webcit/sitemap.c

   1 /*
   2  * XML sitemap generator
   3  *
   4  * Copyright (c) 2010-2012 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  * 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 "webcit.h"
  16 #include "webserver.h"
  17
  18
  19 /*
  20  * XML sitemap generator -- go through the message list for a BBS room
  21  */
  22 void sitemap_do_bbs(void) {
  23         wcsession *WCC = WC;
  24         int num_msgs = 0;
  25         int i;
  26         SharedMessageStatus Stat;
  27         message_summary *Msg = NULL;
  28
  29         memset(&Stat, 0, sizeof Stat);
  30         Stat.maxload = INT_MAX;
  31         Stat.lowest_found = (-1);
  32         Stat.highest_found = (-1);
  33         num_msgs = load_msg_ptrs("MSGS ALL", NULL, &Stat, NULL);
  34         if (num_msgs < 1) return;
  35
  36         for (i=0; i<num_msgs; i+=20) {
  37                 Msg = GetMessagePtrAt(i, WCC->summ);
  38                 if (Msg != NULL) {
  39                         wc_printf("<url><loc>%s/readfwd", ChrPtr(site_prefix));
  40                         wc_printf("?go=");
  41                         urlescputs(ChrPtr(WC->CurRoom.name));
  42                         wc_printf("?start_reading_at=%ld", Msg->msgnum);
  43                         wc_printf("</loc></url>\r\n");
  44                 }
  45         }
  46 }
  47
  48
  49 /*
  50  * XML sitemap generator -- go through the message list for a Blog room
  51  */
  52 void sitemap_do_blog(void) {
  53         wcsession *WCC = WC;
  54         int num_msgs = 0;
  55         int i;
  56         SharedMessageStatus Stat;
  57         message_summary *Msg = NULL;
  58         StrBuf *Buf = NewStrBuf();
  59         StrBuf *FoundCharset = NewStrBuf();
  60
  61         memset(&Stat, 0, sizeof Stat);
  62         Stat.maxload = INT_MAX;
  63         Stat.lowest_found = (-1);
  64         Stat.highest_found = (-1);
  65         num_msgs = load_msg_ptrs("MSGS ALL", NULL, &Stat, NULL);
  66         if (num_msgs < 1) return;
  67
  68         for (i=0; i<num_msgs; ++i) {
  69                 Msg = GetMessagePtrAt(i, WCC->summ);
  70                 if (Msg != NULL) {
  71                         ReadOneMessageSummary(Msg, FoundCharset, Buf);
  72                         /* Show only top level posts, not comments */
  73                         if ((Msg->reply_inreplyto_hash != 0) && (Msg->reply_references_hash == 0)) {
  74                                 WCC->bptlid = Msg->reply_inreplyto_hash;
  75                                 wc_printf("<url><loc>%s", ChrPtr(site_prefix));
  76                                 tmplput_blog_permalink(NULL, NULL);
  77                                 wc_printf("</loc></url>\r\n");
  78                         }
  79                 }
  80         }
  81         FreeStrBuf(&Buf);
  82         FreeStrBuf(&FoundCharset);
  83 }
  84
  85
  86 /*
  87  * XML sitemap generator -- go through the message list for a wiki room
  88  */
  89 void sitemap_do_wiki(void) {
  90         wcsession *WCC = WC;
  91         int num_msgs = 0;
  92         int i;
  93         SharedMessageStatus Stat;
  94         message_summary *Msg = NULL;
  95         char buf[256];
  96
  97         memset(&Stat, 0, sizeof Stat);
  98         Stat.maxload = INT_MAX;
  99         Stat.lowest_found = (-1);
 100         Stat.highest_found = (-1);
 101         num_msgs = load_msg_ptrs("MSGS ALL", NULL, &Stat, NULL);
 102         if (num_msgs < 1) return;
 103
 104         for (i=0; i<num_msgs; ++i) {
 105                 Msg = GetMessagePtrAt(i, WCC->summ);
 106                 if (Msg != NULL) {
 107
 108                         serv_printf("MSG0 %ld|3", Msg->msgnum);
 109                         serv_getln(buf, sizeof buf);
 110                         if (buf[0] == '1') while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 111                                 if (    (!strncasecmp(buf, "exti=", 5))
 112                                         && (!bmstrcasestr(buf, "_HISTORY_"))
 113                                 ) {
 114                                         wc_printf("<url><loc>%s/wiki", ChrPtr(site_prefix));
 115                                         wc_printf("?go=");
 116                                         urlescputs(ChrPtr(WC->CurRoom.name));
 117                                         wc_printf("?page=%s", &buf[5]);
 118                                         wc_printf("</loc></url>\r\n");
 119                                 }
 120                         }
 121                 }
 122         }
 123 }
 124
 125
 126 struct sitemap_room_list {
 127         struct sitemap_room_list *next;
 128         StrBuf *roomname;
 129         int defview;
 130 };
 131
 132
 133 /*
 134  * Load the room list for the sitemap
 135  */
 136 struct sitemap_room_list *sitemap_load_roomlist(void) {
 137         char buf[SIZ];
 138         char roomname_plain[SIZ];
 139         struct sitemap_room_list *roomlist = NULL;
 140
 141         serv_puts("LKRA");
 142         serv_getln(buf, sizeof buf);
 143         if (buf[0] == '1') while(serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 144                 struct sitemap_room_list *ptr = malloc(sizeof(struct sitemap_room_list));
 145                 extract_token(roomname_plain, buf, 0, '|', sizeof roomname_plain);
 146                 ptr->roomname = NewStrBufPlain(roomname_plain, -1);
 147                 ptr->defview = extract_int(buf, 6);
 148                 ptr->next = roomlist;
 149                 roomlist = ptr;
 150         }
 151
 152         return(roomlist);
 153 }
 154
 155
 156 /*
 157  * Entry point for RSS feed generator
 158  */
 159 void sitemap(void) {
 160         struct sitemap_room_list *roomlist = NULL;
 161         output_headers(0, 0, 0, 0, 1, 0);
 162         hprintf("Content-type: text/xml\r\n");
 163         hprintf(
 164                 "Server: %s / %s\r\n"
 165                 "Connection: close\r\n"
 166         ,
 167                 PACKAGE_STRING, ChrPtr(WC->serv_info->serv_software)
 168         );
 169         begin_burst();
 170
 171         wc_printf("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n");
 172         wc_printf("<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\">\r\n");
 173
 174         roomlist = sitemap_load_roomlist();
 175
 176         while (roomlist != NULL)
 177         {
 178                 struct sitemap_room_list *ptr;
 179
 180                 gotoroom(roomlist->roomname);
 181
 182                 /* Output the messages in this room only if it's a room type we can make sense of */
 183                 switch(roomlist->defview) {
 184                 case VIEW_BBS:
 185                         sitemap_do_bbs();
 186                         break;
 187                 case VIEW_WIKI:
 188                 case VIEW_WIKIMD:
 189                         sitemap_do_wiki();
 190                         break;
 191                 case VIEW_BLOG:
 192                         sitemap_do_blog();
 193                         break;
 194                 default:
 195                         break;
 196                 }
 197
 198                 ptr = roomlist;
 199                 roomlist = roomlist->next;
 200                 FreeStrBuf(&ptr->roomname);
 201                 free(ptr);
 202         }
 203
 204         wc_printf("</urlset>\r\n");
 205         wDumpContent(0);
 206 }
 207
 208
 209 void
 210 InitModule_SITEMAP
 211 (void)
 212 {
 213         WebcitAddUrlHandler(HKEY("sitemap"), "", 0, sitemap, ANONYMOUS|COOKIEUNNEEDED);
 214         WebcitAddUrlHandler(HKEY("sitemap.xml"), "", 0, sitemap, ANONYMOUS|COOKIEUNNEEDED);
 215 }