 634: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
635: return entry_exists(fname, 0);
1.118 paf 636: }
637:
1.51 paf 638: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 639: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 640: if(need_dir) {
1.126 paf 641: size_t size=strlen(fname);
1.120 paf 642: while(size) {
1.126 paf 643: char c=fname[size-1];
1.120 paf 644: if(c=='/' || c=='\\')
645: fname[--size]=0;
646: else
647: break;
648: }
649: }
1.51 paf 650: struct stat finfo;
1.118 paf 651: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 652: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 653: return is_dir==need_dir;
654: }
655: return false;
656: }
1.31 paf 657: bool file_readable(const String& file_spec) {
1.126 paf 658: return entry_readable(file_spec, false);
1.51 paf 659: }
660: bool dir_readable(const String& file_spec) {
1.126 paf 661: return entry_readable(file_spec, true);
1.65 parser 662: }
663: String *file_readable(const String& path, const String& name) {
1.126 paf 664: String *result=new(path.pool()) String(path);
665: *result << "/";
1.65 parser 666: *result << name;
667: return file_readable(*result)?result:0;
1.43 paf 668: }
669: bool file_executable(const String& file_spec) {
1.64 parser 670: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 671: }
672:
1.64 parser 673: bool file_stat(const String& file_spec,
1.58 parser 674: size_t& rsize,
1.126 paf 675: time_t& ratime,
676: time_t& rmtime,
677: time_t& rctime,
1.64 parser 678: bool fail_on_read_problem) {
1.126 paf 679: Pool& pool=file_spec.pool();
680: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 681: struct stat finfo;
682: if(stat(fname, &finfo)!=0)
1.64 parser 683: if(fail_on_read_problem)
1.126 paf 684: throw Exception("file.missing",
1.67 parser 685: &file_spec,
686: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 687: strerror(errno), errno, fname);
1.64 parser 688: else
689: return false;
1.58 parser 690: rsize=finfo.st_size;
691: ratime=finfo.st_atime;
692: rmtime=finfo.st_mtime;
693: rctime=finfo.st_ctime;
1.64 parser 694: return true;
1.18 paf 695: }
696:
1.126 paf 697: char* getrow(char* *row_ref, char delim) {
698: char* result=*row_ref;
1.8 paf 699: if(result) {
1.126 paf 700: *row_ref=strchr(result, delim);
1.8 paf 701: if(*row_ref)
702: *((*row_ref)++)=0;
703: else if(!*result)
704: return 0;
705: }
706: return result;
707: }
708:
1.126 paf 709: char* lsplit(char* string, char delim) {
1.23 paf 710: if(string) {
1.126 paf 711: char* v=strchr(string, delim);
1.8 paf 712: if(v) {
713: *v=0;
714: return v+1;
715: }
716: }
717: return 0;
718: }
719:
1.126 paf 720: char* lsplit(char* *string_ref, char delim) {
721: char* result=*string_ref;
722: char* next=lsplit(*string_ref, delim);
1.8 paf 723: *string_ref=next;
724: return result;
1.9 paf 725: }
726:
1.126 paf 727: char* rsplit(char* string, char delim) {
1.18 paf 728: if(string) {
1.126 paf 729: char* v=strrchr(string, delim);
1.18 paf 730: if(v) {
1.9 paf 731: *v=0;
732: return v+1;
733: }
734: }
735: return NULL;
1.10 paf 736: }
737:
1.37 paf 738: /// @todo less stupid type detection
1.126 paf 739: char* format(Pool& pool, double value, char* fmt) {
740: char local_buf[MAX_NUMBER];
1.108 paf 741: size_t size;
742:
1.10 paf 743: if(fmt)
744: if(strpbrk(fmt, "diouxX"))
745: if(strpbrk(fmt, "ouxX"))
1.126 paf 746: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 747: else
1.126 paf 748: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 749: else
1.126 paf 750: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 751: else
1.126 paf 752: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 753:
1.126 paf 754: char* pool_buf=(char *)pool.malloc(size+1, 4);
755: memcpy(pool_buf, local_buf, size+1);
1.108 paf 756: return pool_buf;
1.12 paf 757: }
758:
1.36 paf 759: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 760: #ifdef WIN32
761: do{
1.126 paf 762: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 763: if(chunk_written<=0)
764: break;
765: size-=chunk_written;
1.36 paf 766: buf=((const char*)buf)+chunk_written;
1.126 paf 767: } while(size>0);
1.12 paf 768:
769: return size;
770: #else
1.126 paf 771: return fwrite(buf, 1, size, stdout);
1.12 paf 772: #endif
1.2 paf 773: }
1.14 paf 774:
1.126 paf 775: char* unescape_chars(Pool& pool, const char* cp, int len) {
776: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 777: enum EscapeState {
1.33 paf 778: EscapeRest,
779: EscapeFirst,
1.14 paf 780: EscapeSecond
781: } escapeState=EscapeRest;
782: int escapedValue=0;
783: int srcPos=0;
784: int dstPos=0;
785: while(srcPos < len) {
1.126 paf 786: int ch=cp[srcPos];
1.14 paf 787: switch(escapeState) {
788: case EscapeRest:
789: if(ch=='%') {
790: escapeState=EscapeFirst;
791: } else if(ch=='+') {
1.126 paf 792: s[dstPos++]=' ';
1.14 paf 793: } else {
794: s[dstPos++]=ch;
795: }
796: break;
797: case EscapeFirst:
798: escapedValue=hex_value[ch] << 4;
799: escapeState=EscapeSecond;
800: break;
801: case EscapeSecond:
1.126 paf 802: escapedValue +=hex_value[ch];
1.14 paf 803: s[dstPos++]=escapedValue;
804: escapeState=EscapeRest;
805: break;
806: }
1.126 paf 807: srcPos++;
1.14 paf 808: }
809: s[dstPos]=0;
810: return s;
1.24 paf 811: }
812:
813: #ifdef WIN32
1.126 paf 814: void back_slashes_to_slashes(char* s) {
1.24 paf 815: if(s)
816: for(; *s; s++)
817: if(*s=='\\')
1.126 paf 818: *s='/';
1.24 paf 819: }
1.42 paf 820: /*
1.126 paf 821: void slashes_to_back_slashes(char* s) {
1.42 paf 822: if(s)
823: for(; *s; s++)
824: if(*s=='/')
1.126 paf 825: *s='\\';
1.42 paf 826: }
827: */
1.24 paf 828: #endif
1.41 paf 829:
1.126 paf 830: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 831: while(true) {
832: if(!(*s1)) {
833: if(!(*s2))
834: return true;
835: else
836: return !strict;
837: } else if(!(*s2))
838: return !strict;
839: if(isalpha(*s1)) {
840: if(tolower(*s1) !=tolower(*s2))
841: return false;
842: } else if((*s1) !=(*s2))
843: return false;
1.126 paf 844: s1++;
845: s2++;
1.41 paf 846: }
1.57 parser 847: }
848:
1.84 paf 849: static bool isLeap(int year) {
1.57 parser 850: return !(
851: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 852: );
1.57 parser 853: }
854:
855: int getMonthDays(int year, int month) {
856: int monthDays[]={
1.126 paf 857: 31,
858: isLeap(year) ? 29 : 28,
859: 31,
860: 30,
861: 31,
862: 30,
863: 31,
864: 31,
865: 30,
866: 31,
867: 30,
1.57 parser 868: 31
1.126 paf 869: };
870: return monthDays[month];
1.41 paf 871: }
1.69 parser 872:
1.126 paf 873: void remove_crlf(char* start, char* end) {
874: for(char* p=start; p<end; p++)
1.69 parser 875: switch(*p) {
1.126 paf 876: case '\n': *p='|'; break;
877: case '\r': *p=' '; break;
1.69 parser 878: }
1.91 paf 879: }
880:
881:
882: /// must be last in this file
883: #undef vsnprintf
1.126 paf 884: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 885: if(!s)
886: return 0;
887:
888: int r;
889: // note: on win32& maybe somewhere else
890: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
891: --s;
892: #if _MSC_VER
893: /*
894: win32:
895: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
896:
897: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
898: */
1.126 paf 899: r=_vsnprintf(b, s, f, l);
1.91 paf 900: if(r<0)
901: r=s;
902: #else
1.126 paf 903: r=vsnprintf(b, s, f, l);
1.91 paf 904: /*
905: solaris:
906: man vsnprintf
907:
908: The snprintf() function returns the number of characters
909: formatted, that is, the number of characters that would have
910: been written to the buffer if it were large enough. If the
911: value of n is 0 on a call to snprintf(), an unspecified
912: value less than 1 is returned.
913: */
914:
915: if(r<0)
916: r=0;
917: else if(r>s)
918: r=s;
919: #endif
920: b[r]=0;
921: return r;
922: }
923:
1.126 paf 924: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 925: va_list l;
1.126 paf 926: va_start(l, f);
927: int r=__vsnprintf(b, s, f, l);
928: va_end(l);
1.91 paf 929: return r;
1.98 paf 930: }
931:
932: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 933: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 934:
935: #ifdef WIN32
1.126 paf 936: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 937: return 0;
938: #else
939: struct timeval t;
940: t.tv_sec = secs;
941: t.tv_usec = usecs;
1.126 paf 942: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 943: #endif
1.135 ! paf 944: }
! 945:
! 946:
! 947: // attributed meaning
! 948:
! 949: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
! 950: const char month_names[12][4]={
! 951: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
! 952: const char days[7][4]={
! 953: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
! 954:
! 955: time_t when=vdate.get_time();
! 956: struct tm *tms=gmtime(&when);
! 957: if(!tms)
! 958: throw Exception(0,
! 959: 0,
! 960: "bad time in attribute value (seconds from epoch=%ld)", when);
! 961: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
! 962: days[tms->tm_wday],
! 963: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
! 964: tms->tm_hour,tms->tm_min,tms->tm_sec);
! 965: }
! 966: static void append_attribute_meaning(String& result,
! 967: Value& value, String::Untaint_lang lang) {
! 968: if(const String *string=value.get_string())
! 969: result.append(string->join_chains(result.pool(), 0), lang, true);
! 970: else
! 971: if(Value *vdate=value.as(VDATE_TYPE, false)) {
! 972: char *buf=(char *)result.malloc(MAX_STRING);
! 973: size_t size=date_attribute(*static_cast<VDate *>(vdate),
! 974: buf, MAX_STRING);
! 975:
! 976: result.APPEND_CLEAN(buf, size, "converted from date", 0);
! 977: } else
! 978: throw Exception("parser.runtime",
! 979: &result,
! 980: "trying to append here neither string nor date (%s)",
! 981: value.type());
! 982: }
! 983: #ifndef DOXYGEN
! 984: struct Attributed_meaning_info {
! 985: String *header; // header line being constructed
! 986: String::Untaint_lang lang; // language in which to append to that line
! 987: };
! 988: #endif
! 989: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
! 990: void *info) {
! 991: if(akey==VALUE_NAME)
! 992: return;
! 993:
! 994: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
! 995:
! 996: // ...; charset=windows1251
! 997: *ami.header << "; ";
! 998: ami.header->append(akey, ami.lang, true);
! 999: *ami.header << "=";
! 1000: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang);
! 1001: }
! 1002: const String& attributed_meaning_to_string(Value& meaning,
! 1003: String::Untaint_lang lang) {
! 1004: String &result=*new(meaning.pool()) String(meaning.pool());
! 1005: if(Hash *hash=meaning.get_hash(0)) {
! 1006: // $value(value) $subattribute(subattribute value)
! 1007: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
! 1008: append_attribute_meaning(result, *value, lang);
! 1009:
! 1010: Attributed_meaning_info attributed_meaning_info={&result, lang};
! 1011: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
! 1012: } else // result value
! 1013: append_attribute_meaning(result, meaning, lang);
! 1014:
! 1015: return result;
1.74 parser 1016: }
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