citadel/modules/urldeshortener/serv_expand_shorter_urls.c

   1
   2 /*
   3  *
   4  * Copyright (c) 1998-2009 by the citadel.org team
   5  *
   6  *  This program is free software; you can redistribute it and/or modify
   7  *  it under the terms of the GNU General Public License as published by
   8  *  the Free Software Foundation; either version 3 of the License, or
   9  *  (at your option) any later version.
  10  *
  11  *  This program is distributed in the hope that it will be useful,
  12  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  *  GNU General Public License for more details.
  15  *
  16  *  You should have received a copy of the GNU General Public License
  17  *  along with this program; if not, write to the Free Software
  18  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  */
  20
  21 #include "sysdep.h"
  22 #include <stdlib.h>
  23 #include <unistd.h>
  24 #include <stdio.h>
  25 #include <termios.h>
  26 #include <fcntl.h>
  27 #include <signal.h>
  28 #include <pwd.h>
  29 #include <errno.h>
  30 #include <sys/types.h>
  31 #include <syslog.h>
  32
  33 #if TIME_WITH_SYS_TIME
  34 # include <sys/time.h>
  35 # include <time.h>
  36 #else
  37 # if HAVE_SYS_TIME_H
  38 #  include <sys/time.h>
  39 # else
  40 #  include <time.h>
  41 # endif
  42 #endif
  43 #include <sys/wait.h>
  44 #include <ctype.h>
  45 #include <string.h>
  46 #include <limits.h>
  47 #include <sys/socket.h>
  48 #include <netinet/in.h>
  49 #include <arpa/inet.h>
  50 #include <assert.h>
  51
  52 #include <libcitadel.h>
  53 #include "citadel.h"
  54 #include "server.h"
  55 #include "citserver.h"
  56 #include "support.h"
  57 #include "config.h"
  58 #include "control.h"
  59 #include "user_ops.h"
  60 #include "database.h"
  61 #include "msgbase.h"
  62 #include "internet_addressing.h"
  63 #include "genstamp.h"
  64 #include "domain.h"
  65 #include "ctdl_module.h"
  66 #include "locate_host.h"
  67 #include "citadel_dirs.h"
  68
  69 #include "event_client.h"
  70
  71 HashList *UrlShorteners = NULL;
  72
  73 size_t GetLocationString( void *ptr, size_t size, size_t nmemb, void *userdata)
  74 {
  75 #define LOCATION "location"
  76         if (strncasecmp((char*)ptr, LOCATION, sizeof(LOCATION) - 1) == 0)
  77         {
  78                 StrBuf *pURL = (StrBuf*) userdata;
  79                 char *pch = (char*) ptr;
  80                 char *pche;
  81
  82                 pche = pch + (size * nmemb);
  83                 pch += sizeof(LOCATION);
  84
  85                 while (isspace(*pch) || (*pch == ':'))
  86                         pch ++;
  87
  88                 while (isspace(*pche) || (*pche == '\0'))
  89                         pche--;
  90
  91                 FlushStrBuf(pURL);
  92                 StrBufPlain(pURL, pch, pche - pch + 1);
  93         }
  94         return size * nmemb;
  95 }
  96
  97 eNextState TerminateLookupUrl(AsyncIO *IO)
  98 {
  99 //TOOD
 100         return eAbort;
 101 }
 102 eNextState LookupUrlResult(AsyncIO *IO)
 103 {
 104         return eTerminateConnection; /// /TODO
 105 }
 106
 107 int LookupUrl(StrBuf *ShorterUrlStr)
 108 {
 109         CURLcode sta;
 110         int rc = 0;
 111         CURL *chnd;
 112         AsyncIO *IO;
 113
 114
 115         IO = (AsyncIO*) malloc(sizeof(AsyncIO));
 116         memset(IO, 0, sizeof(AsyncIO));
 117         IO->CitContext = CloneContext(CC);
 118
 119         ParseURL(&IO->ConnectMe, ShorterUrlStr, 80);
 120         CurlPrepareURL(IO->ConnectMe);
 121         if (! evcurl_init(IO,
 122 //                        Ctx,
 123                           NULL,
 124                           "Citadel RSS ShorterURL Expander",
 125                           LookupUrlResult,
 126                           TerminateLookupUrl))
 127         {
 128                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 129                 goto shutdown;
 130         }
 131         chnd = IO->HttpReq.chnd;
 132
 133         OPT(SSL_VERIFYPEER, 0);
 134         OPT(SSL_VERIFYHOST, 0);
 135         OPT(FOLLOWLOCATION, 10);
 136 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 137         OPT(HTTP_CONTENT_DECODING, 1);
 138         OPT(ENCODING, "");
 139 #endif
 140         OPT(HEADERFUNCTION , GetLocationString);
 141         OPT(WRITEHEADER, ShorterUrlStr);
 142
 143
 144         if (server_shutting_down)
 145                 goto shutdown ;
 146
 147         QueueCurlContext(IO);
 148
 149 shutdown:
 150
 151         return rc;
 152
 153 }
 154
 155
 156
 157 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 158 {
 159         int nHits = 0;
 160         void *pv;
 161         int nShorter = 0;
 162         const char *pch;
 163         const char *pUrl;
 164         ConstStr *pCUrl;
 165
 166         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 167         {
 168                 nShorter++;
 169                 pch = ChrPtr(Message);
 170                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 171                 while ((pUrl != NULL) && (nHits < 99))
 172                 {
 173                         pCUrl = malloc(sizeof(ConstStr));
 174
 175                         pCUrl->Key = pUrl;
 176                         pch = pUrl + StrLength((StrBuf*)pv);
 177                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 178                                 pch++;
 179                         pCUrl->len = pch - pCUrl->Key;
 180
 181                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 182                         nHits ++;
 183                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 184                 }
 185         }
 186 }
 187
 188 int SortConstStrByPosition(const void *Item1, const void *Item2)
 189 {
 190         const ConstStr *p1, *p2;
 191         p1 = (const ConstStr*) Item1;
 192         p2 = (const ConstStr*) Item2;
 193         if (p1->Key == p2->Key)
 194                 return 0;
 195         if (p1->Key > p2->Key)
 196                 return 1;
 197         return -1;
 198 }
 199
 200 HashList *GetShorterUrls(StrBuf *Message)
 201 {
 202         HashList *pUrls;
 203         /* we just suspect URL shorteners to be inside of feeds from twitter
 204          * or other short content messages, so don't crawl through real blogs.
 205          */
 206         if (StrLength(Message) > 500)
 207                 return NULL;
 208
 209         pUrls = NewHash(1, Flathash);
 210         CrawlMessageForShorterUrls(pUrls, Message);
 211
 212         if (GetCount(pUrls) > 0)
 213                 return pUrls;
 214         else
 215                 return NULL;
 216
 217 }
 218
 219 void ExpandShortUrls(StrBuf *Message, HashList *pUrls, int Callback)
 220 {
 221         StrBuf *Shadow;
 222         ConstStr *pCUrl;
 223         const char *pch;
 224         const char *pche;
 225
 226         StrBuf *ShorterUrlStr;
 227         HashPos *Pos;
 228         const char *Key;
 229         void *pv;
 230         long len;
 231
 232         Shadow = NewStrBufPlain(NULL, StrLength(Message));
 233         SortByPayload (pUrls, SortConstStrByPosition);
 234
 235         ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 236
 237         pch = ChrPtr(Message);
 238         pche = pch + StrLength(Message);
 239         Pos = GetNewHashPos(pUrls, 1);
 240         while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 241         {
 242                 pCUrl = (ConstStr*) pv;
 243
 244                 if (pch != pCUrl->Key)
 245                         StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 246
 247                 StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 248                 if (LookupUrl(ShorterUrlStr))
 249                 {
 250                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 251                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 252                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 253                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 254                         StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 255                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 256                         StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 257                 }
 258                 else
 259                 {
 260                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 261                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 262                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 263                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 264                         StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 265                 }
 266                 pch = pCUrl->Key + pCUrl->len + 1;
 267
 268         }
 269         if (pch < pche)
 270                 StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 271         FlushStrBuf(Message);
 272         StrBufAppendBuf(Message, Shadow, 0);
 273
 274         FreeStrBuf(&ShorterUrlStr);
 275         FreeStrBuf(&Shadow);
 276         DeleteHashPos(&Pos);
 277
 278
 279         DeleteHash(&pUrls);
 280 }
 281
 282 void LoadUrlShorteners(void)
 283 {
 284         int i = 0;
 285         int fd;
 286         const char *POS = NULL;
 287         const char *Err = NULL;
 288         StrBuf *Content, *Line;
 289
 290
 291         UrlShorteners = NewHash(0, Flathash);
 292
 293         fd = open(file_citadel_urlshorteners, 0);
 294
 295         if (fd != 0)
 296         {
 297                 Content = NewStrBufPlain(NULL, SIZ);
 298                 Line = NewStrBuf();
 299                 while (POS != StrBufNOTNULL)
 300                 {
 301                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
 302                         StrBufTrim(Line);
 303                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
 304                         {
 305                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
 306                                 i++;
 307                                 Line = NewStrBuf();
 308                         }
 309                         else
 310                                 FlushStrBuf(Line);
 311                         if (POS == NULL)
 312                                 POS = StrBufNOTNULL;
 313                 }
 314                 FreeStrBuf(&Line);
 315                 FreeStrBuf(&Content);
 316         }
 317         close(fd);
 318 }
 319
 320 void shorter_url_cleanup(void)
 321 {
 322         DeleteHash(&UrlShorteners);
 323 }
 324
 325
 326 CTDL_MODULE_INIT(urldeshortener)
 327 {
 328         if (threading)
 329         {
 330                 syslog(LOG_INFO, "%s\n", curl_version());
 331         }
 332         else
 333         {
 334                 LoadUrlShorteners ();
 335                 CtdlRegisterCleanupHook(shorter_url_cleanup);
 336         }
 337         return "UrlShortener";
 338 }