Annotation of libwww/Robot/src/HTRobot.c, revision 1.12
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,
1.12 ! frystyk 52: MR_PREEMPTIVE= 0x4,
! 53: MR_TIME = 0x8
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) {
1.11 frystyk 133: HT_FREE (hd);
1.2 frystyk 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);
1.12 ! frystyk 186: if (me->flags & MR_TIME) {
! 187: time_t local = time(NULL);
! 188: TTYPrint(TDEST, "Robot terminated %s\n",HTDateTimeStr(&local,YES));
! 189: }
1.11 frystyk 190: HT_FREE(me->cwd);
191: HT_FREE(me->tv);
192: HT_FREE(me);
1.1 frystyk 193: return YES;
194: }
195: return NO;
196: }
197:
1.2 frystyk 198: /*
199: ** This function creates a new request object and initializes it
200: */
201: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
202: {
203: HTRequest * newreq = HTRequest_new();
204: HTRequest_setContext (newreq, mr);
1.7 frystyk 205: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 206: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 207: HTRequest_setMethod(newreq, method);
208: return newreq;
209: }
210:
211: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
212: {
213: if (mr && request) {
214: HTRequest_delete(request);
215: return YES;
216: }
217: return NO;
218: }
219:
220: /*
221: ** Cleanup and make sure we close all connections including the persistent
222: ** ones
223: */
1.1 frystyk 224: PRIVATE void Cleanup (Robot * me, int status)
225: {
1.2 frystyk 226: HTNet_killAll();
1.1 frystyk 227: Robot_delete(me);
228: HTLibTerminate();
229: #ifdef VMS
230: exit(status ? status : 1);
231: #else
232: exit(status ? status : 0);
233: #endif
234: }
235:
236: #ifdef CATCH_SIG
237: #include <signal.h>
238: /* SetSignal
239: ** This function sets up signal handlers. This might not be necessary to
240: ** call if the application has its own handlers (lossage on SVR4)
241: */
242: PRIVATE void SetSignal (void)
243: {
244: /* On some systems (SYSV) it is necessary to catch the SIGPIPE signal
245: ** when attemting to connect to a remote host where you normally should
246: ** get `connection refused' back
247: */
248: if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) {
249: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Can't catch SIGPIPE\n");
250: } else {
251: if (PROT_TRACE) TTYPrint(TDEST, "HTSignal.... Ignoring SIGPIPE\n");
252: }
253: }
254: #endif /* CATCH_SIG */
255:
256: PRIVATE void VersionInfo (void)
257: {
258: TTYPrint(OUTPUT,"\n\nW3C Reference Software\n\n");
259: TTYPrint(OUTPUT,"\tW3C Mini Robot (%s) version %s.\n",
260: APP_NAME, APP_VERSION);
261: TTYPrint(OUTPUT,"\tW3C Reference Library version %s.\n\n",HTLib_version());
262: TTYPrint(OUTPUT,"Please send feedback to <libwww@w3.org>\n");
263: }
264:
265: /* terminate_handler
266: ** -----------------
1.2 frystyk 267: ** This function is registered to handle the result of the request.
268: ** If no more requests are pending then terminate program
1.1 frystyk 269: */
270: PRIVATE int terminate_handler (HTRequest * request, int status)
271: {
272: Robot * mr = (Robot *) HTRequest_context(request);
273: if (mr->logfile) HTLog_add(request, status);
1.2 frystyk 274: Thread_delete(mr, request);
1.3 frystyk 275: if (HTNet_isEmpty()) Cleanup(mr, 0);
1.1 frystyk 276: return HT_OK;
277: }
278:
279: /* timeout_handler
280: ** ---------------
281: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 282: **
283: ** BUG: This doesn't work as we don't get the right request object
284: ** back from the event loop
1.1 frystyk 285: */
286: PRIVATE int timeout_handler (HTRequest * request)
287: {
1.2 frystyk 288: Robot * mr = (Robot *) HTRequest_context(request);
289: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Request timeout...\n");
1.7 frystyk 290: #if 0
1.1 frystyk 291: HTRequest_kill(request);
1.2 frystyk 292: Thread_delete(mr, request);
1.7 frystyk 293: #endif
294: Cleanup(mr, -1);
1.4 frystyk 295: return HT_OK;
1.1 frystyk 296: }
297:
1.8 frystyk 298: /* proxy_handler
299: ** ---------------
300: ** This function is registered to be called before a request is issued
301: ** We look for redirection for proxies and gateways
302: ** returns HT_LOADED We already have this
303: ** HT_ERROR We can't load this
304: ** HT_OK Success
305: */
306: PRIVATE int proxy_handler (HTRequest * request, int status)
307: {
308: HTParentAnchor *anchor = HTRequest_anchor(request);
309: char * addr = HTAnchor_address((HTAnchor *) anchor);
310: char * newaddr = NULL;
311: if ((newaddr = HTProxy_find(addr))) {
312: StrAllocCat(newaddr, addr);
313: HTRequest_setProxying(request, YES);
314: HTAnchor_setPhysical(anchor, newaddr);
315: } else if ((newaddr = HTGateway_find(addr))) {
316: char * path = HTParse(addr,"",PARSE_HOST+PARSE_PATH+PARSE_PUNCTUATION);
317: /* Chop leading / off to make host into part of path */
318: char * gatewayed = HTParse(path+1, newaddr, PARSE_ALL);
319: HTRequest_setProxying(request, NO);
320: HTAnchor_setPhysical(anchor, gatewayed);
1.11 frystyk 321: HT_FREE(path);
322: HT_FREE(gatewayed);
1.8 frystyk 323: } else
324: HTRequest_setProxying(request, NO);
1.11 frystyk 325: HT_FREE(newaddr);
326: HT_FREE(addr);
1.8 frystyk 327: return HT_OK;
328: }
329:
1.1 frystyk 330: /* ------------------------------------------------------------------------- */
331: /* HTEXT INTERFACE */
332: /* ------------------------------------------------------------------------- */
333:
334: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
335: HTStream * stream)
336: {
337: HText * me;
1.4 frystyk 338: Robot * mr = (Robot *) HTRequest_context(request);
1.1 frystyk 339: if ((me = (HText *) calloc(1, sizeof(HText))) == NULL)
340: outofmem(__FILE__, "HText_new2");
1.4 frystyk 341:
342: /* Bind the HText object together with the Request Object */
1.1 frystyk 343: me->request = request;
1.4 frystyk 344:
345: /* Add this HyperDoc object to our list */
346: if (!mr->htext) mr->htext = HTList_new();
347: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 348: return me;
349: }
350:
1.4 frystyk 351: PUBLIC void HText_free (HText * me) {
1.11 frystyk 352: if (me) HT_FREE (me);
1.4 frystyk 353: }
354:
1.1 frystyk 355: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
356: {
357: if (text && anchor) {
1.2 frystyk 358: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 359: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
360: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 361: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 362: HyperDoc * hd = HTAnchor_document(dest_parent);
363:
1.7 frystyk 364: if (SHOW_MSG) TTYPrint(TDEST, "Robot....... Found `%s\' - ", uri ? uri : "NULL");
365:
1.2 frystyk 366: /* Test whether we already have a hyperdoc for this document */
367: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 368: HTParentAnchor * parent = HTRequest_parent(text->request);
369: HyperDoc * last = HTAnchor_document(parent);
370: int depth = last ? last->depth+1 : 0;
1.2 frystyk 371: HTRequest * newreq = Thread_new(mr, METHOD_GET);
372: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 373: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
374: if (depth >= mr->depth) {
375: if (SHOW_MSG)
376: TTYPrint(TDEST, "loading at depth %d using HEAD\n", depth);
377: HTRequest_setMethod(newreq, METHOD_HEAD);
378: HTRequest_setOutputFormat(newreq, WWW_MIME);
379: } else {
380: if (SHOW_MSG) TTYPrint(TDEST, "loading at depth %d\n", depth);
1.2 frystyk 381: }
382: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.7 frystyk 383: if (SHOW_MSG) TTYPrint(TDEST, "not tested!\n");
1.2 frystyk 384: Thread_delete(mr, newreq);
385: }
1.7 frystyk 386: } else {
387: if (SHOW_MSG) TTYPrint(TDEST, "duplicate\n");
1.2 frystyk 388: }
1.11 frystyk 389: HT_FREE(uri);
1.2 frystyk 390: }
391: }
392:
393: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
394: CONST char *alt, CONST char * align, BOOL isMap)
395: {
396: if (text && anchor) {
397: Robot * mr = (Robot *) HTRequest_context(text->request);
398: HTParentAnchor * dest = (HTParentAnchor *)
399: HTAnchor_followMainLink((HTAnchor *) anchor);
400: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 401:
1.2 frystyk 402: /* Test whether we already have a hyperdoc for this document */
403: if (mr->flags & MR_IMG && dest && !hd) {
404: HTParentAnchor * parent = HTRequest_parent(text->request);
405: HyperDoc * last = HTAnchor_document(parent);
406: int depth = last ? last->depth+1 : 0;
407: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
408: HyperDoc_new(mr, dest, depth);
409: if (SHOW_MSG) {
410: char * uri = HTAnchor_address((HTAnchor *) dest);
411: TTYPrint(TDEST, "Robot....... Checking Image `%s\'\n", uri);
1.11 frystyk 412: HT_FREE(uri);
1.2 frystyk 413: }
414: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
415: if (SHOW_MSG)
416: TTYPrint(TDEST, "Robot....... Image not tested!\n");
417: Thread_delete(mr, newreq);
1.1 frystyk 418: }
419: }
420: }
421: }
422:
423: PUBLIC void HText_endAnchor (HText * text) {}
424: PUBLIC void HText_appendText (HText * text, CONST char * str) {}
425: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
426: PUBLIC void HText_endAppend (HText * text) {}
427: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
428: PUBLIC void HText_beginAppend (HText * text) {}
429: PUBLIC void HText_appendParagraph (HText * text) {}
430:
431: /* ------------------------------------------------------------------------- */
432: /* MAIN PROGRAM */
433: /* ------------------------------------------------------------------------- */
434:
435: int main (int argc, char ** argv)
436: {
437: int status = 0;
438: int arg;
439: HTChunk * keywords = NULL; /* From command line */
440: int keycnt = 0;
1.12 ! frystyk 441: Robot * mr = NULL;
1.1 frystyk 442:
443: /* Starts Mac GUSI socket library */
444: #ifdef GUSI
445: GUSISetup(GUSIwithSIOUXSockets);
446: GUSISetup(GUSIwithInternetSockets);
447: #endif
448:
449: #ifdef __MWERKS__ /* STR */
450: InitGraf((Ptr) &qd.thePort);
451: InitFonts();
452: InitWindows();
453: InitMenus(); TEInit();
454: InitDialogs(nil);
455: InitCursor();
456: SIOUXSettings.asktosaveonclose = false;
457: argc=ccommand(&argv);
458: #endif
459:
460: /* Initiate W3C Reference Library */
461: HTLibInit(APP_NAME, APP_VERSION);
462:
1.12 ! frystyk 463: /* Build a new robot object */
! 464: mr = Robot_new();
! 465:
1.1 frystyk 466: /* Initialize the protocol modules */
467: HTAccessInit();
468:
469: /* Initialize set of converters */
470: mr->converters = HTList_new();
471: HTConverterInit(mr->converters);
472: HTFormat_setConversion(mr->converters);
473:
474: /* Initialize bindings between file suffixes and media types */
475: HTFileInit();
476:
477: /* Get any proxy or gateway environment variables */
478: HTProxy_getEnvVar();
479:
480: /* Scan command Line for parameters */
481: for (arg=1; arg<argc; arg++) {
482: if (*argv[arg] == '-') {
483:
484: /* -? or -help: show the command line help page */
485: if (!strcmp(argv[arg],"-?") || !strcmp(argv[arg],"-help")) {
486: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(HELP);
487: keycnt = 1;
488:
489: /* non-interactive */
490: } else if (!strcmp(argv[arg], "-n")) {
491: HTAlert_setInteractive(NO);
492:
493: /* log file */
494: } else if (!strcmp(argv[arg], "-l")) {
495: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
496: argv[++arg] : DEFAULT_LOG_FILE;
497:
498: /* rule file */
499: } else if (!strcmp(argv[arg], "-r")) {
500: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
501: argv[++arg] : DEFAULT_RULE_FILE;
502:
503: /* output filename */
504: } else if (!strcmp(argv[arg], "-o")) {
505: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
506: argv[++arg] : DEFAULT_OUTPUT_FILE;
507:
508: /* timeout -- Change the default request timeout */
509: } else if (!strcmp(argv[arg], "-timeout")) {
510: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
511: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
512: if (timeout > 0) mr->tv->tv_sec = timeout;
513:
1.7 frystyk 514: /* preemptive or non-preemptive access */
1.1 frystyk 515: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 516: HTRequest_setPreemptive(mr->request, YES);
517: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 518:
519: /* test inlined images */
520: } else if (!strcmp(argv[arg], "-img")) {
521: mr->flags |= MR_IMG;
522:
523: /* load anchors */
524: } else if (!strcmp(argv[arg], "-link")) {
525: mr->flags |= MR_LINK;
1.7 frystyk 526: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
527: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 528:
1.7 frystyk 529: /* preemptive or non-preemptive access */
1.2 frystyk 530: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 531: HTRequest_setPreemptive(mr->request, YES);
532: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 533:
1.12 ! frystyk 534: /* Output start and end time */
! 535: } else if (!strcmp(argv[arg], "-ss")) {
! 536: time_t local = time(NULL);
! 537: TTYPrint(TDEST, "Robot started on %s\n",
! 538: HTDateTimeStr(&local, YES));
! 539: mr->flags |= MR_TIME;
! 540:
1.1 frystyk 541: /* print version and exit */
542: } else if (!strcmp(argv[arg], "-version")) {
543: VersionInfo();
544: Cleanup(mr, 0);
545:
546: #ifdef WWWTRACE
547: /* trace flags */
548: } else if (!strncmp(argv[arg], "-v", 2)) {
549: char *p = argv[arg]+2;
550: WWWTRACE = 0;
551: for(; *p; p++) {
552: switch (*p) {
553: case 'a': WWWTRACE |= SHOW_ANCHOR_TRACE; break;
554: case 'b': WWWTRACE |= SHOW_BIND_TRACE; break;
555: case 'c': WWWTRACE |= SHOW_CACHE_TRACE; break;
556: case 'g': WWWTRACE |= SHOW_SGML_TRACE; break;
557: case 'p': WWWTRACE |= SHOW_PROTOCOL_TRACE; break;
558: case 's': WWWTRACE |= SHOW_STREAM_TRACE; break;
559: case 't': WWWTRACE |= SHOW_THREAD_TRACE; break;
560: case 'u': WWWTRACE |= SHOW_URI_TRACE; break;
561: default:
562: if (SHOW_MSG)
563: TTYPrint(TDEST,"Bad parameter (%s) in -v option\n",
564: argv[arg]);
565: }
566: }
567: if (!WWWTRACE) WWWTRACE = SHOW_ALL_TRACE;
568: #endif
569:
570: } else {
571: if (SHOW_MSG) TTYPrint(TDEST,"Bad Argument (%s)\n", argv[arg]);
572: }
573: } else { /* If no leading `-' then check for URL or keywords */
574: if (!keycnt) {
575: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
576: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 577: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 578: keycnt = 1;
1.11 frystyk 579: HT_FREE(ref);
1.1 frystyk 580: } else { /* Check for successive keyword arguments */
581: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
582: if (keycnt++ <= 1)
1.5 frystyk 583: keywords = HTChunk_new(128);
1.1 frystyk 584: else
1.5 frystyk 585: HTChunk_putc(keywords, ' ');
586: HTChunk_puts(keywords, HTStrip(escaped));
1.11 frystyk 587: HT_FREE(escaped);
1.1 frystyk 588: }
589: }
590: }
591:
592: #ifdef CATCH_SIG
593: SetSignal();
594: #endif
595:
596: if (!keycnt) {
1.2 frystyk 597: if (SHOW_MSG) TTYPrint(TDEST, "Please specify URL to check.\n");
1.1 frystyk 598: Cleanup(mr, -1);
599: }
600:
601: /* Rule file specified? */
602: if (mr->rules) {
603: HTList * list = HTList_new();
604: HTRequest * rr = HTRequest_new();
605: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
606: HTParentAnchor * ra = (HTParentAnchor *) HTAnchor_findAddress(rules);
1.7 frystyk 607: HTRequest_setPreemptive(rr, YES);
1.1 frystyk 608: HTConversion_add(list, "application/x-www-rules", "*/*", HTRules,
609: 1.0, 0.0, 0.0);
610: HTRequest_setConversion(rr, list, YES);
1.8 frystyk 611: HTAlert_add(HTConfirm, HT_A_CONFIRM);
1.1 frystyk 612: if (HTLoadAnchor((HTAnchor *) ra, rr) != YES)
613: if (SHOW_MSG) TTYPrint(TDEST, "Can't access rules\n");
614: HTConversion_deleteAll(list);
615: HTRequest_delete(rr);
1.8 frystyk 616: HTAlert_delete(HTConfirm);
1.11 frystyk 617: HT_FREE(rules);
1.1 frystyk 618: }
619:
620: /* Output file specified? */
621: if (mr->outputfile) {
622: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
623: if (SHOW_MSG) TTYPrint(TDEST, "Can't open `%s'\n", mr->outputfile);
624: mr->output = OUTPUT;
625: }
626: }
627:
628: /* Log file specifed? */
629: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
630:
631: /* Register our User Prompts etc in the Alert Manager */
632: if (HTAlert_interactive()) {
633: HTAlert_add(HTError_print, HT_A_MESSAGE);
634: HTAlert_add(HTConfirm, HT_A_CONFIRM);
635: HTAlert_add(HTPrompt, HT_A_PROMPT);
636: HTAlert_add(HTPromptPassword, HT_A_SECRET);
637: HTAlert_add(HTPromptUsernameAndPassword, HT_A_USER_PW);
638: }
639:
640: /* Register a call back function for the Net Manager */
1.8 frystyk 641: HTNetCall_addBefore(proxy_handler, 0);
1.1 frystyk 642: HTNetCall_addAfter(terminate_handler, HT_ALL);
643:
644: /* Set timeout on sockets */
1.7 frystyk 645: HTEvent_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 646:
647: /* Start the request */
648: if (keywords) /* Search */
1.5 frystyk 649: status = HTSearch(HTChunk_data(keywords), mr->anchor, mr->request);
1.1 frystyk 650: else
651: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
652:
1.5 frystyk 653: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 654: if (status != YES) {
655: if (SHOW_MSG) TTYPrint(TDEST, "Can't access resource\n");
656: Cleanup(mr, -1);
657: }
658:
659: /* Go into the event loop... */
660: HTEvent_Loop(mr->request);
661:
662: /* Only gets here if event loop fails */
663: Cleanup(mr, 0);
664: return 0;
665: }
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