]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | * Copyright (C) 1996-2025 The Squid Software Foundation and contributors | |
3 | * | |
4 | * Squid software is distributed under GPLv2+ license and includes | |
5 | * contributions from numerous individuals and organizations. | |
6 | * Please see the COPYING and CONTRIBUTORS files for details. | |
7 | */ | |
8 | ||
9 | /* DEBUG: section 52 URN Parsing */ | |
10 | ||
11 | #include "squid.h" | |
12 | #include "AccessLogEntry.h" | |
13 | #include "acl/FilledChecklist.h" | |
14 | #include "base/TextException.h" | |
15 | #include "cbdata.h" | |
16 | #include "errorpage.h" | |
17 | #include "FwdState.h" | |
18 | #include "globals.h" | |
19 | #include "HttpReply.h" | |
20 | #include "HttpRequest.h" | |
21 | #include "icmp/net_db.h" | |
22 | #include "MemBuf.h" | |
23 | #include "mime_header.h" | |
24 | #include "RequestFlags.h" | |
25 | #include "Store.h" | |
26 | #include "StoreClient.h" | |
27 | #include "tools.h" | |
28 | #include "urn.h" | |
29 | ||
30 | class UrnState : public StoreClient | |
31 | { | |
32 | CBDATA_CLASS(UrnState); | |
33 | ||
34 | public: | |
35 | explicit UrnState(const AccessLogEntry::Pointer &anAle): ale(anAle) {} | |
36 | ||
37 | void start (HttpRequest *, StoreEntry *); | |
38 | void setUriResFromRequest(HttpRequest *); | |
39 | ||
40 | ~UrnState() override; | |
41 | ||
42 | StoreEntry *entry = nullptr; | |
43 | store_client *sc = nullptr; | |
44 | StoreEntry *urlres_e = nullptr; | |
45 | HttpRequest::Pointer request; | |
46 | HttpRequest::Pointer urlres_r; | |
47 | AccessLogEntry::Pointer ale; ///< details of the requesting transaction | |
48 | ||
49 | /// for receiving a URN resolver reply body from Store and interpreting it | |
50 | Store::ParsingBuffer parsingBuffer; | |
51 | ||
52 | private: | |
53 | /* StoreClient API */ | |
54 | LogTags *loggingTags() const override { return ale ? &ale->cache.code : nullptr; } | |
55 | void fillChecklist(ACLFilledChecklist &) const override; | |
56 | ||
57 | char *urlres = nullptr; | |
58 | }; | |
59 | ||
60 | typedef struct { | |
61 | char *url; | |
62 | char *host; | |
63 | int rtt; | |
64 | ||
65 | struct { | |
66 | int cached; | |
67 | } flags; | |
68 | } url_entry; | |
69 | ||
70 | static STCB urnHandleReply; | |
71 | static url_entry *urnParseReply(const SBuf &, const HttpRequestMethod &); | |
72 | static const char *const crlf = "\r\n"; | |
73 | ||
74 | CBDATA_CLASS_INIT(UrnState); | |
75 | ||
76 | UrnState::~UrnState() | |
77 | { | |
78 | SWALLOW_EXCEPTIONS({ | |
79 | if (urlres_e) { | |
80 | if (sc) | |
81 | storeUnregister(sc, urlres_e, this); | |
82 | urlres_e->unlock("~UrnState+res"); | |
83 | } | |
84 | ||
85 | if (entry) | |
86 | entry->unlock("~UrnState+prime"); | |
87 | ||
88 | safe_free(urlres); | |
89 | }); | |
90 | } | |
91 | ||
92 | static url_entry * | |
93 | urnFindMinRtt(url_entry * urls, const HttpRequestMethod &, int *rtt_ret) | |
94 | { | |
95 | int min_rtt = 0; | |
96 | url_entry *u = nullptr; | |
97 | url_entry *min_u = nullptr; | |
98 | int i; | |
99 | int urlcnt = 0; | |
100 | debugs(52, 3, "urnFindMinRtt"); | |
101 | assert(urls != nullptr); | |
102 | ||
103 | for (i = 0; nullptr != urls[i].url; ++i) | |
104 | ++urlcnt; | |
105 | ||
106 | debugs(53, 3, "urnFindMinRtt: Counted " << i << " URLs"); | |
107 | ||
108 | if (1 == urlcnt) { | |
109 | debugs(52, 3, "urnFindMinRtt: Only one URL - return it!"); | |
110 | return urls; | |
111 | } | |
112 | ||
113 | for (i = 0; i < urlcnt; ++i) { | |
114 | u = &urls[i]; | |
115 | debugs(52, 3, "urnFindMinRtt: " << u->host << " rtt=" << u->rtt); | |
116 | ||
117 | if (u->rtt == 0) | |
118 | continue; | |
119 | ||
120 | if (u->rtt > min_rtt && min_rtt != 0) | |
121 | continue; | |
122 | ||
123 | min_rtt = u->rtt; | |
124 | ||
125 | min_u = u; | |
126 | } | |
127 | ||
128 | if (rtt_ret) | |
129 | *rtt_ret = min_rtt; | |
130 | ||
131 | debugs(52, DBG_IMPORTANT, "urnFindMinRtt: Returning '" << | |
132 | (min_u ? min_u->url : "NONE") << "' RTT " << | |
133 | min_rtt ); | |
134 | ||
135 | return min_u; | |
136 | } | |
137 | ||
138 | void | |
139 | UrnState::setUriResFromRequest(HttpRequest *r) | |
140 | { | |
141 | const auto &query = r->url.absolute(); | |
142 | const auto host = r->url.host(); | |
143 | // TODO: use class AnyP::Uri instead of generating a string and re-parsing | |
144 | LOCAL_ARRAY(char, local_urlres, 4096); | |
145 | snprintf(local_urlres, 4096, "http://%s/uri-res/N2L?" SQUIDSBUFPH, host, SQUIDSBUFPRINT(query)); | |
146 | safe_free(urlres); | |
147 | urlres_r = HttpRequest::FromUrlXXX(local_urlres, r->masterXaction); | |
148 | ||
149 | if (!urlres_r) { | |
150 | debugs(52, 3, "Bad uri-res URL " << local_urlres); | |
151 | const auto err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, r, ale); | |
152 | err->url = xstrdup(local_urlres); | |
153 | errorAppendEntry(entry, err); | |
154 | return; | |
155 | } | |
156 | ||
157 | urlres = xstrdup(local_urlres); | |
158 | urlres_r->header.putStr(Http::HdrType::ACCEPT, "text/plain"); | |
159 | } | |
160 | ||
161 | void | |
162 | UrnState::start(HttpRequest * r, StoreEntry * e) | |
163 | { | |
164 | debugs(52, 3, "urnStart: '" << e->url() << "'" ); | |
165 | entry = e; | |
166 | request = r; | |
167 | ||
168 | entry->lock("UrnState::start"); | |
169 | setUriResFromRequest(r); | |
170 | ||
171 | if (urlres_r == nullptr) | |
172 | return; | |
173 | ||
174 | auto urlEntry = storeGetPublic(urlres, Http::METHOD_GET); | |
175 | ||
176 | if (!urlEntry || (urlEntry->hittingRequiresCollapsing() && !startCollapsingOn(*urlEntry, false))) { | |
177 | urlres_e = storeCreateEntry(urlres, urlres, RequestFlags(), Http::METHOD_GET); | |
178 | sc = storeClientListAdd(urlres_e, this); | |
179 | FwdState::Start(Comm::ConnectionPointer(), urlres_e, urlres_r.getRaw(), ale); | |
180 | if (urlEntry) { | |
181 | urlEntry->abandon(__func__); | |
182 | urlEntry = nullptr; | |
183 | } | |
184 | } else { | |
185 | urlres_e = urlEntry; | |
186 | urlres_e->lock(__func__); | |
187 | sc = storeClientListAdd(urlres_e, this); | |
188 | } | |
189 | ||
190 | storeClientCopy(sc, urlres_e, | |
191 | parsingBuffer.makeInitialSpace(), | |
192 | urnHandleReply, | |
193 | this); | |
194 | } | |
195 | ||
196 | void | |
197 | UrnState::fillChecklist(ACLFilledChecklist &checklist) const | |
198 | { | |
199 | checklist.setRequest(request.getRaw()); | |
200 | checklist.al = ale; | |
201 | } | |
202 | ||
203 | void | |
204 | urnStart(HttpRequest *r, StoreEntry *e, const AccessLogEntryPointer &ale) | |
205 | { | |
206 | const auto anUrn = new UrnState(ale); | |
207 | anUrn->start (r, e); | |
208 | } | |
209 | ||
210 | static int | |
211 | url_entry_sort(const void *A, const void *B) | |
212 | { | |
213 | const url_entry *u1 = (const url_entry *)A; | |
214 | const url_entry *u2 = (const url_entry *)B; | |
215 | ||
216 | if (u2->rtt == u1->rtt) | |
217 | return 0; | |
218 | else if (0 == u1->rtt) | |
219 | return 1; | |
220 | else if (0 == u2->rtt) | |
221 | return -1; | |
222 | else | |
223 | return u1->rtt - u2->rtt; | |
224 | } | |
225 | ||
226 | /* TODO: use the clientStream support for this */ | |
227 | static void | |
228 | urnHandleReply(void *data, StoreIOBuffer result) | |
229 | { | |
230 | UrnState *urnState = static_cast<UrnState *>(data); | |
231 | StoreEntry *e = urnState->entry; | |
232 | StoreEntry *urlres_e = urnState->urlres_e; | |
233 | url_entry *urls; | |
234 | url_entry *u; | |
235 | url_entry *min_u; | |
236 | ErrorState *err; | |
237 | int i; | |
238 | int urlcnt = 0; | |
239 | ||
240 | debugs(52, 3, result << " with " << *e); | |
241 | ||
242 | if (EBIT_TEST(urlres_e->flags, ENTRY_ABORTED) || result.flags.error) { | |
243 | delete urnState; | |
244 | return; | |
245 | } | |
246 | ||
247 | if (!e->isAccepting()) { | |
248 | debugs(52, 3, "terminating due to bad " << *e); | |
249 | delete urnState; | |
250 | return; | |
251 | } | |
252 | ||
253 | urnState->parsingBuffer.appended(result.data, result.length); | |
254 | ||
255 | /* If we haven't received the entire object (urn), copy more */ | |
256 | if (!urnState->sc->atEof()) { | |
257 | const auto bufferedBytes = urnState->parsingBuffer.contentSize(); | |
258 | const auto remainingSpace = urnState->parsingBuffer.space().positionAt(bufferedBytes); | |
259 | ||
260 | if (!remainingSpace.length) { | |
261 | debugs(52, 3, "ran out of buffer space after " << bufferedBytes << " bytes"); | |
262 | // TODO: Here and in other error cases, send ERR_URN_RESOLVE to client. | |
263 | delete urnState; | |
264 | return; | |
265 | } | |
266 | ||
267 | storeClientCopy(urnState->sc, urlres_e, | |
268 | remainingSpace, | |
269 | urnHandleReply, | |
270 | urnState); | |
271 | return; | |
272 | } | |
273 | ||
274 | const auto &peerReply = urlres_e->mem().baseReply(); | |
275 | debugs(52, 3, "got reply, code=" << peerReply.sline.status()); | |
276 | if (peerReply.sline.status() != Http::scOkay) { | |
277 | debugs(52, 3, "urnHandleReply: failed."); | |
278 | err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, urnState->request.getRaw(), urnState->ale); | |
279 | err->url = xstrdup(e->url()); | |
280 | errorAppendEntry(e, err); | |
281 | delete urnState; | |
282 | return; | |
283 | } | |
284 | ||
285 | // XXX: Missing reply freshness checks (e.g., calling refreshCheckHTTP()). | |
286 | ||
287 | urls = urnParseReply(urnState->parsingBuffer.toSBuf(), urnState->request->method); | |
288 | ||
289 | if (!urls) { /* unknown URN error */ | |
290 | debugs(52, 3, "urnTranslateDone: unknown URN " << e->url()); | |
291 | err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, urnState->request.getRaw(), urnState->ale); | |
292 | err->url = xstrdup(e->url()); | |
293 | errorAppendEntry(e, err); | |
294 | delete urnState; | |
295 | return; | |
296 | } | |
297 | ||
298 | for (i = 0; urls[i].url; ++i) | |
299 | ++urlcnt; | |
300 | ||
301 | debugs(53, 3, "urnFindMinRtt: Counted " << i << " URLs"); | |
302 | ||
303 | min_u = urnFindMinRtt(urls, urnState->request->method, nullptr); | |
304 | qsort(urls, urlcnt, sizeof(*urls), url_entry_sort); | |
305 | e->buffer(); | |
306 | SBuf body; | |
307 | SBuf *mb = &body; // diff reduction hack; TODO: Remove | |
308 | mb->appendf( "<TITLE>Select URL for %s</TITLE>\n" | |
309 | "<STYLE type=\"text/css\"><!--BODY{background-color:#ffffff;font-family:verdana,sans-serif}--></STYLE>\n" | |
310 | "<H2>Select URL for %s</H2>\n" | |
311 | "<TABLE BORDER=\"0\" WIDTH=\"100%%\">\n", e->url(), e->url()); | |
312 | ||
313 | for (i = 0; i < urlcnt; ++i) { | |
314 | u = &urls[i]; | |
315 | debugs(52, 3, "URL {" << u->url << "}"); | |
316 | mb->appendf( | |
317 | "<TR><TD><A HREF=\"%s\">%s</A></TD>", u->url, u->url); | |
318 | ||
319 | if (urls[i].rtt > 0) | |
320 | mb->appendf( | |
321 | "<TD align=\"right\">%4d <it>ms</it></TD>", u->rtt); | |
322 | else | |
323 | mb->appendf("<TD align=\"right\">Unknown</TD>"); | |
324 | ||
325 | mb->appendf("<TD>%s</TD></TR>\n", u->flags.cached ? " [cached]" : " "); | |
326 | } | |
327 | ||
328 | mb->appendf( | |
329 | "</TABLE>" | |
330 | "<HR noshade size=\"1px\">\n" | |
331 | "<ADDRESS>\n" | |
332 | "Generated by %s@%s\n" | |
333 | "</ADDRESS>\n", | |
334 | visible_appname_string, getMyHostname()); | |
335 | const auto rep = new HttpReply; | |
336 | rep->setHeaders(Http::scFound, nullptr, "text/html", mb->length(), 0, squid_curtime); | |
337 | ||
338 | if (min_u) { | |
339 | rep->header.putStr(Http::HdrType::LOCATION, min_u->url); | |
340 | } | |
341 | ||
342 | rep->body.set(body); | |
343 | e->replaceHttpReply(rep); | |
344 | e->complete(); | |
345 | ||
346 | for (i = 0; i < urlcnt; ++i) { | |
347 | safe_free(urls[i].url); | |
348 | safe_free(urls[i].host); | |
349 | } | |
350 | ||
351 | safe_free(urls); | |
352 | ||
353 | delete urnState; | |
354 | } | |
355 | ||
356 | static url_entry * | |
357 | urnParseReply(const SBuf &inBuf, const HttpRequestMethod &m) | |
358 | { | |
359 | char *token; | |
360 | url_entry *list; | |
361 | url_entry *old; | |
362 | int n = 32; | |
363 | int i = 0; | |
364 | debugs(52, 3, "urnParseReply"); | |
365 | list = (url_entry *)xcalloc(n + 1, sizeof(*list)); | |
366 | ||
367 | // XXX: Switch to tokenizer-based parsing. | |
368 | const auto allocated = SBufToCstring(inBuf); | |
369 | ||
370 | auto buf = allocated; | |
371 | while (xisspace(*buf)) | |
372 | ++buf; | |
373 | ||
374 | for (token = strtok(buf, crlf); token; token = strtok(nullptr, crlf)) { | |
375 | debugs(52, 3, "urnParseReply: got '" << token << "'"); | |
376 | ||
377 | if (i == n) { | |
378 | old = list; | |
379 | n <<= 2; | |
380 | list = (url_entry *)xcalloc(n + 1, sizeof(*list)); | |
381 | memcpy(list, old, i * sizeof(*list)); | |
382 | safe_free(old); | |
383 | } | |
384 | ||
385 | AnyP::Uri uri; | |
386 | if (!uri.parse(m, SBuf(token)) || !*uri.host()) | |
387 | continue; | |
388 | ||
389 | #if USE_ICMP | |
390 | list[i].rtt = netdbHostRtt(uri.host()); | |
391 | ||
392 | if (0 == list[i].rtt) { | |
393 | debugs(52, 3, "Pinging " << uri.host()); | |
394 | netdbPingSite(uri.host()); | |
395 | } | |
396 | #else | |
397 | list[i].rtt = 0; | |
398 | #endif | |
399 | ||
400 | list[i].url = xstrdup(uri.absolute().c_str()); | |
401 | list[i].host = xstrdup(uri.host()); | |
402 | // TODO: Use storeHas() or lock/unlock entry to avoid creating unlocked | |
403 | // ones. | |
404 | list[i].flags.cached = storeGetPublic(list[i].url, m) ? 1 : 0; | |
405 | ++i; | |
406 | } | |
407 | ||
408 | debugs(52, 3, "urnParseReply: Found " << i << " URLs"); | |
409 | xfree(allocated); | |
410 | return list; | |
411 | } | |
412 |