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
  97 eNextState LookupUrlResult(AsyncIO *IO)
  98 {
  99         return eTerminateConnection; /// /TODO
 100 }
 101
 102 int LookupUrl(StrBuf *ShorterUrlStr)
 103 {
 104         CURLcode sta;
 105         int rc = 0;
 106         CURL *chnd;
 107         AsyncIO *IO;
 108
 109
 110         IO = (AsyncIO*) malloc(sizeof(AsyncIO));
 111         memset(IO, 0, sizeof(AsyncIO));
 112         IO->CitContext = CloneContext(CC);
 113
 114         ParseURL(&IO->ConnectMe, ShorterUrlStr, 80);
 115         CurlPrepareURL(IO->ConnectMe);
 116
 117         if (! evcurl_init(IO,
 118 //                        Ctx,
 119                           NULL,
 120                           "Citadel RSS ShorterURL Expander",
 121                           LookupUrlResult))
 122         {
 123                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 124                 goto shutdown;
 125         }
 126         chnd = IO->HttpReq.chnd;
 127
 128         OPT(SSL_VERIFYPEER, 0);
 129         OPT(SSL_VERIFYHOST, 0);
 130         OPT(FOLLOWLOCATION, 10);
 131 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 132         OPT(HTTP_CONTENT_DECODING, 1);
 133         OPT(ENCODING, "");
 134 #endif
 135         OPT(HEADERFUNCTION , GetLocationString);
 136         OPT(WRITEHEADER, ShorterUrlStr);
 137
 138
 139         if (CtdlThreadCheckStop())
 140                 goto shutdown ;
 141
 142         evcurl_handle_start(IO);
 143
 144 shutdown:
 145
 146         return rc;
 147
 148 }
 149
 150
 151
 152 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 153 {
 154         int nHits = 0;
 155         void *pv;
 156         int nShorter = 0;
 157         const char *pch;
 158         const char *pUrl;
 159         ConstStr *pCUrl;
 160
 161         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 162         {
 163                 nShorter++;
 164                 pch = ChrPtr(Message);
 165                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 166                 while ((pUrl != NULL) && (nHits < 99))
 167                 {
 168                         pCUrl = malloc(sizeof(ConstStr));
 169
 170                         pCUrl->Key = pUrl;
 171                         pch = pUrl + StrLength((StrBuf*)pv);
 172                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 173                                 pch++;
 174                         pCUrl->len = pch - pCUrl->Key;
 175
 176                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 177                         nHits ++;
 178                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 179                 }
 180         }
 181 }
 182
 183 int SortConstStrByPosition(const void *Item1, const void *Item2)
 184 {
 185         const ConstStr *p1, *p2;
 186         p1 = (const ConstStr*) Item1;
 187         p2 = (const ConstStr*) Item2;
 188         if (p1->Key == p2->Key)
 189                 return 0;
 190         if (p1->Key > p2->Key)
 191                 return 1;
 192         return -1;
 193 }
 194
 195 void ExpandShortUrls(StrBuf *Message)
 196 {
 197         StrBuf *Shadow;
 198         HashList *pUrls;
 199         ConstStr *pCUrl;
 200         const char *pch;
 201         const char *pche;
 202
 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;
 208
 209         pUrls = NewHash(1, Flathash);
 210         CrawlMessageForShorterUrls(pUrls, Message);
 211
 212         if (GetCount(pUrls) > 0)
 213         {
 214                 StrBuf *ShorterUrlStr;
 215                 HashPos *Pos;
 216                 const char *Key;
 217                 void *pv;
 218                 long len;
 219
 220                 Shadow = NewStrBufPlain(NULL, StrLength(Message));
 221                 SortByPayload (pUrls, SortConstStrByPosition);
 222
 223                 ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 224
 225                 pch = ChrPtr(Message);
 226                 pche = pch + StrLength(Message);
 227                 Pos = GetNewHashPos(pUrls, 1);
 228                 while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 229                 {
 230                         pCUrl = (ConstStr*) pv;
 231
 232                         if (pch != pCUrl->Key)
 233                                 StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 234
 235                         StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 236                         if (LookupUrl(ShorterUrlStr))
 237                         {
 238                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 239                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 240                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 241                                 StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 242                                 StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 243                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 244                                 StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 245                         }
 246                         else
 247                         {
 248                                 StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 249                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 250                                 StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 251                                 StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 252                                 StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 253                         }
 254                         pch = pCUrl->Key + pCUrl->len + 1;
 255
 256                 }
 257                 if (pch < pche)
 258                         StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 259                 FlushStrBuf(Message);
 260                 StrBufAppendBuf(Message, Shadow, 0);
 261
 262                 FreeStrBuf(&ShorterUrlStr);
 263                 FreeStrBuf(&Shadow);
 264                 DeleteHashPos(&Pos);
 265         }
 266
 267         DeleteHash(&pUrls);
 268 }
 269
 270 void LoadUrlShorteners(void)
 271 {
 272         int i = 0;
 273         int fd;
 274         const char *POS = NULL;
 275         const char *Err = NULL;
 276         StrBuf *Content, *Line;
 277
 278
 279         UrlShorteners = NewHash(0, Flathash);
 280
 281         fd = open(file_citadel_urlshorteners, 0);
 282
 283         if (fd != 0)
 284         {
 285                 Content = NewStrBufPlain(NULL, SIZ);
 286                 Line = NewStrBuf();
 287                 while (POS != StrBufNOTNULL)
 288                 {
 289                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
 290                         StrBufTrim(Line);
 291                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
 292                         {
 293                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
 294                                 i++;
 295                                 Line = NewStrBuf();
 296                         }
 297                         else
 298                                 FlushStrBuf(Line);
 299                         if (POS == NULL)
 300                                 POS = StrBufNOTNULL;
 301                 }
 302                 FreeStrBuf(&Line);
 303                 FreeStrBuf(&Content);
 304         }
 305         close(fd);
 306 }
 307
 308 void shorter_url_cleanup(void)
 309 {
 310         DeleteHash(&UrlShorteners);
 311 }
 312
 313
 314 CTDL_MODULE_INIT(urldeshortener)
 315 {
 316         if (threading)
 317         {
 318                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
 319         }
 320         else
 321         {
 322                 LoadUrlShorteners ();
 323                 CtdlRegisterCleanupHook(shorter_url_cleanup);
 324         }
 325         return "UrlShortener";
 326 }