citadel/modules/urldeshortener/serv_expand_shorter_urls.c

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