Annotation of libwww/Library/src/HTParse.c, revision 2.64
2.41 frystyk 1: /* HTParse.c
2: ** URI MANAGEMENT
3: **
2.46 frystyk 4: ** (c) COPYRIGHT MIT 1995.
2.41 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
2.64 ! frystyk 6: ** @(#) $Id: Date Author State $
2.26 frystyk 7: **
8: ** history:
9: ** May 12 94 TAB added as legal char in HTCleanTelnetString
10: **
1.1 timbl 11: */
2.31 frystyk 12:
2.44 frystyk 13: /* Library include files */
2.63 frystyk 14: #include "sysdep.h"
1.1 timbl 15: #include "HTUtils.h"
2.38 frystyk 16: #include "HTParse.h"
2.44 frystyk 17: #include "HTString.h"
2.64 ! frystyk 18: #include "HTInet.h"
2.6 timbl 19:
1.1 timbl 20: struct struct_parts {
2.20 timbl 21: char * access; /* Now known as "scheme" */
1.1 timbl 22: char * host;
23: char * absolute;
24: char * relative;
25: /* char * search; no - treated as part of path */
26: char * anchor;
27: };
28:
29: /* Scan a filename for its consituents
30: ** -----------------------------------
31: **
32: ** On entry,
33: ** name points to a document name which may be incomplete.
34: ** On exit,
35: ** absolute or relative may be nonzero (but not both).
36: ** host, anchor and access may be nonzero if they were specified.
37: ** Any which are nonzero point to zero terminated strings.
38: */
2.60 frystyk 39: PRIVATE void scan (char * name, struct struct_parts * parts)
1.1 timbl 40: {
41: char * after_access;
42: char * p;
43: int length = strlen(name);
44:
45: parts->access = 0;
46: parts->host = 0;
47: parts->absolute = 0;
48: parts->relative = 0;
49: parts->anchor = 0;
50:
51: after_access = name;
52: for(p=name; *p; p++) {
53: if (*p==':') {
54: *p = 0;
2.20 timbl 55: parts->access = after_access; /* Scheme has been specified */
2.37 howcome 56:
2.44 frystyk 57: /* The combination of gcc, the "-O" flag and the HP platform is
58: unhealthy. The following three lines is a quick & dirty fix, but is
59: not recommended. Rather, turn off "-O". */
60:
2.42 howcome 61: /* after_access = p;*/
2.44 frystyk 62: /* while (*after_access == 0)*/
63: /* after_access++;*/
2.42 howcome 64:
1.1 timbl 65: after_access = p+1;
2.37 howcome 66:
2.22 luotonen 67: if (0==strcasecomp("URL", parts->access)) {
2.20 timbl 68: parts->access = NULL; /* Ignore IETF's URL: pre-prefix */
69: } else break;
1.1 timbl 70: }
2.20 timbl 71: if (*p=='/') break; /* Access has not been specified */
1.1 timbl 72: if (*p=='#') break;
73: }
74:
75: for(p=name+length-1; p>=name; p--) {
76: if (*p =='#') {
77: parts->anchor=p+1;
78: *p=0; /* terminate the rest */
79: }
80: }
81: p = after_access;
82: if (*p=='/'){
83: if (p[1]=='/') {
84: parts->host = p+2; /* host has been specified */
85: *p=0; /* Terminate access */
86: p=strchr(parts->host,'/'); /* look for end of host name if any */
87: if(p) {
88: *p=0; /* Terminate host */
89: parts->absolute = p+1; /* Root has been found */
90: }
91: } else {
92: parts->absolute = p+1; /* Root found but no host */
93: }
94: } else {
2.49 frystyk 95: parts->relative = (*after_access) ? after_access : 0; /* zero for "" */
1.1 timbl 96: }
2.50 frystyk 97: }
1.1 timbl 98:
99:
100: /* Parse a Name relative to another name
101: ** -------------------------------------
102: **
103: ** This returns those parts of a name which are given (and requested)
104: ** substituting bits from the related name where necessary.
105: **
106: ** On entry,
107: ** aName A filename given
2.33 howcome 108: ** relatedName A name relative to which aName is to be parsed. Give
109: ** it an empty string if aName is absolute.
1.1 timbl 110: ** wanted A mask for the bits which are wanted.
111: **
112: ** On exit,
113: ** returns A pointer to a malloc'd string which MUST BE FREED
114: */
2.63 frystyk 115: PUBLIC char * HTParse (const char *aName, const char *relatedName, int wanted)
1.1 timbl 116: {
117: char * result = 0;
118: char * return_value = 0;
119: int len;
120: char * name = 0;
121: char * rel = 0;
122: char * p;
2.12 timbl 123: char * access;
1.1 timbl 124: struct struct_parts given, related;
2.33 howcome 125:
126: if (!relatedName) /* HWL 23/8/94: dont dump due to NULL */
127: relatedName = "";
1.1 timbl 128:
2.50 frystyk 129: /* Make working copies of input strings to cut up: */
1.1 timbl 130: len = strlen(aName)+strlen(relatedName)+10;
2.61 frystyk 131: if ((result=(char *) HT_MALLOC(len)) == NULL) /* Lots of space: more than enough */
132: HT_OUTOFMEM("parse space");
1.1 timbl 133: StrAllocCopy(name, aName);
134: StrAllocCopy(rel, relatedName);
135:
136: scan(name, &given);
137: scan(rel, &related);
138: result[0]=0; /* Clear string */
2.12 timbl 139: access = given.access ? given.access : related.access;
1.1 timbl 140: if (wanted & PARSE_ACCESS)
2.12 timbl 141: if (access) {
142: strcat(result, access);
1.1 timbl 143: if(wanted & PARSE_PUNCTUATION) strcat(result, ":");
144: }
145:
146: if (given.access && related.access) /* If different, inherit nothing. */
147: if (strcmp(given.access, related.access)!=0) {
148: related.host=0;
149: related.absolute=0;
150: related.relative=0;
151: related.anchor=0;
152: }
153:
154: if (wanted & PARSE_HOST)
155: if(given.host || related.host) {
156: if(wanted & PARSE_PUNCTUATION) strcat(result, "//");
157: strcat(result, given.host ? given.host : related.host);
158: }
159:
160: if (given.host && related.host) /* If different hosts, inherit no path. */
161: if (strcmp(given.host, related.host)!=0) {
162: related.absolute=0;
163: related.relative=0;
164: related.anchor=0;
165: }
166:
167: if (wanted & PARSE_PATH) {
168: if(given.absolute) { /* All is given */
169: if(wanted & PARSE_PUNCTUATION) strcat(result, "/");
170: strcat(result, given.absolute);
171: } else if(related.absolute) { /* Adopt path not name */
172: strcat(result, "/");
173: strcat(result, related.absolute);
174: if (given.relative) {
175: p = strchr(result, '?'); /* Search part? */
176: if (!p) p=result+strlen(result)-1;
177: for (; *p!='/'; p--); /* last / */
178: p[1]=0; /* Remove filename */
179: strcat(result, given.relative); /* Add given one */
2.55 frystyk 180: #if 0
2.45 frystyk 181: result = HTSimplify (&result);
2.55 frystyk 182: #endif
1.1 timbl 183: }
184: } else if(given.relative) {
185: strcat(result, given.relative); /* what we've got */
186: } else if(related.relative) {
187: strcat(result, related.relative);
188: } else { /* No inheritance */
189: strcat(result, "/");
190: }
191: }
192:
193: if (wanted & PARSE_ANCHOR)
2.58 neon 194: if(given.anchor || related.anchor) {
195: if(given.absolute && given.anchor) { /*Fixes for relURLs...*/
196: if(wanted & PARSE_PUNCTUATION) strcat(result, "#");
197: strcat(result, given.anchor);
198: } else if (!(given.absolute) && !(given.anchor)) {
199: strcat(result, "");
200: } else {
201: if(wanted & PARSE_PUNCTUATION) strcat(result, "#");
202: strcat(result, given.anchor ? given.anchor : related.anchor);
203: }
1.1 timbl 204: }
2.61 frystyk 205: HT_FREE(rel);
206: HT_FREE(name);
1.1 timbl 207:
208: StrAllocCopy(return_value, result);
2.61 frystyk 209: HT_FREE(result);
1.1 timbl 210: return return_value; /* exactly the right length */
211: }
2.21 frystyk 212:
2.20 timbl 213:
2.53 frystyk 214: /* HTCanon
215: **
216: ** Canonicalizes the URL in the following manner starting from the host
217: ** pointer:
218: **
219: ** 1) The host name is converted to lowercase
220: ** 2) Expands the host name of the URL from a local name to a full
221: ** domain name. A host name is started by `://'.
222: ** 3) Chop off port if `:80' (http), `:70' (gopher), or `:21' (ftp)
223: **
224: ** Return: OK The position of the current path part of the URL
225: ** which might be the old one or a new one.
226: */
2.60 frystyk 227: PRIVATE char *HTCanon (char ** filename, char * host)
2.53 frystyk 228: {
229: char *newname = NULL;
230: char *port;
231: char *strptr;
232: char *path;
233: char *access = host-3;
234:
235: while (access>*filename && *(access-1)!='/') /* Find access method */
236: access--;
237: if ((path = strchr(host, '/')) == NULL) /* Find path */
238: path = host + strlen(host);
239: if ((strptr = strchr(host, '@')) != NULL && strptr<path) /* UserId */
240: host = strptr;
241: if ((port = strchr(host, ':')) != NULL && port>path) /* Port number */
242: port = NULL;
243:
244: strptr = host; /* Convert to lower-case */
245: while (strptr<path) {
246: *strptr = TOLOWER(*strptr);
247: strptr++;
248: }
249:
250: /* Does the URL contain a full domain name? This also works for a
251: numerical host name. The domain name is already made lower-case
252: and without a trailing dot. */
253: if (((strptr = strchr(host, '.')) == NULL || strptr >= path) &&
254: strncasecomp(host, "localhost", 9)) {
2.63 frystyk 255: const char *domain = HTGetDomainName();
2.53 frystyk 256: if (domain && *domain) {
2.61 frystyk 257: if ((newname = (char *) HT_CALLOC(1, strlen(*filename) + strlen(domain)+2)) == NULL)
258: HT_OUTOFMEM("HTCanon");
2.53 frystyk 259: if (port)
260: strncpy(newname, *filename, (int) (port-*filename));
261: else
262: strncpy(newname, *filename, (int) (path-*filename));
263: strcat(newname, ".");
264: strcat(newname, domain);
265: }
266: } else { /* Look for a trailing dot */
267: char *dot = port ? port : path;
268: if (dot > *filename && *--dot=='.') {
269: char *orig=dot, *dest=dot+1;
270: while((*orig++ = *dest++));
271: if (port) port--;
272: path--;
273: }
274: }
275: /* Chop off port if `:', `:80' (http), `:70' (gopher), or `:21' (ftp) */
276: if (port) {
277: if (!*(port+1) || *(port+1)=='/') {
278: if (!newname) {
279: char *orig=port, *dest=port+1;
280: while((*orig++ = *dest++));
281: }
282: } else if ((!strncmp(access, "http", 4) &&
283: (*(port+1)=='8'&&*(port+2)=='0'&&(*(port+3)=='/'||!*(port+3)))) ||
284: (!strncmp(access, "gopher", 6) &&
285: (*(port+1)=='7'&&*(port+2)=='0'&&(*(port+3)=='/'||!*(port+3)))) ||
286: (!strncmp(access, "ftp", 3) &&
287: (*(port+1)=='2'&&*(port+2)=='1'&&(*(port+3)=='/'||!*(port+3))))) {
288: if (!newname) {
289: char *orig=port, *dest=port+3;
290: while((*orig++ = *dest++));
2.54 frystyk 291: path -= 3; /* Update path position, Henry Minsky */
2.53 frystyk 292: }
293: } else if (newname)
294: strncat(newname, port, (int) (path-port));
295: }
296:
297: if (newname) {
298: char *newpath = newname+strlen(newname);
299: strcat(newname, path);
300: path = newpath;
2.61 frystyk 301: HT_FREE(*filename); /* Free old copy */
2.53 frystyk 302: *filename = newname;
303: }
304: return path;
305: }
306:
2.20 timbl 307: /* Simplify a URI
308: // --------------
309: // A URI is allowed to contain the seqeunce xxx/../ which may be
1.1 timbl 310: // replaced by "" , and the seqeunce "/./" which may be replaced by "/".
2.20 timbl 311: // Simplification helps us recognize duplicate URIs.
1.1 timbl 312: //
313: // Thus, /etc/junk/../fred becomes /etc/fred
314: // /etc/junk/./fred becomes /etc/junk/fred
2.11 timbl 315: //
316: // but we should NOT change
317: // http://fred.xxx.edu/../..
318: //
319: // or ../../albert.html
2.26 frystyk 320: //
321: // In order to avoid empty URLs the following URLs become:
322: //
323: // /fred/.. becomes /fred/..
324: // /fred/././.. becomes /fred/..
2.27 frystyk 325: // /fred/.././junk/.././ becomes /fred/..
2.26 frystyk 326: //
2.30 frystyk 327: // If more than one set of `://' is found (several proxies in cascade) then
328: // only the part after the last `://' is simplified.
2.44 frystyk 329: //
330: // Returns: A string which might be the old one or a new one.
1.1 timbl 331: */
2.60 frystyk 332: PUBLIC char *HTSimplify (char ** url)
1.1 timbl 333: {
2.31 frystyk 334: char *path;
335: char *p;
2.55 frystyk 336: if (!url || !*url) {
2.62 eric 337: if (URI_TRACE) HTTrace("HTSimplify.. Nothing done\n");
2.55 frystyk 338: return *url;
2.31 frystyk 339: }
2.62 eric 340: if (URI_TRACE) HTTrace("HTSimplify.. `%s\' ", *url);
2.27 frystyk 341:
2.55 frystyk 342: /* Find any scheme name */
343: if ((path = strstr(*url, "://")) != NULL) { /* Find host name */
2.30 frystyk 344: char *newptr;
2.60 frystyk 345: char *access = *url;
346: while (access<path && (*access=TOLOWER(*access))) access++;
2.31 frystyk 347: path += 3;
2.60 frystyk 348: while ((newptr = strstr(path, "://")) != NULL) /* For proxies */
2.31 frystyk 349: path = newptr+3;
2.55 frystyk 350: path = HTCanon(url, path); /* We have a host name */
351: } else if ((path = strstr(*url, ":/")) != NULL) {
2.31 frystyk 352: path += 2;
2.27 frystyk 353: } else
2.55 frystyk 354: path = *url;
2.31 frystyk 355: if (*path == '/' && *(path+1)=='/') { /* Some URLs start //<foo> */
356: path += 1;
2.34 frystyk 357: } else if (!strncmp(path, "news:", 5)) {
358: char *ptr = strchr(path+5, '@');
359: if (!ptr) ptr = path+5;
360: while (*ptr) { /* Make group or host lower case */
361: *ptr = TOLOWER(*ptr);
362: ptr++;
2.31 frystyk 363: }
364: if (URI_TRACE)
2.62 eric 365: HTTrace("into\n............ `%s'\n", *url);
2.55 frystyk 366: return *url; /* Doesn't need to do any more */
2.31 frystyk 367: }
368: if ((p = path)) {
2.55 frystyk 369: char *end;
370: if (!((end = strchr(path, ';')) || (end = strchr(path, '?')) ||
371: (end = strchr(path, '#'))))
372: end = path+strlen(path);
2.31 frystyk 373:
374: /* Parse string second time to simplify */
375: p = path;
2.55 frystyk 376: while(p<end) {
2.31 frystyk 377: if (*p=='/') {
2.56 frystyk 378: if (p>*url && *(p+1)=='.' && (*(p+2)=='/' || !*(p+2))) {
379: char *orig = p+1;
380: char *dest = (*(p+2)!='/') ? p+2 : p+3;
2.31 frystyk 381: while ((*orig++ = *dest++)); /* Remove a slash and a dot */
2.55 frystyk 382: end = orig-1;
2.56 frystyk 383: } else if (*(p+1)=='.' && *(p+2)=='.' && (*(p+3)=='/' || !*(p+3))) {
2.31 frystyk 384: char *q = p;
385: while (q>path && *--q!='/'); /* prev slash */
2.56 frystyk 386: if (strncmp(q, "/../", 4)) {
387: char *orig = q+1;
388: char *dest = (*(p+3)!='/') ? p+3 : p+4;
2.31 frystyk 389: while ((*orig++ = *dest++)); /* Remove /xxx/.. */
2.55 frystyk 390: end = orig-1;
2.56 frystyk 391: p = q; /* Start again with prev slash */
2.31 frystyk 392: } else
393: p++;
394: } else if (*(p+1)=='/') {
395: while (*(p+1)=='/') {
396: char *orig=p, *dest=p+1;
397: while ((*orig++ = *dest++)); /* Remove multiple /'s */
2.55 frystyk 398: end = orig-1;
2.19 frystyk 399: }
2.56 frystyk 400: } else
401: p++;
402: } else
403: p++;
404: }
2.19 frystyk 405: }
2.31 frystyk 406: if (URI_TRACE)
2.62 eric 407: HTTrace("into\n............ `%s'\n", *url);
2.55 frystyk 408: return *url;
2.19 frystyk 409: }
2.31 frystyk 410:
1.1 timbl 411: /* Make Relative Name
412: ** ------------------
413: **
414: ** This function creates and returns a string which gives an expression of
415: ** one address as related to another. Where there is no relation, an absolute
416: ** address is retured.
417: **
418: ** On entry,
419: ** Both names must be absolute, fully qualified names of nodes
420: ** (no anchor bits)
421: **
422: ** On exit,
423: ** The return result points to a newly allocated name which, if
424: ** parsed by HTParse relative to relatedName, will yield aName.
425: ** The caller is responsible for freeing the resulting name later.
426: **
427: */
2.63 frystyk 428: PUBLIC char * HTRelative (const char * aName, const char * relatedName)
1.1 timbl 429: {
430: char * result = 0;
2.63 frystyk 431: const char *p = aName;
432: const char *q = relatedName;
433: const char * after_access = 0;
434: const char * path = 0;
435: const char * last_slash = 0;
1.1 timbl 436: int slashes = 0;
437:
438: for(;*p; p++, q++) { /* Find extent of match */
439: if (*p!=*q) break;
440: if (*p==':') after_access = p+1;
441: if (*p=='/') {
442: last_slash = p;
443: slashes++;
444: if (slashes==3) path=p;
445: }
446: }
447:
448: /* q, p point to the first non-matching character or zero */
449:
450: if (!after_access) { /* Different access */
451: StrAllocCopy(result, aName);
452: } else if (slashes<3){ /* Different nodes */
453: StrAllocCopy(result, after_access);
454: } else { /* Some path in common */
455: int levels= 0;
456: for(; *q && (*q!='#'); q++) if (*q=='/') levels++;
2.61 frystyk 457: if ((result = (char *) HT_MALLOC(3*levels + strlen(last_slash) + 1)) == NULL)
458: HT_OUTOFMEM("HTRelative");
1.1 timbl 459: result[0]=0;
460: for(;levels; levels--)strcat(result, "../");
461: strcat(result, last_slash+1);
462: }
2.62 eric 463: if (URI_TRACE) HTTrace(
2.21 frystyk 464: "HTRelative.. `%s' expressed relative to `%s' is `%s'\n",
465: aName, relatedName, result);
1.1 timbl 466: return result;
467: }
2.1 timbl 468:
2.24 luotonen 469: /* HTCleanTelnetString()
470: * Make sure that the given string doesn't contain characters that
471: * could cause security holes, such as newlines in ftp, gopher,
472: * news or telnet URLs; more specifically: allows everything between
2.26 frystyk 473: * ASCII 20-7E, and also A0-FE, inclusive. Also TAB ('\t') allowed!
2.24 luotonen 474: *
475: * On entry,
476: * str the string that is *modified* if necessary. The
477: * string will be truncated at the first illegal
478: * character that is encountered.
479: * On exit,
480: * returns YES, if the string was modified.
481: * NO, otherwise.
482: */
2.60 frystyk 483: PUBLIC BOOL HTCleanTelnetString (char * str)
2.24 luotonen 484: {
485: char * cur = str;
486:
487: if (!str) return NO;
488:
489: while (*cur) {
2.52 frystyk 490: int a = TOASCII((unsigned char) *cur);
2.26 frystyk 491: if (a != 0x9 && (a < 0x20 || (a > 0x7E && a < 0xA0) || a > 0xFE)) {
2.31 frystyk 492: if (URI_TRACE)
2.62 eric 493: HTTrace("Illegal..... character in URL: \"%s\"\n",str);
2.24 luotonen 494: *cur = 0;
2.31 frystyk 495: if (URI_TRACE)
2.62 eric 496: HTTrace("Truncated... \"%s\"\n",str);
2.24 luotonen 497: return YES;
498: }
499: cur++;
500: }
501: return NO;
502: }
503:
Webmaster