Annotation of parser3/src/main/pa_common.C, revision 1.138
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:
1.138 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2002/12/15 13:53:31 $";
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) {
1.87 paf 91: //_asm int 3;
1.126 paf 92: const char* const eob=buf+size;
93: char* dest=buf;
1.72 parser 94: // fix DOS: \r\n -> \n
95: // fix Macintosh: \r -> \n
1.126 paf 96: char* bol=buf;
1.137 paf 97: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 98: size_t len=eol-bol;
99: if(dest!=bol)
1.126 paf 100: memcpy(dest, bol, len);
1.72 parser 101: dest+=len;
1.126 paf 102: *dest++='\n';
1.72 parser 103:
1.126 paf 104: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 105: bol=eol+2;
1.126 paf 106: size--;
107: } else // \r, not \n = Macintosh
1.72 parser 108: bol=eol+1;
109: }
110: // last piece without \r, including terminating 0
111: if(dest!=bol)
1.126 paf 112: memcpy(dest, bol, eob-bol);
1.72 parser 113: }
1.18 paf 114:
1.126 paf 115: char* file_read_text(Pool& pool, const String& file_spec,
116: bool fail_on_read_problem,
117: Hash *params, Hash** out_fields) {
1.72 parser 118: void *result; size_t size;
1.126 paf 119: return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;
120: }
121:
122: //http request stuff
123: /* ************************ http stuff *********************** */
124:
125: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
126: memset(addr, 0, sizeof(*addr));
127: addr->sin_family=AF_INET;
128: addr->sin_port=htons(port);
129: if(host) {
130: if(struct hostent *hostIP=gethostbyname(host))
131: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
132: else
133: return false;
134: } else
135: addr->sin_addr.s_addr=INADDR_ANY;
136: return true;
137: }
138:
139: static void http_read_response(String& response, int sock){
1.130 paf 140: bool check_status=true;
141: ssize_t EOLat=0;
1.126 paf 142: while(true) {
143: char *buf=(char *)response.pool().malloc(MAX_STRING);
144: ssize_t size=recv(sock, buf, MAX_STRING, 0);
145: if(size<=0)
146: break;
1.130 paf 147: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
148: if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
149: check_status=false;
150:
151: const String& status_line=response.mid(0, (size_t)EOLat);
152: Array astatus(response.pool());
153: size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
154: const String* status_code=astatus.get_string(1);
155:
156: if(!status_code || *status_code!="200")
157: throw Exception("http.status",
158: status_code,
159: "invalid HTTP response status");
160: }
1.126 paf 161: }
1.130 paf 162: if(check_status)
163: throw Exception("http.response",
164: 0,
165: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 166: }
167:
168: /* ********************** request *************************** */
169:
170: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
171: # define WE_CAN_USE_ALARM
172: #endif
173:
174: #ifdef WE_CAN_USE_ALARM
175: static sigjmp_buf timeout_env;
176: static void timeout_handler(int sig){
177: siglongjmp(timeout_env, 1);
178: }
179: #endif
180:
181: static void http_request(String& response,
182: const String *origin_string,
183: const char* host, int port,
184: const char* request,
185: int timeout){
186: if(!host)
187: throw Exception("http.host",
188: origin_string,
189: "zero hostname"); //never
190:
191: #ifdef WE_CAN_USE_ALARM
192: signal(SIGALRM, timeout_handler);
193: #endif
194: int sock=-1;
195: try {
196: #ifdef WE_CAN_USE_ALARM
197: if(sigsetjmp(timeout_env, 1))
198: throw Exception("http.timeout",
199: origin_string,
200: "timeout occured while retrieving document");
201: else {
202: alarm(timeout);
203: #endif
204: struct sockaddr_in dest;
205:
206: if(!set_addr(&dest, host, port))
207: throw Exception("http.host",
208: origin_string,
1.127 paf 209: "can not resolve hostname \"%s\"", host);
1.126 paf 210:
211: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
212: throw Exception("http.connect",
213: origin_string,
1.127 paf 214: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 215: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
216: throw Exception("http.connect",
217: origin_string,
1.127 paf 218: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 219: size_t request_size=strlen(request);
220: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
221: throw Exception("http.connect",
222: origin_string,
1.127 paf 223: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 224:
225: http_read_response(response, sock);
226: if(sock>=0)
227: closesocket(sock);
228: #ifdef WE_CAN_USE_ALARM
229: }
230: #endif
231: } catch(...) {
232: if(sock>=0)
233: closesocket(sock);
234: #ifdef WE_CAN_USE_ALARM
235: alarm(0);
236: #endif
237: /*re*/throw;
238: }
239: }
240:
1.127 paf 241: #ifndef DOXYGEN
242: struct Http_pass_header_info {
243: String* request;
244: bool user_agent_specified;
245: };
246: #endif
247: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 248: {
1.127 paf 249: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
250: Pool& pool=i.request->pool();
251:
1.135 paf 252: *(i.request)<<key<<": "
1.136 paf 253: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 254: <<"\n";
255:
1.127 paf 256: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
257: i.user_agent_specified=true;
1.126 paf 258: }
259: static void file_read_http(Pool& pool, const String& file_spec,
260: void*& data, size_t& data_size,
1.127 paf 261: Hash *options=0, Hash** out_fields=0) {
1.126 paf 262: char host[MAX_STRING];
1.129 paf 263: const char* uri;
1.126 paf 264: int port;
265: const char* method="GET";
266: int timeout=2;
1.127 paf 267: Value *vheaders=0;
1.126 paf 268:
1.133 paf 269: String connect_string(pool);
270: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
271: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
272:
273: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
274: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 275: throw Exception(0,
1.133 paf 276: &connect_string,
1.126 paf 277: "does not start with http://"); //never
1.133 paf 278: connect_string_cstr+=7;
1.126 paf 279:
1.133 paf 280: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 281: char* host_uri=lsplit(host, '/');
1.133 paf 282: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 283: char* port_cstr=lsplit(host, ':');
284: char* error_pos=0;
285: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
286:
1.127 paf 287: if(options) {
288: int valid_options=0;
289: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
290: valid_options++;
291: method=vmethod->as_string().cstr();
292: }
293: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
294: valid_options++;
295: timeout=vtimeout->as_int();
296: }
297: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
298: valid_options++;
299: }
300: if(valid_options!=options->size())
301: throw Exception("parser.runtime",
302: 0,
303: "invalid option passed");
1.133 paf 304: }
1.126 paf 305:
306: //making request
1.127 paf 307: String request(pool);
1.126 paf 308: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 309: bool user_agent_specified=false;
310: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 311: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 312: Http_pass_header_info info={&request};
313: headers->for_each(http_pass_header, &info);
314: user_agent_specified=info.user_agent_specified;
315: } else
316: throw Exception("parser.runtime",
1.133 paf 317: &connect_string,
1.127 paf 318: "headers param must be hash");
319: };
320: if(!user_agent_specified) // defaulting
321: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 322: request<<"\n";
323:
324: //sending request
325: String response(pool);
326: http_request(response,
1.133 paf 327: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout);
1.126 paf 328:
329: //processing results
330: int pos=response.pos("\r\n\r\n", 4);
331: if(pos<1){
332: throw Exception("http.response",
1.133 paf 333: &connect_string,
1.126 paf 334: "bad response from host - no headers found");
335: }
336: String header_block=response.mid(0, pos);
337: String body=response.mid(pos+4, response.size());
338:
339: Array aheaders(pool);
340: Hash& headers=*new(pool) Hash(pool);
341: size_t pos_after_ref=0;
342: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
343:
344: //processing status code
345: const String *status_line=aheaders.get_string(0);
346: if(!status_line){
347: throw Exception("http.response",
1.133 paf 348: &connect_string,
1.126 paf 349: "bad response from host - no status line ");
350: }
351: //processing headers
352: for(int i=1;i<aheaders.size();i++) {
353: if(const String *line=aheaders.get_string(i)) {
354: pos=line->pos(": ", 2);
355: if(pos<1)
356: throw Exception("http.response",
1.133 paf 357: &connect_string,
1.126 paf 358: "bad response from host - bad header \"%s\"", line->cstr());
359:
1.128 paf 360: headers.put(
361: line->mid(0, pos).change_case(pool, String::CC_UPPER),
362: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 363: } else
364: throw Exception("http.response",
1.133 paf 365: &connect_string,
1.126 paf 366: "bad response from host - bad headers \"%s\"", header_block.cstr());
367: }
368:
369: // output response
370: data=body.cstr(); data_size=body.size();
1.130 paf 371: if(out_fields)
1.126 paf 372: *out_fields=&headers;
1.34 paf 373: }
1.123 paf 374:
375: #ifndef DOXYGEN
376: struct File_read_action_info {
377: void **data; size_t *data_size;
1.126 paf 378: };
1.123 paf 379: #endif
1.126 paf 380: static void file_read_action(Pool& pool,
381: struct stat& finfo,
1.123 paf 382: int f,
1.126 paf 383: const String& file_spec, const char* fname, bool as_text,
1.123 paf 384: void *context) {
1.126 paf 385: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 386: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 387: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
388: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 389:
390: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 391: throw Exception(0,
1.123 paf 392: &file_spec,
393: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 394: *info.data_size, to_read_size);
1.123 paf 395: } else { // empty file
396: if(as_text) {
1.126 paf 397: *info.data=pool.malloc(1);
1.123 paf 398: *(char*)(*info.data)=0;
399: } else
400: *info.data=0;
401: *info.data_size=0;
402: return;
403: }
1.126 paf 404: }
405: bool file_read(Pool& pool, const String& file_spec,
406: void*& data, size_t& data_size,
407: bool as_text, Hash *params, Hash** out_fields,
408: bool fail_on_read_problem) {
409: bool result;
410: if(file_spec.starts_with("http://", 7)) {
411: // fail on read problem
412: file_read_http(pool, file_spec, data, data_size, params, out_fields);
413: result=true;
414: } else {
415: File_read_action_info info={&data, &data_size};
416: result=file_read_action_under_lock(pool, file_spec,
417: "read", file_read_action, &info,
418: as_text, fail_on_read_problem);
419: }
1.123 paf 420:
1.126 paf 421: if(result && as_text) {
1.131 paf 422: // UTF-8 signature: EF BB BF
423: if(data_size>=3) {
424: char *in=(char *)data;
425: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
426: (in[2] == '\xBF')) {
427: data=in+3; data_size-=3;// skip prefix
428: }
429: }
430:
1.126 paf 431: fix_line_breaks((char *)(data), data_size);
1.123 paf 432: // note: after fixing
1.126 paf 433: ((char*&)(data))[data_size]=0;
1.123 paf 434: }
1.126 paf 435:
436: return result;
1.123 paf 437: }
438:
439: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 440: const char* action_name, File_read_action action, void *context,
441: bool as_text,
1.123 paf 442: bool fail_on_read_problem) {
1.126 paf 443: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 444: int f;
445:
446: // first open, next stat:
1.45 paf 447: // directory update of NTFS hard links performed on open.
1.33 paf 448: // ex:
449: // a.html:^test[] and b.html hardlink to a.html
450: // user inserts ! before ^test in a.html
1.126 paf 451: // directory entry of b.html in NTFS not updated at once,
1.35 paf 452: // they delay update till open, so we would receive "!^test[" string
453: // if would do stat, next open.
1.123 paf 454: // later: it seems, even this does not help sometimes
1.98 paf 455: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 456: try {
457: if(lock_shared_blocking(f)!=0)
1.126 paf 458: throw Exception("file.lock",
1.123 paf 459: &file_spec,
460: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 461: strerror(errno), errno, fname);
1.123 paf 462:
1.124 paf 463: struct stat finfo;
464: if(stat(fname, &finfo)!=0)
465: throw Exception("file.missing", // hardly possible: we just opened it OK
466: &file_spec,
467: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 468: strerror(errno), errno, fname);
1.124 paf 469:
1.105 paf 470: #ifdef NO_FOREIGN_GROUP_FILES
1.132 paf 471: if(finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 472: throw Exception("parser.runtime",
473: &file_spec,
1.123 paf 474: "parser reading files of foreign group disabled [recompile parser without --disable-foreign-group-files configure option], actual filename '%s'",
1.126 paf 475: fname);
1.105 paf 476: #endif
1.32 paf 477:
1.126 paf 478: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 479: } catch(...) {
1.126 paf 480: unlock(f);close(f);
1.123 paf 481: if(fail_on_read_problem)
482: /*re*/throw;
483: return false;
484: }
1.87 paf 485:
1.126 paf 486: unlock(f);close(f);
1.72 parser 487: return true;
1.118 paf 488: } else {
489: if(fail_on_read_problem)
1.126 paf 490: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 491: &file_spec,
1.123 paf 492: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 493: action_name, strerror(errno), errno, fname);
1.118 paf 494: return false;
495: }
1.8 paf 496: }
497:
1.63 parser 498: static void create_dir_for_file(const String& file_spec) {
499: size_t pos_after=1;
500: int pos_before;
501: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 502: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 503: pos_after=pos_before+1;
504: }
505: }
506:
1.98 paf 507: bool file_write_action_under_lock(
1.28 paf 508: const String& file_spec,
1.126 paf 509: const char* action_name, File_write_action action, void *context,
510: bool as_text,
511: bool do_append,
512: bool do_block,
1.110 paf 513: bool fail_on_lock_problem) {
1.126 paf 514: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 515: int f;
1.80 paf 516: if(access(fname, W_OK)!=0) // no
1.126 paf 517: create_dir_for_file(file_spec);
1.50 paf 518:
1.80 paf 519: if((f=open(fname,
520: O_CREAT|O_RDWR
521: |(as_text?_O_TEXT:_O_BINARY)
1.138 ! paf 522: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 523: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 524: Exception e("file.lock",
1.110 paf 525: &file_spec,
526: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 527: strerror(errno), errno, fname);
528: close(f);
1.110 paf 529: if(fail_on_lock_problem)
530: throw e;
1.98 paf 531: return false;
532: }
1.96 paf 533:
534: try {
1.126 paf 535: action(f, context);
1.96 paf 536: } catch(...) {
1.138 ! paf 537: #ifdef HAVE_FTRUNCATE
1.104 paf 538: if(!do_append)
1.125 paf 539: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 ! paf 540: #endif
1.126 paf 541: unlock(f);close(f);
1.96 paf 542: /*re*/throw;
543: }
1.80 paf 544:
1.138 ! paf 545: #ifdef HAVE_FTRUNCATE
1.104 paf 546: if(!do_append)
1.125 paf 547: 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 548: #endif
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,
1.136 paf 967: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 968: if(const String *string=value.get_string())
1.136 paf 969: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 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
1.136 paf 987: bool forced; // do they force that lang?
1.135 paf 988: };
989: #endif
990: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
991: void *info) {
992: if(akey==VALUE_NAME)
993: return;
994:
995: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
996:
997: // ...; charset=windows1251
998: *ami.header << "; ";
1.136 paf 999: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1000: *ami.header << "=";
1.136 paf 1001: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1002: }
1003: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1004: String::Untaint_lang lang, bool forced) {
1.135 paf 1005: String &result=*new(meaning.pool()) String(meaning.pool());
1006: if(Hash *hash=meaning.get_hash(0)) {
1007: // $value(value) $subattribute(subattribute value)
1008: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1009: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1010:
1011: Attributed_meaning_info attributed_meaning_info={&result, lang};
1012: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1013: } else // result value
1.136 paf 1014: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1015:
1016: return result;
1.74 parser 1017: }
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