Annotation of libwww/Robot/src/HTRobot.c, revision 1.11
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"
1.10 frystyk 20: #include "WWWInit.h"
1.9 frystyk 21:
1.4 frystyk 22: #include "HText.h"
1.1 frystyk 23:
24: #include "HTRobot.h" /* Implemented here */
25:
26: #ifndef VR
27: #define VR "unspecified"
28: #endif
29:
30: #define APP_NAME "W3CRobot"
31: #define APP_VERSION VR
32:
33: /* Default page for "-help" command line option */
34: #define HELP "http://www.w3.org/pub/WWW/Robot/User/CommandLine.html"
35:
36: #define DEFAULT_OUTPUT_FILE "robot.out"
37: #define DEFAULT_RULE_FILE "robot.conf"
38: #define DEFAULT_LOG_FILE "robot.log"
1.7 frystyk 39: #define DEFAULT_DEPTH 0
1.1 frystyk 40:
41: #define SHOW_MSG (WWWTRACE || HTAlert_interactive())
42:
1.7 frystyk 43: #define DEFAULT_TIMEOUT 10 /* timeout in seconds */
1.1 frystyk 44:
45: #if defined(__svr4__)
46: #define CATCH_SIG
47: #endif
48:
49: typedef enum _MRFlags {
1.2 frystyk 50: MR_IMG = 0x1,
51: MR_LINK = 0x2,
1.7 frystyk 52: MR_PREEMPTIVE= 0x4
1.1 frystyk 53: } MRFlags;
54:
55: typedef struct _Robot {
56: HTRequest * request;
1.7 frystyk 57: HTRequest * timeout; /* Until we get a server eventloop */
1.1 frystyk 58: HTParentAnchor * anchor;
1.2 frystyk 59: int depth; /* How deep is our tree */
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: HTList * converters;
65: char * rules;
66: char * logfile;
67: char * outputfile;
68: FILE * output;
69: MRFlags flags;
70: } Robot;
71:
72: typedef enum _LoadState {
73: L_INVALID = -2,
74: L_LOADING = -1,
75: L_SUCCESS = 0,
76: L_ERROR
77: } LoadState;
78:
79: /*
80: ** The HyperDoc object is bound to the anchor and contains information about
81: ** where we are in the search for recursive searches
82: */
83: typedef struct _HyperDoc {
84: HTParentAnchor * anchor;
85: LoadState state;
86: int depth;
87: } HyperDoc;
88:
89: /*
90: ** This is the HText object that is created every time we start parsing a
91: ** HTML object
92: */
1.4 frystyk 93: struct _HText {
1.1 frystyk 94: HTRequest * request;
1.4 frystyk 95: };
1.1 frystyk 96:
97: PUBLIC HText * HTMainText = NULL;
98: PUBLIC HTParentAnchor * HTMainAnchor = NULL;
99: PUBLIC HTStyleSheet * styleSheet = NULL;
100:
101: /* ------------------------------------------------------------------------- */
102:
1.2 frystyk 103: /* Create a "HyperDoc" object
104: ** --------------------------
105: ** A HyperDoc object contains information about whether we have already
106: ** started checking the anchor and the depth in our search
107: */
108: PRIVATE HyperDoc * HyperDoc_new (Robot * mr,HTParentAnchor * anchor, int depth)
109: {
110: HyperDoc * hd;
111: if ((hd = (HyperDoc *) calloc(1, sizeof(HyperDoc))) == NULL)
112: outofmem(__FILE__, "HyperDoc_new");
113: hd->state = L_INVALID;
114: hd->depth = depth;
115:
116: /* Bind the HyperDoc object together with the Anchor Object */
117: hd->anchor = anchor;
118: HTAnchor_setDocument(anchor, (void *) hd);
119:
120: /* Add this HyperDoc object to our list */
121: if (!mr->hyperdoc) mr->hyperdoc = HTList_new();
122: HTList_addObject(mr->hyperdoc, (void *) hd);
123: return hd;
124: }
125:
126: /* Delete a "HyperDoc" object
127: ** --------------------------
128: */
129: PRIVATE BOOL HyperDoc_delete (HyperDoc * hd)
130: {
131: if (hd) {
1.11 ! frystyk 132: HT_FREE (hd);
1.2 frystyk 133: return YES;
134: }
135: return NO;
136: }
137:
1.1 frystyk 138: /* Create a Command Line Object
139: ** ----------------------------
140: */
141: PRIVATE Robot * Robot_new (void)
142: {
143: Robot * me;
144: if ((me = (Robot *) calloc(1, sizeof(Robot))) == NULL ||
145: (me->tv = (struct timeval*) calloc(1, sizeof(struct timeval))) == NULL)
146: outofmem(__FILE__, "Robot_new");
1.2 frystyk 147: me->hyperdoc = HTList_new();
1.4 frystyk 148: me->htext = HTList_new();
1.1 frystyk 149: me->tv->tv_sec = DEFAULT_TIMEOUT;
150: me->cwd = HTFindRelatedName();
151: me->output = OUTPUT;
152:
1.7 frystyk 153: /* We keep an extra timeout request object for the timeout_handler */
154: me->timeout = HTRequest_new();
155: HTRequest_setContext (me->timeout, me);
156:
1.1 frystyk 157: /* Bind the Robot object together with the Request Object */
158: me->request = HTRequest_new();
159: HTRequest_setContext (me->request, me);
160: return me;
161: }
162:
163: /* Delete a Command Line Object
164: ** ----------------------------
165: */
166: PRIVATE BOOL Robot_delete (Robot * me)
167: {
168: if (me) {
1.2 frystyk 169: if (me->hyperdoc) {
170: HTList * cur = me->hyperdoc;
171: HyperDoc * pres;
172: while ((pres = (HyperDoc *) HTList_nextObject(cur)))
173: HyperDoc_delete(pres);
174: HTList_delete(me->hyperdoc);
175: }
1.4 frystyk 176: if (me->htext) {
177: HTList * cur = me->htext;
178: HText * pres;
179: while ((pres = (HText *) HTList_nextObject(cur)))
180: HText_free(pres);
181: HTList_delete(me->htext);
182: }
1.1 frystyk 183: if (me->logfile) HTLog_close();
184: if (me->output && me->output != STDOUT) fclose(me->output);
1.11 ! frystyk 185: HT_FREE(me->cwd);
! 186: HT_FREE(me->tv);
! 187: HT_FREE(me);
1.1 frystyk 188: return YES;
189: }
190: return NO;
191: }
192:
1.2 frystyk 193: /*
194: ** This function creates a new request object and initializes it
195: */
196: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
197: {
198: HTRequest * newreq = HTRequest_new();
199: HTRequest_setContext (newreq, mr);
1.7 frystyk 200: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 201: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 202: HTRequest_setMethod(newreq, method);
203: return newreq;
204: }
205:
206: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
207: {
208: if (mr && request) {
209: HTRequest_delete(request);
210: return YES;
211: }
212: return NO;
213: }
214:
215: /*
216: ** Cleanup and make sure we close all connections including the persistent
217: ** ones
218: */
1.1 frystyk 219: PRIVATE void Cleanup (Robot * me, int status)
220: {
1.2 frystyk 221: HTNet_killAll();
1.1 frystyk 222: Robot_delete(me);
223: HTLibTerminate();
224: #ifdef VMS
225: exit(status ? status : 1);
226: #else
227: exit(status ? status : 0);
228: #endif
229: }
230:
231: #ifdef CATCH_SIG
232: #include <signal.h>
233: /* SetSignal
234: ** This function sets up signal handlers. This might not be necessary to
235: ** call if the application has its own handlers (lossage on SVR4)
236: */
237: PRIVATE void SetSignal (void)
238: {
239: /* On some systems (SYSV) it is necessary to catch the SIGPIPE signal
240: ** when attemting to connect to a remote host where you normally should
241: ** get `connection refused' back
242: */
243: if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) {
244: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Can't catch SIGPIPE\n");
245: } else {
246: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Ignoring SIGPIPE\n");
247: }
248: }
249: #endif /* CATCH_SIG */
250:
251: PRIVATE void VersionInfo (void)
252: {
253: TTYPrint(OUTPUT,"\n\nW3C Reference Software\n\n");
254: TTYPrint(OUTPUT,"\tW3C Mini Robot (%s) version %s.\n",
255: APP_NAME, APP_VERSION);
256: TTYPrint(OUTPUT,"\tW3C Reference Library version %s.\n\n",HTLib_version());
257: TTYPrint(OUTPUT,"Please send feedback to <libwww@w3.org>\n");
258: }
259:
260: /* terminate_handler
261: ** -----------------
1.2 frystyk 262: ** This function is registered to handle the result of the request.
263: ** If no more requests are pending then terminate program
1.1 frystyk 264: */
265: PRIVATE int terminate_handler (HTRequest * request, int status)
266: {
267: Robot * mr = (Robot *) HTRequest_context(request);
268: if (mr->logfile) HTLog_add(request, status);
1.2 frystyk 269: Thread_delete(mr, request);
1.3 frystyk 270: if (HTNet_isEmpty()) Cleanup(mr, 0);
1.1 frystyk 271: return HT_OK;
272: }
273:
274: /* timeout_handler
275: ** ---------------
276: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 277: **
278: ** BUG: This doesn't work as we don't get the right request object
279: ** back from the event loop
1.1 frystyk 280: */
281: PRIVATE int timeout_handler (HTRequest * request)
282: {
1.2 frystyk 283: Robot * mr = (Robot *) HTRequest_context(request);
284: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Request timeout...\n");
1.7 frystyk 285: #if 0
1.1 frystyk 286: HTRequest_kill(request);
1.2 frystyk 287: Thread_delete(mr, request);
1.7 frystyk 288: #endif
289: Cleanup(mr, -1);
1.4 frystyk 290: return HT_OK;
1.1 frystyk 291: }
292:
1.8 frystyk 293: /* proxy_handler
294: ** ---------------
295: ** This function is registered to be called before a request is issued
296: ** We look for redirection for proxies and gateways
297: ** returns HT_LOADED We already have this
298: ** HT_ERROR We can't load this
299: ** HT_OK Success
300: */
301: PRIVATE int proxy_handler (HTRequest * request, int status)
302: {
303: HTParentAnchor *anchor = HTRequest_anchor(request);
304: char * addr = HTAnchor_address((HTAnchor *) anchor);
305: char * newaddr = NULL;
306: if ((newaddr = HTProxy_find(addr))) {
307: StrAllocCat(newaddr, addr);
308: HTRequest_setProxying(request, YES);
309: HTAnchor_setPhysical(anchor, newaddr);
310: } else if ((newaddr = HTGateway_find(addr))) {
311: char * path = HTParse(addr,"",PARSE_HOST+PARSE_PATH+PARSE_PUNCTUATION);
312: /* Chop leading / off to make host into part of path */
313: char * gatewayed = HTParse(path+1, newaddr, PARSE_ALL);
314: HTRequest_setProxying(request, NO);
315: HTAnchor_setPhysical(anchor, gatewayed);
1.11 ! frystyk 316: HT_FREE(path);
! 317: HT_FREE(gatewayed);
1.8 frystyk 318: } else
319: HTRequest_setProxying(request, NO);
1.11 ! frystyk 320: HT_FREE(newaddr);
! 321: HT_FREE(addr);
1.8 frystyk 322: return HT_OK;
323: }
324:
1.1 frystyk 325: /* ------------------------------------------------------------------------- */
326: /* HTEXT INTERFACE */
327: /* ------------------------------------------------------------------------- */
328:
329: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
330: HTStream * stream)
331: {
332: HText * me;
1.4 frystyk 333: Robot * mr = (Robot *) HTRequest_context(request);
1.1 frystyk 334: if ((me = (HText *) calloc(1, sizeof(HText))) == NULL)
335: outofmem(__FILE__, "HText_new2");
1.4 frystyk 336:
337: /* Bind the HText object together with the Request Object */
1.1 frystyk 338: me->request = request;
1.4 frystyk 339:
340: /* Add this HyperDoc object to our list */
341: if (!mr->htext) mr->htext = HTList_new();
342: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 343: return me;
344: }
345:
1.4 frystyk 346: PUBLIC void HText_free (HText * me) {
1.11 ! frystyk 347: if (me) HT_FREE (me);
1.4 frystyk 348: }
349:
1.1 frystyk 350: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
351: {
352: if (text && anchor) {
1.2 frystyk 353: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 354: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
355: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 356: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 357: HyperDoc * hd = HTAnchor_document(dest_parent);
358:
1.7 frystyk 359: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Found `%s\' - ", uri ? uri : "NULL");
360:
1.2 frystyk 361: /* Test whether we already have a hyperdoc for this document */
362: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 363: HTParentAnchor * parent = HTRequest_parent(text->request);
364: HyperDoc * last = HTAnchor_document(parent);
365: int depth = last ? last->depth+1 : 0;
1.2 frystyk 366: HTRequest * newreq = Thread_new(mr, METHOD_GET);
367: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 368: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
369: if (depth >= mr->depth) {
370: if (SHOW_MSG)
371: TTYPrint(TDEST, "loading at depth %d using HEAD\n", depth);
372: HTRequest_setMethod(newreq, METHOD_HEAD);
373: HTRequest_setOutputFormat(newreq, WWW_MIME);
374: } else {
375: if (SHOW_MSG) TTYPrint(TDEST, "loading at depth %d\n", depth);
1.2 frystyk 376: }
377: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.7 frystyk 378: if (SHOW_MSG) TTYPrint(TDEST, "not tested!\n");
1.2 frystyk 379: Thread_delete(mr, newreq);
380: }
1.7 frystyk 381: } else {
382: if (SHOW_MSG) TTYPrint(TDEST, "duplicate\n");
1.2 frystyk 383: }
1.11 ! frystyk 384: HT_FREE(uri);
1.2 frystyk 385: }
386: }
387:
388: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
389: CONST char *alt, CONST char * align, BOOL isMap)
390: {
391: if (text && anchor) {
392: Robot * mr = (Robot *) HTRequest_context(text->request);
393: HTParentAnchor * dest = (HTParentAnchor *)
394: HTAnchor_followMainLink((HTAnchor *) anchor);
395: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 396:
1.2 frystyk 397: /* Test whether we already have a hyperdoc for this document */
398: if (mr->flags & MR_IMG && dest && !hd) {
399: HTParentAnchor * parent = HTRequest_parent(text->request);
400: HyperDoc * last = HTAnchor_document(parent);
401: int depth = last ? last->depth+1 : 0;
402: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
403: HyperDoc_new(mr, dest, depth);
404: if (SHOW_MSG) {
405: char * uri = HTAnchor_address((HTAnchor *) dest);
406: TTYPrint(TDEST, "Robot....... Checking Image `%s\'\n", uri);
1.11 ! frystyk 407: HT_FREE(uri);
1.2 frystyk 408: }
409: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
410: if (SHOW_MSG)
411: TTYPrint(TDEST, "Robot....... Image not tested!\n");
412: Thread_delete(mr, newreq);
1.1 frystyk 413: }
414: }
415: }
416: }
417:
418: PUBLIC void HText_endAnchor (HText * text) {}
419: PUBLIC void HText_appendText (HText * text, CONST char * str) {}
420: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
421: PUBLIC void HText_endAppend (HText * text) {}
422: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
423: PUBLIC void HText_beginAppend (HText * text) {}
424: PUBLIC void HText_appendParagraph (HText * text) {}
425:
426: /* ------------------------------------------------------------------------- */
427: /* MAIN PROGRAM */
428: /* ------------------------------------------------------------------------- */
429:
430: int main (int argc, char ** argv)
431: {
432: int status = 0;
433: int arg;
434: HTChunk * keywords = NULL; /* From command line */
435: int keycnt = 0;
436: Robot * mr = Robot_new(); /* Create new Robot instance */
437:
438: /* Starts Mac GUSI socket library */
439: #ifdef GUSI
440: GUSISetup(GUSIwithSIOUXSockets);
441: GUSISetup(GUSIwithInternetSockets);
442: #endif
443:
444: #ifdef __MWERKS__ /* STR */
445: InitGraf((Ptr) &qd.thePort);
446: InitFonts();
447: InitWindows();
448: InitMenus(); TEInit();
449: InitDialogs(nil);
450: InitCursor();
451: SIOUXSettings.asktosaveonclose = false;
452: argc=ccommand(&argv);
453: #endif
454:
455: /* Initiate W3C Reference Library */
456: HTLibInit(APP_NAME, APP_VERSION);
457:
458: /* Initialize the protocol modules */
459: HTAccessInit();
460:
461: /* Initialize set of converters */
462: mr->converters = HTList_new();
463: HTConverterInit(mr->converters);
464: HTFormat_setConversion(mr->converters);
465:
466: /* Initialize bindings between file suffixes and media types */
467: HTFileInit();
468:
469: /* Get any proxy or gateway environment variables */
470: HTProxy_getEnvVar();
471:
472: /* Scan command Line for parameters */
473: for (arg=1; arg<argc; arg++) {
474: if (*argv[arg] == '-') {
475:
476: /* -? or -help: show the command line help page */
477: if (!strcmp(argv[arg],"-?") || !strcmp(argv[arg],"-help")) {
478: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(HELP);
479: keycnt = 1;
480:
481: /* non-interactive */
482: } else if (!strcmp(argv[arg], "-n")) {
483: HTAlert_setInteractive(NO);
484:
485: /* log file */
486: } else if (!strcmp(argv[arg], "-l")) {
487: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
488: argv[++arg] : DEFAULT_LOG_FILE;
489:
490: /* rule file */
491: } else if (!strcmp(argv[arg], "-r")) {
492: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
493: argv[++arg] : DEFAULT_RULE_FILE;
494:
495: /* output filename */
496: } else if (!strcmp(argv[arg], "-o")) {
497: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
498: argv[++arg] : DEFAULT_OUTPUT_FILE;
499:
500: /* timeout -- Change the default request timeout */
501: } else if (!strcmp(argv[arg], "-timeout")) {
502: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
503: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
504: if (timeout > 0) mr->tv->tv_sec = timeout;
505:
1.7 frystyk 506: /* preemptive or non-preemptive access */
1.1 frystyk 507: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 508: HTRequest_setPreemptive(mr->request, YES);
509: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 510:
511: /* test inlined images */
512: } else if (!strcmp(argv[arg], "-img")) {
513: mr->flags |= MR_IMG;
514:
515: /* load anchors */
516: } else if (!strcmp(argv[arg], "-link")) {
517: mr->flags |= MR_LINK;
1.7 frystyk 518: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
519: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 520:
1.7 frystyk 521: /* preemptive or non-preemptive access */
1.2 frystyk 522: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 523: HTRequest_setPreemptive(mr->request, YES);
524: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 525:
526: /* print version and exit */
527: } else if (!strcmp(argv[arg], "-version")) {
528: VersionInfo();
529: Cleanup(mr, 0);
530:
531: #ifdef WWWTRACE
532: /* trace flags */
533: } else if (!strncmp(argv[arg], "-v", 2)) {
534: char *p = argv[arg]+2;
535: WWWTRACE = 0;
536: for(; *p; p++) {
537: switch (*p) {
538: case 'a': WWWTRACE |= SHOW_ANCHOR_TRACE; break;
539: case 'b': WWWTRACE |= SHOW_BIND_TRACE; break;
540: case 'c': WWWTRACE |= SHOW_CACHE_TRACE; break;
541: case 'g': WWWTRACE |= SHOW_SGML_TRACE; break;
542: case 'p': WWWTRACE |= SHOW_PROTOCOL_TRACE; break;
543: case 's': WWWTRACE |= SHOW_STREAM_TRACE; break;
544: case 't': WWWTRACE |= SHOW_THREAD_TRACE; break;
545: case 'u': WWWTRACE |= SHOW_URI_TRACE; break;
546: default:
547: if (SHOW_MSG)
548: TTYPrint(TDEST,"Bad parameter (%s) in -v option\n",
549: argv[arg]);
550: }
551: }
552: if (!WWWTRACE) WWWTRACE = SHOW_ALL_TRACE;
553: #endif
554:
555: } else {
556: if (SHOW_MSG) TTYPrint(TDEST,"Bad Argument (%s)\n", argv[arg]);
557: }
558: } else { /* If no leading `-' then check for URL or keywords */
559: if (!keycnt) {
560: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
561: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 562: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 563: keycnt = 1;
1.11 ! frystyk 564: HT_FREE(ref);
1.1 frystyk 565: } else { /* Check for successive keyword arguments */
566: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
567: if (keycnt++ <= 1)
1.5 frystyk 568: keywords = HTChunk_new(128);
1.1 frystyk 569: else
1.5 frystyk 570: HTChunk_putc(keywords, ' ');
571: HTChunk_puts(keywords, HTStrip(escaped));
1.11 ! frystyk 572: HT_FREE(escaped);
1.1 frystyk 573: }
574: }
575: }
576:
577: #ifdef CATCH_SIG
578: SetSignal();
579: #endif
580:
581: if (!keycnt) {
1.2 frystyk 582: if (SHOW_MSG) TTYPrint(TDEST, "Please specify URL to check.\n");
1.1 frystyk 583: Cleanup(mr, -1);
584: }
585:
586: /* Rule file specified? */
587: if (mr->rules) {
588: HTList * list = HTList_new();
589: HTRequest * rr = HTRequest_new();
590: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
591: HTParentAnchor * ra = (HTParentAnchor *) HTAnchor_findAddress(rules);
1.7 frystyk 592: HTRequest_setPreemptive(rr, YES);
1.1 frystyk 593: HTConversion_add(list, "application/x-www-rules", "*/*", HTRules,
594: 1.0, 0.0, 0.0);
595: HTRequest_setConversion(rr, list, YES);
1.8 frystyk 596: HTAlert_add(HTConfirm, HT_A_CONFIRM);
1.1 frystyk 597: if (HTLoadAnchor((HTAnchor *) ra, rr) != YES)
598: if (SHOW_MSG) TTYPrint(TDEST, "Can't access rules\n");
599: HTConversion_deleteAll(list);
600: HTRequest_delete(rr);
1.8 frystyk 601: HTAlert_delete(HTConfirm);
1.11 ! frystyk 602: HT_FREE(rules);
1.1 frystyk 603: }
604:
605: /* Output file specified? */
606: if (mr->outputfile) {
607: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
608: if (SHOW_MSG) TTYPrint(TDEST, "Can't open `%s'\n", mr->outputfile);
609: mr->output = OUTPUT;
610: }
611: }
612:
613: /* Log file specifed? */
614: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
615:
616: /* Register our User Prompts etc in the Alert Manager */
617: if (HTAlert_interactive()) {
618: HTAlert_add(HTError_print, HT_A_MESSAGE);
619: HTAlert_add(HTConfirm, HT_A_CONFIRM);
620: HTAlert_add(HTPrompt, HT_A_PROMPT);
621: HTAlert_add(HTPromptPassword, HT_A_SECRET);
622: HTAlert_add(HTPromptUsernameAndPassword, HT_A_USER_PW);
623: }
624:
625: /* Register a call back function for the Net Manager */
1.8 frystyk 626: HTNetCall_addBefore(proxy_handler, 0);
1.1 frystyk 627: HTNetCall_addAfter(terminate_handler, HT_ALL);
628:
629: /* Set timeout on sockets */
1.7 frystyk 630: HTEvent_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 631:
632: /* Start the request */
633: if (keywords) /* Search */
1.5 frystyk 634: status = HTSearch(HTChunk_data(keywords), mr->anchor, mr->request);
1.1 frystyk 635: else
636: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
637:
1.5 frystyk 638: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 639: if (status != YES) {
640: if (SHOW_MSG) TTYPrint(TDEST, "Can't access resource\n");
641: Cleanup(mr, -1);
642: }
643:
644: /* Go into the event loop... */
645: HTEvent_Loop(mr->request);
646:
647: /* Only gets here if event loop fails */
648: Cleanup(mr, 0);
649: return 0;
650: }
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