citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * Bring external RSS feeds into rooms.
   3  *
   4  * Copyright (c) 2007-2017 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  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
  12  * GNU General Public License for more details.
  13  */
  14
  15 #include <stdlib.h>
  16 #include <unistd.h>
  17 #include <stdio.h>
  18
  19 #if TIME_WITH_SYS_TIME
  20 # include <sys/time.h>
  21 # include <time.h>
  22 #else
  23 # if HAVE_SYS_TIME_H
  24 #include <sys/time.h>
  25 # else
  26 #include <time.h>
  27 # endif
  28 #endif
  29
  30 #include <ctype.h>
  31 #include <string.h>
  32 #include <errno.h>
  33 #include <sys/types.h>
  34 #include <sys/stat.h>
  35 #include <expat.h>
  36 #include <curl/curl.h>
  37 #include <libcitadel.h>
  38 #include "citadel.h"
  39 #include "server.h"
  40 #include "citserver.h"
  41 #include "support.h"
  42 #include "config.h"
  43 #include "threads.h"
  44 #include "ctdl_module.h"
  45 #include "msgbase.h"
  46 #include "parsedate.h"
  47 #include "database.h"
  48 #include "citadel_dirs.h"
  49 #include "md5.h"
  50 #include "context.h"
  51 #include "internet_addressing.h"
  52
  53 struct rssroom {
  54         struct rssroom *next;
  55         char *room;
  56 };
  57
  58 struct rssurl {
  59         struct rssurl *next;
  60         char *url;
  61         struct rssroom *rooms;
  62 };
  63
  64 struct rssparser {
  65         StrBuf *CData;
  66         struct CtdlMessage *msg;
  67         char *link;
  68         char *description;
  69         char *item_id;
  70         struct rssroom *rooms;
  71 };
  72
  73 time_t last_run = 0L;
  74 struct CitContext rss_CC;
  75 struct rssurl *rsstodo = NULL;
  76
  77
  78 // This handler is called whenever an XML tag opens.
  79 //
  80 void rss_start_element(void *data, const char *el, const char **attribute)
  81 {
  82         struct rssparser *r = (struct rssparser *)data;
  83         int i;
  84
  85         if (
  86                 (!strcasecmp(el, "entry"))
  87                 || (!strcasecmp(el, "item"))
  88         ) {
  89                 // this is the start of a new item(rss) or entry(atom)
  90                 if (r->msg != NULL) {
  91                         CM_Free(r->msg);
  92                         r->msg = NULL;
  93                 }
  94                 r->msg = malloc(sizeof(struct CtdlMessage));
  95                 memset(r->msg, 0, sizeof(struct CtdlMessage));
  96                 r->msg->cm_magic = CTDLMESSAGE_MAGIC;
  97                 r->msg->cm_anon_type = MES_NORMAL;
  98                 r->msg->cm_format_type = FMT_RFC822;
  99         }
 100
 101         else if (!strcasecmp(el, "link")) {                     // atom feeds have the link as an attribute
 102                 for(i = 0; attribute[i]; i += 2) {
 103                         if (!strcasecmp(attribute[i], "href")) {
 104                                 if (r->link != NULL) {
 105                                         free(r->link);
 106                                         r->link = NULL;
 107                                 }
 108                                 r->link = strdup(attribute[i+1]);
 109                                 striplt(r->link);
 110                         }
 111                 }
 112         }
 113 }
 114
 115
 116 // This handler is called whenever an XML tag closes.
 117 //
 118 void rss_end_element(void *data, const char *el)
 119 {
 120         struct rssparser *r = (struct rssparser *)data;
 121
 122         if (                                                    // end of a new item(rss) or entry(atom)
 123                 (!strcasecmp(el, "entry"))
 124                 || (!strcasecmp(el, "item"))
 125         ) {
 126
 127                 if (r->msg != NULL) {                           // Save the message to the rooms
 128
 129                         // use the link as an item id if nothing else is available
 130                         if ((r->item_id == NULL) && (r->link != NULL)) {
 131                                 r->item_id = strdup(r->link);
 132                         }
 133
 134                         // check the use table
 135                         StrBuf *u = NewStrBuf();
 136                         StrBufAppendPrintf(u, "rss/%s", r->item_id);
 137                         time_t already_seen = CheckIfAlreadySeen("rss", u, time(NULL), 604800, eUpdate, 0, 0);
 138                         FreeStrBuf(&u);
 139
 140                         if (already_seen == 0) {
 141
 142                                 // Compose the message text
 143                                 StrBuf *TheMessage = NewStrBuf();
 144                                 StrBufAppendPrintf(TheMessage,
 145                                         "Content-type: text/html\n\n"
 146                                         "\n\n"
 147                                         "<html><head></head><body>"
 148                                 );
 149
 150                                 if (r->description != NULL) {
 151                                         StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
 152                                         free(r->description);
 153                                         r->description = NULL;
 154                                 }
 155
 156                                 if (r->link != NULL) {
 157                                         StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
 158                                         free(r->link);
 159                                         r->link = NULL;
 160                                 }
 161
 162                                 StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
 163                                 CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
 164                                 FreeStrBuf(&TheMessage);
 165
 166                                 if (CM_IsEmpty(r->msg, eAuthor)) {
 167                                         CM_SetField(r->msg, eAuthor, HKEY("rss"));
 168                                 }
 169
 170                                 if (CM_IsEmpty(r->msg, eTimestamp)) {
 171                                         CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
 172                                 }
 173
 174                                 // Save it to the room(s)
 175                                 struct rssroom *rr = NULL;
 176                                 long msgnum = (-1);
 177                                 for (rr=r->rooms; rr!=NULL; rr=rr->next) {
 178                                         if (rr == r->rooms) {
 179                                                 msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room, 0);
 180                                         }
 181                                         else {
 182                                                 CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL);
 183                                         }
 184                                         syslog(LOG_DEBUG, "Saved message %ld to %s", msgnum, rr->room);
 185                                 }
 186                         }
 187                         else {
 188                                 syslog(LOG_DEBUG, "%s was already seen %ld seconds ago", r->item_id, already_seen);
 189                         }
 190
 191                         CM_Free(r->msg);
 192                         r->msg = NULL;
 193                 }
 194
 195                 if (r->item_id != NULL) {
 196                         free(r->item_id);
 197                         r->item_id = NULL;
 198                 }
 199         }
 200
 201         else if (!strcasecmp(el, "title")) {                    // item subject (rss and atom)
 202                 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
 203                         CM_SetField(r->msg, eMsgSubject, ChrPtr(r->CData), StrLength(r->CData));
 204                         striplt(r->msg->cm_fields[eMsgSubject]);
 205                 }
 206         }
 207
 208         else if (!strcasecmp(el, "author")) {                   // author of item (rss and maybe atom)
 209                 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
 210                         CM_SetField(r->msg, eAuthor, ChrPtr(r->CData), StrLength(r->CData));
 211                         striplt(r->msg->cm_fields[eAuthor]);
 212                 }
 213         }
 214
 215         else if (!strcasecmp(el, "pubdate")) {                  // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
 216                 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
 217                         CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
 218                 }
 219         }
 220
 221         else if (!strcasecmp(el, "updated")) {                  // date/time stamp (atom) 2003-12-13T18:30:02Z
 222                 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
 223                         struct tm t;
 224                         char zulu;
 225                         memset(&t, 0, sizeof t);
 226                         sscanf(ChrPtr(r->CData), "%d-%d-%dT%d:%d:%d%c", &t.tm_year, &t.tm_mon, &t.tm_mday, &t.tm_hour, &t.tm_min, &t.tm_sec, &zulu);
 227                         t.tm_year -= 1900;
 228                         t.tm_mon -= 1;
 229                         CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
 230                 }
 231         }
 232
 233         else if (!strcasecmp(el, "link")) {                     // link to story (rss)
 234                 if (r->link != NULL) {
 235                         free(r->link);
 236                         r->link = NULL;
 237                 }
 238                 r->link = strdup(ChrPtr(r->CData));
 239                 striplt(r->link);
 240         }
 241
 242         else if (
 243                 (!strcasecmp(el, "guid"))                       // unique item id (rss)
 244                 || (!strcasecmp(el, "id"))                      // unique item id (atom)
 245         ) {
 246                 if (r->item_id != NULL) {
 247                         free(r->item_id);
 248                         r->item_id = NULL;
 249                 }
 250                 r->item_id = strdup(ChrPtr(r->CData));
 251                 striplt(r->item_id);
 252         }
 253
 254         else if (
 255                 (!strcasecmp(el, "description"))                // message text (rss)
 256                 || (!strcasecmp(el, "summary"))                 // message text (atom)
 257         ) {
 258                 if (r->description != NULL) {
 259                         free(r->description);
 260                         r->description = NULL;
 261                 }
 262                 r->description = strdup(ChrPtr(r->CData));
 263                 striplt(r->description);
 264         }
 265
 266         if (r->CData != NULL) {
 267                 FreeStrBuf(&r->CData);
 268                 r->CData = NULL;
 269         }
 270 }
 271
 272
 273 // This handler is called whenever data appears between opening and closing tags.
 274 //
 275 void rss_handle_data(void *data, const char *content, int length)
 276 {
 277         struct rssparser *r = (struct rssparser *)data;
 278
 279         if (r->CData == NULL) {
 280                 r->CData = NewStrBuf();
 281         }
 282
 283         StrBufAppendBufPlain(r->CData, content, length, 0);
 284 }
 285
 286
 287 // Feed has been downloaded, now parse it.
 288 //
 289 void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms)
 290 {
 291         struct rssparser r;
 292
 293         memset(&r, 0, sizeof r);
 294         r.rooms = rooms;
 295         XML_Parser p = XML_ParserCreate("UTF-8");
 296         XML_SetElementHandler(p, rss_start_element, rss_end_element);
 297         XML_SetCharacterDataHandler(p, rss_handle_data);
 298         XML_SetUserData(p, (void *)&r);
 299         XML_Parse(p, ChrPtr(Feed), StrLength(Feed), XML_TRUE);
 300         XML_ParserFree(p);
 301 }
 302
 303
 304 // Add a feed/room pair into the todo list
 305 //
 306 void rssclient_push_todo(char *rssurl, char *roomname)
 307 {
 308         struct rssurl *r = NULL;
 309         struct rssurl *thisone = NULL;
 310         struct rssroom *newroom = NULL;
 311
 312         syslog(LOG_DEBUG, "rssclient_push_todo(%s, %s)", rssurl, roomname);
 313
 314         for (r=rsstodo; r!=NULL; r=r->next) {
 315                 if (!strcasecmp(r->url, rssurl)) {
 316                         thisone = r;
 317                 }
 318         }
 319         if (thisone == NULL) {
 320                 thisone = malloc(sizeof(struct rssurl));
 321                 thisone->url = strdup(rssurl);
 322                 thisone->rooms = NULL;
 323                 thisone->next = rsstodo;
 324                 rsstodo = thisone;
 325         }
 326
 327         newroom = malloc(sizeof(struct rssroom));
 328         newroom->room = strdup(roomname);
 329         newroom->next = thisone->rooms;
 330         thisone->rooms = newroom;
 331 }
 332
 333
 334 // Callback function for curl
 335 //
 336 size_t rss_pof_write_data(void *buffer, size_t size, size_t nmemb, void *userp)
 337 {
 338         StrBuf *Downloaded = (StrBuf *)userp;
 339         size_t bytes = size * nmemb;
 340         StrBufAppendBufPlain(Downloaded, buffer, bytes, 0);
 341         return(bytes);
 342 }
 343
 344
 345 // pull one feed (possibly multiple rooms)
 346 //
 347 void rss_pull_one_feed(struct rssurl *url)
 348 {
 349         CURL *curl;
 350         CURLcode res;
 351         StrBuf *Downloaded = NULL;
 352
 353         syslog(LOG_DEBUG, "rss_pull_one_feed(%s)", url->url);
 354
 355         curl = curl_easy_init();
 356         if (!curl) {
 357                 return;
 358         }
 359
 360         Downloaded = NewStrBuf();
 361
 362         curl_easy_setopt(curl, CURLOPT_URL, url->url);
 363         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L);                     // Follow redirects
 364         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_pof_write_data);      // What to do with downloaded data
 365         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded);                  // Give it our StrBuf to work with
 366         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L);                           // Time out after 20 seconds
 367         res = curl_easy_perform(curl);                                          // Perform the request
 368         if (res != CURLE_OK) {
 369                 syslog(LOG_WARNING, "Failed to load feed: %s", curl_easy_strerror(res));
 370         }
 371         curl_easy_cleanup(curl);
 372
 373         rss_parse_feed(Downloaded, url->rooms);                                 // parse the feed
 374         FreeStrBuf(&Downloaded);                                                // free the downloaded feed data
 375 }
 376
 377
 378 // We have a list, now download the feeds
 379 //
 380 void rss_pull_feeds(void)
 381 {
 382         struct rssurl *r;
 383         struct rssroom *rr;
 384
 385         while (rsstodo != NULL) {
 386                 rss_pull_one_feed(rsstodo);
 387                 r = rsstodo;
 388                 rsstodo = rsstodo->next;
 389                 while (r->rooms != NULL) {
 390                         rr = r->rooms;
 391                         r->rooms = r->rooms->next;
 392                         free(rr->room);
 393                         free(rr);
 394                 }
 395                 free(r->url);
 396                 free(r);
 397         }
 398 }
 399
 400
 401 // Scan a room's netconfig looking for RSS feed parsing requests
 402 //
 403 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 404 {
 405         char *serialized_config = NULL;
 406         int num_configs = 0;
 407         char cfgline[SIZ];
 408         int i = 0;
 409
 410         serialized_config = LoadRoomNetConfigFile(qrbuf->QRnumber);
 411         if (!serialized_config) {
 412                 return;
 413         }
 414
 415         num_configs = num_tokens(serialized_config, '\n');
 416         for (i=0; i<num_configs; ++i) {
 417                 extract_token(cfgline, serialized_config, i, '\n', sizeof cfgline);
 418                 if (!strncasecmp(cfgline, HKEY("rssclient|"))) {
 419                         strcpy(cfgline, &cfgline[10]);
 420                         char *vbar = strchr(cfgline, '|');
 421                         if (vbar != NULL) {
 422                                 *vbar = 0;
 423                         }
 424                         rssclient_push_todo(cfgline, qrbuf->QRname);
 425                 }
 426         }
 427
 428         free(serialized_config);
 429 }
 430
 431
 432 /*
 433  * Scan for rooms that have RSS client requests configured
 434  */
 435 void rssclient_scan(void) {
 436         time_t now = time(NULL);
 437
 438         /* Run no more than once every 15 minutes. */
 439         if ((now - last_run) < 900) {
 440                 syslog(LOG_DEBUG,
 441                         "Client: polling interval not yet reached; last run was %ldm%lds ago",
 442                         ((now - last_run) / 60),
 443                         ((now - last_run) % 60)
 444                 );
 445                 return;
 446         }
 447
 448         become_session(&rss_CC);
 449         syslog(LOG_DEBUG, "rssclient started");
 450         CtdlForEachRoom(rssclient_scan_room, NULL);
 451         rss_pull_feeds();
 452         syslog(LOG_DEBUG, "rssclient ended");
 453         last_run = time(NULL);
 454         return;
 455 }
 456
 457
 458 CTDL_MODULE_INIT(rssclient)
 459 {
 460         if (!threading)
 461         {
 462                 syslog(LOG_INFO, "%s", curl_version());
 463                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
 464         }
 465         else
 466         {
 467                 CtdlFillSystemContext(&rss_CC, "rssclient");
 468         }
 469         return "rssclient";
 470 }
 471