1/* snac - A simple, minimalistic ActivityPub instance */
  2/* copyright (c) 2025 - 2026 grunfink et al. / MIT license */
  3
  4#include "xs.h"
  5#include "xs_html.h"
  6#include "xs_regex.h"
  7#include "xs_time.h"
  8#include "xs_match.h"
  9#include "xs_curl.h"
 10#include "xs_openssl.h"
 11#include "xs_json.h"
 12#include "xs_http.h"
 13#include "xs_unicode.h"
 14
 15#include "snac.h"
 16
 17xs_str *rss_from_timeline(snac *user, const xs_list *timeline,
 18                        const char *title, const char *link, const char *desc)
 19/* converts a timeline to rss */
 20{
 21    xs_html *rss = xs_html_tag("rss",
 22        xs_html_attr("xmlns:content", "http:/" "/purl.org/rss/1.0/modules/content/"),
 23        xs_html_attr("version",       "2.0"),
 24        xs_html_attr("xmlns:atom",    "http:/" "/www.w3.org/2005/Atom"));
 25
 26    xs_html *channel = xs_html_tag("channel",
 27        xs_html_tag("title",
 28            xs_html_text(title)),
 29        xs_html_tag("language",
 30            xs_html_text("en")),
 31        xs_html_tag("link",
 32            xs_html_text(link)),
 33        xs_html_sctag("atom:link",
 34            xs_html_attr("href", link),
 35            xs_html_attr("rel", "self"),
 36            xs_html_attr("type", "application/rss+xml")),
 37        xs_html_tag("generator",
 38            xs_html_text(USER_AGENT)),
 39        xs_html_tag("description",
 40            xs_html_text(desc)));
 41
 42    xs_html_add(rss, channel);
 43
 44    int cnt = 0;
 45    const char *v;
 46
 47    xs_list_foreach(timeline, v) {
 48        xs *msg = NULL;
 49
 50        if (user) {
 51            if (!valid_status(timeline_get_by_md5(user, v, &msg)))
 52                continue;
 53        }
 54        else {
 55            if (!valid_status(object_get_by_md5(v, &msg)))
 56                continue;
 57        }
 58
 59        const char *id = xs_dict_get(msg, "id");
 60        const char *content = xs_dict_get(msg, "content");
 61        const char *published = xs_dict_get(msg, "published");
 62
 63        if (user && !is_msg_mine(user, id))
 64            continue;
 65
 66        if (!id || !content || !published)
 67            continue;
 68
 69        if (!is_msg_public(msg))
 70            continue;
 71
 72        /* create a title with the first line of the content */
 73        xs *title = xs_replace(content, "<br>", "\n");
 74        title = xs_regex_replace_i(title, "<[^>]+>", " ");
 75        title = xs_regex_replace_i(title, "&[^;]+;", " ");
 76        int i;
 77
 78        for (i = 0; title[i] && title[i] != '\n' && i < 50; ) {
 79            const char *p = &title[i];
 80            unsigned int cp = xs_utf8_dec(&p);
 81            int n = p - title;
 82            if (cp == 0xfffd || n > 50)
 83                break;
 84            i = n;
 85        }
 86
 87        if (title[i] != '\0') {
 88            title[i] = '\0';
 89            title = xs_str_cat(title, "...");
 90        }
 91
 92        title = xs_strip_i(title);
 93
 94        /* convert the date */
 95        time_t t = xs_parse_iso_date(published, 0);
 96        xs *rss_date = xs_str_utctime(t, "%a, %d %b %Y %T +0000");
 97
 98        /* if it's the first one, add it to the header */
 99        if (cnt == 0)
100            xs_html_add(channel,
101                xs_html_tag("lastBuildDate",
102                    xs_html_text(rss_date)));
103
104        xs *att_content = xs_dup(content);
105        xs *attach_list = get_attachments(msg);
106
107        /* add possible attachments */
108        if (xs_list_len(attach_list)) {
109            const xs_val *a;
110            xs_html *content_attachments = xs_html_tag("div",
111                xs_html_attr("class", "snac-content-attachments"));
112
113            xs_list_foreach(attach_list, a) {
114                const char *type = xs_dict_get(a, "type");
115                const char *o_href = xs_dict_get(a, "href");
116                const char *name = xs_dict_get_def(a, "name", "");
117
118                if (!xs_is_string(type) || !xs_is_string(o_href))
119                    continue;
120
121                if (xs_startswith(type, "image/")) {
122                    xs_html_add(content_attachments,
123                        xs_html_tag("figure",
124                            xs_html_sctag("img",
125                                xs_html_attr("src", o_href)),
126                            xs_html_tag("figcaption",
127                                xs_html_text(name))));
128                }
129                else {
130                    xs_html_add(content_attachments,
131                        xs_html_tag("a",
132                            xs_html_attr("href", o_href),
133                            xs_html_text(name),
134                            xs_html_text(" "),
135                            xs_html_text(o_href),
136                        xs_html_sctag("br", NULL)));
137                }
138            }
139
140            xs *s1 = xs_html_render(content_attachments);
141            att_content = xs_str_cat(att_content, s1);
142        }
143
144        xs_html_add(channel,
145            xs_html_tag("item",
146                xs_html_tag("title",
147                    xs_html_text(title)),
148                xs_html_tag("link",
149                    xs_html_text(id)),
150                xs_html_tag("guid",
151                    xs_html_text(id)),
152                xs_html_tag("pubDate",
153                    xs_html_text(rss_date)),
154                xs_html_tag("description",
155                    xs_html_text(att_content))));
156
157        cnt++;
158    }
159
160    return xs_html_render_s(rss, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
161}
162
163
164void rss_to_timeline(snac *user, const char *url)
165/* reads an RSS and inserts all ActivityPub posts into the user's timeline */
166{
167    if (!xs_startswith(url, "https:/") && !xs_startswith(url, "http:/"))
168        return;
169
170    xs *hdrs = xs_dict_new();
171    hdrs = xs_dict_set(hdrs, "accept",     "application/rss+xml");
172    hdrs = xs_dict_set(hdrs, "user-agent", USER_AGENT);
173
174    /* get the RSS metadata */
175    xs *md5 = xs_md5_hex(url, strlen(url));
176    xs *rss_md_fn = xs_fmt("%s/rss", user->basedir);
177    mkdirx(rss_md_fn);
178    rss_md_fn = xs_str_cat(rss_md_fn, "/", md5, ".json");
179
180    xs *rss_md = NULL;
181    const char *etag = NULL;
182
183    FILE *f;
184    if ((f = fopen(rss_md_fn, "r")) != NULL) {
185        rss_md = xs_json_load(f);
186        fclose(f);
187
188        etag = xs_dict_get(rss_md, "etag");
189
190        if (xs_is_string(etag))
191            hdrs = xs_dict_set(hdrs, "if-none-match", etag);
192    }
193
194    if (rss_md == NULL)
195        rss_md = xs_dict_new();
196
197    xs *payload = NULL;
198    int status;
199    int p_size;
200
201    xs *rsp = xs_http_request("GET", url, hdrs, NULL, 0, &status, &payload, &p_size, 0);
202
203    snac_log(user, xs_fmt("parsing RSS %s %d", url, status));
204
205    if (!valid_status(status) || !xs_is_string(payload))
206        return;
207
208    /* not an RSS? done */
209    const char *ctype = xs_dict_get(rsp, "content-type");
210    if (!xs_is_string(ctype) || xs_str_in(ctype, "application/rss+xml") == -1)
211        return;
212
213    /* yes, parsing is done with regexes (now I have two problems blah blah blah) */
214    xs *links = xs_regex_select(payload, "<link>[^<]+</link>");
215    const char *link;
216
217    xs_list_foreach(links, link) {
218        xs *l = xs_replace(link, "<link>", "");
219        char *p = strchr(l, '<');
220
221        if (p == NULL)
222            continue;
223        *p = '\0';
224
225        /* skip this same URL */
226        if (strcmp(l, url) == 0)
227            continue;
228
229        /* skip crap */
230        if (!xs_startswith(l, "https:/") && !xs_startswith(l, "http:/"))
231            continue;
232
233        snac_debug(user, 1, xs_fmt("RSS link: %s", l));
234
235        if (timeline_here(user, l)) {
236            snac_debug(user, 1, xs_fmt("RSS entry already in timeline %s", l));
237            continue;
238        }
239
240        /* special trick for Mastodon: convert from the alternate format */
241        if (strchr(l, '@') != NULL) {
242            xs *l2 = xs_split(l, "/");
243
244            if (xs_list_len(l2) == 5) {
245                const char *uid = xs_list_get(l2, 3);
246                if (*uid == '@') {
247                    xs *guessed_id = xs_fmt("https:/" "/%s/users/%s/statuses/%s",
248                        xs_list_get(l2, 2), uid + 1, xs_list_get(l2, -1));
249
250                    if (timeline_here(user, guessed_id)) {
251                        snac_debug(user, 1, xs_fmt("RSS entry already in timeline (alt) %s", guessed_id));
252                        continue;
253                    }
254                }
255            }
256        }
257
258        xs *obj = NULL;
259
260        if (!valid_status(object_get(l, &obj))) {
261            /* object is not here: bring it */
262            if (!valid_status(activitypub_request(user, l, &obj)))
263                continue;
264        }
265
266        if (xs_is_dict(obj)) {
267            const char *id      = xs_dict_get(obj, "id");
268            const char *type    = xs_dict_get(obj, "type");
269            const char *attr_to = get_atto(obj);
270
271            if (!xs_is_string(id) || !xs_is_string(type) || !xs_is_string(attr_to))
272                continue;
273
274            if (!xs_match(type, POSTLIKE_OBJECT_TYPE))
275                continue;
276
277            if (timeline_here(user, id)) {
278                snac_debug(user, 1, xs_fmt("RSS entry already in timeline (id) %s", id));
279                continue;
280            }
281
282            enqueue_actor_refresh(user, attr_to, 0);
283
284            timeline_add(user, id, obj);
285
286            snac_log(user, xs_fmt("new '%s' (RSS) %s %s", type, attr_to, id));
287        }
288    }
289
290    /* update the RSS metadata */
291    etag = xs_dict_get(rsp, "etag");
292
293    if (xs_is_string(etag)) {
294        rss_md = xs_dict_set(rss_md, "etag", etag);
295        rss_md = xs_dict_set(rss_md, "url", url);
296        if ((f = fopen(rss_md_fn, "w")) != NULL) {
297            xs_json_dump(rss_md, 4, f);
298            fclose(f);
299        }
300    }
301}
302
303
304void rss_poll_hashtags(void)
305/* parses all RSS from all users */
306{
307    xs *list = user_list();
308    const char *uid;
309
310    xs_list_foreach(list, uid) {
311        snac user;
312
313        if (user_open(&user, uid)) {
314            const xs_list *rss = xs_dict_get(user.config, "followed_hashtags");
315
316            if (xs_is_list(rss)) {
317                const char *url;
318
319                xs_list_foreach(rss, url)
320                    rss_to_timeline(&user, url);
321            }
322
323            user_free(&user);
324        }
325    }
326}