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         return eAbort;
 100 }
 101 eNextState LookupUrlResult(AsyncIO *IO)
 102 {
 103         return eTerminateConnection; /// /TODO
 104 }
 105
 106 int LookupUrl(StrBuf *ShorterUrlStr)
 107 {
 108         CURLcode sta;
 109         int rc = 0;
 110         CURL *chnd;
 111         AsyncIO *IO;
 112
 113
 114         IO = (AsyncIO*) malloc(sizeof(AsyncIO));
 115         memset(IO, 0, sizeof(AsyncIO));
 116         IO->CitContext = CloneContext(CC);
 117
 118         ParseURL(&IO->ConnectMe, ShorterUrlStr, 80);
 119         CurlPrepareURL(IO->ConnectMe);
 120         if (! evcurl_init(IO,
 121 //                        Ctx,
 122                           NULL,
 123                           "Citadel RSS ShorterURL Expander",
 124                           LookupUrlResult,
 125                           TerminateLookupUrl))
 126         {
 127                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 128                 goto shutdown;
 129         }
 130         chnd = IO->HttpReq.chnd;
 131
 132         OPT(SSL_VERIFYPEER, 0);
 133         OPT(SSL_VERIFYHOST, 0);
 134         OPT(FOLLOWLOCATION, 10);
 135 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 136         OPT(HTTP_CONTENT_DECODING, 1);
 137         OPT(ENCODING, "");
 138 #endif
 139         OPT(HEADERFUNCTION , GetLocationString);
 140         OPT(WRITEHEADER, ShorterUrlStr);
 141
 142
 143         if (server_shutting_down)
 144                 goto shutdown ;
 145
 146         evcurl_handle_start(IO);
 147
 148 shutdown:
 149
 150         return rc;
 151
 152 }
 153
 154
 155
 156 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 157 {
 158         int nHits = 0;
 159         void *pv;
 160         int nShorter = 0;
 161         const char *pch;
 162         const char *pUrl;
 163         ConstStr *pCUrl;
 164
 165         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 166         {
 167                 nShorter++;
 168                 pch = ChrPtr(Message);
 169                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 170                 while ((pUrl != NULL) && (nHits < 99))
 171                 {
 172                         pCUrl = malloc(sizeof(ConstStr));
 173
 174                         pCUrl->Key = pUrl;
 175                         pch = pUrl + StrLength((StrBuf*)pv);
 176                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 177                                 pch++;
 178                         pCUrl->len = pch - pCUrl->Key;
 179
 180                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 181                         nHits ++;
 182                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 183                 }
 184         }
 185 }
 186
 187 int SortConstStrByPosition(const void *Item1, const void *Item2)
 188 {
 189         const ConstStr *p1, *p2;
 190         p1 = (const ConstStr*) Item1;
 191         p2 = (const ConstStr*) Item2;
 192         if (p1->Key == p2->Key)
 193                 return 0;
 194         if (p1->Key > p2->Key)
 195                 return 1;
 196         return -1;
 197 }
 198
 199 HashList *GetShorterUrls(StrBuf *Message)
 200 {
 201         HashList *pUrls;
 202         /* we just suspect URL shorteners to be inside of feeds from twitter
 203          * or other short content messages, so don't crawl through real blogs.
 204          */
 205         if (StrLength(Message) > 500)
 206                 return NULL;
 207
 208         pUrls = NewHash(1, Flathash);
 209         CrawlMessageForShorterUrls(pUrls, Message);
 210
 211         if (GetCount(pUrls) > 0)
 212                 return pUrls;
 213         else
 214                 return NULL;
 215
 216 }
 217
 218 void ExpandShortUrls(StrBuf *Message, HashList *pUrls, int Callback)
 219 {
 220         StrBuf *Shadow;
 221         ConstStr *pCUrl;
 222         const char *pch;
 223         const char *pche;
 224
 225         StrBuf *ShorterUrlStr;
 226         HashPos *Pos;
 227         const char *Key;
 228         void *pv;
 229         long len;
 230
 231         Shadow = NewStrBufPlain(NULL, StrLength(Message));
 232         SortByPayload (pUrls, SortConstStrByPosition);
 233
 234         ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 235
 236         pch = ChrPtr(Message);
 237         pche = pch + StrLength(Message);
 238         Pos = GetNewHashPos(pUrls, 1);
 239         while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 240         {
 241                 pCUrl = (ConstStr*) pv;
 242
 243                 if (pch != pCUrl->Key)
 244                         StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 245
 246                 StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 247                 if (LookupUrl(ShorterUrlStr))
 248                 {
 249                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 250                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 251                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 252                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 253                         StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 254                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 255                         StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 256                 }
 257                 else
 258                 {
 259                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 260                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 261                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 262                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 263                         StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 264                 }
 265                 pch = pCUrl->Key + pCUrl->len + 1;
 266
 267         }
 268         if (pch < pche)
 269                 StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 270         FlushStrBuf(Message);
 271         StrBufAppendBuf(Message, Shadow, 0);
 272
 273         FreeStrBuf(&ShorterUrlStr);
 274         FreeStrBuf(&Shadow);
 275         DeleteHashPos(&Pos);
 276
 277
 278         DeleteHash(&pUrls);
 279 }
 280
 281 void LoadUrlShorteners(void)
 282 {
 283         int i = 0;
 284         int fd;
 285         const char *POS = NULL;
 286         const char *Err = NULL;
 287         StrBuf *Content, *Line;
 288
 289
 290         UrlShorteners = NewHash(0, Flathash);
 291
 292         fd = open(file_citadel_urlshorteners, 0);
 293
 294         if (fd != 0)
 295         {
 296                 Content = NewStrBufPlain(NULL, SIZ);
 297                 Line = NewStrBuf();
 298                 while (POS != StrBufNOTNULL)
 299                 {
 300                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
 301                         StrBufTrim(Line);
 302                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
 303                         {
 304                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
 305                                 i++;
 306                                 Line = NewStrBuf();
 307                         }
 308                         else
 309                                 FlushStrBuf(Line);
 310                         if (POS == NULL)
 311                                 POS = StrBufNOTNULL;
 312                 }
 313                 FreeStrBuf(&Line);
 314                 FreeStrBuf(&Content);
 315         }
 316         close(fd);
 317 }
 318
 319 void shorter_url_cleanup(void)
 320 {
 321         DeleteHash(&UrlShorteners);
 322 }
 323
 324
 325 CTDL_MODULE_INIT(urldeshortener)
 326 {
 327         if (threading)
 328         {
 329                 syslog(LOG_INFO, "%s\n", curl_version());
 330         }
 331         else
 332         {
 333                 LoadUrlShorteners ();
 334                 CtdlRegisterCleanupHook(shorter_url_cleanup);
 335         }
 336         return "UrlShortener";
 337 }