citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * $Id$
   3  *
   4  * Bring external RSS feeds into rooms.
   5  *
   6  * Copyright (c) 2007-2009 by the citadel.org team
   7  *
   8  *  This program is free software; you can redistribute it and/or modify
   9  *  it under the terms of the GNU General Public License as published by
  10  *  the Free Software Foundation; either version 3 of the License, or
  11  *  (at your option) any later version.
  12  *
  13  *  This program is distributed in the hope that it will be useful,
  14  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  *  GNU General Public License for more details.
  17  *
  18  *  You should have received a copy of the GNU General Public License
  19  *  along with this program; if not, write to the Free Software
  20  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  */
  22
  23 #include <stdlib.h>
  24 #include <unistd.h>
  25 #include <stdio.h>
  26
  27 #if TIME_WITH_SYS_TIME
  28 # include <sys/time.h>
  29 # include <time.h>
  30 #else
  31 # if HAVE_SYS_TIME_H
  32 #  include <sys/time.h>
  33 # else
  34 #  include <time.h>
  35 # endif
  36 #endif
  37
  38 #include <ctype.h>
  39 #include <string.h>
  40 #include <errno.h>
  41 #include <sys/types.h>
  42 #include <sys/stat.h>
  43 #include <expat.h>
  44 #include <curl/curl.h>
  45 #include <libcitadel.h>
  46 #include "citadel.h"
  47 #include "server.h"
  48 #include "citserver.h"
  49 #include "support.h"
  50 #include "config.h"
  51 #include "threads.h"
  52 #include "room_ops.h"
  53 #include "ctdl_module.h"
  54 #include "clientsocket.h"
  55 #include "msgbase.h"
  56 #include "parsedate.h"
  57 #include "database.h"
  58 #include "citadel_dirs.h"
  59 #include "md5.h"
  60
  61
  62 struct rssnetcfg {
  63         struct rssnetcfg *next;
  64         char url[256];
  65         char *rooms;
  66 };
  67
  68 struct rss_item {
  69         char *chardata;
  70         int chardata_len;
  71         char *roomlist;
  72         int done_parsing;
  73         char *guid;
  74         char *title;
  75         char *link;
  76         char *description;
  77         time_t pubdate;
  78         char channel_title[256];
  79         int item_tag_nesting;
  80         char *author_or_creator;
  81 };
  82
  83 struct rssnetcfg *rnclist = NULL;
  84
  85
  86 /*
  87  * Commit a fetched and parsed RSS item to disk
  88  */
  89 void rss_save_item(struct rss_item *ri) {
  90
  91         struct MD5Context md5context;
  92         u_char rawdigest[MD5_DIGEST_LEN];
  93         int i;
  94         char utmsgid[SIZ];
  95         struct cdbdata *cdbut;
  96         struct UseTable ut;
  97         struct CtdlMessage *msg;
  98         struct recptypes *recp = NULL;
  99         int msglen = 0;
 100
 101         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
 102         if (recp == NULL) return;
 103         memset(recp, 0, sizeof(struct recptypes));
 104         recp->recp_room = strdup(ri->roomlist);
 105         recp->num_room = num_tokens(ri->roomlist, '|');
 106         recp->recptypes_magic = RECPTYPES_MAGIC;
 107
 108         /* Construct a GUID to use in the S_USETABLE table.
 109          * If one is not present in the item itself, make one up.
 110          */
 111         if (ri->guid != NULL) {
 112                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
 113         }
 114         else {
 115                 MD5Init(&md5context);
 116                 if (ri->title != NULL) {
 117                         MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
 118                 }
 119                 if (ri->link != NULL) {
 120                         MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
 121                 }
 122                 MD5Final(rawdigest, &md5context);
 123                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 124                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 125                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 126                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 127                 }
 128                 strcat(utmsgid, "_rss2ctdl");
 129         }
 130
 131         /* Find out if we've already seen this item */
 132         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 133         if (cdbut != NULL) {
 134                 /* Item has already been seen */
 135                 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
 136                 cdb_free(cdbut);
 137
 138                 /* rewrite the record anyway, to update the timestamp */
 139                 strcpy(ut.ut_msgid, utmsgid);
 140                 ut.ut_timestamp = time(NULL);
 141                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 142         }
 143         else {
 144                 /* Item has not been seen, so save it. */
 145
 146                 if (ri->description == NULL) ri->description = strdup("");
 147                 for (i=strlen(ri->description); i>=0; --i) {
 148                         if (isspace(ri->description[i])) {
 149                                 ri->description[i] = ' ';
 150                         }
 151                 }
 152
 153                 msg = malloc(sizeof(struct CtdlMessage));
 154                 memset(msg, 0, sizeof(struct CtdlMessage));
 155                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 156                 msg->cm_anon_type = MES_NORMAL;
 157                 msg->cm_format_type = FMT_RFC822;
 158
 159                 if (ri->author_or_creator != NULL) {
 160                         msg->cm_fields['A'] = html_to_ascii(ri->author_or_creator,
 161                                 strlen(ri->author_or_creator), 512, 0);
 162                         striplt(msg->cm_fields['A']);
 163                 }
 164                 else {
 165                         msg->cm_fields['A'] = strdup("rss");
 166                 }
 167
 168                 msg->cm_fields['N'] = strdup(NODENAME);
 169                 if (ri->title != NULL) {
 170                         msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
 171                         striplt(msg->cm_fields['U']);
 172                 }
 173                 msg->cm_fields['T'] = malloc(64);
 174                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 175                 if (ri->channel_title != NULL) {
 176                         if (!IsEmptyStr(ri->channel_title)) {
 177                                 msg->cm_fields['O'] = strdup(ri->channel_title);
 178                         }
 179                 }
 180                 if (ri->link == NULL)
 181                         ri->link = strdup("");
 182                 msglen += 1024 + strlen(ri->link) + strlen(ri->description) ;
 183                 msg->cm_fields['M'] = malloc(msglen);
 184                 snprintf(msg->cm_fields['M'], msglen,
 185                         "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 186                         "<html><body>\n"
 187                         "%s<br><br>\n"
 188                         "<a href=\"%s\">%s</a>\n"
 189                         "</body></html>\n"
 190                         ,
 191                         ri->description,
 192                         ri->link, ri->link
 193                 );
 194
 195                 CtdlSubmitMsg(msg, recp, NULL, 0);
 196                 CtdlFreeMessage(msg);
 197
 198                 /* write the uidl to the use table so we don't store this item again */
 199                 strcpy(ut.ut_msgid, utmsgid);
 200                 ut.ut_timestamp = time(NULL);
 201                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 202         }
 203         free_recipients(recp);
 204 }
 205
 206
 207
 208 /*
 209  * Convert an RDF/RSS datestamp into a time_t
 210  */
 211 time_t rdf_parsedate(char *p)
 212 {
 213         struct tm tm;
 214         time_t t = 0;
 215
 216         if (!p) return 0L;
 217         if (strlen(p) < 10) return 0L;
 218
 219         memset(&tm, 0, sizeof tm);
 220
 221         /* YYYY-MM-DDTHH:MM format...
 222          */
 223         if ( (p[4] == '-') && (p[7] == '-') ) {
 224                 tm.tm_year = atoi(&p[0]) - 1900;
 225                 tm.tm_mon = atoi(&p[5]) - 1;
 226                 tm.tm_mday = atoi(&p[8]);
 227                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 228                         tm.tm_hour = atoi(&p[11]);
 229                         tm.tm_min = atoi(&p[14]);
 230                 }
 231                 return mktime(&tm);
 232         }
 233
 234         /* hmm... try RFC822 date stamp format */
 235
 236         t = parsedate(p);
 237         if (t > 0) return(t);
 238
 239         /* yeesh.  ok, just return the current date and time. */
 240         return(time(NULL));
 241 }
 242
 243
 244
 245 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
 246         struct rss_item *ri = (struct rss_item *) data;
 247         char el[256];
 248         char *sep = NULL;
 249
 250         /* Axe the namespace, we don't care about it */
 251         safestrncpy(el, supplied_el, sizeof el);
 252         while (sep = strchr(el, ':'), sep) {
 253                 strcpy(el, ++sep);
 254         }
 255
 256         if (!strcasecmp(el, "item")) {
 257                 ++ri->item_tag_nesting;
 258
 259                 /* Initialize the feed item data structure */
 260                 if (ri->guid != NULL) free(ri->guid);
 261                 ri->guid = NULL;
 262                 if (ri->title != NULL) free(ri->title);
 263                 ri->title = NULL;
 264                 if (ri->link != NULL) free(ri->link);
 265                 ri->link = NULL;
 266                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 267                 ri->author_or_creator = NULL;
 268                 if (ri->description != NULL) free(ri->description);
 269                 ri->description = NULL;
 270
 271                 /* Throw away any existing character data */
 272                 if (ri->chardata_len > 0) {
 273                         free(ri->chardata);
 274                         ri->chardata = 0;
 275                         ri->chardata_len = 0;
 276                 }
 277         }
 278
 279
 280
 281 }
 282
 283 void rss_xml_end(void *data, const char *supplied_el) {
 284         struct rss_item *ri = (struct rss_item *) data;
 285         char el[256];
 286         char *sep = NULL;
 287
 288         /* Axe the namespace, we don't care about it */
 289         safestrncpy(el, supplied_el, sizeof el);
 290         while (sep = strchr(el, ':'), sep) {
 291                 strcpy(el, ++sep);
 292         }
 293
 294         if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
 295                 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
 296                 striplt(ri->channel_title);
 297         }
 298
 299         if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
 300                 if (ri->guid != NULL) free(ri->guid);
 301                 striplt(ri->chardata);
 302                 ri->guid = strdup(ri->chardata);
 303         }
 304
 305         if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
 306                 if (ri->title != NULL) free(ri->title);
 307                 striplt(ri->chardata);
 308                 ri->title = strdup(ri->chardata);
 309         }
 310
 311         if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
 312                 if (ri->link != NULL) free(ri->link);
 313                 striplt(ri->chardata);
 314                 ri->link = strdup(ri->chardata);
 315         }
 316
 317         if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
 318                 if (ri->description != NULL) free(ri->description);
 319                 ri->description = strdup(ri->chardata);
 320         }
 321
 322         if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
 323                 striplt(ri->chardata);
 324                 ri->pubdate = rdf_parsedate(ri->chardata);
 325         }
 326
 327         if ( ((!strcasecmp(el, "author")) || (!strcasecmp(el, "creator"))) && (ri->chardata != NULL) ) {
 328                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 329                 striplt(ri->chardata);
 330                 ri->author_or_creator = strdup(ri->chardata);
 331         }
 332
 333         if (!strcasecmp(el, "item")) {
 334                 --ri->item_tag_nesting;
 335                 rss_save_item(ri);
 336         }
 337
 338         if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
 339                 CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 340                 ri->done_parsing = 1;
 341         }
 342
 343         if (ri->chardata_len > 0) {
 344                 free(ri->chardata);
 345                 ri->chardata = 0;
 346                 ri->chardata_len = 0;
 347         }
 348
 349 }
 350
 351
 352 /*
 353  * This callback stores up the data which appears in between tags.
 354  */
 355 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
 356         struct rss_item *ri = (struct rss_item *) data;
 357         int old_len;
 358         int new_len;
 359         char *new_buffer;
 360
 361         old_len = ri->chardata_len;
 362         new_len = old_len + len;
 363         new_buffer = realloc(ri->chardata, new_len + 1);
 364         if (new_buffer != NULL) {
 365                 memcpy(&new_buffer[old_len], s, len);
 366                 new_buffer[new_len] = 0;
 367                 ri->chardata = new_buffer;
 368                 ri->chardata_len = new_len;
 369         }
 370 }
 371
 372
 373
 374 /*
 375  * Callback function for passing libcurl's output to expat for parsing
 376  */
 377 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
 378 {
 379         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
 380         return (size*nmemb);
 381 }
 382
 383
 384
 385 /*
 386  * Begin a feed parse
 387  */
 388 void rss_do_fetching(char *url, char *rooms) {
 389         struct rss_item ri;
 390         XML_Parser xp;
 391
 392         CURL *curl;
 393         CURLcode res;
 394         char errmsg[1024] = "";
 395
 396         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
 397
 398         curl = curl_easy_init();
 399         if (!curl) {
 400                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 401                 return;
 402         }
 403
 404         xp = XML_ParserCreateNS("UTF-8", ':');
 405         if (!xp) {
 406                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
 407                 curl_easy_cleanup(curl);
 408                 return;
 409         }
 410
 411         curl_easy_setopt(curl, CURLOPT_URL, url);
 412         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 413         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 414         curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
 415         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 416         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 417         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 418 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 419         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
 420         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
 421 #endif
 422         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 423         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 424         if (!IsEmptyStr(config.c_ip_addr)) {
 425                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 426         }
 427
 428         memset(&ri, 0, sizeof(struct rss_item));
 429         ri.roomlist = rooms;
 430         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
 431         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
 432         XML_SetUserData(xp, &ri);
 433
 434         if (CtdlThreadCheckStop())
 435         {
 436                 XML_ParserFree(xp);
 437                 curl_easy_cleanup(curl);
 438                 return;
 439         }
 440
 441         if (CtdlThreadCheckStop())
 442                 goto shutdown ;
 443
 444         res = curl_easy_perform(curl);
 445         if (res) {
 446                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
 447         }
 448
 449         if (CtdlThreadCheckStop())
 450                 goto shutdown ;
 451
 452         if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
 453
 454
 455 shutdown:
 456         curl_easy_cleanup(curl);
 457         XML_ParserFree(xp);
 458
 459         /* Free the feed item data structure */
 460         if (ri.guid != NULL) free(ri.guid);
 461         ri.guid = NULL;
 462         if (ri.title != NULL) free(ri.title);
 463         ri.title = NULL;
 464         if (ri.link != NULL) free(ri.link);
 465         ri.link = NULL;
 466         if (ri.author_or_creator != NULL) free(ri.author_or_creator);
 467         ri.author_or_creator = NULL;
 468         if (ri.description != NULL) free(ri.description);
 469         ri.description = NULL;
 470         if (ri.chardata_len > 0) {
 471                 free(ri.chardata);
 472                 ri.chardata = 0;
 473                 ri.chardata_len = 0;
 474         }
 475 }
 476
 477
 478 /*
 479  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 480  */
 481 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 482 {
 483         char filename[PATH_MAX];
 484         char buf[1024];
 485         char instr[32];
 486         FILE *fp;
 487         char feedurl[256];
 488         struct rssnetcfg *rncptr = NULL;
 489         struct rssnetcfg *use_this_rncptr = NULL;
 490         int len = 0;
 491         char *ptr = NULL;
 492
 493         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
 494
 495         if (CtdlThreadCheckStop())
 496                 return;
 497
 498         /* Only do net processing for rooms that have netconfigs */
 499         fp = fopen(filename, "r");
 500         if (fp == NULL) {
 501                 return;
 502         }
 503
 504         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
 505                 buf[strlen(buf)-1] = 0;
 506
 507                 extract_token(instr, buf, 0, '|', sizeof instr);
 508                 if (!strcasecmp(instr, "rssclient")) {
 509
 510                         use_this_rncptr = NULL;
 511
 512                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
 513
 514                         /* If any other rooms have requested the same feed, then we will just add this
 515                          * room to the target list for that client request.
 516                          */
 517                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
 518                                 if (!strcmp(rncptr->url, feedurl)) {
 519                                         use_this_rncptr = rncptr;
 520                                 }
 521                         }
 522
 523                         /* Otherwise create a new client request */
 524                         if (use_this_rncptr == NULL) {
 525                                 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
 526                                 if (rncptr != NULL) {
 527                                         rncptr->next = rnclist;
 528                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
 529                                         rncptr->rooms = NULL;
 530                                         rnclist = rncptr;
 531                                         use_this_rncptr = rncptr;
 532                                 }
 533                         }
 534
 535                         /* Add the room name to the request */
 536                         if (use_this_rncptr != NULL) {
 537                                 if (use_this_rncptr->rooms == NULL) {
 538                                         rncptr->rooms = strdup(qrbuf->QRname);
 539                                 }
 540                                 else {
 541                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
 542                                         ptr = realloc(use_this_rncptr->rooms, len);
 543                                         if (ptr != NULL) {
 544                                                 strcat(ptr, "|");
 545                                                 strcat(ptr, qrbuf->QRname);
 546                                                 use_this_rncptr->rooms = ptr;
 547                                         }
 548                                 }
 549                         }
 550                 }
 551
 552         }
 553
 554         fclose(fp);
 555
 556 }
 557
 558 /*
 559  * Scan for rooms that have RSS client requests configured
 560  */
 561 void *rssclient_scan(void *args) {
 562         static time_t last_run = 0L;
 563         static int doing_rssclient = 0;
 564         struct rssnetcfg *rptr = NULL;
 565         struct CitContext rssclientCC;
 566
 567         /* Give this thread its own private CitContext */
 568         CtdlFillSystemContext(&rssclientCC, "rssclient");
 569         citthread_setspecific(MyConKey, (void *)&rssclientCC );
 570
 571         CtdlThreadAllocTSD();
 572
 573         /*
 574          * This is a simple concurrency check to make sure only one rssclient run
 575          * is done at a time.  We could do this with a mutex, but since we
 576          * don't really require extremely fine granularity here, we'll do it
 577          * with a static variable instead.
 578          */
 579         if (doing_rssclient) return NULL;
 580         doing_rssclient = 1;
 581
 582         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
 583         ForEachRoom(rssclient_scan_room, NULL);
 584
 585         while (rnclist != NULL && !CtdlThreadCheckStop()) {
 586                 rss_do_fetching(rnclist->url, rnclist->rooms);
 587                 rptr = rnclist;
 588                 rnclist = rnclist->next;
 589                 if (rptr->rooms != NULL) free(rptr->rooms);
 590                 free(rptr);
 591         }
 592
 593         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
 594         last_run = time(NULL);
 595         doing_rssclient = 0;
 596         if (!CtdlThreadCheckStop())
 597                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
 598         else
 599                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
 600         return NULL;
 601 }
 602
 603
 604 CTDL_MODULE_INIT(rssclient)
 605 {
 606         if (threading)
 607         {
 608                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
 609                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
 610         }
 611         /* return our Subversion id for the Log */
 612         return "$Id$";
 613 }