cave/store: gather 5 images per tag

cave/src/feed.rs

@@ -51,17 +51,29 @@ impl Mention {
     }
 }
 
+#[derive(Debug, serde::Serialize, serde::Deserialize)]
+pub struct MediaAttachment {
+    #[serde(rename = "type")]
+    pub media_type: String,
+    pub remote_url: Option<String>,
+}
+
 #[derive(Debug, serde::Serialize, serde::Deserialize)]
 pub struct Post {
     pub created_at: String,
     pub uri: String,
+    #[serde(default = "String::new")]
     pub content: String,
     pub account: Account,
+    #[serde(default)]
     pub tags: Vec<Tag>,
     pub application: Option<Application>,
     pub sensitive: Option<bool>,
+    #[serde(default)]
     pub mentions: Vec<Mention>,
     pub language: Option<String>,
+    #[serde(default)]
+    pub media_attachments: Vec<MediaAttachment>,
 }
 
 impl Post {

cave/src/store.rs

@@ -9,6 +9,7 @@ const POST_EXPIRE: usize = 86400;
 const TAG_EXPIRE: u64 = 30 * 24;
 
 pub const TREND_POOL_SIZE: usize = 20;
+pub const IMAGES_PER_TAG: usize = 5;
 
 pub type Error = RedisError;
 
@@ -168,7 +169,7 @@ impl Store {
         let language = post.lang();
 
         let mut cmd = redis::pipe();
-        let mut store_tags = |spellings, tag_key, user_key| {
+        let store_tags = |cmd: &mut redis::Pipeline, spellings, tag_key, user_key| {
             // by spelling
             for spelling in spellings {
                 cmd.hincr(
@@ -186,29 +187,62 @@ impl Store {
             if let Some(user_id) = &user_id {
                 // users by tag/hour
                 cmd.sadd(&user_key, &user_id).ignore()
-                    .expire(&user_key, TAG_EXPIRE as usize * 86400);
+                    .expire(&user_key, TAG_EXPIRE as usize * 86400)
+                    .ignore();
             }
         };
+        let images = if post.sensitive == Some(false) {
+            post.media_attachments.iter()
+                .filter(|a| a.media_type == "image")
+                .filter_map(|a| a.remote_url.as_ref())
+                .take(2)
+                .collect::<Vec<&String>>()
+        } else {
+            // ignore disturbing porn images from sensitive posts
+            vec![]
+        };
+        let mut image_keys = vec![];
         for (name, spellings) in post.tags_set() {
             log::debug!("tag {}", name);
             // global
-            store_tags(
+            store_tags(&mut cmd,
                 spellings.clone(),
                 format!("g:{}", name),
                 format!("u::{}:{}", hour, name),
             );
             // by language
             if let Some(language) = &language {
-                store_tags(
+                store_tags(&mut cmd,
                     spellings,
                     format!("l:{}:{}", language, name),
                     format!("u:{}:{}:{}", language, hour, name),
                 );
             }
+
+            for image in &images {
+                let image_key = format!("i:{}", name);
+                cmd.sadd(&image_key, image)
+                    .ignore()
+                    .expire(&image_key, TAG_EXPIRE as usize * 86400)
+                    .ignore()
+                    .scard(&image_key);
+                image_keys.push(image_key);
+            }
         }
 
-        match cmd.query_async(self).await {
-            Ok(()) => {}
+        match cmd.query_async::<_, Vec<usize>>(self).await {
+            Ok(image_key_sizes) => {
+                assert_eq!(image_keys.len(), image_key_sizes.len());
+                let mut cmd = redis::pipe();
+                for (image_key, size) in image_keys.into_iter().zip(image_key_sizes.into_iter()) {
+                    let excess = size.saturating_sub(IMAGES_PER_TAG);
+                    if excess > 0 {
+                        cmd.spop(image_key).arg(excess)
+                            .ignore();
+                    }
+                }
+                let _  = cmd.query_async::<_, ()>(self).await;
+            }
             Err(e) => {
                 log::error!("redis error: {:?}", e);
             }