Annotation of libwww/Robot/src/HTRobot.c, revision 1.9
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"
! 20:
! 21: #include "HTInit.c"
! 22: #include "HTBInit.c"
1.4 frystyk 23: #include "HText.h"
1.1 frystyk 24:
25: #include "HTRobot.h" /* Implemented here */
26:
27: #ifndef VR
28: #define VR "unspecified"
29: #endif
30:
31: #define APP_NAME "W3CRobot"
32: #define APP_VERSION VR
33:
34: /* Default page for "-help" command line option */
35: #define HELP "http://www.w3.org/pub/WWW/Robot/User/CommandLine.html"
36:
37: #define DEFAULT_OUTPUT_FILE "robot.out"
38: #define DEFAULT_RULE_FILE "robot.conf"
39: #define DEFAULT_LOG_FILE "robot.log"
1.7 frystyk 40: #define DEFAULT_DEPTH 0
1.1 frystyk 41:
42: #define SHOW_MSG (WWWTRACE || HTAlert_interactive())
43:
1.7 frystyk 44: #define DEFAULT_TIMEOUT 10 /* timeout in seconds */
1.1 frystyk 45:
46: #if defined(__svr4__)
47: #define CATCH_SIG
48: #endif
49:
50: typedef enum _MRFlags {
1.2 frystyk 51: MR_IMG = 0x1,
52: MR_LINK = 0x2,
1.7 frystyk 53: MR_PREEMPTIVE= 0x4
1.1 frystyk 54: } MRFlags;
55:
56: typedef struct _Robot {
57: HTRequest * request;
1.7 frystyk 58: HTRequest * timeout; /* Until we get a server eventloop */
1.1 frystyk 59: HTParentAnchor * anchor;
1.2 frystyk 60: int depth; /* How deep is our tree */
61: HTList * hyperdoc; /* List of our HyperDoc Objects */
1.4 frystyk 62: HTList * htext; /* List of our HText Objects */
1.1 frystyk 63: struct timeval * tv; /* Timeout on socket */
64: char * cwd; /* Current dir URL */
65: HTList * converters;
66: char * rules;
67: char * logfile;
68: char * outputfile;
69: FILE * output;
70: MRFlags flags;
71: } Robot;
72:
73: typedef enum _LoadState {
74: L_INVALID = -2,
75: L_LOADING = -1,
76: L_SUCCESS = 0,
77: L_ERROR
78: } LoadState;
79:
80: /*
81: ** The HyperDoc object is bound to the anchor and contains information about
82: ** where we are in the search for recursive searches
83: */
84: typedef struct _HyperDoc {
85: HTParentAnchor * anchor;
86: LoadState state;
87: int depth;
88: } HyperDoc;
89:
90: /*
91: ** This is the HText object that is created every time we start parsing a
92: ** HTML object
93: */
1.4 frystyk 94: struct _HText {
1.1 frystyk 95: HTRequest * request;
1.4 frystyk 96: };
1.1 frystyk 97:
98: PUBLIC HText * HTMainText = NULL;
99: PUBLIC HTParentAnchor * HTMainAnchor = NULL;
100: PUBLIC HTStyleSheet * styleSheet = NULL;
101:
102: /* ------------------------------------------------------------------------- */
103:
1.2 frystyk 104: /* Create a "HyperDoc" object
105: ** --------------------------
106: ** A HyperDoc object contains information about whether we have already
107: ** started checking the anchor and the depth in our search
108: */
109: PRIVATE HyperDoc * HyperDoc_new (Robot * mr,HTParentAnchor * anchor, int depth)
110: {
111: HyperDoc * hd;
112: if ((hd = (HyperDoc *) calloc(1, sizeof(HyperDoc))) == NULL)
113: outofmem(__FILE__, "HyperDoc_new");
114: hd->state = L_INVALID;
115: hd->depth = depth;
116:
117: /* Bind the HyperDoc object together with the Anchor Object */
118: hd->anchor = anchor;
119: HTAnchor_setDocument(anchor, (void *) hd);
120:
121: /* Add this HyperDoc object to our list */
122: if (!mr->hyperdoc) mr->hyperdoc = HTList_new();
123: HTList_addObject(mr->hyperdoc, (void *) hd);
124: return hd;
125: }
126:
127: /* Delete a "HyperDoc" object
128: ** --------------------------
129: */
130: PRIVATE BOOL HyperDoc_delete (HyperDoc * hd)
131: {
132: if (hd) {
133: free (hd);
134: return YES;
135: }
136: return NO;
137: }
138:
1.1 frystyk 139: /* Create a Command Line Object
140: ** ----------------------------
141: */
142: PRIVATE Robot * Robot_new (void)
143: {
144: Robot * me;
145: if ((me = (Robot *) calloc(1, sizeof(Robot))) == NULL ||
146: (me->tv = (struct timeval*) calloc(1, sizeof(struct timeval))) == NULL)
147: outofmem(__FILE__, "Robot_new");
1.2 frystyk 148: me->hyperdoc = HTList_new();
1.4 frystyk 149: me->htext = HTList_new();
1.1 frystyk 150: me->tv->tv_sec = DEFAULT_TIMEOUT;
151: me->cwd = HTFindRelatedName();
152: me->output = OUTPUT;
153:
1.7 frystyk 154: /* We keep an extra timeout request object for the timeout_handler */
155: me->timeout = HTRequest_new();
156: HTRequest_setContext (me->timeout, me);
157:
1.1 frystyk 158: /* Bind the Robot object together with the Request Object */
159: me->request = HTRequest_new();
160: HTRequest_setContext (me->request, me);
161: return me;
162: }
163:
164: /* Delete a Command Line Object
165: ** ----------------------------
166: */
167: PRIVATE BOOL Robot_delete (Robot * me)
168: {
169: if (me) {
1.2 frystyk 170: if (me->hyperdoc) {
171: HTList * cur = me->hyperdoc;
172: HyperDoc * pres;
173: while ((pres = (HyperDoc *) HTList_nextObject(cur)))
174: HyperDoc_delete(pres);
175: HTList_delete(me->hyperdoc);
176: }
1.4 frystyk 177: if (me->htext) {
178: HTList * cur = me->htext;
179: HText * pres;
180: while ((pres = (HText *) HTList_nextObject(cur)))
181: HText_free(pres);
182: HTList_delete(me->htext);
183: }
1.1 frystyk 184: if (me->logfile) HTLog_close();
185: if (me->output && me->output != STDOUT) fclose(me->output);
186: FREE(me->cwd);
187: free(me->tv);
188: free(me);
189: return YES;
190: }
191: return NO;
192: }
193:
1.2 frystyk 194: /*
195: ** This function creates a new request object and initializes it
196: */
197: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
198: {
199: HTRequest * newreq = HTRequest_new();
200: HTRequest_setContext (newreq, mr);
1.7 frystyk 201: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 202: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 203: HTRequest_setMethod(newreq, method);
204: return newreq;
205: }
206:
207: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
208: {
209: if (mr && request) {
210: HTRequest_delete(request);
211: return YES;
212: }
213: return NO;
214: }
215:
216: /*
217: ** Cleanup and make sure we close all connections including the persistent
218: ** ones
219: */
1.1 frystyk 220: PRIVATE void Cleanup (Robot * me, int status)
221: {
1.2 frystyk 222: HTNet_killAll();
1.1 frystyk 223: Robot_delete(me);
224: HTLibTerminate();
225: #ifdef VMS
226: exit(status ? status : 1);
227: #else
228: exit(status ? status : 0);
229: #endif
230: }
231:
232: #ifdef CATCH_SIG
233: #include <signal.h>
234: /* SetSignal
235: ** This function sets up signal handlers. This might not be necessary to
236: ** call if the application has its own handlers (lossage on SVR4)
237: */
238: PRIVATE void SetSignal (void)
239: {
240: /* On some systems (SYSV) it is necessary to catch the SIGPIPE signal
241: ** when attemting to connect to a remote host where you normally should
242: ** get `connection refused' back
243: */
244: if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) {
245: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Can't catch SIGPIPE\n");
246: } else {
247: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Ignoring SIGPIPE\n");
248: }
249: }
250: #endif /* CATCH_SIG */
251:
252: PRIVATE void VersionInfo (void)
253: {
254: TTYPrint(OUTPUT,"\n\nW3C Reference Software\n\n");
255: TTYPrint(OUTPUT,"\tW3C Mini Robot (%s) version %s.\n",
256: APP_NAME, APP_VERSION);
257: TTYPrint(OUTPUT,"\tW3C Reference Library version %s.\n\n",HTLib_version());
258: TTYPrint(OUTPUT,"Please send feedback to <libwww@w3.org>\n");
259: }
260:
261: /* terminate_handler
262: ** -----------------
1.2 frystyk 263: ** This function is registered to handle the result of the request.
264: ** If no more requests are pending then terminate program
1.1 frystyk 265: */
266: PRIVATE int terminate_handler (HTRequest * request, int status)
267: {
268: Robot * mr = (Robot *) HTRequest_context(request);
269: if (mr->logfile) HTLog_add(request, status);
1.2 frystyk 270: Thread_delete(mr, request);
1.3 frystyk 271: if (HTNet_isEmpty()) Cleanup(mr, 0);
1.1 frystyk 272: return HT_OK;
273: }
274:
275: /* timeout_handler
276: ** ---------------
277: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 278: **
279: ** BUG: This doesn't work as we don't get the right request object
280: ** back from the event loop
1.1 frystyk 281: */
282: PRIVATE int timeout_handler (HTRequest * request)
283: {
1.2 frystyk 284: Robot * mr = (Robot *) HTRequest_context(request);
285: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Request timeout...\n");
1.7 frystyk 286: #if 0
1.1 frystyk 287: HTRequest_kill(request);
1.2 frystyk 288: Thread_delete(mr, request);
1.7 frystyk 289: #endif
290: Cleanup(mr, -1);
1.4 frystyk 291: return HT_OK;
1.1 frystyk 292: }
293:
1.8 frystyk 294: /* proxy_handler
295: ** ---------------
296: ** This function is registered to be called before a request is issued
297: ** We look for redirection for proxies and gateways
298: ** returns HT_LOADED We already have this
299: ** HT_ERROR We can't load this
300: ** HT_OK Success
301: */
302: PRIVATE int proxy_handler (HTRequest * request, int status)
303: {
304: HTParentAnchor *anchor = HTRequest_anchor(request);
305: char * addr = HTAnchor_address((HTAnchor *) anchor);
306: char * newaddr = NULL;
307: if ((newaddr = HTProxy_find(addr))) {
308: StrAllocCat(newaddr, addr);
309: HTRequest_setProxying(request, YES);
310: HTAnchor_setPhysical(anchor, newaddr);
311: } else if ((newaddr = HTGateway_find(addr))) {
312: char * path = HTParse(addr,"",PARSE_HOST+PARSE_PATH+PARSE_PUNCTUATION);
313: /* Chop leading / off to make host into part of path */
314: char * gatewayed = HTParse(path+1, newaddr, PARSE_ALL);
315: HTRequest_setProxying(request, NO);
316: HTAnchor_setPhysical(anchor, gatewayed);
317: free(path);
318: free(gatewayed);
319: } else
320: HTRequest_setProxying(request, NO);
321: FREE(newaddr);
322: FREE(addr);
323: return HT_OK;
324: }
325:
1.1 frystyk 326: /* ------------------------------------------------------------------------- */
327: /* HTEXT INTERFACE */
328: /* ------------------------------------------------------------------------- */
329:
330: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
331: HTStream * stream)
332: {
333: HText * me;
1.4 frystyk 334: Robot * mr = (Robot *) HTRequest_context(request);
1.1 frystyk 335: if ((me = (HText *) calloc(1, sizeof(HText))) == NULL)
336: outofmem(__FILE__, "HText_new2");
1.4 frystyk 337:
338: /* Bind the HText object together with the Request Object */
1.1 frystyk 339: me->request = request;
1.4 frystyk 340:
341: /* Add this HyperDoc object to our list */
342: if (!mr->htext) mr->htext = HTList_new();
343: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 344: return me;
345: }
346:
1.4 frystyk 347: PUBLIC void HText_free (HText * me) {
348: if (me) free (me);
349: }
350:
1.1 frystyk 351: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
352: {
353: if (text && anchor) {
1.2 frystyk 354: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 355: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
356: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 357: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 358: HyperDoc * hd = HTAnchor_document(dest_parent);
359:
1.7 frystyk 360: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Found `%s\' - ", uri ? uri : "NULL");
361:
1.2 frystyk 362: /* Test whether we already have a hyperdoc for this document */
363: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 364: HTParentAnchor * parent = HTRequest_parent(text->request);
365: HyperDoc * last = HTAnchor_document(parent);
366: int depth = last ? last->depth+1 : 0;
1.2 frystyk 367: HTRequest * newreq = Thread_new(mr, METHOD_GET);
368: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 369: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
370: if (depth >= mr->depth) {
371: if (SHOW_MSG)
372: TTYPrint(TDEST, "loading at depth %d using HEAD\n", depth);
373: HTRequest_setMethod(newreq, METHOD_HEAD);
374: HTRequest_setOutputFormat(newreq, WWW_MIME);
375: } else {
376: if (SHOW_MSG) TTYPrint(TDEST, "loading at depth %d\n", depth);
1.2 frystyk 377: }
378: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.7 frystyk 379: if (SHOW_MSG) TTYPrint(TDEST, "not tested!\n");
1.2 frystyk 380: Thread_delete(mr, newreq);
381: }
1.7 frystyk 382: } else {
383: if (SHOW_MSG) TTYPrint(TDEST, "duplicate\n");
1.2 frystyk 384: }
1.7 frystyk 385: FREE(uri);
1.2 frystyk 386: }
387: }
388:
389: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
390: CONST char *alt, CONST char * align, BOOL isMap)
391: {
392: if (text && anchor) {
393: Robot * mr = (Robot *) HTRequest_context(text->request);
394: HTParentAnchor * dest = (HTParentAnchor *)
395: HTAnchor_followMainLink((HTAnchor *) anchor);
396: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 397:
1.2 frystyk 398: /* Test whether we already have a hyperdoc for this document */
399: if (mr->flags & MR_IMG && dest && !hd) {
400: HTParentAnchor * parent = HTRequest_parent(text->request);
401: HyperDoc * last = HTAnchor_document(parent);
402: int depth = last ? last->depth+1 : 0;
403: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
404: HyperDoc_new(mr, dest, depth);
405: if (SHOW_MSG) {
406: char * uri = HTAnchor_address((HTAnchor *) dest);
407: TTYPrint(TDEST, "Robot....... Checking Image `%s\'\n", uri);
408: free(uri);
409: }
410: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
411: if (SHOW_MSG)
412: TTYPrint(TDEST, "Robot....... Image not tested!\n");
413: Thread_delete(mr, newreq);
1.1 frystyk 414: }
415: }
416: }
417: }
418:
419: PUBLIC void HText_endAnchor (HText * text) {}
420: PUBLIC void HText_appendText (HText * text, CONST char * str) {}
421: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
422: PUBLIC void HText_endAppend (HText * text) {}
423: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
424: PUBLIC void HText_beginAppend (HText * text) {}
425: PUBLIC void HText_appendParagraph (HText * text) {}
426:
427: /* ------------------------------------------------------------------------- */
428: /* MAIN PROGRAM */
429: /* ------------------------------------------------------------------------- */
430:
431: int main (int argc, char ** argv)
432: {
433: int status = 0;
434: int arg;
435: HTChunk * keywords = NULL; /* From command line */
436: int keycnt = 0;
437: Robot * mr = Robot_new(); /* Create new Robot instance */
438:
439: /* Starts Mac GUSI socket library */
440: #ifdef GUSI
441: GUSISetup(GUSIwithSIOUXSockets);
442: GUSISetup(GUSIwithInternetSockets);
443: #endif
444:
445: #ifdef __MWERKS__ /* STR */
446: InitGraf((Ptr) &qd.thePort);
447: InitFonts();
448: InitWindows();
449: InitMenus(); TEInit();
450: InitDialogs(nil);
451: InitCursor();
452: SIOUXSettings.asktosaveonclose = false;
453: argc=ccommand(&argv);
454: #endif
455:
456: /* Initiate W3C Reference Library */
457: HTLibInit(APP_NAME, APP_VERSION);
458:
459: /* Initialize the protocol modules */
460: HTAccessInit();
461:
462: /* Initialize set of converters */
463: mr->converters = HTList_new();
464: HTConverterInit(mr->converters);
465: HTFormat_setConversion(mr->converters);
466:
467: /* Initialize bindings between file suffixes and media types */
468: HTFileInit();
469:
470: /* Get any proxy or gateway environment variables */
471: HTProxy_getEnvVar();
472:
473: /* Scan command Line for parameters */
474: for (arg=1; arg<argc; arg++) {
475: if (*argv[arg] == '-') {
476:
477: /* -? or -help: show the command line help page */
478: if (!strcmp(argv[arg],"-?") || !strcmp(argv[arg],"-help")) {
479: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(HELP);
480: keycnt = 1;
481:
482: /* non-interactive */
483: } else if (!strcmp(argv[arg], "-n")) {
484: HTAlert_setInteractive(NO);
485:
486: /* log file */
487: } else if (!strcmp(argv[arg], "-l")) {
488: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
489: argv[++arg] : DEFAULT_LOG_FILE;
490:
491: /* rule file */
492: } else if (!strcmp(argv[arg], "-r")) {
493: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
494: argv[++arg] : DEFAULT_RULE_FILE;
495:
496: /* output filename */
497: } else if (!strcmp(argv[arg], "-o")) {
498: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
499: argv[++arg] : DEFAULT_OUTPUT_FILE;
500:
501: /* timeout -- Change the default request timeout */
502: } else if (!strcmp(argv[arg], "-timeout")) {
503: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
504: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
505: if (timeout > 0) mr->tv->tv_sec = timeout;
506:
1.7 frystyk 507: /* preemptive or non-preemptive access */
1.1 frystyk 508: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 509: HTRequest_setPreemptive(mr->request, YES);
510: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 511:
512: /* test inlined images */
513: } else if (!strcmp(argv[arg], "-img")) {
514: mr->flags |= MR_IMG;
515:
516: /* load anchors */
517: } else if (!strcmp(argv[arg], "-link")) {
518: mr->flags |= MR_LINK;
1.7 frystyk 519: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
520: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 521:
1.7 frystyk 522: /* preemptive or non-preemptive access */
1.2 frystyk 523: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 524: HTRequest_setPreemptive(mr->request, YES);
525: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 526:
527: /* print version and exit */
528: } else if (!strcmp(argv[arg], "-version")) {
529: VersionInfo();
530: Cleanup(mr, 0);
531:
532: #ifdef WWWTRACE
533: /* trace flags */
534: } else if (!strncmp(argv[arg], "-v", 2)) {
535: char *p = argv[arg]+2;
536: WWWTRACE = 0;
537: for(; *p; p++) {
538: switch (*p) {
539: case 'a': WWWTRACE |= SHOW_ANCHOR_TRACE; break;
540: case 'b': WWWTRACE |= SHOW_BIND_TRACE; break;
541: case 'c': WWWTRACE |= SHOW_CACHE_TRACE; break;
542: case 'g': WWWTRACE |= SHOW_SGML_TRACE; break;
543: case 'p': WWWTRACE |= SHOW_PROTOCOL_TRACE; break;
544: case 's': WWWTRACE |= SHOW_STREAM_TRACE; break;
545: case 't': WWWTRACE |= SHOW_THREAD_TRACE; break;
546: case 'u': WWWTRACE |= SHOW_URI_TRACE; break;
547: default:
548: if (SHOW_MSG)
549: TTYPrint(TDEST,"Bad parameter (%s) in -v option\n",
550: argv[arg]);
551: }
552: }
553: if (!WWWTRACE) WWWTRACE = SHOW_ALL_TRACE;
554: #endif
555:
556: } else {
557: if (SHOW_MSG) TTYPrint(TDEST,"Bad Argument (%s)\n", argv[arg]);
558: }
559: } else { /* If no leading `-' then check for URL or keywords */
560: if (!keycnt) {
561: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
562: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 563: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 564: keycnt = 1;
565: FREE(ref);
566: } else { /* Check for successive keyword arguments */
567: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
568: if (keycnt++ <= 1)
1.5 frystyk 569: keywords = HTChunk_new(128);
1.1 frystyk 570: else
1.5 frystyk 571: HTChunk_putc(keywords, ' ');
572: HTChunk_puts(keywords, HTStrip(escaped));
1.1 frystyk 573: free(escaped);
574: }
575: }
576: }
577:
578: #ifdef CATCH_SIG
579: SetSignal();
580: #endif
581:
582: if (!keycnt) {
1.2 frystyk 583: if (SHOW_MSG) TTYPrint(TDEST, "Please specify URL to check.\n");
1.1 frystyk 584: Cleanup(mr, -1);
585: }
586:
587: /* Rule file specified? */
588: if (mr->rules) {
589: HTList * list = HTList_new();
590: HTRequest * rr = HTRequest_new();
591: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
592: HTParentAnchor * ra = (HTParentAnchor *) HTAnchor_findAddress(rules);
1.7 frystyk 593: HTRequest_setPreemptive(rr, YES);
1.1 frystyk 594: HTConversion_add(list, "application/x-www-rules", "*/*", HTRules,
595: 1.0, 0.0, 0.0);
596: HTRequest_setConversion(rr, list, YES);
1.8 frystyk 597: HTAlert_add(HTConfirm, HT_A_CONFIRM);
1.1 frystyk 598: if (HTLoadAnchor((HTAnchor *) ra, rr) != YES)
599: if (SHOW_MSG) TTYPrint(TDEST, "Can't access rules\n");
600: HTConversion_deleteAll(list);
601: HTRequest_delete(rr);
1.8 frystyk 602: HTAlert_delete(HTConfirm);
1.1 frystyk 603: FREE(rules);
604: }
605:
606: /* Output file specified? */
607: if (mr->outputfile) {
608: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
609: if (SHOW_MSG) TTYPrint(TDEST, "Can't open `%s'\n", mr->outputfile);
610: mr->output = OUTPUT;
611: }
612: }
613:
614: /* Log file specifed? */
615: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
616:
617: /* Register our User Prompts etc in the Alert Manager */
618: if (HTAlert_interactive()) {
619: HTAlert_add(HTError_print, HT_A_MESSAGE);
620: HTAlert_add(HTConfirm, HT_A_CONFIRM);
621: HTAlert_add(HTPrompt, HT_A_PROMPT);
622: HTAlert_add(HTPromptPassword, HT_A_SECRET);
623: HTAlert_add(HTPromptUsernameAndPassword, HT_A_USER_PW);
624: }
625:
626: /* Register a call back function for the Net Manager */
1.8 frystyk 627: HTNetCall_addBefore(proxy_handler, 0);
1.1 frystyk 628: HTNetCall_addAfter(terminate_handler, HT_ALL);
629:
630: /* Set timeout on sockets */
1.7 frystyk 631: HTEvent_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 632:
633: /* Start the request */
634: if (keywords) /* Search */
1.5 frystyk 635: status = HTSearch(HTChunk_data(keywords), mr->anchor, mr->request);
1.1 frystyk 636: else
637: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
638:
1.5 frystyk 639: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 640: if (status != YES) {
641: if (SHOW_MSG) TTYPrint(TDEST, "Can't access resource\n");
642: Cleanup(mr, -1);
643: }
644:
645: /* Go into the event loop... */
646: HTEvent_Loop(mr->request);
647:
648: /* Only gets here if event loop fails */
649: Cleanup(mr, 0);
650: return 0;
651: }
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