Annotation of libwww/Library/src/HTParse.c, revision 2.39
1.1 timbl 1: /* Parse HyperText Document Address HTParse.c
2: ** ================================
2.26 frystyk 3: **
4: ** history:
5: ** May 12 94 TAB added as legal char in HTCleanTelnetString
6: **
1.1 timbl 7: */
2.31 frystyk 8:
2.38 frystyk 9: /* Platform dependent stuff */
1.1 timbl 10: #include "HTUtils.h"
2.38 frystyk 11: #include "tcp.h"
12:
13: /* Library Includes */
14: #include "HTParse.h"
2.31 frystyk 15: #include "HTTCP.h"
2.6 timbl 16:
1.1 timbl 17: struct struct_parts {
2.20 timbl 18: char * access; /* Now known as "scheme" */
1.1 timbl 19: char * host;
20: char * absolute;
21: char * relative;
22: /* char * search; no - treated as part of path */
23: char * anchor;
24: };
25:
26: /* Strip white space off a string
27: ** ------------------------------
28: **
29: ** On exit,
30: ** Return value points to first non-white character, or to 0 if none.
31: ** All trailing white space is OVERWRITTEN with zero.
32: */
33:
2.13 luotonen 34: PUBLIC char * HTStrip ARGS1(char *, s)
1.1 timbl 35: {
36: #define SPACE(c) ((c==' ')||(c=='\t')||(c=='\n'))
37: char * p=s;
2.13 luotonen 38: if (!s) return NULL; /* Doesn't dump core if NULL */
39: for(p=s;*p;p++); /* Find end of string */
1.1 timbl 40: for(p--;p>=s;p--) {
41: if(SPACE(*p)) *p=0; /* Zap trailing blanks */
42: else break;
43: }
44: while(SPACE(*s))s++; /* Strip leading blanks */
45: return s;
46: }
47:
48:
49: /* Scan a filename for its consituents
50: ** -----------------------------------
51: **
52: ** On entry,
53: ** name points to a document name which may be incomplete.
54: ** On exit,
55: ** absolute or relative may be nonzero (but not both).
56: ** host, anchor and access may be nonzero if they were specified.
57: ** Any which are nonzero point to zero terminated strings.
58: */
2.32 frystyk 59: PRIVATE void scan ARGS2(char *, name, struct struct_parts *, parts)
1.1 timbl 60: {
61: char * after_access;
62: char * p;
63: int length = strlen(name);
64:
65: parts->access = 0;
66: parts->host = 0;
67: parts->absolute = 0;
68: parts->relative = 0;
69: parts->anchor = 0;
70:
71: after_access = name;
72: for(p=name; *p; p++) {
73: if (*p==':') {
74: *p = 0;
2.20 timbl 75: parts->access = after_access; /* Scheme has been specified */
2.37 howcome 76:
1.1 timbl 77: after_access = p+1;
2.37 howcome 78:
79: /*
80: after_access = p;
81: while (*after_access == 0) /* HWL 15/10/94: weird bug on HP *//*
82: after_access++; /* after_access = p + 1 *//*
83: */
2.22 luotonen 84: if (0==strcasecomp("URL", parts->access)) {
2.20 timbl 85: parts->access = NULL; /* Ignore IETF's URL: pre-prefix */
86: } else break;
1.1 timbl 87: }
2.20 timbl 88: if (*p=='/') break; /* Access has not been specified */
1.1 timbl 89: if (*p=='#') break;
90: }
91:
92: for(p=name+length-1; p>=name; p--) {
93: if (*p =='#') {
94: parts->anchor=p+1;
95: *p=0; /* terminate the rest */
96: }
97: }
98: p = after_access;
99: if (*p=='/'){
100: if (p[1]=='/') {
101: parts->host = p+2; /* host has been specified */
102: *p=0; /* Terminate access */
103: p=strchr(parts->host,'/'); /* look for end of host name if any */
104: if(p) {
105: *p=0; /* Terminate host */
106: parts->absolute = p+1; /* Root has been found */
107: }
108: } else {
109: parts->absolute = p+1; /* Root found but no host */
110: }
111: } else {
112: parts->relative = (*after_access) ? after_access : 0; /* zero for "" */
113: }
114:
2.16 timbl 115: #ifdef OLD_CODE
1.1 timbl 116: /* Access specified but no host: the anchor was not really one
2.16 timbl 117: e.g. news:j462#36487@foo.bar -- JFG 10/jul/92, from bug report */
118: /* This kludge doesn't work for example when coming across
119: file:/usr/local/www/fred#123
120: which loses its anchor. Correct approach in news is to
121: escape weird characters not allowed in URL. TBL 21/dec/93
122: */
1.1 timbl 123: if (parts->access && ! parts->host && parts->anchor) {
124: *(parts->anchor - 1) = '#'; /* Restore the '#' in the address */
125: parts->anchor = 0;
126: }
2.16 timbl 127: #endif
1.1 timbl 128:
129: #ifdef NOT_DEFINED /* search is just treated as part of path */
130: {
131: char *p = relative ? relative : absolute;
132: if (p) {
133: char * q = strchr(p, '?'); /* Any search string? */
134: if (q) {
135: *q = 0; /* If so, chop that off. */
136: parts->search = q+1;
137: }
138: }
139: }
140: #endif
141: } /*scan */
142:
143:
144: /* Parse a Name relative to another name
145: ** -------------------------------------
146: **
147: ** This returns those parts of a name which are given (and requested)
148: ** substituting bits from the related name where necessary.
149: **
150: ** On entry,
151: ** aName A filename given
2.33 howcome 152: ** relatedName A name relative to which aName is to be parsed. Give
153: ** it an empty string if aName is absolute.
1.1 timbl 154: ** wanted A mask for the bits which are wanted.
155: **
156: ** On exit,
157: ** returns A pointer to a malloc'd string which MUST BE FREED
158: */
2.32 frystyk 159: char * HTParse ARGS3(CONST char *, aName, CONST char *, relatedName,
160: int, wanted)
1.1 timbl 161: {
162: char * result = 0;
163: char * return_value = 0;
164: int len;
165: char * name = 0;
166: char * rel = 0;
167: char * p;
2.12 timbl 168: char * access;
1.1 timbl 169: struct struct_parts given, related;
2.33 howcome 170:
171: if (!relatedName) /* HWL 23/8/94: dont dump due to NULL */
172: relatedName = "";
1.1 timbl 173:
174: /* Make working copies of input strings to cut up:
175: */
176: len = strlen(aName)+strlen(relatedName)+10;
177: result=(char *)malloc(len); /* Lots of space: more than enough */
178: if (result == NULL) outofmem(__FILE__, "HTParse");
179:
180: StrAllocCopy(name, aName);
181: StrAllocCopy(rel, relatedName);
182:
183: scan(name, &given);
184: scan(rel, &related);
185: result[0]=0; /* Clear string */
2.12 timbl 186: access = given.access ? given.access : related.access;
1.1 timbl 187: if (wanted & PARSE_ACCESS)
2.12 timbl 188: if (access) {
189: strcat(result, access);
1.1 timbl 190: if(wanted & PARSE_PUNCTUATION) strcat(result, ":");
191: }
192:
193: if (given.access && related.access) /* If different, inherit nothing. */
194: if (strcmp(given.access, related.access)!=0) {
195: related.host=0;
196: related.absolute=0;
197: related.relative=0;
198: related.anchor=0;
199: }
200:
201: if (wanted & PARSE_HOST)
202: if(given.host || related.host) {
203: if(wanted & PARSE_PUNCTUATION) strcat(result, "//");
204: strcat(result, given.host ? given.host : related.host);
205: }
206:
207: if (given.host && related.host) /* If different hosts, inherit no path. */
208: if (strcmp(given.host, related.host)!=0) {
209: related.absolute=0;
210: related.relative=0;
211: related.anchor=0;
212: }
213:
214: if (wanted & PARSE_PATH) {
215: if(given.absolute) { /* All is given */
216: if(wanted & PARSE_PUNCTUATION) strcat(result, "/");
217: strcat(result, given.absolute);
218: } else if(related.absolute) { /* Adopt path not name */
219: strcat(result, "/");
220: strcat(result, related.absolute);
221: if (given.relative) {
222: p = strchr(result, '?'); /* Search part? */
223: if (!p) p=result+strlen(result)-1;
224: for (; *p!='/'; p--); /* last / */
225: p[1]=0; /* Remove filename */
226: strcat(result, given.relative); /* Add given one */
2.31 frystyk 227: result = HTSimplify (result);
1.1 timbl 228: }
229: } else if(given.relative) {
230: strcat(result, given.relative); /* what we've got */
231: } else if(related.relative) {
232: strcat(result, related.relative);
233: } else { /* No inheritance */
234: strcat(result, "/");
235: }
236: }
237:
238: if (wanted & PARSE_ANCHOR)
239: if(given.anchor || related.anchor) {
240: if(wanted & PARSE_PUNCTUATION) strcat(result, "#");
241: strcat(result, given.anchor ? given.anchor : related.anchor);
242: }
243: free(rel);
244: free(name);
245:
246: StrAllocCopy(return_value, result);
247: free(result);
248: return return_value; /* exactly the right length */
249: }
250:
2.11 timbl 251:
2.21 frystyk 252: #if 0 /* NOT USED FOR THE MOMENT */
2.15 luotonen 253: /*
254: ** As strcpy() but guaranteed to work correctly
255: ** with overlapping parameters. AL 7 Feb 1994
256: */
257: PRIVATE void ari_strcpy ARGS2(char *, to,
258: char *, from)
259: {
260: char * tmp;
261:
262: if (!to || !from) return;
263:
264: tmp = (char*)malloc(strlen(from)+1);
265: if (!tmp) outofmem(__FILE__, "my_strcpy");
266:
267: strcpy(tmp, from);
268: strcpy(to, tmp);
269: free(tmp);
270: }
2.21 frystyk 271: #endif
272:
2.20 timbl 273:
274: /* Simplify a URI
275: // --------------
276: // A URI is allowed to contain the seqeunce xxx/../ which may be
1.1 timbl 277: // replaced by "" , and the seqeunce "/./" which may be replaced by "/".
2.20 timbl 278: // Simplification helps us recognize duplicate URIs.
1.1 timbl 279: //
280: // Thus, /etc/junk/../fred becomes /etc/fred
281: // /etc/junk/./fred becomes /etc/junk/fred
2.11 timbl 282: //
283: // but we should NOT change
284: // http://fred.xxx.edu/../..
285: //
286: // or ../../albert.html
2.26 frystyk 287: //
288: // In the same manner, the following prefixed are preserved:
289: //
290: // ./<etc>
291: // //<etc>
292: //
293: // In order to avoid empty URLs the following URLs become:
294: //
295: // /fred/.. becomes /fred/..
296: // /fred/././.. becomes /fred/..
2.27 frystyk 297: // /fred/.././junk/.././ becomes /fred/..
2.26 frystyk 298: //
2.30 frystyk 299: // If more than one set of `://' is found (several proxies in cascade) then
300: // only the part after the last `://' is simplified.
1.1 timbl 301: */
2.31 frystyk 302: PUBLIC char *HTSimplify ARGS1(char *, filename)
1.1 timbl 303: {
2.31 frystyk 304: char *path;
305: char *p;
2.19 frystyk 306:
2.31 frystyk 307: if (!filename) {
308: if (URI_TRACE)
309: fprintf(stderr, "HTSimplify.. Bad argument\n");
310: return filename;
311: }
312: if (URI_TRACE)
2.27 frystyk 313: fprintf(stderr, "HTSimplify.. `%s\' ", filename);
314:
2.31 frystyk 315: if ((path = strstr(filename, "://")) != NULL) { /* Find host name */
2.30 frystyk 316: char *newptr;
2.31 frystyk 317: path += 3;
318: while ((newptr = strstr(path, "://")) != NULL)
319: path = newptr+3;
320: path = HTCanon(&filename, path); /* We have a host name */
321: } else if ((path = strstr(filename, ":/")) != NULL) {
322: path += 2;
2.27 frystyk 323: } else
2.31 frystyk 324: path = filename;
325: if (*path == '/' && *(path+1)=='/') { /* Some URLs start //<foo> */
326: path += 1;
2.34 frystyk 327: } else if (!strncmp(path, "news:", 5)) {
328: char *ptr = strchr(path+5, '@');
329: if (!ptr) ptr = path+5;
330: while (*ptr) { /* Make group or host lower case */
331: *ptr = TOLOWER(*ptr);
332: ptr++;
2.31 frystyk 333: }
334: if (URI_TRACE)
335: fprintf(stderr, "into\n............ `%s'\n", filename);
336: return filename; /* Doesn't need to do any more */
337: }
338: if ((p = path)) {
339: int segments = 0;
340:
341: /* Parse string first time to find number of `real' tokens */
342: while (*p) {
343: if (*p=='/' || p==path) {
344: if (!((*(p+1)=='/' || !*(p+1)) ||
345: (*(p+1)=='.' && (*(p+2)=='/' || !*(p+2))) ||
346: (*(p+1)=='.' && *(p+2)=='.' &&(*(p+3)=='/' || !*(p+3)))))
347: segments++;
348: }
349: p++;
350: }
2.19 frystyk 351:
2.31 frystyk 352: /* Parse string second time to simplify */
353: p = path;
354: while(*p) {
355: if (*p=='/') {
356: if (p>path && *(p+1)=='.' && (*(p+2)=='/' || !*(p+2))) {
357: char *orig=p, *dest=p+2;
358: while ((*orig++ = *dest++)); /* Remove a slash and a dot */
359: p--;
360: } else if (segments>1 && *(p+1)=='.' && *(p+2)=='.' &&
361: (*(p+3)=='/' || !*(p+3))) {
362: char *q = p;
363: while (q>path && *--q!='/'); /* prev slash */
364: if (strncmp(q, "/../", 4) && strncmp(q, "/./", 3) &&
365: strncmp(q, "./", 2)) {
366: char *orig=q, *dest=p+3;
367: if (*q!='/') dest++;
368: while ((*orig++ = *dest++)); /* Remove /xxx/.. */
369: segments--;
370: p = q-1; /* Start again with prev slash */
371: } else
372: p++;
373: } else if (*(p+1)=='/') {
374: while (*(p+1)=='/') {
375: char *orig=p, *dest=p+1;
376: while ((*orig++ = *dest++)); /* Remove multiple /'s */
2.19 frystyk 377: }
378: }
379: }
2.31 frystyk 380: p++;
381: } /* end while (*p) */
2.19 frystyk 382: }
2.31 frystyk 383: if (URI_TRACE)
2.27 frystyk 384: fprintf(stderr, "into\n............ `%s'\n", filename);
2.31 frystyk 385: return filename;
2.19 frystyk 386: }
2.31 frystyk 387:
2.19 frystyk 388: #ifdef OLD_CODE
2.17 frystyk 389: char * p = filename;
1.1 timbl 390: char * q;
2.17 frystyk 391:
392: if (p) {
393: while (*p && (*p == '/' || *p == '.')) /* Pass starting / or .'s */
394: p++;
395: while(*p) {
396: if (*p=='/') {
1.1 timbl 397: if ((p[1]=='.') && (p[2]=='.') && (p[3]=='/' || !p[3] )) {
2.11 timbl 398: for (q=p-1; (q>=filename) && (*q!='/'); q--); /* prev slash */
399: if (q[0]=='/' && 0!=strncmp(q, "/../", 4)
400: &&!(q-1>filename && q[-1]=='/')) {
2.15 luotonen 401: ari_strcpy(q, p+3); /* Remove /xxx/.. */
1.1 timbl 402: if (!*filename) strcpy(filename, "/");
403: p = q-1; /* Start again with prev slash */
2.11 timbl 404: } else { /* xxx/.. leave it! */
2.9 timbl 405: #ifdef BUG_CODE
2.15 luotonen 406: ari_strcpy(filename, p[3] ? p+4 : p+3); /* rm xxx/../ */
1.1 timbl 407: p = filename; /* Start again */
2.9 timbl 408: #endif
1.1 timbl 409: }
410: } else if ((p[1]=='.') && (p[2]=='/' || !p[2])) {
2.15 luotonen 411: ari_strcpy(p, p+2); /* Remove a slash and a dot */
2.13 luotonen 412: } else if (p[-1] != ':') {
413: while (p[1] == '/') {
2.15 luotonen 414: ari_strcpy(p, p+1); /* Remove multiple slashes */
2.13 luotonen 415: }
1.1 timbl 416: }
2.17 frystyk 417: }
418: p++;
419: } /* end while (*p) */
420: } /* end if (p) */
1.1 timbl 421: }
2.19 frystyk 422: #endif /* OLD_CODE */
1.1 timbl 423:
424:
425: /* Make Relative Name
426: ** ------------------
427: **
428: ** This function creates and returns a string which gives an expression of
429: ** one address as related to another. Where there is no relation, an absolute
430: ** address is retured.
431: **
432: ** On entry,
433: ** Both names must be absolute, fully qualified names of nodes
434: ** (no anchor bits)
435: **
436: ** On exit,
437: ** The return result points to a newly allocated name which, if
438: ** parsed by HTParse relative to relatedName, will yield aName.
439: ** The caller is responsible for freeing the resulting name later.
440: **
441: */
2.32 frystyk 442: char * HTRelative ARGS2(CONST char *, aName, CONST char *, relatedName)
1.1 timbl 443: {
444: char * result = 0;
445: CONST char *p = aName;
446: CONST char *q = relatedName;
447: CONST char * after_access = 0;
448: CONST char * path = 0;
449: CONST char * last_slash = 0;
450: int slashes = 0;
451:
452: for(;*p; p++, q++) { /* Find extent of match */
453: if (*p!=*q) break;
454: if (*p==':') after_access = p+1;
455: if (*p=='/') {
456: last_slash = p;
457: slashes++;
458: if (slashes==3) path=p;
459: }
460: }
461:
462: /* q, p point to the first non-matching character or zero */
463:
464: if (!after_access) { /* Different access */
465: StrAllocCopy(result, aName);
466: } else if (slashes<3){ /* Different nodes */
467: StrAllocCopy(result, after_access);
2.29 frystyk 468: #if 0 /* Henrik */
1.1 timbl 469: } else if (slashes==3){ /* Same node, different path */
470: StrAllocCopy(result, path);
2.21 frystyk 471: #endif
1.1 timbl 472: } else { /* Some path in common */
473: int levels= 0;
474: for(; *q && (*q!='#'); q++) if (*q=='/') levels++;
475: result = (char *)malloc(3*levels + strlen(last_slash) + 1);
476: if (result == NULL) outofmem(__FILE__, "HTRelative");
477: result[0]=0;
478: for(;levels; levels--)strcat(result, "../");
479: strcat(result, last_slash+1);
480: }
2.31 frystyk 481: if (URI_TRACE) fprintf(stderr,
2.21 frystyk 482: "HTRelative.. `%s' expressed relative to `%s' is `%s'\n",
483: aName, relatedName, result);
1.1 timbl 484: return result;
485: }
2.1 timbl 486:
487:
2.31 frystyk 488: /* HTCanon
489: **
490: ** Canonicalizes the URL in the following manner starting from the host
491: ** pointer:
492: **
493: ** 1) The host name is converted to lowercase
494: ** 2) Expands the host name of the URL from a local name to a full
495: ** domain name. A host name is started by `://'.
2.38 frystyk 496: ** 3) Chop off port if `:80' (http), `:70' (gopher), or `:21' (ftp)
2.31 frystyk 497: **
498: ** Return: OK The position of the current path part of the URL
499: */
500: PUBLIC char *HTCanon ARGS2 (char **, filename, char *, host)
501: {
2.32 frystyk 502: char *newname = NULL;
2.31 frystyk 503: char *port;
504: char *strptr;
505: char *path;
2.36 frystyk 506: char *access = host-3;
2.31 frystyk 507:
2.36 frystyk 508: while (access>*filename && *(access-1)!='/') /* Find access method */
509: access--;
2.31 frystyk 510: if ((path = strchr(host, '/')) == NULL) /* Find path */
511: path = host + strlen(host);
512: if ((strptr = strchr(host, '@')) != NULL && strptr<path) /* UserId */
513: host = strptr;
2.39 ! frystyk 514: if ((port = strchr(host, ':')) != NULL && port>path) /* Port number */
! 515: port = NULL;
2.31 frystyk 516:
517: strptr = host; /* Convert to lower-case */
518: while (strptr<path) {
519: *strptr = TOLOWER(*strptr);
520: strptr++;
521: }
522:
523: /* Does the URL contain a full domain name? This also works for a
524: numerical host name. The domain name is already made lower-case
525: and without a trailing dot. */
2.35 frystyk 526: if (((strptr = strchr(host, '.')) == NULL || strptr >= path) &&
527: strncasecomp(host, "localhost", 9)) {
2.31 frystyk 528: CONST char *domain = HTGetDomainName();
529: if (domain) {
2.32 frystyk 530: if ((newname = (char *) calloc(1, strlen(*filename) +
2.31 frystyk 531: strlen(domain)+2)) == NULL)
532: outofmem(__FILE__, "HTCanon");
533: if (port)
2.32 frystyk 534: strncpy(newname, *filename, (int) (port-*filename));
2.31 frystyk 535: else
2.32 frystyk 536: strncpy(newname, *filename, (int) (path-*filename));
537: strcat(newname, ".");
538: strcat(newname, domain);
2.31 frystyk 539: }
540: } else { /* Look for a trailing dot */
541: char *dot = port ? port : path;
542: if (dot > *filename && *--dot=='.') {
543: char *orig=dot, *dest=dot+1;
544: while((*orig++ = *dest++));
545: if (port) port--;
546: path--;
547: }
548: }
2.36 frystyk 549: /* Chop off port if `:80' (http), `:70' (gopher), or `:21' (ftp) */
550: if (port) {
551: if ((!strncmp(access, "http", 4) &&
552: (*(port+1)=='8'&&*(port+2)=='0'&&(*(port+3)=='/'||!*(port+3)))) ||
553: (!strncmp(access, "gopher", 6) &&
554: (*(port+1)=='7'&&*(port+2)=='0'&&(*(port+3)=='/'||!*(port+3)))) ||
555: (!strncmp(access, "ftp", 3) &&
556: (*(port+1)=='2'&&*(port+2)=='1'&&(*(port+3)=='/'||!*(port+3))))) {
557: if (!newname) {
558: char *orig=port, *dest=port+3;
559: while((*orig++ = *dest++));
560: }
561: } else if (newname)
562: strncat(newname, port, (int) (path-port));
563: }
564:
2.32 frystyk 565: if (newname) {
566: char *newpath = newname+strlen(newname);
567: strcat(newname, path);
2.31 frystyk 568: path = newpath;
569: free(*filename); /* Free old copy */
2.32 frystyk 570: *filename = newname;
2.31 frystyk 571: }
572: return path;
573: }
2.1 timbl 574:
575:
2.24 luotonen 576: /* HTCleanTelnetString()
577: * Make sure that the given string doesn't contain characters that
578: * could cause security holes, such as newlines in ftp, gopher,
579: * news or telnet URLs; more specifically: allows everything between
2.26 frystyk 580: * ASCII 20-7E, and also A0-FE, inclusive. Also TAB ('\t') allowed!
2.24 luotonen 581: *
582: * On entry,
583: * str the string that is *modified* if necessary. The
584: * string will be truncated at the first illegal
585: * character that is encountered.
586: * On exit,
587: * returns YES, if the string was modified.
588: * NO, otherwise.
589: */
590: PUBLIC BOOL HTCleanTelnetString ARGS1(char *, str)
591: {
592: char * cur = str;
593:
594: if (!str) return NO;
595:
596: while (*cur) {
597: int a = TOASCII(*cur);
2.26 frystyk 598: if (a != 0x9 && (a < 0x20 || (a > 0x7E && a < 0xA0) || a > 0xFE)) {
2.31 frystyk 599: if (URI_TRACE)
600: fprintf(stderr, "Illegal..... character in URL: \"%s\"\n",str);
2.24 luotonen 601: *cur = 0;
2.31 frystyk 602: if (URI_TRACE)
603: fprintf(stderr, "Truncated... \"%s\"\n",str);
2.24 luotonen 604: return YES;
605: }
606: cur++;
607: }
608: return NO;
609: }
610:
Webmaster