3 files changed, 127 insertions, 0 deletions

diff --git a/data.c b/data.c
index 4e1682c..d9449d8 100644
--- a/data.c
+++ b/data.c
@@ -2488,6 +2488,113 @@ void notify_clear(snac *snac)
 }
 
 
+/** searches **/
+
+xs_list *content_search(snac *user, const char *regex, int priv, int max_secs, int *timeout)
+/* returns a list of posts which content matches the regex */
+{
+    xs_set seen;
+
+    xs_set_init(&seen);
+
+    if (max_secs == 0)
+        max_secs = 3;
+
+    time_t t = time(NULL) + max_secs;
+    *timeout = 0;
+
+    /* iterate both timelines simultaneously */
+    xs *pub_tl    = timeline_simple_list(user, "public", 0, XS_ALL);
+    int pub_c     = 0;
+    char *pub_md5 = NULL;
+
+    xs *priv_tl    = priv ? timeline_simple_list(user, "private", 0, XS_ALL) : xs_list_new();
+    int priv_c     = 0;
+    char *priv_md5 = NULL;
+
+    /* first positioning */
+    xs_list_next(pub_tl,  &pub_md5,  &pub_c);
+    xs_list_next(priv_tl, &priv_md5, &priv_c);
+
+    for (;;) {
+        char *md5 = NULL;
+        enum { NONE, PUBLIC, PRIVATE } from = NONE;
+
+        /* timeout? */
+        if (time(NULL) > t) {
+            *timeout = 1;
+            break;
+        }
+
+        if (pub_md5 == NULL) {
+            /* out of both lists? done */
+            if (priv_md5 == NULL)
+                break;
+
+            /* out of public: take element from the private timeline and advance */
+            from = PRIVATE;
+        }
+        else
+        if (priv_md5 == NULL) {
+            /* out of private: take element from the public timeline and advance */
+            from = PUBLIC;
+        }
+        else {
+            /* candidates from both: choose one from the file dates */
+            xs *pub_fn  = xs_fmt("%s/public/%s.json",  user->basedir, pub_md5);
+            xs *priv_fn = xs_fmt("%s/private/%s.json", user->basedir, priv_md5);
+
+            if (mtime(pub_fn) < mtime(priv_fn))
+                from = PRIVATE;
+            else
+                from = PUBLIC;
+        }
+
+        if (from == PUBLIC) { /* public */
+            md5 = pub_md5;
+            if (!xs_list_next(pub_tl, &pub_md5, &pub_c))
+                pub_md5 = NULL;
+        }
+        else
+        if (from == PRIVATE) { /* private */
+            md5 = priv_md5;
+            if (!xs_list_next(priv_tl, &priv_md5, &priv_c))
+                priv_md5 = NULL;
+        }
+
+        if (md5 == NULL)
+            break;
+
+        xs *post = NULL;
+
+        if (!valid_status(timeline_get_by_md5(user, md5, &post)))
+            continue;
+
+        /* must be a Note */
+        if (strcmp(xs_dict_get_def(post, "type", ""), "Note"))
+            continue;
+
+        char *content = xs_dict_get(post, "content");
+
+        if (xs_is_null(content))
+            continue;
+
+        /* strip HTML */
+        xs *c = xs_regex_replace(content, "<[^>]+>", " ");
+        c = xs_regex_replace_i(c, " {2,}", " ");
+        c = xs_tolower_i(c);
+
+        /* apply regex */
+        xs *l = xs_regex_select_n(c, regex, 1);
+
+        if (xs_list_len(l))
+            xs_set_add(&seen, md5);
+    }
+
+    return xs_set_result(&seen);
+}
+
+
 /** the queue **/
 
 static xs_dict *_enqueue_put(const char *fn, xs_dict *msg)
diff --git a/main.c b/main.c
index 6a38412..bd52b19 100644
--- a/main.c
+++ b/main.c
@@ -44,6 +44,7 @@ int usage(void)
     printf("limit {basedir} {uid} {actor}        Limits an actor (drops their announces)\n");
     printf("unlimit {basedir} {uid} {actor}      Unlimits an actor\n");
     printf("verify_links {basedir} {uid}         Verifies a user's links (in the metadata)\n");
+    printf("search {basedir} {uid} {regex}       Searches posts by content\n");
 
     return 1;
 }
@@ -374,6 +375,23 @@ int main(int argc, char *argv[])
         return 0;
     }
 
+    if (strcmp(cmd, "search") == 0) { /** **/
+        int to;
+
+        /* 'url' contains the regex */
+        xs *r = content_search(&snac, url, 1, 10, &to);
+
+        int c = 0;
+        char *v;
+
+        /* print results as standalone links */
+        while (xs_list_next(r, &v, &c)) {
+            printf("%s/admin/p/%s\n", snac.actor, v);
+        }
+
+        return 0;
+    }
+
     if (strcmp(cmd, "ping") == 0) { /** **/
         xs *actor_o = NULL;
 
diff --git a/snac.h b/snac.h
index 2a988f1..e04dc65 100644
--- a/snac.h
+++ b/snac.h
@@ -179,6 +179,8 @@ xs_list *list_timeline(snac *user, const char *list, int skip, int show);
 xs_val *list_content(snac *user, const char *list_id, const char *actor_md5, int op);
 void list_distribute(snac *user, const char *who, const xs_dict *post);
 
+xs_list *content_search(snac *user, const char *regex, int priv, int max_secs, int *timeout);
+
 int actor_add(const char *actor, xs_dict *msg);
 int actor_get(const char *actor, xs_dict **data);
 int actor_get_refresh(snac *user, const char *actor, xs_dict **data);