Annotation of libwww/Library/src/HTFilter.c, revision 2.38
2.1 frystyk 1: /*
2: ** BEFORE AND AFTER FILTERS
3: **
4: ** (c) COPYRIGHT MIT 1995.
5: ** Please first read the full copyright statement in the file COPYRIGH.
2.38 ! frystyk 6: ** @(#) $Id: HTFilter.c,v 2.37 1999/03/09 18:59:08 frystyk Exp $
2.1 frystyk 7: **
8: ** This module implrments a set of default filters that can be registerd
9: ** as BEFORE and AFTER filters to the Net manager
10: ** Authors
11: ** HFN Henrik Frystyk, frystyk@w.org
12: ** History
13: ** Jul 4, 96 Written
14: */
15:
16: /* Library include files */
17: #include "WWWLib.h"
18: #include "WWWCache.h"
19: #include "WWWHTTP.h"
20: #include "HTLog.h"
21: #include "HTAccess.h"
2.10 frystyk 22: #include "HTProxy.h"
23: #include "HTRules.h"
2.1 frystyk 24: #include "HTFilter.h" /* Implemented here */
25:
26: /* ------------------------------------------------------------------------- */
27:
28: /*
29: ** Proxy and Gateway BEFORE filter
30: ** -------------------------------
31: ** Checks for registerd proxy servers or gateways and sees whether this
32: ** request should be redirected to a proxy or a gateway. Proxies have
33: ** higher priority than gateways so we look for them first!
34: ** For HTTP/1.0 and HTTP/1.1 we may only send a full URL (including the
35: ** host portion) to proxy servers. Therefore, we tell the Library whether
36: ** to use the full URL or the traditional HTTP one without the host part.
37: */
2.15 frystyk 38: PUBLIC int HTProxyFilter (HTRequest * request, void * param, int mode)
2.1 frystyk 39: {
40: HTParentAnchor * anchor = HTRequest_anchor(request);
2.2 frystyk 41: char * addr = HTAnchor_physical(anchor);
2.1 frystyk 42: char * physical = NULL;
43: if ((physical = HTProxy_find(addr))) {
2.6 frystyk 44: HTRequest_setFullURI(request, YES); /* For now */
2.5 frystyk 45: HTRequest_setProxy(request, physical);
2.8 frystyk 46: HT_FREE(physical);
2.6 frystyk 47: #if 0
48: /* Don't paste the URLs together anymore */
2.1 frystyk 49: StrAllocCat(physical, addr);
2.5 frystyk 50: HTAnchor_setPhysical(anchor, physical);
2.6 frystyk 51: #endif
2.1 frystyk 52: } else if ((physical = HTGateway_find(addr))) {
53: /*
54: ** A gateway URL is crated by chopping off any leading "/" to make the
55: ** host into part of path
56: */
57: char * path =
58: HTParse(addr, "", PARSE_HOST + PARSE_PATH + PARSE_PUNCTUATION);
2.26 frystyk 59: char * gatewayed = HTParse((*path=='/') ? path+1 : path, physical, PARSE_ALL);
2.1 frystyk 60: HTAnchor_setPhysical(anchor, gatewayed);
61: HT_FREE(path);
62: HT_FREE(gatewayed);
63: HTRequest_setFullURI(request, NO);
2.6 frystyk 64: HTRequest_deleteProxy(request);
2.1 frystyk 65: } else {
2.6 frystyk 66: HTRequest_setFullURI(request, NO); /* For now */
67: HTRequest_deleteProxy(request);
2.1 frystyk 68: }
69: return HT_OK;
70: }
71:
72: /*
73: ** Rule Translation BEFORE Filter
74: ** ------------------------------
75: ** If we have a set of rules loaded (see the Rule manager) then check
76: ** before each request whether how that should be translated. The trick
77: ** is that a parent anchor has a "address" which is the part from the URL
78: ** we used when we created the anchor. However, it also have a "physical
79: ** address" which is the place we are actually going to look for the
2.2 frystyk 80: ** resource. Hence this filter translates the physical address
81: ** (if any translations are found)
2.1 frystyk 82: */
2.15 frystyk 83: PUBLIC int HTRuleFilter (HTRequest * request, void * param, int mode)
2.1 frystyk 84: {
85: HTList * list = HTRule_global();
86: HTParentAnchor * anchor = HTRequest_anchor(request);
2.2 frystyk 87: char * addr = HTAnchor_physical(anchor);
2.1 frystyk 88: char * physical = HTRule_translate(list, addr, NO);
89: if (!physical) {
90: HTRequest_addError(request, ERR_FATAL, NO, HTERR_FORBIDDEN,
91: NULL, 0, "HTRuleFilter");
92: return HT_ERROR;
93: }
94: HTAnchor_setPhysical(anchor, physical);
95: HT_FREE(physical);
2.16 frystyk 96: return HT_OK;
97: }
98:
99: /*
2.15 frystyk 100: ** Check the Memory Cache (History list) BEFORE filter
101: ** ---------------------------------------------------
2.11 frystyk 102: ** Check if document is already loaded. The user can define whether
103: ** the history list should follow normal expiration or work as a
104: ** traditional history list where expired documents are not updated.
105: ** We don't check for anything but existence proof of a document
106: ** associated with the anchor as the definition is left to the application
107: */
2.15 frystyk 108: PUBLIC int HTMemoryCacheFilter (HTRequest * request, void * param, int mode)
2.11 frystyk 109: {
110: HTReload validation = HTRequest_reloadMode(request);
111: HTParentAnchor * anchor = HTRequest_anchor(request);
112: void * document = HTAnchor_document(anchor);
2.1 frystyk 113:
2.11 frystyk 114: /*
2.14 frystyk 115: ** We only check the memory cache if it's a GET method
116: */
117: if (HTRequest_method(request) != METHOD_GET) {
2.35 frystyk 118: HTTRACE(CACHE_TRACE, "Mem Cache... We only check GET methods\n");
2.14 frystyk 119: return HT_OK;
120: }
121:
122: /*
2.11 frystyk 123: ** If we are asked to flush the persistent cache then there is no reason
124: ** to do anything here - we're flushing it anyway. Also if no document
125: ** then just exit from this filter.
126: */
127: if (!document || validation > HT_CACHE_FLUSH_MEM) {
2.35 frystyk 128: HTTRACE(CACHE_TRACE, "Mem Cache... No fresh document...\n");
2.11 frystyk 129: return HT_OK;
130: }
2.1 frystyk 131:
2.11 frystyk 132: /*
133: ** If we have a document object associated with this anchor then we also
134: ** have the object in the history list. Depending on what the user asked,
135: ** we can add a cache validator
136: */
2.14 frystyk 137: if (document && validation != HT_CACHE_FLUSH_MEM) {
2.35 frystyk 138: HTTRACE(CACHE_TRACE, "Mem Cache... Document already in memory\n");
2.14 frystyk 139: return HT_LOADED;
2.15 frystyk 140: }
2.17 frystyk 141: return HT_OK;
2.15 frystyk 142: }
143:
144: /*
2.1 frystyk 145: ** Error and Information AFTER filter
146: ** ----------------------------------
147: ** It checks the status code from a request and generates an
148: ** error/information message if required.
149: */
2.15 frystyk 150: PUBLIC int HTInfoFilter (HTRequest * request, HTResponse * response,
151: void * param, int status)
2.1 frystyk 152: {
153: HTParentAnchor * anchor = HTRequest_anchor(request);
154: char * uri = HTAnchor_address((HTAnchor*) anchor);
155: switch (status) {
2.28 frystyk 156: case HT_RETRY: {
157: HTAlertCallback *cbf = HTAlert_find(HT_A_MESSAGE);
158: if (cbf) (*cbf)(request, HT_A_MESSAGE, HT_MSG_NULL, NULL,
159: HTRequest_error(request), NULL);
2.35 frystyk 160: HTTRACE(PROT_TRACE, "Load End.... NOT AVAILABLE, RETRY AT %ld\n" _
2.15 frystyk 161: HTResponse_retryTime(response));
2.28 frystyk 162: }
163: break;
2.1 frystyk 164:
2.7 frystyk 165: case HT_NO_DATA:
166: {
167: /*
168: ** The document was empty
169: */
170: HTAlertCallback *cbf = HTAlert_find(HT_A_MESSAGE);
171: if (cbf) (*cbf)(request, HT_A_MESSAGE, HT_MSG_NULL, NULL,
172: HTRequest_error(request), NULL);
2.35 frystyk 173: HTTRACE(PROT_TRACE, "Load End.... EMPTY: No content `%s\'\n" _
2.7 frystyk 174: uri ? uri : "<UNKNOWN>");
175: break;
176: }
2.3 frystyk 177:
178: case HT_LOADED:
2.35 frystyk 179: HTTRACE(PROT_TRACE, "Load End.... OK: `%s\'\n" _ uri);
2.16 frystyk 180: break;
181:
182: default:
2.3 frystyk 183: {
184: /*
2.16 frystyk 185: ** See if we have a function registered for outputting errors.
186: ** If so then call it and present the message to the user
2.3 frystyk 187: */
188: HTAlertCallback *cbf = HTAlert_find(HT_A_MESSAGE);
189: if (cbf) (*cbf)(request, HT_A_MESSAGE, HT_MSG_NULL, NULL,
190: HTRequest_error(request), NULL);
2.35 frystyk 191: HTTRACE(PROT_TRACE, "Load End.... Request ended with code %d\n" _ status);
2.1 frystyk 192: break;
193: }
2.16 frystyk 194: }
2.1 frystyk 195: HT_FREE(uri);
196: return HT_OK;
197: }
198:
199: /*
200: ** Redirection AFTER filter
201: ** ------------------------
202: ** The redirection handler only handles redirections
203: ** on the GET or HEAD method (or any other safe method)
204: */
2.15 frystyk 205: PUBLIC int HTRedirectFilter (HTRequest * request, HTResponse * response,
206: void * param, int status)
2.1 frystyk 207: {
208: HTMethod method = HTRequest_method(request);
2.15 frystyk 209: HTAnchor * new_anchor = HTResponse_redirection(response);
2.37 frystyk 210:
211: /* Check for destination */
2.7 frystyk 212: if (!new_anchor) {
2.35 frystyk 213: HTTRACE(PROT_TRACE, "Redirection. No destination\n");
2.7 frystyk 214: return HT_OK;
215: }
216:
2.1 frystyk 217: /*
2.21 frystyk 218: ** Only do automatic redirect on GET and HEAD. Ask for all
2.27 frystyk 219: ** other methods.
2.1 frystyk 220: */
2.21 frystyk 221: if (!HTMethod_isSafe(method)) {
222:
223: /*
224: ** If we got a 303 See Other then change the method to GET.
225: ** Otherwise ask the user whether we should continue.
226: */
227: if (status == HT_SEE_OTHER) {
2.35 frystyk 228: HTTRACE(PROT_TRACE, "Redirection. Changing method from %s to GET\n" _
2.21 frystyk 229: HTMethod_name(method));
230: HTRequest_setMethod(request, METHOD_GET);
231: } else {
232: HTAlertCallback * prompt = HTAlert_find(HT_A_CONFIRM);
233: if (prompt) {
234: if ((*prompt)(request, HT_A_CONFIRM, HT_MSG_REDIRECTION,
235: NULL, NULL, NULL) != YES)
2.27 frystyk 236: return HT_OK;
2.21 frystyk 237: }
2.4 frystyk 238: }
2.1 frystyk 239: }
2.36 frystyk 240:
241: /* Register the redirection as a link relationship */
242: {
243: HTLinkType ltype = status==HT_PERM_REDIRECT ? HT_LR_PERM_REDIRECT :
244: (status==HT_TEMP_REDIRECT || status==HT_FOUND) ? HT_LR_TEMP_REDIRECT :
245: status==HT_SEE_OTHER ? HT_LR_SEE_OTHER : NULL;
246: if (ltype) {
247: HTLink_add((HTAnchor *) HTRequest_anchor(request), new_anchor,
248: ltype, method);
249: }
250: }
251:
2.34 frystyk 252: /* Delete any auth credendials as they get regenerated */
253: HTRequest_deleteCredentialsAll(request);
254:
2.1 frystyk 255: /*
256: ** Start new request with the redirect anchor found in the headers.
257: ** Note that we reuse the same request object which means that we must
258: ** keep this around until the redirected request has terminated. It also
259: ** allows us in an easy way to keep track of the number of redirections
260: ** so that we can detect endless loops.
261: */
2.4 frystyk 262: if (HTRequest_doRetry(request)) {
2.1 frystyk 263: HTLoadAnchor(new_anchor, request);
2.9 frystyk 264: } else {
265: HTRequest_addError(request, ERR_FATAL, NO, HTERR_MAX_REDIRECT,
266: NULL, 0, "HTRedirectFilter");
2.37 frystyk 267: return HT_OK; /* Wanna fall through */
2.9 frystyk 268: }
269:
270: /*
271: ** By returning HT_ERROR we make sure that this is the last handler to be
272: ** called. We do this as we don't want any other filter to delete the
273: ** request object now when we have just started a new one ourselves
274: */
275: return HT_ERROR;
276: }
277:
278: /*
2.15 frystyk 279: ** Retry through Proxy AFTER Filter
280: ** --------------------------------
2.9 frystyk 281: ** This filter handles a 305 Use Proxy response and retries the request
282: ** through the proxy
283: */
2.15 frystyk 284: PUBLIC int HTUseProxyFilter (HTRequest * request, HTResponse * response,
285: void * param, int status)
2.9 frystyk 286: {
2.20 frystyk 287: HTAlertCallback * cbf = HTAlert_find(HT_A_CONFIRM);
2.15 frystyk 288: HTAnchor * proxy_anchor = HTResponse_redirection(response);
2.9 frystyk 289: if (!proxy_anchor) {
2.35 frystyk 290: HTTRACE(PROT_TRACE, "Use Proxy... No proxy location\n");
2.9 frystyk 291: return HT_OK;
292: }
293:
294: /*
295: ** Add the proxy to the list. Assume HTTP access method only!
2.20 frystyk 296: ** Because evil servers may rediret the client to an untrusted
297: ** proxy, we can only accept redirects for this particular
298: ** server. Also, we do not know whether this is for HTTP or all
299: ** other requests as well
2.9 frystyk 300: */
2.20 frystyk 301: if ((cbf && (*cbf)(request, HT_A_CONFIRM, HT_MSG_PROXY, NULL,NULL,NULL))) {
2.9 frystyk 302: char * addr = HTAnchor_address(proxy_anchor);
303: HTProxy_add("http", addr);
304: HT_FREE(addr);
305:
2.20 frystyk 306: /*
307: ** Start new request through the proxy if we haven't reached the max
308: ** number of redirections for this request
309: */
310: if (HTRequest_doRetry(request)) {
311: HTLoadAnchor(proxy_anchor, request);
312: } else {
313: HTRequest_addError(request, ERR_FATAL, NO, HTERR_MAX_REDIRECT,
314: NULL, 0, "HTRedirectFilter");
315: }
316:
317: /*
318: ** By returning HT_ERROR we make sure that this is the last handler to be
319: ** called. We do this as we don't want any other filter to delete the
320: ** request object now when we have just started a new one ourselves
321: */
322: return HT_ERROR;
323:
2.1 frystyk 324: } else {
2.20 frystyk 325: HTRequest_addError(request, ERR_FATAL, NO, HTERR_NO_AUTO_PROXY,
326: NULL, 0, "HTUseProxyFilter");
327: return HT_OK;
2.1 frystyk 328: }
329: }
330:
331: /*
332: ** Client side authentication BEFORE filter
333: ** ----------------------------------------
334: ** The filter generates the credentials required to access a document
335: ** Getting the credentials may involve asking the user
336: */
2.15 frystyk 337: PUBLIC int HTCredentialsFilter (HTRequest * request, void * param, int mode)
2.1 frystyk 338: {
339: /*
340: ** Ask the authentication module to call the right credentials generator
341: ** that understands this scheme
342: */
2.15 frystyk 343: if (HTAA_beforeFilter(request, param, mode) == HT_OK) {
2.35 frystyk 344: HTTRACE(PROT_TRACE, "Credentials. verified\n");
2.1 frystyk 345: return HT_OK;
346: } else {
347: HTRequest_addError(request, ERR_FATAL, NO, HTERR_UNAUTHORIZED,
348: NULL, 0, "HTCredentialsFilter");
349: return HT_ERROR;
350: }
351: }
352:
353: /*
354: ** Client side authentication AFTER filter
355: ** ---------------------------------------
356: ** The client side authentication filter uses the
357: ** user dialog messages registered in the HTAlert module.
358: ** By default these are the ones used by the line mode browser but you can
359: ** just register something else.
360: */
2.15 frystyk 361: PUBLIC int HTAuthFilter (HTRequest * request, HTResponse * response,
362: void * param, int status)
2.1 frystyk 363: {
364: /*
365: ** Ask the authentication module to call the right challenge parser
366: ** that understands this scheme
367: */
2.15 frystyk 368: if (HTAA_afterFilter(request, response, param, status) == HT_OK) {
2.1 frystyk 369:
370: /*
371: ** Start request with new credentials. As with the redirection filter
372: ** we reuse the same request object which means that we must
373: ** keep this around until the redirected request has terminated
374: */
375: HTLoad(request, NO);
376:
377: /*
378: ** We return HT_ERROR to make sure that this is the last handler to be
379: ** called. We do this as we don't want any other filter to delete the
380: ** request object now when we have just started a new one ourselves
381: */
382: return HT_ERROR;
383: }
384: return HT_OK;
2.32 kahan 385: }
386:
387: /*
388: ** Client side authentication info AFTER filter
389: ** ---------------------------------------
390: */
391: PUBLIC int HTAuthInfoFilter (HTRequest * request, HTResponse * response,
392: void * param, int status)
393: {
394: /*
395: ** Ask the authentication module to call the right authentication info
396: ** parser
397: */
398: if (! HTResponse_challenge (response))
399: return HT_OK;
400: else if (HTAA_updateFilter(request, response, param, status) == HT_OK)
401: return HT_OK;
402: else
403: return HT_ERROR;
2.1 frystyk 404: }
405:
406: /*
2.24 frystyk 407: ** Request Logging AFTER filter
408: ** ----------------------------
2.1 frystyk 409: ** Default Logging filter using the log manager provided by HTLog.c
410: */
2.15 frystyk 411: PUBLIC int HTLogFilter (HTRequest * request, HTResponse * response,
412: void * param, int status)
2.1 frystyk 413: {
414: if (request) {
2.23 frystyk 415: HTLog * log = (HTLog *) param;
416: if (log) HTLog_addCLF(log, request, status);
2.24 frystyk 417: return HT_OK;
418: }
419: return HT_ERROR;
420: }
421:
422: /*
423: ** Request Referer AFTER filter
424: ** ----------------------------
425: ** Default Referer Log filter using the log manager provided by HTLog.c
426: */
427: PUBLIC int HTRefererFilter (HTRequest * request, HTResponse * response,
428: void * param, int status)
429: {
430: if (request) {
431: HTLog * log = (HTLog *) param;
432: if (log) HTLog_addReferer(log, request, status);
2.1 frystyk 433: return HT_OK;
434: }
435: return HT_ERROR;
436: }
Webmaster