Annotation of libwww/Robot/src/HTRobot.c, revision 1.8
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 */
1.4 frystyk 18: #include "HText.h"
1.1 frystyk 19:
20: #include "HTRobot.h" /* Implemented here */
21:
22: #ifndef VR
23: #define VR "unspecified"
24: #endif
25:
26: #define APP_NAME "W3CRobot"
27: #define APP_VERSION VR
28:
29: /* Default page for "-help" command line option */
30: #define HELP "http://www.w3.org/pub/WWW/Robot/User/CommandLine.html"
31:
32: #define DEFAULT_OUTPUT_FILE "robot.out"
33: #define DEFAULT_RULE_FILE "robot.conf"
34: #define DEFAULT_LOG_FILE "robot.log"
1.7 frystyk 35: #define DEFAULT_DEPTH 0
1.1 frystyk 36:
37: #define SHOW_MSG (WWWTRACE || HTAlert_interactive())
38:
1.7 frystyk 39: #define DEFAULT_TIMEOUT 10 /* timeout in seconds */
1.1 frystyk 40:
41: #if defined(__svr4__)
42: #define CATCH_SIG
43: #endif
44:
45: typedef enum _MRFlags {
1.2 frystyk 46: MR_IMG = 0x1,
47: MR_LINK = 0x2,
1.7 frystyk 48: MR_PREEMPTIVE= 0x4
1.1 frystyk 49: } MRFlags;
50:
51: typedef struct _Robot {
52: HTRequest * request;
1.7 frystyk 53: HTRequest * timeout; /* Until we get a server eventloop */
1.1 frystyk 54: HTParentAnchor * anchor;
1.2 frystyk 55: int depth; /* How deep is our tree */
56: HTList * hyperdoc; /* List of our HyperDoc Objects */
1.4 frystyk 57: HTList * htext; /* List of our HText Objects */
1.1 frystyk 58: struct timeval * tv; /* Timeout on socket */
59: char * cwd; /* Current dir URL */
60: HTList * converters;
61: char * rules;
62: char * logfile;
63: char * outputfile;
64: FILE * output;
65: MRFlags flags;
66: } Robot;
67:
68: typedef enum _LoadState {
69: L_INVALID = -2,
70: L_LOADING = -1,
71: L_SUCCESS = 0,
72: L_ERROR
73: } LoadState;
74:
75: /*
76: ** The HyperDoc object is bound to the anchor and contains information about
77: ** where we are in the search for recursive searches
78: */
79: typedef struct _HyperDoc {
80: HTParentAnchor * anchor;
81: LoadState state;
82: int depth;
83: } HyperDoc;
84:
85: /*
86: ** This is the HText object that is created every time we start parsing a
87: ** HTML object
88: */
1.4 frystyk 89: struct _HText {
1.1 frystyk 90: HTRequest * request;
1.4 frystyk 91: };
1.1 frystyk 92:
93: PUBLIC HText * HTMainText = NULL;
94: PUBLIC HTParentAnchor * HTMainAnchor = NULL;
95: PUBLIC HTStyleSheet * styleSheet = NULL;
96:
97: /* ------------------------------------------------------------------------- */
98:
1.2 frystyk 99: /* Create a "HyperDoc" object
100: ** --------------------------
101: ** A HyperDoc object contains information about whether we have already
102: ** started checking the anchor and the depth in our search
103: */
104: PRIVATE HyperDoc * HyperDoc_new (Robot * mr,HTParentAnchor * anchor, int depth)
105: {
106: HyperDoc * hd;
107: if ((hd = (HyperDoc *) calloc(1, sizeof(HyperDoc))) == NULL)
108: outofmem(__FILE__, "HyperDoc_new");
109: hd->state = L_INVALID;
110: hd->depth = depth;
111:
112: /* Bind the HyperDoc object together with the Anchor Object */
113: hd->anchor = anchor;
114: HTAnchor_setDocument(anchor, (void *) hd);
115:
116: /* Add this HyperDoc object to our list */
117: if (!mr->hyperdoc) mr->hyperdoc = HTList_new();
118: HTList_addObject(mr->hyperdoc, (void *) hd);
119: return hd;
120: }
121:
122: /* Delete a "HyperDoc" object
123: ** --------------------------
124: */
125: PRIVATE BOOL HyperDoc_delete (HyperDoc * hd)
126: {
127: if (hd) {
128: free (hd);
129: return YES;
130: }
131: return NO;
132: }
133:
1.1 frystyk 134: /* Create a Command Line Object
135: ** ----------------------------
136: */
137: PRIVATE Robot * Robot_new (void)
138: {
139: Robot * me;
140: if ((me = (Robot *) calloc(1, sizeof(Robot))) == NULL ||
141: (me->tv = (struct timeval*) calloc(1, sizeof(struct timeval))) == NULL)
142: outofmem(__FILE__, "Robot_new");
1.2 frystyk 143: me->hyperdoc = HTList_new();
1.4 frystyk 144: me->htext = HTList_new();
1.1 frystyk 145: me->tv->tv_sec = DEFAULT_TIMEOUT;
146: me->cwd = HTFindRelatedName();
147: me->output = OUTPUT;
148:
1.7 frystyk 149: /* We keep an extra timeout request object for the timeout_handler */
150: me->timeout = HTRequest_new();
151: HTRequest_setContext (me->timeout, me);
152:
1.1 frystyk 153: /* Bind the Robot object together with the Request Object */
154: me->request = HTRequest_new();
155: HTRequest_setContext (me->request, me);
156: return me;
157: }
158:
159: /* Delete a Command Line Object
160: ** ----------------------------
161: */
162: PRIVATE BOOL Robot_delete (Robot * me)
163: {
164: if (me) {
1.2 frystyk 165: if (me->hyperdoc) {
166: HTList * cur = me->hyperdoc;
167: HyperDoc * pres;
168: while ((pres = (HyperDoc *) HTList_nextObject(cur)))
169: HyperDoc_delete(pres);
170: HTList_delete(me->hyperdoc);
171: }
1.4 frystyk 172: if (me->htext) {
173: HTList * cur = me->htext;
174: HText * pres;
175: while ((pres = (HText *) HTList_nextObject(cur)))
176: HText_free(pres);
177: HTList_delete(me->htext);
178: }
1.1 frystyk 179: if (me->logfile) HTLog_close();
180: if (me->output && me->output != STDOUT) fclose(me->output);
181: FREE(me->cwd);
182: free(me->tv);
183: free(me);
184: return YES;
185: }
186: return NO;
187: }
188:
1.2 frystyk 189: /*
190: ** This function creates a new request object and initializes it
191: */
192: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
193: {
194: HTRequest * newreq = HTRequest_new();
195: HTRequest_setContext (newreq, mr);
1.7 frystyk 196: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 197: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 198: HTRequest_setMethod(newreq, method);
199: return newreq;
200: }
201:
202: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
203: {
204: if (mr && request) {
205: HTRequest_delete(request);
206: return YES;
207: }
208: return NO;
209: }
210:
211: /*
212: ** Cleanup and make sure we close all connections including the persistent
213: ** ones
214: */
1.1 frystyk 215: PRIVATE void Cleanup (Robot * me, int status)
216: {
1.2 frystyk 217: HTNet_killAll();
1.1 frystyk 218: Robot_delete(me);
219: HTLibTerminate();
220: #ifdef VMS
221: exit(status ? status : 1);
222: #else
223: exit(status ? status : 0);
224: #endif
225: }
226:
227: #ifdef CATCH_SIG
228: #include <signal.h>
229: /* SetSignal
230: ** This function sets up signal handlers. This might not be necessary to
231: ** call if the application has its own handlers (lossage on SVR4)
232: */
233: PRIVATE void SetSignal (void)
234: {
235: /* On some systems (SYSV) it is necessary to catch the SIGPIPE signal
236: ** when attemting to connect to a remote host where you normally should
237: ** get `connection refused' back
238: */
239: if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) {
240: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Can't catch SIGPIPE\n");
241: } else {
242: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Ignoring SIGPIPE\n");
243: }
244: }
245: #endif /* CATCH_SIG */
246:
247: PRIVATE void VersionInfo (void)
248: {
249: TTYPrint(OUTPUT,"\n\nW3C Reference Software\n\n");
250: TTYPrint(OUTPUT,"\tW3C Mini Robot (%s) version %s.\n",
251: APP_NAME, APP_VERSION);
252: TTYPrint(OUTPUT,"\tW3C Reference Library version %s.\n\n",HTLib_version());
253: TTYPrint(OUTPUT,"Please send feedback to <libwww@w3.org>\n");
254: }
255:
256: /* terminate_handler
257: ** -----------------
1.2 frystyk 258: ** This function is registered to handle the result of the request.
259: ** If no more requests are pending then terminate program
1.1 frystyk 260: */
261: PRIVATE int terminate_handler (HTRequest * request, int status)
262: {
263: Robot * mr = (Robot *) HTRequest_context(request);
264: if (mr->logfile) HTLog_add(request, status);
1.2 frystyk 265: Thread_delete(mr, request);
1.3 frystyk 266: if (HTNet_isEmpty()) Cleanup(mr, 0);
1.1 frystyk 267: return HT_OK;
268: }
269:
270: /* timeout_handler
271: ** ---------------
272: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 273: **
274: ** BUG: This doesn't work as we don't get the right request object
275: ** back from the event loop
1.1 frystyk 276: */
277: PRIVATE int timeout_handler (HTRequest * request)
278: {
1.2 frystyk 279: Robot * mr = (Robot *) HTRequest_context(request);
280: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Request timeout...\n");
1.7 frystyk 281: #if 0
1.1 frystyk 282: HTRequest_kill(request);
1.2 frystyk 283: Thread_delete(mr, request);
1.7 frystyk 284: #endif
285: Cleanup(mr, -1);
1.4 frystyk 286: return HT_OK;
1.1 frystyk 287: }
288:
1.8 ! frystyk 289: /* proxy_handler
! 290: ** ---------------
! 291: ** This function is registered to be called before a request is issued
! 292: ** We look for redirection for proxies and gateways
! 293: ** returns HT_LOADED We already have this
! 294: ** HT_ERROR We can't load this
! 295: ** HT_OK Success
! 296: */
! 297: PRIVATE int proxy_handler (HTRequest * request, int status)
! 298: {
! 299: HTParentAnchor *anchor = HTRequest_anchor(request);
! 300: char * addr = HTAnchor_address((HTAnchor *) anchor);
! 301: char * newaddr = NULL;
! 302: if ((newaddr = HTProxy_find(addr))) {
! 303: StrAllocCat(newaddr, addr);
! 304: HTRequest_setProxying(request, YES);
! 305: HTAnchor_setPhysical(anchor, newaddr);
! 306: } else if ((newaddr = HTGateway_find(addr))) {
! 307: char * path = HTParse(addr,"",PARSE_HOST+PARSE_PATH+PARSE_PUNCTUATION);
! 308: /* Chop leading / off to make host into part of path */
! 309: char * gatewayed = HTParse(path+1, newaddr, PARSE_ALL);
! 310: HTRequest_setProxying(request, NO);
! 311: HTAnchor_setPhysical(anchor, gatewayed);
! 312: free(path);
! 313: free(gatewayed);
! 314: } else
! 315: HTRequest_setProxying(request, NO);
! 316: FREE(newaddr);
! 317: FREE(addr);
! 318: return HT_OK;
! 319: }
! 320:
1.1 frystyk 321: /* ------------------------------------------------------------------------- */
322: /* HTEXT INTERFACE */
323: /* ------------------------------------------------------------------------- */
324:
325: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
326: HTStream * stream)
327: {
328: HText * me;
1.4 frystyk 329: Robot * mr = (Robot *) HTRequest_context(request);
1.1 frystyk 330: if ((me = (HText *) calloc(1, sizeof(HText))) == NULL)
331: outofmem(__FILE__, "HText_new2");
1.4 frystyk 332:
333: /* Bind the HText object together with the Request Object */
1.1 frystyk 334: me->request = request;
1.4 frystyk 335:
336: /* Add this HyperDoc object to our list */
337: if (!mr->htext) mr->htext = HTList_new();
338: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 339: return me;
340: }
341:
1.4 frystyk 342: PUBLIC void HText_free (HText * me) {
343: if (me) free (me);
344: }
345:
1.1 frystyk 346: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
347: {
348: if (text && anchor) {
1.2 frystyk 349: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 350: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
351: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 352: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 353: HyperDoc * hd = HTAnchor_document(dest_parent);
354:
1.7 frystyk 355: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Found `%s\' - ", uri ? uri : "NULL");
356:
1.2 frystyk 357: /* Test whether we already have a hyperdoc for this document */
358: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 359: HTParentAnchor * parent = HTRequest_parent(text->request);
360: HyperDoc * last = HTAnchor_document(parent);
361: int depth = last ? last->depth+1 : 0;
1.2 frystyk 362: HTRequest * newreq = Thread_new(mr, METHOD_GET);
363: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 364: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
365: if (depth >= mr->depth) {
366: if (SHOW_MSG)
367: TTYPrint(TDEST, "loading at depth %d using HEAD\n", depth);
368: HTRequest_setMethod(newreq, METHOD_HEAD);
369: HTRequest_setOutputFormat(newreq, WWW_MIME);
370: } else {
371: if (SHOW_MSG) TTYPrint(TDEST, "loading at depth %d\n", depth);
1.2 frystyk 372: }
373: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.7 frystyk 374: if (SHOW_MSG) TTYPrint(TDEST, "not tested!\n");
1.2 frystyk 375: Thread_delete(mr, newreq);
376: }
1.7 frystyk 377: } else {
378: if (SHOW_MSG) TTYPrint(TDEST, "duplicate\n");
1.2 frystyk 379: }
1.7 frystyk 380: FREE(uri);
1.2 frystyk 381: }
382: }
383:
384: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
385: CONST char *alt, CONST char * align, BOOL isMap)
386: {
387: if (text && anchor) {
388: Robot * mr = (Robot *) HTRequest_context(text->request);
389: HTParentAnchor * dest = (HTParentAnchor *)
390: HTAnchor_followMainLink((HTAnchor *) anchor);
391: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 392:
1.2 frystyk 393: /* Test whether we already have a hyperdoc for this document */
394: if (mr->flags & MR_IMG && dest && !hd) {
395: HTParentAnchor * parent = HTRequest_parent(text->request);
396: HyperDoc * last = HTAnchor_document(parent);
397: int depth = last ? last->depth+1 : 0;
398: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
399: HyperDoc_new(mr, dest, depth);
400: if (SHOW_MSG) {
401: char * uri = HTAnchor_address((HTAnchor *) dest);
402: TTYPrint(TDEST, "Robot....... Checking Image `%s\'\n", uri);
403: free(uri);
404: }
405: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
406: if (SHOW_MSG)
407: TTYPrint(TDEST, "Robot....... Image not tested!\n");
408: Thread_delete(mr, newreq);
1.1 frystyk 409: }
410: }
411: }
412: }
413:
414: PUBLIC void HText_endAnchor (HText * text) {}
415: PUBLIC void HText_appendText (HText * text, CONST char * str) {}
416: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
417: PUBLIC void HText_endAppend (HText * text) {}
418: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
419: PUBLIC void HText_beginAppend (HText * text) {}
420: PUBLIC void HText_appendParagraph (HText * text) {}
421:
422: /* ------------------------------------------------------------------------- */
423: /* MAIN PROGRAM */
424: /* ------------------------------------------------------------------------- */
425:
426: int main (int argc, char ** argv)
427: {
428: int status = 0;
429: int arg;
430: HTChunk * keywords = NULL; /* From command line */
431: int keycnt = 0;
432: Robot * mr = Robot_new(); /* Create new Robot instance */
433:
434: /* Starts Mac GUSI socket library */
435: #ifdef GUSI
436: GUSISetup(GUSIwithSIOUXSockets);
437: GUSISetup(GUSIwithInternetSockets);
438: #endif
439:
440: #ifdef __MWERKS__ /* STR */
441: InitGraf((Ptr) &qd.thePort);
442: InitFonts();
443: InitWindows();
444: InitMenus(); TEInit();
445: InitDialogs(nil);
446: InitCursor();
447: SIOUXSettings.asktosaveonclose = false;
448: argc=ccommand(&argv);
449: #endif
450:
451: /* Initiate W3C Reference Library */
452: HTLibInit(APP_NAME, APP_VERSION);
453:
454: /* Initialize the protocol modules */
455: HTAccessInit();
456:
457: /* Initialize set of converters */
458: mr->converters = HTList_new();
459: HTConverterInit(mr->converters);
460: HTFormat_setConversion(mr->converters);
461:
462: /* Initialize bindings between file suffixes and media types */
463: HTFileInit();
464:
465: /* Get any proxy or gateway environment variables */
466: HTProxy_getEnvVar();
467:
468: /* Scan command Line for parameters */
469: for (arg=1; arg<argc; arg++) {
470: if (*argv[arg] == '-') {
471:
472: /* -? or -help: show the command line help page */
473: if (!strcmp(argv[arg],"-?") || !strcmp(argv[arg],"-help")) {
474: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(HELP);
475: keycnt = 1;
476:
477: /* non-interactive */
478: } else if (!strcmp(argv[arg], "-n")) {
479: HTAlert_setInteractive(NO);
480:
481: /* log file */
482: } else if (!strcmp(argv[arg], "-l")) {
483: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
484: argv[++arg] : DEFAULT_LOG_FILE;
485:
486: /* rule file */
487: } else if (!strcmp(argv[arg], "-r")) {
488: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
489: argv[++arg] : DEFAULT_RULE_FILE;
490:
491: /* output filename */
492: } else if (!strcmp(argv[arg], "-o")) {
493: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
494: argv[++arg] : DEFAULT_OUTPUT_FILE;
495:
496: /* timeout -- Change the default request timeout */
497: } else if (!strcmp(argv[arg], "-timeout")) {
498: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
499: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
500: if (timeout > 0) mr->tv->tv_sec = timeout;
501:
1.7 frystyk 502: /* preemptive or non-preemptive access */
1.1 frystyk 503: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 504: HTRequest_setPreemptive(mr->request, YES);
505: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 506:
507: /* test inlined images */
508: } else if (!strcmp(argv[arg], "-img")) {
509: mr->flags |= MR_IMG;
510:
511: /* load anchors */
512: } else if (!strcmp(argv[arg], "-link")) {
513: mr->flags |= MR_LINK;
1.7 frystyk 514: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
515: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 516:
1.7 frystyk 517: /* preemptive or non-preemptive access */
1.2 frystyk 518: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 519: HTRequest_setPreemptive(mr->request, YES);
520: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 521:
522: /* print version and exit */
523: } else if (!strcmp(argv[arg], "-version")) {
524: VersionInfo();
525: Cleanup(mr, 0);
526:
527: #ifdef WWWTRACE
528: /* trace flags */
529: } else if (!strncmp(argv[arg], "-v", 2)) {
530: char *p = argv[arg]+2;
531: WWWTRACE = 0;
532: for(; *p; p++) {
533: switch (*p) {
534: case 'a': WWWTRACE |= SHOW_ANCHOR_TRACE; break;
535: case 'b': WWWTRACE |= SHOW_BIND_TRACE; break;
536: case 'c': WWWTRACE |= SHOW_CACHE_TRACE; break;
537: case 'g': WWWTRACE |= SHOW_SGML_TRACE; break;
538: case 'p': WWWTRACE |= SHOW_PROTOCOL_TRACE; break;
539: case 's': WWWTRACE |= SHOW_STREAM_TRACE; break;
540: case 't': WWWTRACE |= SHOW_THREAD_TRACE; break;
541: case 'u': WWWTRACE |= SHOW_URI_TRACE; break;
542: default:
543: if (SHOW_MSG)
544: TTYPrint(TDEST,"Bad parameter (%s) in -v option\n",
545: argv[arg]);
546: }
547: }
548: if (!WWWTRACE) WWWTRACE = SHOW_ALL_TRACE;
549: #endif
550:
551: } else {
552: if (SHOW_MSG) TTYPrint(TDEST,"Bad Argument (%s)\n", argv[arg]);
553: }
554: } else { /* If no leading `-' then check for URL or keywords */
555: if (!keycnt) {
556: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
557: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 558: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 559: keycnt = 1;
560: FREE(ref);
561: } else { /* Check for successive keyword arguments */
562: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
563: if (keycnt++ <= 1)
1.5 frystyk 564: keywords = HTChunk_new(128);
1.1 frystyk 565: else
1.5 frystyk 566: HTChunk_putc(keywords, ' ');
567: HTChunk_puts(keywords, HTStrip(escaped));
1.1 frystyk 568: free(escaped);
569: }
570: }
571: }
572:
573: #ifdef CATCH_SIG
574: SetSignal();
575: #endif
576:
577: if (!keycnt) {
1.2 frystyk 578: if (SHOW_MSG) TTYPrint(TDEST, "Please specify URL to check.\n");
1.1 frystyk 579: Cleanup(mr, -1);
580: }
581:
582: /* Rule file specified? */
583: if (mr->rules) {
584: HTList * list = HTList_new();
585: HTRequest * rr = HTRequest_new();
586: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
587: HTParentAnchor * ra = (HTParentAnchor *) HTAnchor_findAddress(rules);
1.7 frystyk 588: HTRequest_setPreemptive(rr, YES);
1.1 frystyk 589: HTConversion_add(list, "application/x-www-rules", "*/*", HTRules,
590: 1.0, 0.0, 0.0);
591: HTRequest_setConversion(rr, list, YES);
1.8 ! frystyk 592: HTAlert_add(HTConfirm, HT_A_CONFIRM);
1.1 frystyk 593: if (HTLoadAnchor((HTAnchor *) ra, rr) != YES)
594: if (SHOW_MSG) TTYPrint(TDEST, "Can't access rules\n");
595: HTConversion_deleteAll(list);
596: HTRequest_delete(rr);
1.8 ! frystyk 597: HTAlert_delete(HTConfirm);
1.1 frystyk 598: FREE(rules);
599: }
600:
601: /* Output file specified? */
602: if (mr->outputfile) {
603: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
604: if (SHOW_MSG) TTYPrint(TDEST, "Can't open `%s'\n", mr->outputfile);
605: mr->output = OUTPUT;
606: }
607: }
608:
609: /* Log file specifed? */
610: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
611:
612: /* Register our User Prompts etc in the Alert Manager */
613: if (HTAlert_interactive()) {
614: HTAlert_add(HTError_print, HT_A_MESSAGE);
615: HTAlert_add(HTConfirm, HT_A_CONFIRM);
616: HTAlert_add(HTPrompt, HT_A_PROMPT);
617: HTAlert_add(HTPromptPassword, HT_A_SECRET);
618: HTAlert_add(HTPromptUsernameAndPassword, HT_A_USER_PW);
619: }
620:
621: /* Register a call back function for the Net Manager */
1.8 ! frystyk 622: HTNetCall_addBefore(proxy_handler, 0);
1.1 frystyk 623: HTNetCall_addAfter(terminate_handler, HT_ALL);
624:
625: /* Set timeout on sockets */
1.7 frystyk 626: HTEvent_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 627:
628: /* Start the request */
629: if (keywords) /* Search */
1.5 frystyk 630: status = HTSearch(HTChunk_data(keywords), mr->anchor, mr->request);
1.1 frystyk 631: else
632: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
633:
1.5 frystyk 634: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 635: if (status != YES) {
636: if (SHOW_MSG) TTYPrint(TDEST, "Can't access resource\n");
637: Cleanup(mr, -1);
638: }
639:
640: /* Go into the event loop... */
641: HTEvent_Loop(mr->request);
642:
643: /* Only gets here if event loop fails */
644: Cleanup(mr, 0);
645: return 0;
646: }
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