Annotation of libwww/Robot/src/HTRobot.c, revision 1.30
1.1 frystyk 1: /* HTRobot.c
2: ** W3C MINI ROBOT
3: **
4: ** (c) COPRIGHT MIT 1995.
5: ** Please first read the full copyright statement in the file COPYRIGH.
6: **
7: ** This program illustrates how to travers links using the Anchor object
8: **
9: ** Authors:
10: ** HFN Henrik Frystyk Nielsen, (frystyk@w3.org)
11: **
12: ** History:
13: ** Dec 04 95 First version
14: */
15:
16: #include "WWWLib.h" /* Global Library Include file */
17: #include "WWWApp.h" /* Application stuff */
1.9 frystyk 18: #include "WWWRules.h"
19: #include "WWWApp.h"
1.17 frystyk 20: #include "WWWTrans.h"
1.10 frystyk 21: #include "WWWInit.h"
1.9 frystyk 22:
1.4 frystyk 23: #include "HText.h"
1.1 frystyk 24:
25: #include "HTRobot.h" /* Implemented here */
26:
1.14 frystyk 27: #ifndef W3C_VERSION
28: #define W3C_VERSION "unspecified"
1.1 frystyk 29: #endif
30:
31: #define APP_NAME "W3CRobot"
1.14 frystyk 32: #define APP_VERSION W3C_VERSION
1.1 frystyk 33:
34: #define DEFAULT_OUTPUT_FILE "robot.out"
35: #define DEFAULT_RULE_FILE "robot.conf"
36: #define DEFAULT_LOG_FILE "robot.log"
1.7 frystyk 37: #define DEFAULT_DEPTH 0
1.1 frystyk 38:
39: #define SHOW_MSG (WWWTRACE || HTAlert_interactive())
40:
1.7 frystyk 41: #define DEFAULT_TIMEOUT 10 /* timeout in seconds */
1.1 frystyk 42:
43: #if defined(__svr4__)
44: #define CATCH_SIG
45: #endif
46:
47: typedef enum _MRFlags {
1.2 frystyk 48: MR_IMG = 0x1,
49: MR_LINK = 0x2,
1.12 frystyk 50: MR_PREEMPTIVE= 0x4,
51: MR_TIME = 0x8
1.1 frystyk 52: } MRFlags;
53:
54: typedef struct _Robot {
55: HTRequest * request;
1.7 frystyk 56: HTRequest * timeout; /* Until we get a server eventloop */
1.1 frystyk 57: HTParentAnchor * anchor;
1.2 frystyk 58: int depth; /* How deep is our tree */
1.30 ! frystyk 59: int cnt; /* Count of requests */
1.2 frystyk 60: HTList * hyperdoc; /* List of our HyperDoc Objects */
1.4 frystyk 61: HTList * htext; /* List of our HText Objects */
1.1 frystyk 62: struct timeval * tv; /* Timeout on socket */
63: char * cwd; /* Current dir URL */
64: char * rules;
65: char * logfile;
66: char * outputfile;
67: FILE * output;
68: MRFlags flags;
69: } Robot;
70:
71: typedef enum _LoadState {
72: L_INVALID = -2,
73: L_LOADING = -1,
74: L_SUCCESS = 0,
75: L_ERROR
76: } LoadState;
77:
78: /*
79: ** The HyperDoc object is bound to the anchor and contains information about
80: ** where we are in the search for recursive searches
81: */
82: typedef struct _HyperDoc {
83: HTParentAnchor * anchor;
84: LoadState state;
85: int depth;
86: } HyperDoc;
87:
88: /*
89: ** This is the HText object that is created every time we start parsing a
90: ** HTML object
91: */
1.4 frystyk 92: struct _HText {
1.1 frystyk 93: HTRequest * request;
1.4 frystyk 94: };
1.1 frystyk 95:
96: PUBLIC HText * HTMainText = NULL;
97: PUBLIC HTParentAnchor * HTMainAnchor = NULL;
98: PUBLIC HTStyleSheet * styleSheet = NULL;
99:
100: /* ------------------------------------------------------------------------- */
101:
1.13 eric 102: /* Standard (non-error) Output
103: ** ---------------------------
104: */
105: PUBLIC int OutputData(const char * fmt, ...)
106: {
107: int ret;
108: va_list pArgs;
109: va_start(pArgs, fmt);
110: ret = vfprintf(stdout, fmt, pArgs);
111: va_end(pArgs);
112: return ret;
113: }
114:
115: /* ------------------------------------------------------------------------- */
116:
1.2 frystyk 117: /* Create a "HyperDoc" object
118: ** --------------------------
119: ** A HyperDoc object contains information about whether we have already
120: ** started checking the anchor and the depth in our search
121: */
122: PRIVATE HyperDoc * HyperDoc_new (Robot * mr,HTParentAnchor * anchor, int depth)
123: {
124: HyperDoc * hd;
1.14 frystyk 125: if ((hd = (HyperDoc *) HT_CALLOC(1, sizeof(HyperDoc))) == NULL)
126: HT_OUTOFMEM("HyperDoc_new");
1.2 frystyk 127: hd->state = L_INVALID;
128: hd->depth = depth;
129:
130: /* Bind the HyperDoc object together with the Anchor Object */
131: hd->anchor = anchor;
132: HTAnchor_setDocument(anchor, (void *) hd);
133:
134: /* Add this HyperDoc object to our list */
135: if (!mr->hyperdoc) mr->hyperdoc = HTList_new();
136: HTList_addObject(mr->hyperdoc, (void *) hd);
137: return hd;
138: }
139:
140: /* Delete a "HyperDoc" object
141: ** --------------------------
142: */
143: PRIVATE BOOL HyperDoc_delete (HyperDoc * hd)
144: {
145: if (hd) {
1.11 frystyk 146: HT_FREE (hd);
1.2 frystyk 147: return YES;
148: }
149: return NO;
150: }
151:
1.1 frystyk 152: /* Create a Command Line Object
153: ** ----------------------------
154: */
155: PRIVATE Robot * Robot_new (void)
156: {
157: Robot * me;
1.14 frystyk 158: if ((me = (Robot *) HT_CALLOC(1, sizeof(Robot))) == NULL ||
159: (me->tv = (struct timeval*) HT_CALLOC(1, sizeof(struct timeval))) == NULL)
160: HT_OUTOFMEM("Robot_new");
1.2 frystyk 161: me->hyperdoc = HTList_new();
1.4 frystyk 162: me->htext = HTList_new();
1.1 frystyk 163: me->tv->tv_sec = DEFAULT_TIMEOUT;
1.25 frystyk 164: me->cwd = HTGetCurrentDirectoryURL();
1.1 frystyk 165: me->output = OUTPUT;
1.30 ! frystyk 166: me->cnt = 1;
1.1 frystyk 167:
1.7 frystyk 168: /* We keep an extra timeout request object for the timeout_handler */
169: me->timeout = HTRequest_new();
170: HTRequest_setContext (me->timeout, me);
171:
1.1 frystyk 172: /* Bind the Robot object together with the Request Object */
173: me->request = HTRequest_new();
174: HTRequest_setContext (me->request, me);
175: return me;
176: }
177:
178: /* Delete a Command Line Object
179: ** ----------------------------
180: */
181: PRIVATE BOOL Robot_delete (Robot * me)
182: {
183: if (me) {
1.2 frystyk 184: if (me->hyperdoc) {
185: HTList * cur = me->hyperdoc;
186: HyperDoc * pres;
187: while ((pres = (HyperDoc *) HTList_nextObject(cur)))
188: HyperDoc_delete(pres);
189: HTList_delete(me->hyperdoc);
190: }
1.4 frystyk 191: if (me->htext) {
192: HTList * cur = me->htext;
193: HText * pres;
194: while ((pres = (HText *) HTList_nextObject(cur)))
195: HText_free(pres);
196: HTList_delete(me->htext);
197: }
1.1 frystyk 198: if (me->logfile) HTLog_close();
199: if (me->output && me->output != STDOUT) fclose(me->output);
1.12 frystyk 200: if (me->flags & MR_TIME) {
201: time_t local = time(NULL);
1.13 eric 202: HTTrace("Robot terminated %s\n",HTDateTimeStr(&local,YES));
1.12 frystyk 203: }
1.11 frystyk 204: HT_FREE(me->cwd);
205: HT_FREE(me->tv);
206: HT_FREE(me);
1.1 frystyk 207: return YES;
208: }
209: return NO;
210: }
211:
1.2 frystyk 212: /*
213: ** This function creates a new request object and initializes it
214: */
215: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
216: {
217: HTRequest * newreq = HTRequest_new();
218: HTRequest_setContext (newreq, mr);
1.7 frystyk 219: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 220: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 221: HTRequest_setMethod(newreq, method);
1.30 ! frystyk 222: mr->cnt++;
1.2 frystyk 223: return newreq;
224: }
225:
226: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
227: {
228: if (mr && request) {
229: HTRequest_delete(request);
1.30 ! frystyk 230: mr->cnt--;
1.2 frystyk 231: return YES;
232: }
233: return NO;
234: }
235:
236: /*
237: ** Cleanup and make sure we close all connections including the persistent
238: ** ones
239: */
1.1 frystyk 240: PRIVATE void Cleanup (Robot * me, int status)
241: {
242: Robot_delete(me);
1.29 eric 243: HTProfile_delete();
1.1 frystyk 244: #ifdef VMS
245: exit(status ? status : 1);
246: #else
247: exit(status ? status : 0);
248: #endif
249: }
250:
251: #ifdef CATCH_SIG
252: #include <signal.h>
253: /* SetSignal
254: ** This function sets up signal handlers. This might not be necessary to
255: ** call if the application has its own handlers (lossage on SVR4)
256: */
257: PRIVATE void SetSignal (void)
258: {
259: /* On some systems (SYSV) it is necessary to catch the SIGPIPE signal
260: ** when attemting to connect to a remote host where you normally should
261: ** get `connection refused' back
262: */
263: if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) {
1.13 eric 264: if (PROT_TRACE) HTTrace("HTSignal.... Can't catch SIGPIPE\n");
1.1 frystyk 265: } else {
1.13 eric 266: if (PROT_TRACE) HTTrace("HTSignal.... Ignoring SIGPIPE\n");
1.1 frystyk 267: }
268: }
269: #endif /* CATCH_SIG */
270:
271: PRIVATE void VersionInfo (void)
272: {
1.13 eric 273: OutputData("\n\nW3C Reference Software\n\n");
274: OutputData("\tW3C Mini Robot (%s) version %s.\n",
1.1 frystyk 275: APP_NAME, APP_VERSION);
1.13 eric 276: OutputData("\tW3C Reference Library version %s.\n\n",HTLib_version());
277: OutputData("Please send feedback to <libwww@w3.org>\n");
1.1 frystyk 278: }
279:
280: /* terminate_handler
281: ** -----------------
1.2 frystyk 282: ** This function is registered to handle the result of the request.
283: ** If no more requests are pending then terminate program
1.1 frystyk 284: */
1.15 frystyk 285: PRIVATE int terminate_handler (HTRequest * request, void * param, int status)
1.1 frystyk 286: {
287: Robot * mr = (Robot *) HTRequest_context(request);
1.2 frystyk 288: Thread_delete(mr, request);
1.30 ! frystyk 289: if (HTNet_isEmpty()) {
! 290: if (SHOW_MSG) HTTrace("Robot....... Everything is finished...\n");
! 291: Cleanup(mr, 0);
! 292: }
! 293: if (SHOW_MSG) HTTrace("Robot....... %d outstanding requests\n", mr->cnt);
1.1 frystyk 294: return HT_OK;
295: }
296:
297: /* timeout_handler
298: ** ---------------
299: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 300: **
301: ** BUG: This doesn't work as we don't get the right request object
302: ** back from the event loop
1.1 frystyk 303: */
304: PRIVATE int timeout_handler (HTRequest * request)
305: {
1.27 frystyk 306: #if 0
1.2 frystyk 307: Robot * mr = (Robot *) HTRequest_context(request);
1.27 frystyk 308: #endif
1.25 frystyk 309: if (SHOW_MSG) HTTrace("Robot....... We don't know how to handle timeout...\n");
1.7 frystyk 310: #if 0
1.1 frystyk 311: HTRequest_kill(request);
1.2 frystyk 312: Thread_delete(mr, request);
1.7 frystyk 313: #endif
1.4 frystyk 314: return HT_OK;
1.1 frystyk 315: }
316:
317: /* ------------------------------------------------------------------------- */
318: /* HTEXT INTERFACE */
319: /* ------------------------------------------------------------------------- */
320:
321: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
322: HTStream * stream)
323: {
324: HText * me;
1.4 frystyk 325: Robot * mr = (Robot *) HTRequest_context(request);
1.14 frystyk 326: if ((me = (HText *) HT_CALLOC(1, sizeof(HText))) == NULL)
327: HT_OUTOFMEM("HText_new2");
1.4 frystyk 328:
329: /* Bind the HText object together with the Request Object */
1.1 frystyk 330: me->request = request;
1.4 frystyk 331:
332: /* Add this HyperDoc object to our list */
333: if (!mr->htext) mr->htext = HTList_new();
334: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 335: return me;
336: }
337:
1.4 frystyk 338: PUBLIC void HText_free (HText * me) {
1.11 frystyk 339: if (me) HT_FREE (me);
1.4 frystyk 340: }
341:
1.1 frystyk 342: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
343: {
344: if (text && anchor) {
1.2 frystyk 345: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 346: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
347: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 348: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 349: HyperDoc * hd = HTAnchor_document(dest_parent);
350:
1.13 eric 351: if (SHOW_MSG) HTTrace("Robot....... Found `%s\' - ", uri ? uri : "NULL");
1.7 frystyk 352:
1.2 frystyk 353: /* Test whether we already have a hyperdoc for this document */
354: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 355: HTParentAnchor * parent = HTRequest_parent(text->request);
356: HyperDoc * last = HTAnchor_document(parent);
357: int depth = last ? last->depth+1 : 0;
1.2 frystyk 358: HTRequest * newreq = Thread_new(mr, METHOD_GET);
359: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 360: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
361: if (depth >= mr->depth) {
362: if (SHOW_MSG)
1.13 eric 363: HTTrace("loading at depth %d using HEAD\n", depth);
1.7 frystyk 364: HTRequest_setMethod(newreq, METHOD_HEAD);
1.30 ! frystyk 365: HTRequest_setOutputFormat(newreq, WWW_DEBUG);
1.7 frystyk 366: } else {
1.13 eric 367: if (SHOW_MSG) HTTrace("loading at depth %d\n", depth);
1.2 frystyk 368: }
369: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.13 eric 370: if (SHOW_MSG) HTTrace("not tested!\n");
1.2 frystyk 371: Thread_delete(mr, newreq);
372: }
1.7 frystyk 373: } else {
1.18 frystyk 374: if (SHOW_MSG) HTTrace("duplicate or max depth reached\n");
1.2 frystyk 375: }
1.11 frystyk 376: HT_FREE(uri);
1.2 frystyk 377: }
378: }
379:
380: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
1.14 frystyk 381: const char *alt, const char * align, BOOL isMap)
1.2 frystyk 382: {
383: if (text && anchor) {
384: Robot * mr = (Robot *) HTRequest_context(text->request);
385: HTParentAnchor * dest = (HTParentAnchor *)
386: HTAnchor_followMainLink((HTAnchor *) anchor);
387: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 388:
1.2 frystyk 389: /* Test whether we already have a hyperdoc for this document */
390: if (mr->flags & MR_IMG && dest && !hd) {
391: HTParentAnchor * parent = HTRequest_parent(text->request);
392: HyperDoc * last = HTAnchor_document(parent);
393: int depth = last ? last->depth+1 : 0;
394: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
395: HyperDoc_new(mr, dest, depth);
396: if (SHOW_MSG) {
397: char * uri = HTAnchor_address((HTAnchor *) dest);
1.13 eric 398: HTTrace("Robot....... Checking Image `%s\'\n", uri);
1.11 frystyk 399: HT_FREE(uri);
1.2 frystyk 400: }
401: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
402: if (SHOW_MSG)
1.13 eric 403: HTTrace("Robot....... Image not tested!\n");
1.2 frystyk 404: Thread_delete(mr, newreq);
1.1 frystyk 405: }
406: }
407: }
408: }
409:
410: PUBLIC void HText_endAnchor (HText * text) {}
1.14 frystyk 411: PUBLIC void HText_appendText (HText * text, const char * str) {}
1.1 frystyk 412: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
413: PUBLIC void HText_endAppend (HText * text) {}
414: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
415: PUBLIC void HText_beginAppend (HText * text) {}
416: PUBLIC void HText_appendParagraph (HText * text) {}
417:
418: /* ------------------------------------------------------------------------- */
419: /* MAIN PROGRAM */
420: /* ------------------------------------------------------------------------- */
421:
422: int main (int argc, char ** argv)
423: {
424: int status = 0;
425: int arg;
426: HTChunk * keywords = NULL; /* From command line */
427: int keycnt = 0;
1.12 frystyk 428: Robot * mr = NULL;
1.1 frystyk 429:
430: /* Starts Mac GUSI socket library */
431: #ifdef GUSI
432: GUSISetup(GUSIwithSIOUXSockets);
433: GUSISetup(GUSIwithInternetSockets);
434: #endif
435:
436: #ifdef __MWERKS__ /* STR */
437: InitGraf((Ptr) &qd.thePort);
438: InitFonts();
439: InitWindows();
440: InitMenus(); TEInit();
441: InitDialogs(nil);
442: InitCursor();
443: SIOUXSettings.asktosaveonclose = false;
444: argc=ccommand(&argv);
445: #endif
446:
1.27 frystyk 447: /* Initiate W3C Reference Library with a robot profile */
448: HTProfile_newRobot(APP_NAME, APP_VERSION);
449:
450: /* Add the default HTML parser to the set of converters */
451: {
452: HTList * converters = HTFormat_conversion();
453: HTMLInit(converters);
454: }
1.1 frystyk 455:
1.12 frystyk 456: /* Build a new robot object */
457: mr = Robot_new();
458:
1.1 frystyk 459: /* Scan command Line for parameters */
460: for (arg=1; arg<argc; arg++) {
461: if (*argv[arg] == '-') {
462:
463: /* non-interactive */
1.17 frystyk 464: if (!strcmp(argv[arg], "-n")) {
1.1 frystyk 465: HTAlert_setInteractive(NO);
466:
467: /* log file */
468: } else if (!strcmp(argv[arg], "-l")) {
469: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
470: argv[++arg] : DEFAULT_LOG_FILE;
471:
472: /* rule file */
473: } else if (!strcmp(argv[arg], "-r")) {
474: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
475: argv[++arg] : DEFAULT_RULE_FILE;
476:
477: /* output filename */
478: } else if (!strcmp(argv[arg], "-o")) {
479: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
480: argv[++arg] : DEFAULT_OUTPUT_FILE;
481:
482: /* timeout -- Change the default request timeout */
483: } else if (!strcmp(argv[arg], "-timeout")) {
484: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
485: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
486: if (timeout > 0) mr->tv->tv_sec = timeout;
487:
1.7 frystyk 488: /* preemptive or non-preemptive access */
1.1 frystyk 489: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 490: HTRequest_setPreemptive(mr->request, YES);
491: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 492:
493: /* test inlined images */
494: } else if (!strcmp(argv[arg], "-img")) {
495: mr->flags |= MR_IMG;
496:
497: /* load anchors */
498: } else if (!strcmp(argv[arg], "-link")) {
499: mr->flags |= MR_LINK;
1.7 frystyk 500: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
501: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 502:
1.7 frystyk 503: /* preemptive or non-preemptive access */
1.2 frystyk 504: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 505: HTRequest_setPreemptive(mr->request, YES);
506: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 507:
1.12 frystyk 508: /* Output start and end time */
509: } else if (!strcmp(argv[arg], "-ss")) {
510: time_t local = time(NULL);
1.13 eric 511: HTTrace("Robot started on %s\n",
1.12 frystyk 512: HTDateTimeStr(&local, YES));
513: mr->flags |= MR_TIME;
514:
1.1 frystyk 515: /* print version and exit */
516: } else if (!strcmp(argv[arg], "-version")) {
517: VersionInfo();
518: Cleanup(mr, 0);
519:
520: #ifdef WWWTRACE
521: /* trace flags */
522: } else if (!strncmp(argv[arg], "-v", 2)) {
1.24 frystyk 523: HTSetTraceMessageMask(argv[arg]+2);
1.1 frystyk 524: #endif
525:
526: } else {
1.13 eric 527: if (SHOW_MSG) HTTrace("Bad Argument (%s)\n", argv[arg]);
1.1 frystyk 528: }
1.17 frystyk 529: } else { /* If no leading `-' then check for URL or keywords */
1.1 frystyk 530: if (!keycnt) {
531: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
532: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 533: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 534: keycnt = 1;
1.11 frystyk 535: HT_FREE(ref);
1.1 frystyk 536: } else { /* Check for successive keyword arguments */
537: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
538: if (keycnt++ <= 1)
1.5 frystyk 539: keywords = HTChunk_new(128);
1.1 frystyk 540: else
1.5 frystyk 541: HTChunk_putc(keywords, ' ');
542: HTChunk_puts(keywords, HTStrip(escaped));
1.11 frystyk 543: HT_FREE(escaped);
1.1 frystyk 544: }
545: }
546: }
547:
548: #ifdef CATCH_SIG
549: SetSignal();
550: #endif
551:
552: if (!keycnt) {
1.13 eric 553: if (SHOW_MSG) HTTrace("Please specify URL to check.\n");
1.1 frystyk 554: Cleanup(mr, -1);
555: }
556:
1.23 manoli 557: /* Testing that HTTrace is working */
558: HTTrace ("Welcome to the W3C mini Robot\n");
559:
1.1 frystyk 560: /* Rule file specified? */
561: if (mr->rules) {
562: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
1.27 frystyk 563: if (!HTLoadRules(rules))
1.13 eric 564: if (SHOW_MSG) HTTrace("Can't access rules\n");
1.11 frystyk 565: HT_FREE(rules);
1.1 frystyk 566: }
567:
568: /* Output file specified? */
569: if (mr->outputfile) {
570: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
1.13 eric 571: if (SHOW_MSG) HTTrace("Can't open `%s'\n", mr->outputfile);
1.1 frystyk 572: mr->output = OUTPUT;
573: }
574: }
575:
576: /* Log file specifed? */
577: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
578:
1.27 frystyk 579: /* Register our own someterminater filter */
1.15 frystyk 580: HTNetCall_addAfter(terminate_handler, NULL, HT_ALL);
1.1 frystyk 581:
582: /* Set timeout on sockets */
1.16 frystyk 583: HTEventrg_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 584:
585: /* Start the request */
586: if (keywords) /* Search */
1.28 frystyk 587: status = HTSearchAnchor(keywords, (HTAnchor *)mr->anchor, mr->request);
1.1 frystyk 588: else
589: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
590:
1.5 frystyk 591: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 592: if (status != YES) {
1.13 eric 593: if (SHOW_MSG) HTTrace("Can't access resource\n");
1.1 frystyk 594: Cleanup(mr, -1);
595: }
596:
597: /* Go into the event loop... */
1.16 frystyk 598: HTEventrg_loop(mr->request);
1.1 frystyk 599:
600: /* Only gets here if event loop fails */
601: Cleanup(mr, 0);
602: return 0;
603: }
Webmaster
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to HTRobot.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Public] / libwww / Robot / src