Annotation of libwww/Robot/src/HTRobot.c, revision 1.15
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:
1.14 frystyk 26: #ifndef W3C_VERSION
27: #define W3C_VERSION "unspecified"
1.1 frystyk 28: #endif
29:
30: #define APP_NAME "W3CRobot"
1.14 frystyk 31: #define APP_VERSION W3C_VERSION
1.1 frystyk 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:
1.13 eric 104: /* Standard (non-error) Output
105: ** ---------------------------
106: */
107: PUBLIC int OutputData(const char * fmt, ...)
108: {
109: int ret;
110: va_list pArgs;
111: va_start(pArgs, fmt);
112: ret = vfprintf(stdout, fmt, pArgs);
113: va_end(pArgs);
114: return ret;
115: }
116:
117: /* ------------------------------------------------------------------------- */
118:
1.2 frystyk 119: /* Create a "HyperDoc" object
120: ** --------------------------
121: ** A HyperDoc object contains information about whether we have already
122: ** started checking the anchor and the depth in our search
123: */
124: PRIVATE HyperDoc * HyperDoc_new (Robot * mr,HTParentAnchor * anchor, int depth)
125: {
126: HyperDoc * hd;
1.14 frystyk 127: if ((hd = (HyperDoc *) HT_CALLOC(1, sizeof(HyperDoc))) == NULL)
128: HT_OUTOFMEM("HyperDoc_new");
1.2 frystyk 129: hd->state = L_INVALID;
130: hd->depth = depth;
131:
132: /* Bind the HyperDoc object together with the Anchor Object */
133: hd->anchor = anchor;
134: HTAnchor_setDocument(anchor, (void *) hd);
135:
136: /* Add this HyperDoc object to our list */
137: if (!mr->hyperdoc) mr->hyperdoc = HTList_new();
138: HTList_addObject(mr->hyperdoc, (void *) hd);
139: return hd;
140: }
141:
142: /* Delete a "HyperDoc" object
143: ** --------------------------
144: */
145: PRIVATE BOOL HyperDoc_delete (HyperDoc * hd)
146: {
147: if (hd) {
1.11 frystyk 148: HT_FREE (hd);
1.2 frystyk 149: return YES;
150: }
151: return NO;
152: }
153:
1.1 frystyk 154: /* Create a Command Line Object
155: ** ----------------------------
156: */
157: PRIVATE Robot * Robot_new (void)
158: {
159: Robot * me;
1.14 frystyk 160: if ((me = (Robot *) HT_CALLOC(1, sizeof(Robot))) == NULL ||
161: (me->tv = (struct timeval*) HT_CALLOC(1, sizeof(struct timeval))) == NULL)
162: HT_OUTOFMEM("Robot_new");
1.2 frystyk 163: me->hyperdoc = HTList_new();
1.4 frystyk 164: me->htext = HTList_new();
1.1 frystyk 165: me->tv->tv_sec = DEFAULT_TIMEOUT;
166: me->cwd = HTFindRelatedName();
167: me->output = OUTPUT;
168:
1.7 frystyk 169: /* We keep an extra timeout request object for the timeout_handler */
170: me->timeout = HTRequest_new();
171: HTRequest_setContext (me->timeout, me);
172:
1.1 frystyk 173: /* Bind the Robot object together with the Request Object */
174: me->request = HTRequest_new();
175: HTRequest_setContext (me->request, me);
176: return me;
177: }
178:
179: /* Delete a Command Line Object
180: ** ----------------------------
181: */
182: PRIVATE BOOL Robot_delete (Robot * me)
183: {
184: if (me) {
1.2 frystyk 185: if (me->hyperdoc) {
186: HTList * cur = me->hyperdoc;
187: HyperDoc * pres;
188: while ((pres = (HyperDoc *) HTList_nextObject(cur)))
189: HyperDoc_delete(pres);
190: HTList_delete(me->hyperdoc);
191: }
1.4 frystyk 192: if (me->htext) {
193: HTList * cur = me->htext;
194: HText * pres;
195: while ((pres = (HText *) HTList_nextObject(cur)))
196: HText_free(pres);
197: HTList_delete(me->htext);
198: }
1.1 frystyk 199: if (me->logfile) HTLog_close();
200: if (me->output && me->output != STDOUT) fclose(me->output);
1.12 frystyk 201: if (me->flags & MR_TIME) {
202: time_t local = time(NULL);
1.13 eric 203: HTTrace("Robot terminated %s\n",HTDateTimeStr(&local,YES));
1.12 frystyk 204: }
1.11 frystyk 205: HT_FREE(me->cwd);
206: HT_FREE(me->tv);
207: HT_FREE(me);
1.1 frystyk 208: return YES;
209: }
210: return NO;
211: }
212:
1.2 frystyk 213: /*
214: ** This function creates a new request object and initializes it
215: */
216: PRIVATE HTRequest * Thread_new (Robot * mr, HTMethod method)
217: {
218: HTRequest * newreq = HTRequest_new();
219: HTRequest_setContext (newreq, mr);
1.7 frystyk 220: if (mr->flags & MR_PREEMPTIVE) HTRequest_setPreemptive(newreq, YES);
1.5 frystyk 221: HTRequest_addRqHd(newreq, HT_C_HOST);
1.2 frystyk 222: HTRequest_setMethod(newreq, method);
223: return newreq;
224: }
225:
226: PRIVATE BOOL Thread_delete (Robot * mr, HTRequest * request)
227: {
228: if (mr && request) {
229: HTRequest_delete(request);
230: return YES;
231: }
232: return NO;
233: }
234:
235: /*
236: ** Cleanup and make sure we close all connections including the persistent
237: ** ones
238: */
1.1 frystyk 239: PRIVATE void Cleanup (Robot * me, int status)
240: {
1.2 frystyk 241: HTNet_killAll();
1.1 frystyk 242: Robot_delete(me);
243: HTLibTerminate();
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);
288: if (mr->logfile) HTLog_add(request, status);
1.2 frystyk 289: Thread_delete(mr, request);
1.3 frystyk 290: if (HTNet_isEmpty()) Cleanup(mr, 0);
1.1 frystyk 291: return HT_OK;
292: }
293:
294: /* timeout_handler
295: ** ---------------
296: ** This function is registered to handle timeout in select eventloop
1.7 frystyk 297: **
298: ** BUG: This doesn't work as we don't get the right request object
299: ** back from the event loop
1.1 frystyk 300: */
301: PRIVATE int timeout_handler (HTRequest * request)
302: {
1.2 frystyk 303: Robot * mr = (Robot *) HTRequest_context(request);
1.13 eric 304: if (SHOW_MSG) HTTrace("Robot....... Request timeout...\n");
1.7 frystyk 305: #if 0
1.1 frystyk 306: HTRequest_kill(request);
1.2 frystyk 307: Thread_delete(mr, request);
1.7 frystyk 308: #endif
309: Cleanup(mr, -1);
1.4 frystyk 310: return HT_OK;
1.1 frystyk 311: }
312:
1.8 frystyk 313: /* proxy_handler
314: ** ---------------
315: ** This function is registered to be called before a request is issued
316: ** We look for redirection for proxies and gateways
317: ** returns HT_LOADED We already have this
318: ** HT_ERROR We can't load this
319: ** HT_OK Success
320: */
1.15 ! frystyk 321: PRIVATE int proxy_handler (HTRequest * request, void * param, int status)
1.8 frystyk 322: {
323: HTParentAnchor *anchor = HTRequest_anchor(request);
324: char * addr = HTAnchor_address((HTAnchor *) anchor);
325: char * newaddr = NULL;
326: if ((newaddr = HTProxy_find(addr))) {
327: StrAllocCat(newaddr, addr);
328: HTRequest_setProxying(request, YES);
329: HTAnchor_setPhysical(anchor, newaddr);
330: } else if ((newaddr = HTGateway_find(addr))) {
331: char * path = HTParse(addr,"",PARSE_HOST+PARSE_PATH+PARSE_PUNCTUATION);
332: /* Chop leading / off to make host into part of path */
333: char * gatewayed = HTParse(path+1, newaddr, PARSE_ALL);
334: HTRequest_setProxying(request, NO);
335: HTAnchor_setPhysical(anchor, gatewayed);
1.11 frystyk 336: HT_FREE(path);
337: HT_FREE(gatewayed);
1.8 frystyk 338: } else
339: HTRequest_setProxying(request, NO);
1.11 frystyk 340: HT_FREE(newaddr);
341: HT_FREE(addr);
1.8 frystyk 342: return HT_OK;
343: }
344:
1.1 frystyk 345: /* ------------------------------------------------------------------------- */
346: /* HTEXT INTERFACE */
347: /* ------------------------------------------------------------------------- */
348:
349: PUBLIC HText * HText_new2 (HTRequest * request, HTParentAnchor * anchor,
350: HTStream * stream)
351: {
352: HText * me;
1.4 frystyk 353: Robot * mr = (Robot *) HTRequest_context(request);
1.14 frystyk 354: if ((me = (HText *) HT_CALLOC(1, sizeof(HText))) == NULL)
355: HT_OUTOFMEM("HText_new2");
1.4 frystyk 356:
357: /* Bind the HText object together with the Request Object */
1.1 frystyk 358: me->request = request;
1.4 frystyk 359:
360: /* Add this HyperDoc object to our list */
361: if (!mr->htext) mr->htext = HTList_new();
362: HTList_addObject(mr->htext, (void *) me);
1.1 frystyk 363: return me;
364: }
365:
1.4 frystyk 366: PUBLIC void HText_free (HText * me) {
1.11 frystyk 367: if (me) HT_FREE (me);
1.4 frystyk 368: }
369:
1.1 frystyk 370: PUBLIC void HText_beginAnchor (HText * text, HTChildAnchor * anchor)
371: {
372: if (text && anchor) {
1.2 frystyk 373: Robot * mr = (Robot *) HTRequest_context(text->request);
1.1 frystyk 374: HTAnchor * dest = HTAnchor_followMainLink((HTAnchor *) anchor);
375: HTParentAnchor * dest_parent = HTAnchor_parent(dest);
1.7 frystyk 376: char * uri = HTAnchor_address((HTAnchor *) dest_parent);
1.1 frystyk 377: HyperDoc * hd = HTAnchor_document(dest_parent);
378:
1.13 eric 379: if (SHOW_MSG) HTTrace("Robot....... Found `%s\' - ", uri ? uri : "NULL");
1.7 frystyk 380:
1.2 frystyk 381: /* Test whether we already have a hyperdoc for this document */
382: if (mr->flags & MR_LINK && dest_parent && !hd) {
1.1 frystyk 383: HTParentAnchor * parent = HTRequest_parent(text->request);
384: HyperDoc * last = HTAnchor_document(parent);
385: int depth = last ? last->depth+1 : 0;
1.2 frystyk 386: HTRequest * newreq = Thread_new(mr, METHOD_GET);
387: HyperDoc_new(mr, dest_parent, depth);
1.7 frystyk 388: HTRequest_setParent(newreq, HTRequest_anchor(text->request));
389: if (depth >= mr->depth) {
390: if (SHOW_MSG)
1.13 eric 391: HTTrace("loading at depth %d using HEAD\n", depth);
1.7 frystyk 392: HTRequest_setMethod(newreq, METHOD_HEAD);
393: HTRequest_setOutputFormat(newreq, WWW_MIME);
394: } else {
1.13 eric 395: if (SHOW_MSG) HTTrace("loading at depth %d\n", depth);
1.2 frystyk 396: }
397: if (HTLoadAnchor((HTAnchor *) dest_parent, newreq) != YES) {
1.13 eric 398: if (SHOW_MSG) HTTrace("not tested!\n");
1.2 frystyk 399: Thread_delete(mr, newreq);
400: }
1.7 frystyk 401: } else {
1.13 eric 402: if (SHOW_MSG) HTTrace("duplicate\n");
1.2 frystyk 403: }
1.11 frystyk 404: HT_FREE(uri);
1.2 frystyk 405: }
406: }
407:
408: PUBLIC void HText_appendImage (HText * text, HTChildAnchor * anchor,
1.14 frystyk 409: const char *alt, const char * align, BOOL isMap)
1.2 frystyk 410: {
411: if (text && anchor) {
412: Robot * mr = (Robot *) HTRequest_context(text->request);
413: HTParentAnchor * dest = (HTParentAnchor *)
414: HTAnchor_followMainLink((HTAnchor *) anchor);
415: HyperDoc * hd = HTAnchor_document(dest);
1.1 frystyk 416:
1.2 frystyk 417: /* Test whether we already have a hyperdoc for this document */
418: if (mr->flags & MR_IMG && dest && !hd) {
419: HTParentAnchor * parent = HTRequest_parent(text->request);
420: HyperDoc * last = HTAnchor_document(parent);
421: int depth = last ? last->depth+1 : 0;
422: HTRequest * newreq = Thread_new(mr, METHOD_HEAD);
423: HyperDoc_new(mr, dest, depth);
424: if (SHOW_MSG) {
425: char * uri = HTAnchor_address((HTAnchor *) dest);
1.13 eric 426: HTTrace("Robot....... Checking Image `%s\'\n", uri);
1.11 frystyk 427: HT_FREE(uri);
1.2 frystyk 428: }
429: if (HTLoadAnchor((HTAnchor *) dest, newreq) != YES) {
430: if (SHOW_MSG)
1.13 eric 431: HTTrace("Robot....... Image not tested!\n");
1.2 frystyk 432: Thread_delete(mr, newreq);
1.1 frystyk 433: }
434: }
435: }
436: }
437:
438: PUBLIC void HText_endAnchor (HText * text) {}
1.14 frystyk 439: PUBLIC void HText_appendText (HText * text, const char * str) {}
1.1 frystyk 440: PUBLIC void HText_appendCharacter (HText * text, char ch) {}
441: PUBLIC void HText_endAppend (HText * text) {}
442: PUBLIC void HText_setStyle (HText * text, HTStyle * style) {}
443: PUBLIC void HText_beginAppend (HText * text) {}
444: PUBLIC void HText_appendParagraph (HText * text) {}
445:
446: /* ------------------------------------------------------------------------- */
447: /* MAIN PROGRAM */
448: /* ------------------------------------------------------------------------- */
449:
450: int main (int argc, char ** argv)
451: {
452: int status = 0;
453: int arg;
454: HTChunk * keywords = NULL; /* From command line */
455: int keycnt = 0;
1.12 frystyk 456: Robot * mr = NULL;
1.1 frystyk 457:
458: /* Starts Mac GUSI socket library */
459: #ifdef GUSI
460: GUSISetup(GUSIwithSIOUXSockets);
461: GUSISetup(GUSIwithInternetSockets);
462: #endif
463:
464: #ifdef __MWERKS__ /* STR */
465: InitGraf((Ptr) &qd.thePort);
466: InitFonts();
467: InitWindows();
468: InitMenus(); TEInit();
469: InitDialogs(nil);
470: InitCursor();
471: SIOUXSettings.asktosaveonclose = false;
472: argc=ccommand(&argv);
473: #endif
474:
475: /* Initiate W3C Reference Library */
476: HTLibInit(APP_NAME, APP_VERSION);
477:
1.12 frystyk 478: /* Build a new robot object */
479: mr = Robot_new();
480:
1.1 frystyk 481: /* Initialize the protocol modules */
482: HTAccessInit();
483:
484: /* Initialize set of converters */
485: mr->converters = HTList_new();
486: HTConverterInit(mr->converters);
487: HTFormat_setConversion(mr->converters);
488:
489: /* Initialize bindings between file suffixes and media types */
490: HTFileInit();
491:
492: /* Get any proxy or gateway environment variables */
493: HTProxy_getEnvVar();
494:
495: /* Scan command Line for parameters */
496: for (arg=1; arg<argc; arg++) {
497: if (*argv[arg] == '-') {
498:
499: /* -? or -help: show the command line help page */
500: if (!strcmp(argv[arg],"-?") || !strcmp(argv[arg],"-help")) {
501: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(HELP);
502: keycnt = 1;
503:
504: /* non-interactive */
505: } else if (!strcmp(argv[arg], "-n")) {
506: HTAlert_setInteractive(NO);
507:
508: /* log file */
509: } else if (!strcmp(argv[arg], "-l")) {
510: mr->logfile = (arg+1 < argc && *argv[arg+1] != '-') ?
511: argv[++arg] : DEFAULT_LOG_FILE;
512:
513: /* rule file */
514: } else if (!strcmp(argv[arg], "-r")) {
515: mr->rules = (arg+1 < argc && *argv[arg+1] != '-') ?
516: argv[++arg] : DEFAULT_RULE_FILE;
517:
518: /* output filename */
519: } else if (!strcmp(argv[arg], "-o")) {
520: mr->outputfile = (arg+1 < argc && *argv[arg+1] != '-') ?
521: argv[++arg] : DEFAULT_OUTPUT_FILE;
522:
523: /* timeout -- Change the default request timeout */
524: } else if (!strcmp(argv[arg], "-timeout")) {
525: int timeout = (arg+1 < argc && *argv[arg+1] != '-') ?
526: atoi(argv[++arg]) : DEFAULT_TIMEOUT;
527: if (timeout > 0) mr->tv->tv_sec = timeout;
528:
1.7 frystyk 529: /* preemptive or non-preemptive access */
1.1 frystyk 530: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 531: HTRequest_setPreemptive(mr->request, YES);
532: mr->flags |= MR_PREEMPTIVE;
1.2 frystyk 533:
534: /* test inlined images */
535: } else if (!strcmp(argv[arg], "-img")) {
536: mr->flags |= MR_IMG;
537:
538: /* load anchors */
539: } else if (!strcmp(argv[arg], "-link")) {
540: mr->flags |= MR_LINK;
1.7 frystyk 541: mr->depth = (arg+1 < argc && *argv[arg+1] != '-') ?
542: atoi(argv[++arg]) : DEFAULT_DEPTH;
1.2 frystyk 543:
1.7 frystyk 544: /* preemptive or non-preemptive access */
1.2 frystyk 545: } else if (!strcmp(argv[arg], "-single")) {
1.7 frystyk 546: HTRequest_setPreemptive(mr->request, YES);
547: mr->flags |= MR_PREEMPTIVE;
1.1 frystyk 548:
1.12 frystyk 549: /* Output start and end time */
550: } else if (!strcmp(argv[arg], "-ss")) {
551: time_t local = time(NULL);
1.13 eric 552: HTTrace("Robot started on %s\n",
1.12 frystyk 553: HTDateTimeStr(&local, YES));
554: mr->flags |= MR_TIME;
555:
1.1 frystyk 556: /* print version and exit */
557: } else if (!strcmp(argv[arg], "-version")) {
558: VersionInfo();
559: Cleanup(mr, 0);
560:
561: #ifdef WWWTRACE
562: /* trace flags */
563: } else if (!strncmp(argv[arg], "-v", 2)) {
564: char *p = argv[arg]+2;
565: WWWTRACE = 0;
566: for(; *p; p++) {
567: switch (*p) {
568: case 'a': WWWTRACE |= SHOW_ANCHOR_TRACE; break;
569: case 'b': WWWTRACE |= SHOW_BIND_TRACE; break;
570: case 'c': WWWTRACE |= SHOW_CACHE_TRACE; break;
571: case 'g': WWWTRACE |= SHOW_SGML_TRACE; break;
572: case 'p': WWWTRACE |= SHOW_PROTOCOL_TRACE; break;
573: case 's': WWWTRACE |= SHOW_STREAM_TRACE; break;
574: case 't': WWWTRACE |= SHOW_THREAD_TRACE; break;
575: case 'u': WWWTRACE |= SHOW_URI_TRACE; break;
576: default:
577: if (SHOW_MSG)
1.13 eric 578: HTTrace("Bad parameter (%s) in -v option\n",
1.1 frystyk 579: argv[arg]);
580: }
581: }
582: if (!WWWTRACE) WWWTRACE = SHOW_ALL_TRACE;
583: #endif
584:
585: } else {
1.13 eric 586: if (SHOW_MSG) HTTrace("Bad Argument (%s)\n", argv[arg]);
1.1 frystyk 587: }
588: } else { /* If no leading `-' then check for URL or keywords */
589: if (!keycnt) {
590: char * ref = HTParse(argv[arg], mr->cwd, PARSE_ALL);
591: mr->anchor = (HTParentAnchor *) HTAnchor_findAddress(ref);
1.7 frystyk 592: HyperDoc_new(mr, mr->anchor, 0);
1.1 frystyk 593: keycnt = 1;
1.11 frystyk 594: HT_FREE(ref);
1.1 frystyk 595: } else { /* Check for successive keyword arguments */
596: char *escaped = HTEscape(argv[arg], URL_XALPHAS);
597: if (keycnt++ <= 1)
1.5 frystyk 598: keywords = HTChunk_new(128);
1.1 frystyk 599: else
1.5 frystyk 600: HTChunk_putc(keywords, ' ');
601: HTChunk_puts(keywords, HTStrip(escaped));
1.11 frystyk 602: HT_FREE(escaped);
1.1 frystyk 603: }
604: }
605: }
606:
607: #ifdef CATCH_SIG
608: SetSignal();
609: #endif
610:
611: if (!keycnt) {
1.13 eric 612: if (SHOW_MSG) HTTrace("Please specify URL to check.\n");
1.1 frystyk 613: Cleanup(mr, -1);
614: }
615:
616: /* Rule file specified? */
617: if (mr->rules) {
618: HTList * list = HTList_new();
619: HTRequest * rr = HTRequest_new();
620: char * rules = HTParse(mr->rules, mr->cwd, PARSE_ALL);
621: HTParentAnchor * ra = (HTParentAnchor *) HTAnchor_findAddress(rules);
1.7 frystyk 622: HTRequest_setPreemptive(rr, YES);
1.1 frystyk 623: HTConversion_add(list, "application/x-www-rules", "*/*", HTRules,
624: 1.0, 0.0, 0.0);
625: HTRequest_setConversion(rr, list, YES);
1.8 frystyk 626: HTAlert_add(HTConfirm, HT_A_CONFIRM);
1.1 frystyk 627: if (HTLoadAnchor((HTAnchor *) ra, rr) != YES)
1.13 eric 628: if (SHOW_MSG) HTTrace("Can't access rules\n");
1.1 frystyk 629: HTConversion_deleteAll(list);
630: HTRequest_delete(rr);
1.8 frystyk 631: HTAlert_delete(HTConfirm);
1.11 frystyk 632: HT_FREE(rules);
1.1 frystyk 633: }
634:
635: /* Output file specified? */
636: if (mr->outputfile) {
637: if ((mr->output = fopen(mr->outputfile, "wb")) == NULL) {
1.13 eric 638: if (SHOW_MSG) HTTrace("Can't open `%s'\n", mr->outputfile);
1.1 frystyk 639: mr->output = OUTPUT;
640: }
641: }
642:
643: /* Log file specifed? */
644: if (mr->logfile) HTLog_open(mr->logfile, YES, YES);
645:
646: /* Register our User Prompts etc in the Alert Manager */
647: if (HTAlert_interactive()) {
648: HTAlert_add(HTError_print, HT_A_MESSAGE);
649: HTAlert_add(HTConfirm, HT_A_CONFIRM);
650: HTAlert_add(HTPrompt, HT_A_PROMPT);
651: HTAlert_add(HTPromptPassword, HT_A_SECRET);
652: HTAlert_add(HTPromptUsernameAndPassword, HT_A_USER_PW);
653: }
654:
655: /* Register a call back function for the Net Manager */
1.15 ! frystyk 656: HTNetCall_addBefore(proxy_handler, NULL, 0);
! 657: HTNetCall_addAfter(terminate_handler, NULL, HT_ALL);
1.1 frystyk 658:
659: /* Set timeout on sockets */
1.7 frystyk 660: HTEvent_registerTimeout(mr->tv, mr->timeout, timeout_handler, NO);
1.1 frystyk 661:
662: /* Start the request */
663: if (keywords) /* Search */
1.5 frystyk 664: status = HTSearch(HTChunk_data(keywords), mr->anchor, mr->request);
1.1 frystyk 665: else
666: status = HTLoadAnchor((HTAnchor *) mr->anchor, mr->request);
667:
1.5 frystyk 668: if (keywords) HTChunk_delete(keywords);
1.1 frystyk 669: if (status != YES) {
1.13 eric 670: if (SHOW_MSG) HTTrace("Can't access resource\n");
1.1 frystyk 671: Cleanup(mr, -1);
672: }
673:
674: /* Go into the event loop... */
675: HTEvent_Loop(mr->request);
676:
677: /* Only gets here if event loop fails */
678: Cleanup(mr, 0);
679: return 0;
680: }
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