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