citadel/modules/urldeshortener/serv_expand_shorter_urls.c

   1
   2 /*
   3  *
   4  * Copyright (c) 1998-2012 by the citadel.org team
   5  *
   6  *  This program is open source software; you can redistribute it and/or modify
   7  *  it under the terms of the GNU General Public License version 3.
   8  *
   9  *
  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  *
  17  *
  18  *
  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 eNextState ShutdownLookuUrl(AsyncIO *IO)
  97 {
  98 //TOOD
  99         return eAbort;
 100 }
 101 eNextState TerminateLookupUrl(AsyncIO *IO)
 102 {
 103 //TOOD
 104         return eAbort;
 105 }
 106 eNextState LookupUrlResult(AsyncIO *IO)
 107 {
 108         return eTerminateConnection; /// /TODO
 109 }
 110
 111 int LookupUrl(StrBuf *ShorterUrlStr)
 112 {
 113         CURLcode sta;
 114         int rc = 0;
 115         CURL *chnd;
 116         AsyncIO *IO;
 117
 118
 119         IO = (AsyncIO*) malloc(sizeof(AsyncIO));
 120         memset(IO, 0, sizeof(AsyncIO));
 121         IO->CitContext = CloneContext(CC);
 122
 123         ParseURL(&IO->ConnectMe, ShorterUrlStr, 80);
 124         CurlPrepareURL(IO->ConnectMe);
 125         if (! InitcURLIOStruct(IO,
 126 //                        Ctx,
 127                           NULL,
 128                           "Citadel RSS ShorterURL Expander",
 129                           LookupUrlResult,
 130                           TerminateLookupUrl,
 131                           ShutdownLookuUrl))
 132         {
 133                 syslog(LOG_ALERT, "Unable to initialize libcurl.\n");
 134                 goto shutdown;
 135         }
 136         chnd = IO->HttpReq.chnd;
 137
 138         OPT(SSL_VERIFYPEER, 0);
 139         OPT(SSL_VERIFYHOST, 0);
 140         OPT(FOLLOWLOCATION, 10);
 141 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 142         OPT(HTTP_CONTENT_DECODING, 1);
 143         OPT(ENCODING, "");
 144 #endif
 145         OPT(HEADERFUNCTION , GetLocationString);
 146         OPT(WRITEHEADER, ShorterUrlStr);
 147
 148
 149         if (server_shutting_down)
 150                 goto shutdown ;
 151
 152         QueueCurlContext(IO);
 153
 154 shutdown:
 155
 156         return rc;
 157
 158 }
 159
 160
 161
 162 void CrawlMessageForShorterUrls(HashList *pUrls, StrBuf *Message)
 163 {
 164         int nHits = 0;
 165         void *pv;
 166         int nShorter = 0;
 167         const char *pch;
 168         const char *pUrl;
 169         ConstStr *pCUrl;
 170
 171         while (GetHash(UrlShorteners, IKEY(nShorter), &pv))
 172         {
 173                 nShorter++;
 174                 pch = ChrPtr(Message);
 175                 pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 176                 while ((pUrl != NULL) && (nHits < 99))
 177                 {
 178                         pCUrl = malloc(sizeof(ConstStr));
 179
 180                         pCUrl->Key = pUrl;
 181                         pch = pUrl + StrLength((StrBuf*)pv);
 182                         while (isalnum(*pch)||(*pch == '-')||(*pch == '/'))
 183                                 pch++;
 184                         pCUrl->len = pch - pCUrl->Key;
 185
 186                         Put(pUrls, IKEY(nHits), pCUrl, NULL);
 187                         nHits ++;
 188                         pUrl = strstr(pch, ChrPtr((StrBuf*)pv));
 189                 }
 190         }
 191 }
 192
 193 int SortConstStrByPosition(const void *Item1, const void *Item2)
 194 {
 195         const ConstStr *p1, *p2;
 196         p1 = (const ConstStr*) Item1;
 197         p2 = (const ConstStr*) Item2;
 198         if (p1->Key == p2->Key)
 199                 return 0;
 200         if (p1->Key > p2->Key)
 201                 return 1;
 202         return -1;
 203 }
 204
 205 HashList *GetShorterUrls(StrBuf *Message)
 206 {
 207         HashList *pUrls;
 208         /* we just suspect URL shorteners to be inside of feeds from twitter
 209          * or other short content messages, so don't crawl through real blogs.
 210          */
 211         if (StrLength(Message) > 500)
 212                 return NULL;
 213
 214         pUrls = NewHash(1, Flathash);
 215         CrawlMessageForShorterUrls(pUrls, Message);
 216
 217         if (GetCount(pUrls) > 0)
 218                 return pUrls;
 219         else
 220                 return NULL;
 221
 222 }
 223
 224 void ExpandShortUrls(StrBuf *Message, HashList *pUrls, int Callback)
 225 {
 226         StrBuf *Shadow;
 227         ConstStr *pCUrl;
 228         const char *pch;
 229         const char *pche;
 230
 231         StrBuf *ShorterUrlStr;
 232         HashPos *Pos;
 233         const char *Key;
 234         void *pv;
 235         long len;
 236
 237         Shadow = NewStrBufPlain(NULL, StrLength(Message));
 238         SortByPayload (pUrls, SortConstStrByPosition);
 239
 240         ShorterUrlStr = NewStrBufPlain(NULL, StrLength(Message));
 241
 242         pch = ChrPtr(Message);
 243         pche = pch + StrLength(Message);
 244         Pos = GetNewHashPos(pUrls, 1);
 245         while (GetNextHashPos(pUrls, Pos, &len, &Key, &pv))
 246         {
 247                 pCUrl = (ConstStr*) pv;
 248
 249                 if (pch != pCUrl->Key)
 250                         StrBufAppendBufPlain(Shadow, pch, pCUrl->Key - pch, 0);
 251
 252                 StrBufPlain(ShorterUrlStr, CKEY(*pCUrl));
 253                 if (LookupUrl(ShorterUrlStr))
 254                 {
 255                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 256                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 257                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 258                         StrBufAppendBuf(Shadow, ShorterUrlStr, 0);
 259                         StrBufAppendBufPlain(Shadow, HKEY("["), 0);
 260                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 261                         StrBufAppendBufPlain(Shadow, HKEY("]</a>"), 0);
 262                 }
 263                 else
 264                 {
 265                         StrBufAppendBufPlain(Shadow, HKEY("<a href=\""), 0);
 266                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 267                         StrBufAppendBufPlain(Shadow, HKEY("\">"), 0);
 268                         StrBufAppendBufPlain(Shadow, pCUrl->Key, pCUrl->len, 0);
 269                         StrBufAppendBufPlain(Shadow, HKEY("</a>"), 0);
 270                 }
 271                 pch = pCUrl->Key + pCUrl->len + 1;
 272
 273         }
 274         if (pch < pche)
 275                 StrBufAppendBufPlain(Shadow, pch, pche - pch, 0);
 276         FlushStrBuf(Message);
 277         StrBufAppendBuf(Message, Shadow, 0);
 278
 279         FreeStrBuf(&ShorterUrlStr);
 280         FreeStrBuf(&Shadow);
 281         DeleteHashPos(&Pos);
 282
 283
 284         DeleteHash(&pUrls);
 285 }
 286
 287 void LoadUrlShorteners(void)
 288 {
 289         int i = 0;
 290         int fd;
 291         const char *POS = NULL;
 292         const char *Err = NULL;
 293         StrBuf *Content, *Line;
 294
 295
 296         UrlShorteners = NewHash(0, Flathash);
 297
 298         fd = open(file_citadel_urlshorteners, 0);
 299
 300         if (fd != 0)
 301         {
 302                 Content = NewStrBufPlain(NULL, SIZ);
 303                 Line = NewStrBuf();
 304                 while (POS != StrBufNOTNULL)
 305                 {
 306                         StrBufTCP_read_buffered_line_fast (Line, Content, &POS, &fd, 1, 1, &Err);
 307                         StrBufTrim(Line);
 308                         if ((*ChrPtr(Line) != '#') && (StrLength(Line) > 0))
 309                         {
 310                                 Put(UrlShorteners, IKEY(i), Line, HFreeStrBuf);
 311                                 i++;
 312                                 Line = NewStrBuf();
 313                         }
 314                         else
 315                                 FlushStrBuf(Line);
 316                         if (POS == NULL)
 317                                 POS = StrBufNOTNULL;
 318                 }
 319                 FreeStrBuf(&Line);
 320                 FreeStrBuf(&Content);
 321         }
 322         close(fd);
 323 }
 324
 325 void shorter_url_cleanup(void)
 326 {
 327         DeleteHash(&UrlShorteners);
 328 }
 329
 330
 331 CTDL_MODULE_INIT(urldeshortener)
 332 {
 333         if (threading)
 334         {
 335                 syslog(LOG_INFO, "%s\n", curl_version());
 336         }
 337         else
 338         {
 339                 LoadUrlShorteners ();
 340                 CtdlRegisterCleanupHook(shorter_url_cleanup);
 341         }
 342         return "UrlShortener";
 343 }