Annotation of parser3/src/main/pa_common.C, revision 1.143.2.21.2.8
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.143.2.10 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:
1.143.2.21.2. (paf 8:): static const char* IDENT_COMMON_C="$Date: 2003/03/20 14:19:09 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.126 paf 12: #include "pa_hash.h"
1.143.2.21.2. (paf 13:): #include "pa_globals.h"
14:):
15:): //#define PA_HTTP
16:):
17:): #ifdef PA_HTTP
1.143.2.3 paf 18: #include "pa_vstring.h"
1.143.2.4 paf 19: #include "pa_vint.h"
1.1 paf 20:
1.143.2.12 paf 21: #ifdef CYGWIN
22: #define _GNU_H_WINDOWS32_SOCKETS
23: // for PASCAL
24: #include <windows.h>
25: // SOCKET
26: typedef u_int SOCKET;
27: int PASCAL closesocket(SOCKET);
1.126 paf 28: #else
1.143.2.12 paf 29: # if defined(WIN32)
30: # include <windows.h>
31: # else
32: # define closesocket close
33: # endif
1.98 paf 34: #endif
35:
1.143.2.21.2. (paf 36:): #else
37:):
38:): # if defined(WIN32)
39:): # include <windows.h>
40:): # endif
41:):
42:): #endif
43:):
1.93 paf 44: // some maybe-undefined constants
45:
1.82 paf 46: #ifndef _O_TEXT
47: # define _O_TEXT 0
48: #endif
49: #ifndef _O_BINARY
50: # define _O_BINARY 0
1.47 paf 51: #endif
1.80 paf 52:
1.138 paf 53: #ifdef HAVE_FTRUNCATE
54: # define PA_O_TRUNC 0
55: #else
56: # ifdef _O_TRUNC
57: # define PA_O_TRUNC _O_TRUNC
58: # else
59: # error you must have either ftruncate function or _O_TRUNC bit declared
60: # endif
61: #endif
62:
1.93 paf 63: // locking constants
64:
1.99 paf 65: #ifdef HAVE_FLOCK
66:
67: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
68: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
69: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
70: static int unlock(int fd) { return flock(fd, LOCK_UN); }
71:
1.98 paf 72: #else
1.99 paf 73: #ifdef HAVE__LOCKING
1.98 paf 74:
1.126 paf 75: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 76: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
77: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
78: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
79: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 80:
1.99 paf 81: #else
82: #ifdef HAVE_FCNTL
1.93 paf 83:
1.126 paf 84: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 85: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
86: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
87: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
88: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 89:
90: #else
91: #ifdef HAVE_LOCKF
1.99 paf 92:
1.126 paf 93: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 94: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
95: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
96: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
97: static int unlock(int fd) { FLOCK(F_TLOCK); }
98:
1.93 paf 99: #else
1.99 paf 100:
101: #error unable to find file locking func
102:
103: #endif
1.93 paf 104: #endif
105: #endif
106: #endif
107:
1.143.2.16 paf 108: // defines for globals
109:
110: #define FILE_STATUS_NAME "status"
111:
112: // globals
113:
1.143.2.21.2. (paf 114:): const String file_status_name(FILE_STATUS_NAME);
1.143.2.16 paf 115:
1.143.2.14 paf 116: // defines for statics
117:
118: #define HTTP_METHOD_NAME "method"
119: #define HTTP_TIMEOUT_NAME "timeout"
120: #define HTTP_HEADERS_NAME "headers"
121: #define HTTP_ANY_STATUS_NAME "any-status"
122:
123: // statics
124:
1.143.2.21.2. (paf 125:): static const String http_method_name(HTTP_METHOD_NAME);
126:): static const String http_timeout_name(HTTP_TIMEOUT_NAME);
127:): static const String http_headers_name(HTTP_HEADERS_NAME);
128:): static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
1.143.2.14 paf 129:
130: // defines
131:
1.127 paf 132: #define DEFAULT_USER_AGENT "parser3"
133:
1.143.2.14 paf 134: // functions
1.127 paf 135:
1.126 paf 136: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 137: if(size==0)
138: return;
139:
1.87 paf 140: //_asm int 3;
1.126 paf 141: const char* const eob=buf+size;
142: char* dest=buf;
1.72 parser 143: // fix DOS: \r\n -> \n
144: // fix Macintosh: \r -> \n
1.126 paf 145: char* bol=buf;
1.137 paf 146: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 147: size_t len=eol-bol;
148: if(dest!=bol)
1.126 paf 149: memcpy(dest, bol, len);
1.72 parser 150: dest+=len;
1.126 paf 151: *dest++='\n';
1.72 parser 152:
1.126 paf 153: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 154: bol=eol+2;
1.126 paf 155: size--;
156: } else // \r, not \n = Macintosh
1.72 parser 157: bol=eol+1;
158: }
159: // last piece without \r, including terminating 0
160: if(dest!=bol)
1.126 paf 161: memcpy(dest, bol, eob-bol);
1.72 parser 162: }
1.18 paf 163:
1.143.2.21.2. (paf 164:): char* file_read_text(Charset& charset,
165:): const String& file_spec,
1.126 paf 166: bool fail_on_read_problem,
1.143.2.21.2. (paf 167:): HashStringValue* params/*, HashStringValue* * out_fields*/) {
1.143.2.18 paf 168: File_read_result file=
1.143.2.21.2. (paf 169:): file_read(charset, file_spec, true, params, fail_on_read_problem);
170:): return file.success?file.str:0;
1.126 paf 171: }
172:
1.143.2.21.2. (paf 173:): #ifdef PA_HTTP
1.126 paf 174: //http request stuff
175:
176: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
177: memset(addr, 0, sizeof(*addr));
178: addr->sin_family=AF_INET;
179: addr->sin_port=htons(port);
180: if(host) {
181: if(struct hostent *hostIP=gethostbyname(host))
182: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
183: else
184: return false;
185: } else
186: addr->sin_addr.s_addr=INADDR_ANY;
187: return true;
188: }
189:
1.143.2.21.2. (paf 190:): static int http_read_responseString& response, int sock, bool fail_on_status_ne_200){
1.143.2.2 paf 191: int result=0;
1.130 paf 192: ssize_t EOLat=0;
1.126 paf 193: while(true) {
1.143.2.21.2. (paf 194:): char *buf=new(PointerFreeGC) char[MAX_STRING];
1.126 paf 195: ssize_t size=recv(sock, buf, MAX_STRING, 0);
196: if(size<=0)
197: break;
1.130 paf 198: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
1.143.2.2 paf 199: if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
1.143.2.21.2. (paf 200:): const String& status_line=response.mid(0, (size_t)EOLat);
1.143.2.15 paf 201: ArrayString astatus;
1.143.2.3 paf 202: status_line->split(astatus, 0, " ", 1);
1.143.2.21.2. (paf 203:): const String& status_code=astatus.get(1);
1.143.2.3 paf 204: result=status_code->as_int();
1.142 paf 205:
1.143.2.3 paf 206: if(fail_on_status_ne_200 && result!=200)
1.142 paf 207: throw Exception("http.status",
208: status_code,
209: "invalid HTTP response status");
210: }
211: }
1.143.2.2 paf 212: if(result)
213: return result;
1.142 paf 214: else
215: throw Exception("http.response",
1.143.2.21.2. (paf 216:): 0,
217:): "bad response from host - no status found (size=%lu)", response.length());
1.126 paf 218: }
219:
220: /* ********************** request *************************** */
221:
222: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
223: # define WE_CAN_USE_ALARM
224: #endif
225:
226: #ifdef WE_CAN_USE_ALARM
227: static sigjmp_buf timeout_env;
228: static void timeout_handler(int sig){
229: siglongjmp(timeout_env, 1);
230: }
231: #endif
232:
1.143.2.21.2. (paf 233:): static int http_request(
1.143.2.3 paf 234: String& response,
1.143.2.21.2. (paf 235:): const String* origin_string,
1.143.2.3 paf 236: const char* host, int port,
237: const char* request,
238: int timeout,
239: bool fail_on_status_ne_200) {
1.126 paf 240: if(!host)
241: throw Exception("http.host",
242: origin_string,
243: "zero hostname"); //never
244:
245: #ifdef WE_CAN_USE_ALARM
246: signal(SIGALRM, timeout_handler);
247: #endif
248: int sock=-1;
249: try {
1.142 paf 250: int result;
1.126 paf 251: #ifdef WE_CAN_USE_ALARM
252: if(sigsetjmp(timeout_env, 1))
253: throw Exception("http.timeout",
254: origin_string,
255: "timeout occured while retrieving document");
256: else {
257: alarm(timeout);
258: #endif
259: struct sockaddr_in dest;
260:
261: if(!set_addr(&dest, host, port))
262: throw Exception("http.host",
263: origin_string,
1.127 paf 264: "can not resolve hostname \"%s\"", host);
1.126 paf 265:
266: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
267: throw Exception("http.connect",
268: origin_string,
1.127 paf 269: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 270: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
271: throw Exception("http.connect",
272: origin_string,
1.127 paf 273: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 274: size_t request_size=strlen(request);
275: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
276: throw Exception("http.connect",
277: origin_string,
1.127 paf 278: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 279:
1.143.2.21.2. (paf 280:): result=http_read_response(response, sock, fail_on_status_ne_200);
1.142 paf 281: closesocket(sock);
1.126 paf 282: #ifdef WE_CAN_USE_ALARM
1.142 paf 283: alarm(0);
1.126 paf 284: }
285: #endif
1.142 paf 286: return result;
1.126 paf 287: } catch(...) {
288: if(sock>=0)
289: closesocket(sock);
290: #ifdef WE_CAN_USE_ALARM
291: alarm(0);
292: #endif
1.143.2.11 paf 293: rethrow;
1.126 paf 294: }
295: }
296:
1.127 paf 297: #ifndef DOXYGEN
298: struct Http_pass_header_info {
1.143.2.21.2. (paf 299:): ;
1.143.2.3 paf 300: Charset* charset;
1.127 paf 301: String* request;
302: bool user_agent_specified;
303: };
304: #endif
1.143.2.3 paf 305: static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value,
306: Http_pass_header_info *info) {
307: *(info->request)<<*key<<": "
1.143.2.21.2. (paf 308:): << *attributed_meaning_to_string(*info->value, String::L_HTTP_HEADER, false)
1.135 paf 309: <<"\n";
310:
1.143.2.21.2. (paf 311:): if(*key->change_case(*info->*info->charset, String::CC_UPPER)=="USER-AGENT")
1.143.2.3 paf 312: info->user_agent_specified=true;
1.126 paf 313: }
1.143.2.16 paf 314: #ifndef DOXYGEN
1.143.2.21 paf 315: struct File_read_http_result {
316: char *data; size_t size;
1.143.2.21.2. (paf 317:): HashStringValue* headers;
1.143.2.16 paf 318: };
319: #endif
1.143.2.21.2. (paf 320:): static File_read_http_result file_read_httpCharset& charset, const String& file_spec,
1.143.2.16 paf 321: HashStringValue *options=0) {
1.143.2.21 paf 322: File_read_http_result result;
1.126 paf 323: char host[MAX_STRING];
1.129 paf 324: const char* uri;
1.126 paf 325: int port;
1.143.2.21.2. (paf 326:): const char* method(0);
1.126 paf 327: int timeout=2;
1.142 paf 328: bool fail_on_status_ne_200=true;
1.143.2.21.2. (paf 329:): Value* vheaders(0);
1.126 paf 330:
1.143.2.21.2. (paf 331:): const String& connect_string(new String());
332:): // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
333:): connect_string->append(*file_spec, String::L_URI); // tainted pieces -> URI pieces
1.133 paf 334:
1.143.2.21.2. (paf 335:): const char* connect_string_cstr=connect_string->cstr(String::L_UNSPECIFIED);
1.143.2.10 paf 336: const char* current=connect_string_cstr;
1.143.2.3 paf 337: if(strncmp(current, "http://", 7)!=0)
1.126 paf 338: throw Exception(0,
1.143.2.3 paf 339: connect_string,
1.126 paf 340: "does not start with http://"); //never
1.143.2.3 paf 341: current+=7;
1.126 paf 342:
1.143.2.3 paf 343: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 344: char* host_uri=lsplit(host, '/');
1.143.2.3 paf 345: uri=host_uri?current+(host_uri-1-host):"/";
1.126 paf 346: char* port_cstr=lsplit(host, ':');
347: char* error_pos=0;
348: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
349:
1.127 paf 350: if(options) {
351: int valid_options=0;
1.143.2.21.2. (paf 352:): if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 353: valid_options++;
1.143.2.21.2. (paf 354:): method=vmethod->as_string()->cstr();
1.127 paf 355: }
1.143.2.21.2. (paf 356:): if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 357: valid_options++;
358: timeout=vtimeout->as_int();
359: }
1.143.2.3 paf 360: if(vheaders=options->get(http_headers_name)) {
1.127 paf 361: valid_options++;
362: }
1.143.2.21.2. (paf 363:): if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 364: valid_options++;
365: fail_on_status_ne_200=!vany_status->as_bool();
366: }
367:
1.143.2.3 paf 368: if(valid_options!=options->count())
1.127 paf 369: throw Exception("parser.runtime",
1.143.2.21.2. (paf 370:): 0,
1.127 paf 371: "invalid option passed");
1.133 paf 372: }
1.126 paf 373:
374: //making request
1.143.2.3 paf 375: String request;
376: if(method)
1.143.2.4 paf 377: request<<method;
1.143.2.3 paf 378: else
379: request<<"GET";
380: request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 381: bool user_agent_specified=false;
382: if(vheaders && !vheaders->is_string()) { // allow empty
1.143.2.8 paf 383: if(HashStringValue *headers=vheaders->get_hash(connect_string)) {
1.143.2.21.2. (paf 384:): Http_pass_header_info info={&&charset, &request};
1.127 paf 385: headers->for_each(http_pass_header, &info);
386: user_agent_specified=info.user_agent_specified;
387: } else
388: throw Exception("parser.runtime",
1.143.2.3 paf 389: connect_string,
1.127 paf 390: "headers param must be hash");
391: };
392: if(!user_agent_specified) // defaulting
393: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 394: request<<"\n";
395:
396: //sending request
1.143.2.3 paf 397: String response;
1.143.2.21.2. (paf 398:): const char* request_cstr=request.cstr(String::L_UNSPECIFIED);
399:): int status_code=http_request(response,
1.143.2.4 paf 400: connect_string, host, port, request_cstr,
1.142 paf 401: timeout, fail_on_status_ne_200);
1.126 paf 402:
403: //processing results
404: int pos=response.pos("\r\n\r\n", 4);
405: if(pos<1){
406: throw Exception("http.response",
1.143.2.3 paf 407: connect_string,
1.126 paf 408: "bad response from host - no headers found");
409: }
1.143.2.21.2. (paf 410:): const String& header_block=response.mid(0, pos);
411:): const String& body=response.mid(pos+4, response.length());
1.126 paf 412:
1.143.2.15 paf 413: ArrayString aheaders;
1.143.2.21.2. (paf 414:): result.headers=HashStringValue* (new HashStringValue);
1.143.2.3 paf 415: header_block->split(aheaders, 0, "\r\n", 2);
1.126 paf 416:
417: //processing headers
1.143.2.3 paf 418: for(int i=1; i<aheaders.count(); i++) {
1.143.2.21.2. (paf 419:): const String& line=aheaders.get(i);
1.143.2.3 paf 420: pos=line->pos(": ", 2);
421: if(pos<1)
1.126 paf 422: throw Exception("http.response",
1.143.2.3 paf 423: connect_string,
1.143.2.12 paf 424: "bad response from host - bad header \"%s\"", line->cstr().get());
1.143.2.3 paf 425:
1.143.2.16 paf 426: result.headers->put(
1.143.2.21.2. (paf 427:): line->mid(0, pos)->change_case(charset, String::CC_UPPER),
428:): Value*(new VString(line->mid(pos+2, line->length()))));
1.126 paf 429: }
430:
431: // output response
1.143.2.21.2. (paf 432:): result.str=new(PointerFreeGC) char[1/*as text puts terminating zero later*/+(result.length=body->length())];
433:): body->store_to(result.str, String::L_AS_IS);
434:): result.headers->put(file_status_name, Value*(new VInt(status_code)));
1.143.2.16 paf 435: return result;
1.34 paf 436: }
1.123 paf 437:
1.143.2.21.2. (paf 438:): #endif
439:):
1.123 paf 440: #ifndef DOXYGEN
441: struct File_read_action_info {
1.143.2.4 paf 442: char **data; size_t *data_size;
1.126 paf 443: };
1.123 paf 444: #endif
1.143.2.21.2. (paf 445:): static void file_read_action(
1.126 paf 446: struct stat& finfo,
1.123 paf 447: int f,
1.143.2.21.2. (paf 448:): const String& file_spec, const char* fname, bool as_text,
1.123 paf 449: void *context) {
1.126 paf 450: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 451: if(size_t to_read_size=(size_t)finfo.st_size) {
1.143.2.21.2. (paf 452:): *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 453: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 454:
455: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 456: throw Exception(0,
1.143.2.21.2. (paf 457:): &file_spec,
1.123 paf 458: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 459: *info.data_size, to_read_size);
1.123 paf 460: } else { // empty file
461: if(as_text) {
1.143.2.21.2. (paf 462:): *info.data=new(PointerFreeGC) char[1];
1.123 paf 463: *(char*)(*info.data)=0;
464: } else
465: *info.data=0;
466: *info.data_size=0;
467: return;
468: }
1.126 paf 469: }
1.143.2.21.2. (paf 470:): File_read_result file_read(Charset& charset, const String& file_spec,
1.143.2.16 paf 471: bool as_text, HashStringValue *params,
1.126 paf 472: bool fail_on_read_problem) {
1.143.2.4 paf 473: File_read_result result;
1.143.2.21.2. (paf 474:): #ifdef PA_HTTP
475:): if(file_spec.starts_with("http://", 7)) {
1.126 paf 476: // fail on read problem
1.143.2.21.2. (paf 477:): File_read_http_result http=file_read_http(charset, file_spec, params);
1.143.2.4 paf 478: result.success=true;
1.143.2.21.2. (paf 479:): result.str=http.data;
480:): result.length=http.size;
1.143.2.21 paf 481: result.headers=http.headers;
1.126 paf 482: } else {
1.143.2.21.2. (paf 483:): #endif
484:): File_read_action_info info={&result.str, &result.length};
485:): result.success=file_read_action_under_lock(file_spec,
1.126 paf 486: "read", file_read_action, &info,
487: as_text, fail_on_read_problem);
1.143.2.21.2. (paf 488:): #ifdef PA_HTTP
1.126 paf 489: }
1.143.2.21.2. (paf 490:): #endif
1.123 paf 491:
1.143.2.4 paf 492: if(result.success && as_text) {
1.131 paf 493: // UTF-8 signature: EF BB BF
1.143.2.21.2. (paf 494:): if(result.length>=3) {
495:): char *in=(char *)result.str;
1.131 paf 496: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
497: (in[2] == '\xBF')) {
1.143.2.21.2. (paf 498:): result.str=in+3; result.length-=3;// skip prefix
1.131 paf 499: }
500: }
501:
1.143.2.21.2. (paf 502:): fix_line_breaks((char *)(result.str), result.length);
1.123 paf 503: // note: after fixing
1.143.2.21.2. (paf 504:): ((char*&)(result.str))[result.length]=0;
1.123 paf 505: }
1.126 paf 506:
507: return result;
1.123 paf 508: }
509:
1.143.2.21.2. (paf 510:): bool file_read_action_under_lock(const String& file_spec,
1.126 paf 511: const char* action_name, File_read_action action, void *context,
512: bool as_text,
1.123 paf 513: bool fail_on_read_problem) {
1.143.2.21.2. (paf 514:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 515: int f;
516:
517: // first open, next stat:
1.45 paf 518: // directory update of NTFS hard links performed on open.
1.33 paf 519: // ex:
520: // a.html:^test[] and b.html hardlink to a.html
521: // user inserts ! before ^test in a.html
1.126 paf 522: // directory entry of b.html in NTFS not updated at once,
1.35 paf 523: // they delay update till open, so we would receive "!^test[" string
524: // if would do stat, next open.
1.123 paf 525: // later: it seems, even this does not help sometimes
1.98 paf 526: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 527: try {
528: if(lock_shared_blocking(f)!=0)
1.126 paf 529: throw Exception("file.lock",
1.143.2.21.2. (paf 530:): &file_spec,
1.123 paf 531: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 532:): strerror(errno), errno, fname);
1.123 paf 533:
1.124 paf 534: struct stat finfo;
535: if(stat(fname, &finfo)!=0)
536: throw Exception("file.missing", // hardly possible: we just opened it OK
1.143.2.21.2. (paf 537:): &file_spec,
1.124 paf 538: "stat failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 539:): strerror(errno), errno, fname);
1.124 paf 540:
1.140 paf 541: #ifdef PA_SAFE_MODE
542: if(finfo.st_uid/*foreign?*/!=geteuid()
543: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 544: throw Exception("parser.runtime",
1.143.2.21.2. (paf 545:): &file_spec,
1.140 paf 546: "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
1.143.2.21.2. (paf 547:): fname);
1.105 paf 548: #endif
1.32 paf 549:
1.143.2.21.2. (paf 550:): action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 551: } catch(...) {
1.126 paf 552: unlock(f);close(f);
1.123 paf 553: if(fail_on_read_problem)
1.143.2.11 paf 554: rethrow;
1.123 paf 555: return false;
556: }
1.87 paf 557:
1.126 paf 558: unlock(f);close(f);
1.72 parser 559: return true;
1.118 paf 560: } else {
561: if(fail_on_read_problem)
1.126 paf 562: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 563:): &file_spec,
1.123 paf 564: "%s failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 565:): action_name, strerror(errno), errno, fname);
1.118 paf 566: return false;
567: }
1.8 paf 568: }
569:
1.143.2.21.2. (paf 570:): static void create_dir_for_file(const String& file_spec) {
1.63 parser 571: size_t pos_after=1;
572: int pos_before;
1.143.2.21.2. (paf 573:): while((pos_before=file_spec.pos("/", pos_after))>=0) {
574:): mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 575: pos_after=pos_before+1;
576: }
577: }
578:
1.98 paf 579: bool file_write_action_under_lock(
1.143.2.21.2. (paf 580:): const String& file_spec,
1.126 paf 581: const char* action_name, File_write_action action, void *context,
582: bool as_text,
583: bool do_append,
584: bool do_block,
1.110 paf 585: bool fail_on_lock_problem) {
1.143.2.21.2. (paf 586:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 587: int f;
1.80 paf 588: if(access(fname, W_OK)!=0) // no
1.126 paf 589: create_dir_for_file(file_spec);
1.50 paf 590:
1.80 paf 591: if((f=open(fname,
592: O_CREAT|O_RDWR
593: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 594: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 595: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 596: Exception e("file.lock",
1.143.2.21.2. (paf 597:): &file_spec,
1.110 paf 598: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 599:): strerror(errno), errno, fname);
1.126 paf 600: close(f);
1.110 paf 601: if(fail_on_lock_problem)
602: throw e;
1.98 paf 603: return false;
604: }
1.96 paf 605:
606: try {
1.126 paf 607: action(f, context);
1.96 paf 608: } catch(...) {
1.138 paf 609: #ifdef HAVE_FTRUNCATE
1.104 paf 610: if(!do_append)
1.125 paf 611: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 612: #endif
1.126 paf 613: unlock(f);close(f);
1.143.2.11 paf 614: rethrow;
1.96 paf 615: }
1.80 paf 616:
1.138 paf 617: #ifdef HAVE_FTRUNCATE
1.104 paf 618: if(!do_append)
1.125 paf 619: 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 620: #endif
1.126 paf 621: unlock(f);close(f);
1.98 paf 622: return true;
1.80 paf 623: } else
1.126 paf 624: throw Exception(errno==EACCES?"file.access":0,
1.143.2.21.2. (paf 625:): &file_spec,
1.96 paf 626: "%s failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 627:): action_name, strerror(errno), errno, fname);
1.96 paf 628: // here should be nothing, see rethrow above
629: }
630:
631: #ifndef DOXYGEN
632: struct File_write_action_info {
1.143.2.21.2. (paf 633:): const char* str; size_t length;
1.126 paf 634: };
1.96 paf 635: #endif
636: static void file_write_action(int f, void *context) {
1.126 paf 637: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.143.2.21.2. (paf 638:): if(info.length) {
639:): int written=write(f, info.str, info.length);
1.116 paf 640: if(written<0)
1.126 paf 641: throw Exception(0,
1.143.2.21.2. (paf 642:): 0,
1.126 paf 643: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 644: }
1.96 paf 645: }
646: void file_write(
1.143.2.21.2. (paf 647:): const String& file_spec,
648:): const char* data, size_t size,
1.126 paf 649: bool as_text,
1.96 paf 650: bool do_append) {
1.126 paf 651: File_write_action_info info={data, size};
1.98 paf 652: file_write_action_under_lock(
1.143.2.4 paf 653: file_spec,
654: "write", file_write_action, &info,
655: as_text,
656: do_append);
1.30 paf 657: }
658:
1.63 parser 659: // throws nothing! [this is required in file_move & file_delete]
1.143.2.21.2. (paf 660:): static void rmdir(const String& file_spec, size_t pos_after) {
1.50 paf 661: int pos_before;
1.143.2.21.2. (paf 662:): if((pos_before=file_spec.pos("/", pos_after))>=0)
1.126 paf 663: rmdir(file_spec, pos_before+1);
1.50 paf 664:
1.143.2.21.2. (paf 665:): rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 666: }
1.143.2.21.2. (paf 667:): bool file_delete(const String& file_spec, bool fail_on_read_problem) {
668:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 669: if(unlink(fname)!=0)
1.93 paf 670: if(fail_on_read_problem)
1.126 paf 671: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 672:): &file_spec,
1.93 paf 673: "unlink failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 674:): strerror(errno), errno, fname);
1.93 paf 675: else
676: return false;
1.50 paf 677:
1.126 paf 678: rmdir(file_spec, 1);
1.93 paf 679: return true;
1.60 parser 680: }
1.143.2.21.2. (paf 681:): void file_move(const String& old_spec, const String& new_spec) {
682:): const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
683:): const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 684:
1.126 paf 685: create_dir_for_file(new_spec);
1.63 parser 686:
1.60 parser 687: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 688: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 689:): &old_spec,
1.60 parser 690: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.143.2.21.2. (paf 691:): strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 692:
1.126 paf 693: rmdir(old_spec, 1);
1.31 paf 694: }
695:
1.51 paf 696:
1.126 paf 697: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 698: struct stat lfinfo;
699: bool result=stat(fname, &lfinfo)==0;
700: if(afinfo)
701: *afinfo=lfinfo;
702: return result;
1.119 paf 703: }
704:
1.143.2.21.2. (paf 705:): bool entry_exists(const String& file_spec) {
706:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 707: return entry_exists(fname, 0);
1.118 paf 708: }
709:
1.143.2.21.2. (paf 710:): static bool entry_readable(const String& file_spec, bool need_dir) {
711:): char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 712: if(need_dir) {
1.126 paf 713: size_t size=strlen(fname);
1.120 paf 714: while(size) {
1.126 paf 715: char c=fname[size-1];
1.120 paf 716: if(c=='/' || c=='\\')
717: fname[--size]=0;
718: else
719: break;
720: }
721: }
1.51 paf 722: struct stat finfo;
1.118 paf 723: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 724: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 725: return is_dir==need_dir;
726: }
727: return false;
728: }
1.143.2.21.2. (paf 729:): bool file_readable(const String& file_spec) {
1.126 paf 730: return entry_readable(file_spec, false);
1.51 paf 731: }
1.143.2.21.2. (paf 732:): bool dir_readable(const String& file_spec) {
1.126 paf 733: return entry_readable(file_spec, true);
1.65 parser 734: }
1.143.2.21.2. (paf 735:): const String* file_readable(String& path, String& name) {
736:): String& result=*new String(path);
737:): result << "/";
738:): result << name;
739:): return file_readable(result)?&result:0;
1.43 paf 740: }
1.143.2.21.2. (paf 741:): bool file_executable(const String& file_spec) {
742:): return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 743: }
744:
1.143.2.21.2. (paf 745:): bool file_stat(const String& file_spec,
1.58 parser 746: size_t& rsize,
1.126 paf 747: time_t& ratime,
748: time_t& rmtime,
749: time_t& rctime,
1.64 parser 750: bool fail_on_read_problem) {
1.143.2.21.2. (paf 751:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.44 paf 752: struct stat finfo;
753: if(stat(fname, &finfo)!=0)
1.64 parser 754: if(fail_on_read_problem)
1.126 paf 755: throw Exception("file.missing",
1.143.2.21.2. (paf 756:): &file_spec,
1.67 parser 757: "getting file size failed: %s (%d), real filename '%s'",
1.143.2.21.2. (paf 758:): strerror(errno), errno, fname);
1.64 parser 759: else
760: return false;
1.58 parser 761: rsize=finfo.st_size;
762: ratime=finfo.st_atime;
763: rmtime=finfo.st_mtime;
764: rctime=finfo.st_ctime;
1.64 parser 765: return true;
1.18 paf 766: }
767:
1.126 paf 768: char* getrow(char* *row_ref, char delim) {
769: char* result=*row_ref;
1.8 paf 770: if(result) {
1.126 paf 771: *row_ref=strchr(result, delim);
1.8 paf 772: if(*row_ref)
773: *((*row_ref)++)=0;
774: else if(!*result)
775: return 0;
776: }
777: return result;
778: }
779:
1.126 paf 780: char* lsplit(char* string, char delim) {
1.23 paf 781: if(string) {
1.126 paf 782: char* v=strchr(string, delim);
1.8 paf 783: if(v) {
784: *v=0;
785: return v+1;
786: }
787: }
788: return 0;
789: }
790:
1.126 paf 791: char* lsplit(char* *string_ref, char delim) {
792: char* result=*string_ref;
793: char* next=lsplit(*string_ref, delim);
1.8 paf 794: *string_ref=next;
795: return result;
1.9 paf 796: }
797:
1.126 paf 798: char* rsplit(char* string, char delim) {
1.18 paf 799: if(string) {
1.126 paf 800: char* v=strrchr(string, delim);
1.18 paf 801: if(v) {
1.9 paf 802: *v=0;
803: return v+1;
804: }
805: }
806: return NULL;
1.10 paf 807: }
808:
1.37 paf 809: /// @todo less stupid type detection
1.143.2.21.2. (paf 810:): const char* format(double value, char* fmt) {
1.126 paf 811: char local_buf[MAX_NUMBER];
1.108 paf 812: size_t size;
813:
1.10 paf 814: if(fmt)
815: if(strpbrk(fmt, "diouxX"))
816: if(strpbrk(fmt, "ouxX"))
1.126 paf 817: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 818: else
1.126 paf 819: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 820: else
1.126 paf 821: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 822: else
1.126 paf 823: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 824:
1.143.2.21.2. (paf 825:): return pa_strdup(local_buf, size);
1.12 paf 826: }
827:
1.36 paf 828: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 829: #ifdef WIN32
830: do{
1.143.2.21.2. (paf 831:): int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 832: if(chunk_written<=0)
833: break;
834: size-=chunk_written;
1.36 paf 835: buf=((const char*)buf)+chunk_written;
1.126 paf 836: } while(size>0);
1.12 paf 837:
838: return size;
839: #else
1.126 paf 840: return fwrite(buf, 1, size, stdout);
1.12 paf 841: #endif
1.2 paf 842: }
1.14 paf 843:
1.143.2.21.2. (paf 844:): char* unescape_chars(const char* cp, int len) {
845:): char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 846: enum EscapeState {
1.33 paf 847: EscapeRest,
848: EscapeFirst,
1.14 paf 849: EscapeSecond
850: } escapeState=EscapeRest;
851: int escapedValue=0;
852: int srcPos=0;
853: int dstPos=0;
854: while(srcPos < len) {
1.126 paf 855: int ch=cp[srcPos];
1.14 paf 856: switch(escapeState) {
857: case EscapeRest:
858: if(ch=='%') {
859: escapeState=EscapeFirst;
860: } else if(ch=='+') {
1.126 paf 861: s[dstPos++]=' ';
1.14 paf 862: } else {
863: s[dstPos++]=ch;
864: }
865: break;
866: case EscapeFirst:
867: escapedValue=hex_value[ch] << 4;
868: escapeState=EscapeSecond;
869: break;
870: case EscapeSecond:
1.126 paf 871: escapedValue +=hex_value[ch];
1.14 paf 872: s[dstPos++]=escapedValue;
873: escapeState=EscapeRest;
874: break;
875: }
1.126 paf 876: srcPos++;
1.14 paf 877: }
878: s[dstPos]=0;
879: return s;
1.24 paf 880: }
881:
882: #ifdef WIN32
1.126 paf 883: void back_slashes_to_slashes(char* s) {
1.24 paf 884: if(s)
885: for(; *s; s++)
886: if(*s=='\\')
1.126 paf 887: *s='/';
1.24 paf 888: }
1.42 paf 889: /*
1.126 paf 890: void slashes_to_back_slashes(char* s) {
1.42 paf 891: if(s)
892: for(; *s; s++)
893: if(*s=='/')
1.126 paf 894: *s='\\';
1.42 paf 895: }
896: */
1.24 paf 897: #endif
1.41 paf 898:
1.126 paf 899: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 900: while(true) {
901: if(!(*s1)) {
902: if(!(*s2))
903: return true;
904: else
905: return !strict;
906: } else if(!(*s2))
907: return !strict;
908: if(isalpha(*s1)) {
909: if(tolower(*s1) !=tolower(*s2))
910: return false;
911: } else if((*s1) !=(*s2))
912: return false;
1.126 paf 913: s1++;
914: s2++;
1.41 paf 915: }
1.57 parser 916: }
917:
1.84 paf 918: static bool isLeap(int year) {
1.57 parser 919: return !(
920: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 921: );
1.57 parser 922: }
923:
924: int getMonthDays(int year, int month) {
925: int monthDays[]={
1.126 paf 926: 31,
927: isLeap(year) ? 29 : 28,
928: 31,
929: 30,
930: 31,
931: 30,
932: 31,
933: 31,
934: 30,
935: 31,
936: 30,
1.57 parser 937: 31
1.126 paf 938: };
939: return monthDays[month];
1.41 paf 940: }
1.69 parser 941:
1.126 paf 942: void remove_crlf(char* start, char* end) {
943: for(char* p=start; p<end; p++)
1.69 parser 944: switch(*p) {
1.126 paf 945: case '\n': *p='|'; break;
946: case '\r': *p=' '; break;
1.69 parser 947: }
1.91 paf 948: }
949:
950:
951: /// must be last in this file
952: #undef vsnprintf
1.126 paf 953: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 954: if(!s)
955: return 0;
956:
957: int r;
958: // note: on win32& maybe somewhere else
959: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
960: --s;
961: #if _MSC_VER
962: /*
963: win32:
964: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
965:
1.143.2.12 paf 966: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 967: */
1.126 paf 968: r=_vsnprintf(b, s, f, l);
1.91 paf 969: if(r<0)
970: r=s;
971: #else
1.126 paf 972: r=vsnprintf(b, s, f, l);
1.91 paf 973: /*
974: solaris:
975: man vsnprintf
976:
977: The snprintf() function returns the number of characters
978: formatted, that is, the number of characters that would have
979: been written to the buffer if it were large enough. If the
980: value of n is 0 on a call to snprintf(), an unspecified
981: value less than 1 is returned.
982: */
983:
984: if(r<0)
985: r=0;
986: else if(r>s)
987: r=s;
988: #endif
989: b[r]=0;
990: return r;
991: }
992:
1.126 paf 993: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 994: va_list l;
1.126 paf 995: va_start(l, f);
996: int r=__vsnprintf(b, s, f, l);
997: va_end(l);
1.91 paf 998: return r;
1.98 paf 999: }
1000:
1001: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 1002: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 1003:
1004: #ifdef WIN32
1.126 paf 1005: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 1006: return 0;
1007: #else
1008: struct timeval t;
1009: t.tv_sec = secs;
1010: t.tv_usec = usecs;
1.126 paf 1011: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 1012: #endif
1.135 paf 1013: }
1014:
1015:
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