gatherer: ignore trends from <2 hosts

gatherer/src/tag.rs

@@ -139,7 +139,7 @@ impl Tag {
             .unwrap_or(&self.name)
     }
 
-    pub fn hosts(&self) -> BTreeSet<(usize, &str)> {
+    pub fn hosts(&self) -> impl Iterator<Item = (usize, &str)> {
         self.other.iter()
             .filter_map(|(key, value)| {
                 if &key[..2] != "h:" {
@@ -152,6 +152,10 @@ impl Tag {
 
                 None
             })
-            .collect()
+    }
+
+    /// ordered by count
+    pub fn hosts_set(&self) -> BTreeSet<(usize, &str)> {
+        self.hosts().collect()
     }
 }

gatherer/src/trends.rs

@@ -78,6 +78,11 @@ impl TrendAnalyzer {
     }
 
     pub fn process_tag(&mut self, tag: &Arc<Tag>) {
+        if tag.hosts().skip(1).next().is_none() {
+            // ignore spam that is posted by accounts on <1 host
+            return;
+        }
+
         let score = tag.score(self.period, self.until);
 
         if score <= 1.0 {
@@ -97,7 +102,7 @@ impl TrendAnalyzer {
             self.result.remove(&least);
             least = self.result.keys().cloned().next().unwrap().clone();
         }
-        
+
         self.score_threshold = Some(least.score);
     }
 }

gatherer/templates/trends.html

@@ -27,11 +27,12 @@
               <h3>#{{ tag.spelling() }}</h3>
               <p class="score">{{ format!("{:.0}%", 100. * score) }}</p>
               <ul class="hosts">
-                {% for (count, host) in tag.hosts().into_iter().rev().take(5) %}
+                {% for (count, host) in tag.hosts_set().into_iter().rev().take(5) %}
                   <li>
                     <a href="https://{{ host }}/tags/{{ tag.name }}">
                       {{ host }}
                     </a>
+                    ({{ count }})
                   </li>
                 {% endfor %}
               </ul>