3 files changed, 55 insertions, 1 deletions

diff --git a/format.c b/format.c
index 69efbdb..1c2a97a 100644
--- a/format.c
+++ b/format.c
@@ -150,3 +150,54 @@ d_char *not_really_markdown(char *content, d_char **f_content)
 
     return *f_content;
 }
+
+
+const char *valid_tags[] = {
+    "a", "p", "br", "img", "blockquote", "ul", "li", "span", NULL
+};
+
+d_char *sanitize(d_char *content)
+/* cleans dangerous HTML output */
+{
+    d_char *s = xs_str_new(NULL);
+    xs *sl;
+    int n = 0;
+    char *p, *v;
+
+    sl = xs_regex_split(content, "</?[^>]+>");
+
+    p = sl;
+
+    while (xs_list_iter(&p, &v)) {
+        if (n & 0x1) {
+            xs *s1  = xs_strip(xs_crop(xs_dup(v), v[1] == '/' ? 2 : 1, -1));
+            xs *l1  = xs_split_n(s1, " ", 1);
+            xs *tag = xs_tolower(xs_dup(xs_list_get(l1, 0)));
+            int i;
+
+            /* check if it's one of the valid tags */
+            for (i = 0; valid_tags[i]; i++) {
+                if (strcmp(tag, valid_tags[i]) == 0)
+                    break;
+            }
+
+            if (valid_tags[i]) {
+                /* accepted tag */
+                s = xs_str_cat(s, v);
+            }
+            else {
+                /* bad tag */
+                xs *s2 = xs_replace(v, "<", "&lt;");
+                s = xs_str_cat(s, s2);
+            }
+        }
+        else {
+            /* non-tag */
+            s = xs_str_cat(s, v);
+        }
+
+        n++;
+    }
+
+    return s;
+}
diff --git a/html.c b/html.c
index 7fd1294..43638d5 100644
--- a/html.c
+++ b/html.c
@@ -600,8 +600,9 @@ d_char *html_entry(snac *snac, d_char *os, char *msg, xs_set *seen, int local, i
             }
         }
 
+        xs *sc = sanitize(c);
 
-        s = xs_str_cat(s, c);
+        s = xs_str_cat(s, sc);
     }
 
     s = xs_str_cat(s, "\n");
diff --git a/snac.h b/snac.h
index f11dcc4..7a3a5f3 100644
--- a/snac.h
+++ b/snac.h
@@ -134,6 +134,8 @@ int activitypub_post_handler(d_char *req, char *q_path,
                              char **body, int *b_size, char **ctype);
 
 d_char *not_really_markdown(char *content, d_char **f_content);
+d_char *sanitize(d_char *str);
+
 int html_get_handler(d_char *req, char *q_path, char **body, int *b_size, char **ctype);
 int html_post_handler(d_char *req, char *q_path, d_char *payload, int p_size,
                       char **body, int *b_size, char **ctype);