src/urn.cc

   1 /*
   2  * Copyright (C) 1996-2016 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 "cbdata.h"
  13 #include "errorpage.h"
  14 #include "FwdState.h"
  15 #include "globals.h"
  16 #include "HttpReply.h"
  17 #include "HttpRequest.h"
  18 #include "icmp/net_db.h"
  19 #include "MemBuf.h"
  20 #include "mime_header.h"
  21 #include "RequestFlags.h"
  22 #include "SquidTime.h"
  23 #include "Store.h"
  24 #include "StoreClient.h"
  25 #include "tools.h"
  26 #include "URL.h"
  27 #include "urn.h"
  28
  29 #define URN_REQBUF_SZ   4096
  30
  31 class UrnState : public StoreClient
  32 {
  33     CBDATA_CLASS(UrnState);
  34
  35 public:
  36     void created (StoreEntry *newEntry);
  37     void start (HttpRequest *, StoreEntry *);
  38     char *getHost(const SBuf &urlpath);
  39     void setUriResFromRequest(HttpRequest *);
  40
  41     virtual ~UrnState();
  42
  43     StoreEntry *entry;
  44     store_client *sc;
  45     StoreEntry *urlres_e;
  46     HttpRequest::Pointer request;
  47     HttpRequest::Pointer urlres_r;
  48
  49     struct {
  50         bool force_menu;
  51     } flags;
  52     char reqbuf[URN_REQBUF_SZ];
  53     int reqofs;
  54
  55 private:
  56     char *urlres;
  57 };
  58
  59 typedef struct {
  60     char *url;
  61     char *host;
  62     int rtt;
  63
  64     struct {
  65         int cached;
  66     } flags;
  67 } url_entry;
  68
  69 static STCB urnHandleReply;
  70 static url_entry *urnParseReply(const char *inbuf, const HttpRequestMethod&);
  71 static const char *const crlf = "\r\n";
  72
  73 CBDATA_CLASS_INIT(UrnState);
  74
  75 UrnState::~UrnState()
  76 {
  77     safe_free(urlres);
  78 }
  79
  80 static url_entry *
  81 urnFindMinRtt(url_entry * urls, const HttpRequestMethod &, int *rtt_ret)
  82 {
  83     int min_rtt = 0;
  84     url_entry *u = NULL;
  85     url_entry *min_u = NULL;
  86     int i;
  87     int urlcnt = 0;
  88     debugs(52, 3, "urnFindMinRtt");
  89     assert(urls != NULL);
  90
  91     for (i = 0; NULL != urls[i].url; ++i)
  92         ++urlcnt;
  93
  94     debugs(53, 3, "urnFindMinRtt: Counted " << i << " URLs");
  95
  96     if (1 == urlcnt) {
  97         debugs(52, 3, "urnFindMinRtt: Only one URL - return it!");
  98         return urls;
  99     }
 100
 101     for (i = 0; i < urlcnt; ++i) {
 102         u = &urls[i];
 103         debugs(52, 3, "urnFindMinRtt: " << u->host << " rtt=" << u->rtt);
 104
 105         if (u->rtt == 0)
 106             continue;
 107
 108         if (u->rtt > min_rtt && min_rtt != 0)
 109             continue;
 110
 111         min_rtt = u->rtt;
 112
 113         min_u = u;
 114     }
 115
 116     if (rtt_ret)
 117         *rtt_ret = min_rtt;
 118
 119     debugs(52, DBG_IMPORTANT, "urnFindMinRtt: Returning '" <<
 120            (min_u ? min_u->url : "NONE") << "' RTT " <<
 121            min_rtt  );
 122
 123     return min_u;
 124 }
 125
 126 char *
 127 UrnState::getHost(const SBuf &urlpath)
 128 {
 129     /** FIXME: this appears to be parsing the URL. *very* badly. */
 130     /*   a proper encapsulated URI/URL type needs to clear this up. */
 131     size_t p;
 132     if ((p = urlpath.find(':')) != SBuf::npos)
 133         return SBufToCstring(urlpath.substr(0, p-1));
 134
 135     return SBufToCstring(urlpath);
 136 }
 137
 138 void
 139 UrnState::setUriResFromRequest(HttpRequest *r)
 140 {
 141     static const SBuf menu(".menu");
 142     if (r->url.path().startsWith(menu)) {
 143         r->url.path(r->url.path().substr(5)); // strip prefix "menu."
 144         flags.force_menu = true;
 145     }
 146
 147     SBuf uri = r->url.path();
 148     LOCAL_ARRAY(char, local_urlres, 4096);
 149     char *host = getHost(uri);
 150     snprintf(local_urlres, 4096, "http://%s/uri-res/N2L?urn:" SQUIDSBUFPH, host, SQUIDSBUFPRINT(uri));
 151     safe_free(host);
 152     safe_free(urlres);
 153     urlres = xstrdup(local_urlres);
 154     urlres_r = HttpRequest::CreateFromUrl(urlres);
 155
 156     if (urlres_r == NULL) {
 157         debugs(52, 3, "urnStart: Bad uri-res URL " << urlres);
 158         ErrorState *err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, r);
 159         err->url = urlres;
 160         urlres = NULL;
 161         errorAppendEntry(entry, err);
 162         return;
 163     }
 164
 165     urlres_r->header.putStr(Http::HdrType::ACCEPT, "text/plain");
 166 }
 167
 168 void
 169 UrnState::start(HttpRequest * r, StoreEntry * e)
 170 {
 171     debugs(52, 3, "urnStart: '" << e->url() << "'" );
 172     entry = e;
 173     request = r;
 174
 175     entry->lock("UrnState::start");
 176     setUriResFromRequest(r);
 177
 178     if (urlres_r == NULL)
 179         return;
 180
 181     StoreEntry::getPublic (this, urlres, Http::METHOD_GET);
 182 }
 183
 184 void
 185 UrnState::created(StoreEntry *newEntry)
 186 {
 187     urlres_e = newEntry;
 188
 189     if (urlres_e->isNull()) {
 190         urlres_e = storeCreateEntry(urlres, urlres, RequestFlags(), Http::METHOD_GET);
 191         sc = storeClientListAdd(urlres_e, this);
 192         FwdState::fwdStart(Comm::ConnectionPointer(), urlres_e, urlres_r.getRaw());
 193     } else {
 194         urlres_e->lock("UrnState::created");
 195         sc = storeClientListAdd(urlres_e, this);
 196     }
 197
 198     reqofs = 0;
 199     StoreIOBuffer tempBuffer;
 200     tempBuffer.offset = reqofs;
 201     tempBuffer.length = URN_REQBUF_SZ;
 202     tempBuffer.data = reqbuf;
 203     storeClientCopy(sc, urlres_e,
 204                     tempBuffer,
 205                     urnHandleReply,
 206                     this);
 207 }
 208
 209 void
 210 urnStart(HttpRequest * r, StoreEntry * e)
 211 {
 212     UrnState *anUrn = new UrnState();
 213     anUrn->start (r, e);
 214 }
 215
 216 static int
 217 url_entry_sort(const void *A, const void *B)
 218 {
 219     const url_entry *u1 = (const url_entry *)A;
 220     const url_entry *u2 = (const url_entry *)B;
 221
 222     if (u2->rtt == u1->rtt)
 223         return 0;
 224     else if (0 == u1->rtt)
 225         return 1;
 226     else if (0 == u2->rtt)
 227         return -1;
 228     else
 229         return u1->rtt - u2->rtt;
 230 }
 231
 232 static void
 233 urnHandleReplyError(UrnState *urnState, StoreEntry *urlres_e)
 234 {
 235     urlres_e->unlock("urnHandleReplyError+res");
 236     urnState->entry->unlock("urnHandleReplyError+prime");
 237     delete urnState;
 238 }
 239
 240 /* TODO: use the clientStream support for this */
 241 static void
 242 urnHandleReply(void *data, StoreIOBuffer result)
 243 {
 244     UrnState *urnState = static_cast<UrnState *>(data);
 245     StoreEntry *e = urnState->entry;
 246     StoreEntry *urlres_e = urnState->urlres_e;
 247     char *s = NULL;
 248     size_t k;
 249     HttpReply *rep;
 250     url_entry *urls;
 251     url_entry *u;
 252     url_entry *min_u;
 253     MemBuf *mb = NULL;
 254     ErrorState *err;
 255     int i;
 256     int urlcnt = 0;
 257     char *buf = urnState->reqbuf;
 258     StoreIOBuffer tempBuffer;
 259
 260     debugs(52, 3, "urnHandleReply: Called with size=" << result.length << ".");
 261
 262     if (EBIT_TEST(urlres_e->flags, ENTRY_ABORTED) || result.length == 0 || result.flags.error) {
 263         urnHandleReplyError(urnState, urlres_e);
 264         return;
 265     }
 266
 267     /* Update reqofs to point to where in the buffer we'd be */
 268     urnState->reqofs += result.length;
 269
 270     /* Handle reqofs being bigger than normal */
 271     if (urnState->reqofs >= URN_REQBUF_SZ) {
 272         urnHandleReplyError(urnState, urlres_e);
 273         return;
 274     }
 275
 276     /* If we haven't received the entire object (urn), copy more */
 277     if (urlres_e->store_status == STORE_PENDING &&
 278             urnState->reqofs < URN_REQBUF_SZ) {
 279         tempBuffer.offset = urnState->reqofs;
 280         tempBuffer.length = URN_REQBUF_SZ;
 281         tempBuffer.data = urnState->reqbuf + urnState->reqofs;
 282         storeClientCopy(urnState->sc, urlres_e,
 283                         tempBuffer,
 284                         urnHandleReply,
 285                         urnState);
 286         return;
 287     }
 288
 289     /* we know its STORE_OK */
 290     k = headersEnd(buf, urnState->reqofs);
 291
 292     if (0 == k) {
 293         debugs(52, DBG_IMPORTANT, "urnHandleReply: didn't find end-of-headers for " << e->url()  );
 294         urnHandleReplyError(urnState, urlres_e);
 295         return;
 296     }
 297
 298     s = buf + k;
 299     assert(urlres_e->getReply());
 300     rep = new HttpReply;
 301     rep->parseCharBuf(buf, k);
 302     debugs(52, 3, "reply exists, code=" << rep->sline.status() << ".");
 303
 304     if (rep->sline.status() != Http::scOkay) {
 305         debugs(52, 3, "urnHandleReply: failed.");
 306         err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, urnState->request.getRaw());
 307         err->url = xstrdup(e->url());
 308         errorAppendEntry(e, err);
 309         delete rep;
 310         urnHandleReplyError(urnState, urlres_e);
 311         return;
 312     }
 313
 314     delete rep;
 315
 316     while (xisspace(*s))
 317         ++s;
 318
 319     urls = urnParseReply(s, urnState->request->method);
 320
 321     if (!urls) {     /* unknown URN error */
 322         debugs(52, 3, "urnTranslateDone: unknown URN " << e->url());
 323         err = new ErrorState(ERR_URN_RESOLVE, Http::scNotFound, urnState->request.getRaw());
 324         err->url = xstrdup(e->url());
 325         errorAppendEntry(e, err);
 326         urnHandleReplyError(urnState, urlres_e);
 327         return;
 328     }
 329
 330     for (i = 0; urls[i].url; ++i)
 331         ++urlcnt;
 332
 333     debugs(53, 3, "urnFindMinRtt: Counted " << i << " URLs");
 334
 335     min_u = urnFindMinRtt(urls, urnState->request->method, NULL);
 336     qsort(urls, urlcnt, sizeof(*urls), url_entry_sort);
 337     e->buffer();
 338     mb = new MemBuf;
 339     mb->init();
 340     mb->appendf( "<TITLE>Select URL for %s</TITLE>\n"
 341                  "<STYLE type=\"text/css\"><!--BODY{background-color:#ffffff;font-family:verdana,sans-serif}--></STYLE>\n"
 342                  "<H2>Select URL for %s</H2>\n"
 343                  "<TABLE BORDER=\"0\" WIDTH=\"100%%\">\n", e->url(), e->url());
 344
 345     for (i = 0; i < urlcnt; ++i) {
 346         u = &urls[i];
 347         debugs(52, 3, "URL {" << u->url << "}");
 348         mb->appendf(
 349             "<TR><TD><A HREF=\"%s\">%s</A></TD>", u->url, u->url);
 350
 351         if (urls[i].rtt > 0)
 352             mb->appendf(
 353                 "<TD align=\"right\">%4d <it>ms</it></TD>", u->rtt);
 354         else
 355             mb->appendf("<TD align=\"right\">Unknown</TD>");
 356
 357         mb->appendf("<TD>%s</TD></TR>\n", u->flags.cached ? "    [cached]" : " ");
 358     }
 359
 360     mb->appendf(
 361         "</TABLE>"
 362         "<HR noshade size=\"1px\">\n"
 363         "<ADDRESS>\n"
 364         "Generated by %s@%s\n"
 365         "</ADDRESS>\n",
 366         APP_FULLNAME, getMyHostname());
 367     rep = new HttpReply;
 368     rep->setHeaders(Http::scFound, NULL, "text/html", mb->contentSize(), 0, squid_curtime);
 369
 370     if (urnState->flags.force_menu) {
 371         debugs(51, 3, "urnHandleReply: forcing menu");
 372     } else if (min_u) {
 373         rep->header.putStr(Http::HdrType::LOCATION, min_u->url);
 374     }
 375
 376     rep->body.setMb(mb);
 377     /* don't clean or delete mb; rep->body owns it now */
 378     e->replaceHttpReply(rep);
 379     e->complete();
 380
 381     for (i = 0; i < urlcnt; ++i) {
 382         safe_free(urls[i].url);
 383         safe_free(urls[i].host);
 384     }
 385
 386     safe_free(urls);
 387     /* mb was absorbed in httpBodySet call, so we must not clean it */
 388     storeUnregister(urnState->sc, urlres_e, urnState);
 389
 390     urnHandleReplyError(urnState, urlres_e);
 391 }
 392
 393 static url_entry *
 394 urnParseReply(const char *inbuf, const HttpRequestMethod& m)
 395 {
 396     char *buf = xstrdup(inbuf);
 397     char *token;
 398     char *url;
 399     char *host;
 400     url_entry *list;
 401     url_entry *old;
 402     int n = 32;
 403     int i = 0;
 404     debugs(52, 3, "urnParseReply");
 405     list = (url_entry *)xcalloc(n + 1, sizeof(*list));
 406
 407     for (token = strtok(buf, crlf); token; token = strtok(NULL, crlf)) {
 408         debugs(52, 3, "urnParseReply: got '" << token << "'");
 409
 410         if (i == n) {
 411             old = list;
 412             n <<= 2;
 413             list = (url_entry *)xcalloc(n + 1, sizeof(*list));
 414             memcpy(list, old, i * sizeof(*list));
 415             safe_free(old);
 416         }
 417
 418         url = xstrdup(token);
 419         host = urlHostname(url);
 420
 421         if (NULL == host)
 422             continue;
 423
 424 #if USE_ICMP
 425         list[i].rtt = netdbHostRtt(host);
 426
 427         if (0 == list[i].rtt) {
 428             debugs(52, 3, "urnParseReply: Pinging " << host);
 429             netdbPingSite(host);
 430         }
 431 #else
 432         list[i].rtt = 0;
 433 #endif
 434
 435         list[i].url = url;
 436         list[i].host = xstrdup(host);
 437         // TODO: Use storeHas() or lock/unlock entry to avoid creating unlocked
 438         // ones.
 439         list[i].flags.cached = storeGetPublic(url, m) ? 1 : 0;
 440         ++i;
 441     }
 442
 443     debugs(52, 3, "urnParseReply: Found " << i << " URLs");
 444     return list;
 445 }
 446