1 /* ====================================================================
2 * The Apache Software License, Version 1.1
4 * Copyright (c) 2000 The Apache Software Foundation. All rights
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in
16 * the documentation and/or other materials provided with the
19 * 3. The end-user documentation included with the redistribution,
20 * if any, must include the following acknowledgment:
21 * "This product includes software developed by the
22 * Apache Software Foundation (http://www.apache.org/)."
23 * Alternately, this acknowledgment may appear in the software itself,
24 * if and wherever such third-party acknowledgments normally appear.
26 * 4. The names "Apache" and "Apache Software Foundation" must
27 * not be used to endorse or promote products derived from this
28 * software without prior written permission. For written
29 * permission, please contact apache@apache.org.
31 * 5. Products derived from this software may not be called "Apache",
32 * nor may "Apache" appear in their name, without prior written
33 * permission of the Apache Software Foundation.
35 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED
36 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
37 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
38 * DISCLAIMED. IN NO EVENT SHALL THE APACHE SOFTWARE FOUNDATION OR
39 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
40 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
41 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
42 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
43 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
44 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
45 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
47 * ====================================================================
49 * This software consists of voluntary contributions made by many
50 * individuals on behalf of the Apache Software Foundation. For more
51 * information on the Apache Software Foundation, please see
52 * <http://www.apache.org/>.
54 * Portions of this software are based upon public domain software
55 * originally written at the National Center for Supercomputing Applications,
56 * University of Illinois, Urbana-Champaign.
59 #define WANT_BASENAME_MATCH
62 #include "http_core.h"
63 #include "http_config.h"
64 #include "http_request.h"
66 #include "apr_file_io.h"
67 #include "apr_strings.h"
73 /* mod_speling.c - by Alexei Kosut <akosut@organic.com> June, 1996
75 * This module is transparent, and simple. It attempts to correct
76 * misspellings of URLs that users might have entered, namely by checking
77 * capitalizations. If it finds a match, it sends a redirect.
79 * 08-Aug-1997 <Martin.Kraemer@Mch.SNI.De>
80 * o Upgraded module interface to apache_1.3a2-dev API (more NULL's in
82 * o Integrated tcsh's "spelling correction" routine which allows one
83 * misspelling (character insertion/omission/typo/transposition).
84 * Rewrote it to ignore case as well. This ought to catch the majority
85 * of misspelled requests.
86 * o Commented out the second pass where files' suffixes are stripped.
87 * Given the better hit rate of the first pass, this rather ugly
88 * (request index.html, receive index.db ?!?!) solution can be
90 * o wrote a "kind of" html page for mod_speling
92 * Activate it with "CheckSpelling On"
95 AP_MODULE_DECLARE_DATA module speling_module;
102 * Create a configuration specific to this module for a server or directory
103 * location, and fill it with the default settings.
105 * The API says that in the absence of a merge function, the record for the
106 * closest ancestor is used exclusively. That's what we want, so we don't
107 * bother to have such a function.
110 static void *mkconfig(apr_pool_t *p)
112 spconfig *cfg = apr_pcalloc(p, sizeof(spconfig));
119 * Respond to a callback to create configuration record for a server or
122 static void *create_mconfig_for_server(apr_pool_t *p, server_rec *s)
128 * Respond to a callback to create a config record for a specific directory.
130 static void *create_mconfig_for_directory(apr_pool_t *p, char *dir)
136 * Handler for the CheckSpelling directive, which is FLAG.
138 static const char *set_speling(cmd_parms *cmd, void *mconfig, int arg)
140 spconfig *cfg = (spconfig *) mconfig;
147 * Define the directives specific to this module. This structure is referenced
148 * later by the 'module' structure.
150 static const command_rec speling_cmds[] =
152 AP_INIT_FLAG("CheckSpelling", set_speling, NULL, OR_OPTIONS,
153 "whether or not to fix miscapitalized/misspelled requests"),
159 SP_MISCAPITALIZED = 1,
160 SP_TRANSPOSITION = 2,
167 static const char *sp_reason_str[] =
171 "transposed characters",
174 "mistyped character",
184 * spdist() is taken from Kernighan & Pike,
185 * _The_UNIX_Programming_Environment_
186 * and adapted somewhat to correspond better to psychological reality.
187 * (Note the changes to the return values)
189 * According to Pollock and Zamora, CACM April 1984 (V. 27, No. 4),
190 * page 363, the correct order for this is:
191 * OMISSION = TRANSPOSITION > INSERTION > SUBSTITUTION
192 * thus, it was exactly backwards in the old version. -- PWP
194 * This routine was taken out of tcsh's spelling correction code
195 * (tcsh-6.07.04) and re-converted to apache data types ("char" type
196 * instead of tcsh's NLS'ed "Char"). Plus it now ignores the case
197 * during comparisons, so is a "approximate strcasecmp()".
198 * NOTE that is still allows only _one_ real "typo",
199 * it does NOT try to correct multiple errors.
202 static sp_reason spdist(const char *s, const char *t)
204 for (; apr_tolower(*s) == apr_tolower(*t); t++, s++) {
206 return SP_MISCAPITALIZED; /* exact match (sans case) */
211 if (s[1] && t[1] && apr_tolower(*s) == apr_tolower(t[1])
212 && apr_tolower(*t) == apr_tolower(s[1])
213 && strcasecmp(s + 2, t + 2) == 0) {
214 return SP_TRANSPOSITION; /* transposition */
216 if (strcasecmp(s + 1, t + 1) == 0) {
217 return SP_SIMPLETYPO; /* 1 char mismatch */
220 if (strcasecmp(s + 1, t) == 0) {
221 return SP_EXTRACHAR; /* extra character */
224 if (*t && strcasecmp(s, t + 1) == 0) {
225 return SP_MISSINGCHAR; /* missing character */
227 return SP_VERYDIFFERENT; /* distance too large to fix. */
230 static int sort_by_quality(const void *left, const void *rite)
232 return (int) (((misspelled_file *) left)->quality)
233 - (int) (((misspelled_file *) rite)->quality);
236 static int check_speling(request_rec *r)
239 char *good, *bad, *postgood, *url;
241 int filoc, dotloc, urlen, pglen;
242 apr_array_header_t *candidates = NULL;
245 cfg = ap_get_module_config(r->per_dir_config, &speling_module);
250 /* We only want to worry about GETs */
251 if (r->method_number != M_GET) {
255 /* We've already got a file of some kind or another */
256 if (r->proxyreq || (r->finfo.protection != 0)) {
260 /* This is a sub request - don't mess with it */
266 * The request should end up looking like this:
267 * r->uri: /correct-url/mispelling/more
268 * r->filename: /correct-file/mispelling r->path_info: /more
270 * So we do this in steps. First break r->filename into two pieces
273 filoc = ap_rind(r->filename, '/');
275 * Don't do anything if the request doesn't contain a slash, or
278 if (filoc == -1 || strcmp(r->uri, "/") == 0) {
282 /* good = /correct-file */
283 good = apr_pstrndup(r->pool, r->filename, filoc);
284 /* bad = mispelling */
285 bad = apr_pstrdup(r->pool, r->filename + filoc + 1);
286 /* postgood = mispelling/more */
287 postgood = apr_pstrcat(r->pool, bad, r->path_info, NULL);
289 urlen = strlen(r->uri);
290 pglen = strlen(postgood);
292 /* Check to see if the URL pieces add up */
293 if (strcmp(postgood, r->uri + (urlen - pglen))) {
297 /* url = /correct-url */
298 url = apr_pstrndup(r->pool, r->uri, (urlen - pglen));
300 /* Now open the directory and do ourselves a check... */
301 if (apr_dir_open(&dir, good, r->pool) != APR_SUCCESS) {
302 /* Oops, not a directory... */
306 candidates = apr_make_array(r->pool, 2, sizeof(misspelled_file));
308 dotloc = ap_ind(bad, '.');
310 dotloc = strlen(bad);
313 while (apr_dir_read(&dirent, APR_FINFO_DIRENT, dir) == APR_SUCCESS) {
317 * If we end up with a "fixed" URL which is identical to the
318 * requested one, we must have found a broken symlink or some such.
319 * Do _not_ try to redirect this, it causes a loop!
321 if (strcmp(bad, dirent.name) == 0) {
327 * miscapitalization errors are checked first (like, e.g., lower case
328 * file, upper case request)
330 else if (strcasecmp(bad, dirent.name) == 0) {
331 misspelled_file *sp_new;
333 sp_new = (misspelled_file *) apr_push_array(candidates);
334 sp_new->name = apr_pstrdup(r->pool, dirent.name);
335 sp_new->quality = SP_MISCAPITALIZED;
339 * simple typing errors are checked next (like, e.g.,
340 * missing/extra/transposed char)
342 else if ((q = spdist(bad, dirent.name)) != SP_VERYDIFFERENT) {
343 misspelled_file *sp_new;
345 sp_new = (misspelled_file *) apr_push_array(candidates);
346 sp_new->name = apr_pstrdup(r->pool, dirent.name);
351 * The spdist() should have found the majority of the misspelled
352 * requests. It is of questionable use to continue looking for
353 * files with the same base name, but potentially of totally wrong
354 * type (index.html <-> index.db).
355 * I would propose to not set the WANT_BASENAME_MATCH define.
356 * 08-Aug-1997 <Martin.Kraemer@Mch.SNI.De>
358 * However, Alexei replied giving some reasons to add it anyway:
359 * > Oh, by the way, I remembered why having the
360 * > extension-stripping-and-matching stuff is a good idea:
362 * > If you're using MultiViews, and have a file named foobar.html,
363 * > which you refer to as "foobar", and someone tried to access
364 * > "Foobar", mod_speling won't find it, because it won't find
365 * > anything matching that spelling. With the extension-munging,
366 * > it would locate "foobar.html". Not perfect, but I ran into
367 * > that problem when I first wrote the module.
370 #ifdef WANT_BASENAME_MATCH
372 * Okay... we didn't find anything. Now we take out the hard-core
373 * power tools. There are several cases here. Someone might have
374 * entered a wrong extension (.htm instead of .html or vice
375 * versa) or the document could be negotiated. At any rate, now
376 * we just compare stuff before the first dot. If it matches, we
377 * figure we got us a match. This can result in wrong things if
378 * there are files of different content types but the same prefix
379 * (e.g. foo.gif and foo.html) This code will pick the first one
380 * it finds. Better than a Not Found, though.
382 int entloc = ap_ind(dirent.name, '.');
384 entloc = strlen(dirent.name);
387 if ((dotloc == entloc)
388 && !strncasecmp(bad, dirent.name, dotloc)) {
389 misspelled_file *sp_new;
391 sp_new = (misspelled_file *) apr_push_array(candidates);
392 sp_new->name = apr_pstrdup(r->pool, dirent.name);
393 sp_new->quality = SP_VERYDIFFERENT;
400 if (candidates->nelts != 0) {
401 /* Wow... we found us a mispelling. Construct a fixed url */
404 misspelled_file *variant = (misspelled_file *) candidates->elts;
407 ref = apr_table_get(r->headers_in, "Referer");
409 qsort((void *) candidates->elts, candidates->nelts,
410 sizeof(misspelled_file), sort_by_quality);
413 * Conditions for immediate redirection:
414 * a) the first candidate was not found by stripping the suffix
415 * AND b) there exists only one candidate OR the best match is not
417 * then return a redirection right away.
419 if (variant[0].quality != SP_VERYDIFFERENT
420 && (candidates->nelts == 1
421 || variant[0].quality != variant[1].quality)) {
423 nuri = ap_escape_uri(r->pool, apr_pstrcat(r->pool, url,
425 r->path_info, NULL));
426 if (r->parsed_uri.query)
427 nuri = apr_pstrcat(r->pool, nuri, "?", r->parsed_uri.query, NULL);
429 apr_table_setn(r->headers_out, "Location",
430 ap_construct_url(r->pool, nuri, r));
432 ap_log_rerror(APLOG_MARK, APLOG_NOERRNO | APLOG_INFO, APR_SUCCESS,
434 ref ? "Fixed spelling: %s to %s from %s"
435 : "Fixed spelling: %s to %s",
438 return HTTP_MOVED_PERMANENTLY;
441 * Otherwise, a "[300] Multiple Choices" list with the variants is
447 apr_pool_t *sub_pool;
448 apr_array_header_t *t;
449 apr_array_header_t *v;
452 if (r->main == NULL) {
458 notes = r->main->notes;
461 if (apr_create_pool(&sub_pool, p) != APR_SUCCESS)
464 t = apr_make_array(sub_pool, candidates->nelts * 8 + 8,
466 v = apr_make_array(sub_pool, candidates->nelts * 5,
469 /* Generate the response text. */
471 *(const char **)apr_push_array(t) =
472 "The document name you requested (<code>";
473 *(const char **)apr_push_array(t) = ap_escape_html(sub_pool, r->uri);
474 *(const char **)apr_push_array(t) =
475 "</code>) could not be found on this server.\n"
476 "However, we found documents with names similar "
477 "to the one you requested.<p>"
478 "Available documents:\n<ul>\n";
480 for (i = 0; i < candidates->nelts; ++i) {
484 reason = sp_reason_str[(int) (variant[i].quality)];
485 /* The format isn't very neat... */
486 vuri = apr_pstrcat(sub_pool, url, variant[i].name, r->path_info,
487 (r->parsed_uri.query != NULL) ? "?" : "",
488 (r->parsed_uri.query != NULL)
489 ? r->parsed_uri.query : "",
491 *(const char **)apr_push_array(v) = "\"";
492 *(const char **)apr_push_array(v) = ap_escape_uri(sub_pool, vuri);
493 *(const char **)apr_push_array(v) = "\";\"";
494 *(const char **)apr_push_array(v) = reason;
495 *(const char **)apr_push_array(v) = "\"";
497 *(const char **)apr_push_array(t) = "<li><a href=\"";
498 *(const char **)apr_push_array(t) = ap_escape_uri(sub_pool, vuri);
499 *(const char **)apr_push_array(t) = "\">";
500 *(const char **)apr_push_array(t) = ap_escape_html(sub_pool, vuri);
501 *(const char **)apr_push_array(t) = "</a> (";
502 *(const char **)apr_push_array(t) = reason;
503 *(const char **)apr_push_array(t) = ")\n";
506 * when we have printed the "close matches" and there are
507 * more "distant matches" (matched by stripping the suffix),
508 * then we insert an additional separator text to suggest
509 * that the user LOOK CLOSELY whether these are really the
512 if (i > 0 && i < candidates->nelts - 1
513 && variant[i].quality != SP_VERYDIFFERENT
514 && variant[i + 1].quality == SP_VERYDIFFERENT) {
515 *(const char **)apr_push_array(t) =
516 "</ul>\nFurthermore, the following related "
517 "documents were found:\n<ul>\n";
520 *(const char **)apr_push_array(t) = "</ul>\n";
522 /* If we know there was a referring page, add a note: */
524 *(const char **)apr_push_array(t) =
525 "Please consider informing the owner of the "
527 *(const char **)apr_push_array(t) = ap_escape_uri(sub_pool, ref);
528 *(const char **)apr_push_array(t) = "\">referring page</a> "
529 "about the broken link.\n";
533 /* Pass our apr_table_t to http_protocol.c (see mod_negotiation): */
534 apr_table_setn(notes, "variant-list", apr_array_pstrcat(p, t, 0));
536 apr_table_mergen(r->subprocess_env, "VARIANTS",
537 apr_array_pstrcat(p, v, ','));
539 apr_destroy_pool(sub_pool);
541 ap_log_rerror(APLOG_MARK, APLOG_NOERRNO | APLOG_INFO, 0, r,
542 ref ? "Spelling fix: %s: %d candidates from %s"
543 : "Spelling fix: %s: %d candidates",
544 r->uri, candidates->nelts, ref);
546 return HTTP_MULTIPLE_CHOICES;
553 static void register_hooks(apr_pool_t *p)
555 ap_hook_fixups(check_speling,NULL,NULL,APR_HOOK_LAST);
558 module AP_MODULE_DECLARE_DATA speling_module =
560 STANDARD20_MODULE_STUFF,
561 create_mconfig_for_directory, /* create per-dir config */
562 NULL, /* merge per-dir config */
563 create_mconfig_for_server, /* server config */
564 NULL, /* merge server config */
565 speling_cmds, /* command apr_table_t */
566 register_hooks /* register hooks */