From 6901aa3d66c363bd69b4666e87ebf54ae4274838 Mon Sep 17 00:00:00 2001
From: Joakim Soderlund <joakim.soderlund@gmail.com>
Date: Wed, 29 May 2024 17:44:40 +0200
Subject: [PATCH] Save on RAM by interning author meta

---
 src/archive/fetcher.rs  | 10 +++++++++-
 src/archive/interner.rs | 28 ++++++++++++++++++----------
 src/archive/story.rs    | 29 +++++++++++++++++++++--------
 3 files changed, 48 insertions(+), 19 deletions(-)

diff --git a/src/archive/fetcher.rs b/src/archive/fetcher.rs
index 508143e..b27a03a 100644
--- a/src/archive/fetcher.rs
+++ b/src/archive/fetcher.rs
@@ -15,6 +15,8 @@ use zip::result::ZipError;
 
 use super::parser::parse;
 use super::story::Story;
+use crate::archive::AUTHORS;
+use crate::archive::TAGS;
 use crate::error::Error;
 use crate::error::Result;
 
@@ -63,7 +65,13 @@ impl<T: Read + Seek> Fetcher<T> {
             _ => Error::archive("Could not open story index"),
         })?;
 
-        parse(BufReader::with_capacity(1048576, file)).map_err(Error::index)
+        let reader = BufReader::with_capacity(1048576, file);
+        let result = parse(reader).map_err(Error::index);
+
+        AUTHORS.clear();
+        TAGS.clear();
+
+        result
     }
 
     pub fn fetch(&self, key: i64) -> Option<&Story> {
diff --git a/src/archive/interner.rs b/src/archive/interner.rs
index 3bc0424..76c24e3 100644
--- a/src/archive/interner.rs
+++ b/src/archive/interner.rs
@@ -1,35 +1,43 @@
 //! Interner module.
 
 use std::collections::HashSet;
+use std::hash::Hash;
+use std::sync::Arc;
 use std::sync::RwLock;
 
-pub struct Interner<T: 'static>(RwLock<HashSet<&'static T>>);
+pub struct Interner<T>(RwLock<HashSet<Arc<T>>>);
 
 impl<T> Interner<T>
 where
-    T: Eq + std::hash::Hash,
+    T: Eq + Hash,
 {
     pub fn new() -> Self {
         Self(RwLock::new(HashSet::new()))
     }
 
-    fn get(&self, value: &T) -> Option<&'static T> {
+    fn get(&self, value: &T) -> Option<Arc<T>> {
         let store = self.0.read().unwrap();
 
-        store.get(value).copied()
+        store.get(value).cloned()
     }
 
-    fn set(&self, value: T) -> &'static T {
-        let boxed: Box<T> = Box::new(value);
-        let leaked: &'static T = Box::leak(boxed);
+    fn set(&self, value: T) -> Arc<T> {
         let mut store = self.0.write().unwrap();
 
-        store.insert(leaked);
+        let arc = Arc::new(value);
+        store.insert(arc.clone());
 
-        leaked
+        arc
     }
 
-    pub fn intern(&self, value: T) -> &'static T {
+    pub fn clear(&self) {
+        let mut store = self.0.write().unwrap();
+
+        store.clear();
+        store.shrink_to_fit();
+    }
+
+    pub fn intern(&self, value: T) -> Arc<T> {
         self.get(&value).unwrap_or_else(|| self.set(value))
     }
 }
diff --git a/src/archive/story.rs b/src/archive/story.rs
index 171915a..e759d5c 100644
--- a/src/archive/story.rs
+++ b/src/archive/story.rs
@@ -1,5 +1,7 @@
 //! Story meta.
 
+use std::sync::Arc;
+
 use chrono::prelude::*;
 use lazy_static::lazy_static;
 use serde::de::Error;
@@ -10,14 +12,16 @@ use serde_json::Value;
 use super::interner::Interner;
 
 lazy_static! {
-    static ref TAGS: Interner<Tag> = Interner::new();
+    pub(crate) static ref AUTHORS: Interner<Author> = Interner::new();
+    pub(crate) static ref TAGS: Interner<Tag> = Interner::new();
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
 pub struct Story {
     pub archive: Archive,
-    pub author: Author,
+    #[serde(deserialize_with = "author_as_static")]
+    pub author: Arc<Author>,
     pub chapters: Vec<Chapter>,
     pub color: Option<Color>,
     pub completion_status: CompletionStatus,
@@ -43,7 +47,7 @@ pub struct Story {
     pub status: Status,
     pub submitted: bool,
     #[serde(deserialize_with = "tags_as_static")]
-    pub tags: Vec<&'static Tag>,
+    pub tags: Vec<Arc<Tag>>,
     #[serde(deserialize_with = "null_to_text")]
     pub title: String,
     pub total_num_views: i32,
@@ -60,7 +64,7 @@ pub struct Archive {
     pub path: String,
 }
 
-#[derive(Clone, Debug, Deserialize)]
+#[derive(Clone, Debug, Deserialize, PartialEq, Eq, Hash)]
 #[serde(deny_unknown_fields)]
 pub struct Author {
     pub avatar: Option<Avatar>,
@@ -75,7 +79,7 @@ pub struct Author {
     pub url: String,
 }
 
-#[derive(Clone, Debug, Deserialize)]
+#[derive(Clone, Debug, Deserialize, PartialEq, Eq, Hash)]
 #[serde(deny_unknown_fields)]
 pub struct Avatar {
     #[serde(rename = "16")]
@@ -209,13 +213,22 @@ where
     }
 }
 
-fn tags_as_static<'de, D>(d: D) -> Result<Vec<&'static Tag>, D::Error>
+fn author_as_static<'de, D>(d: D) -> Result<Arc<Author>, D::Error>
 where
     D: Deserializer<'de>,
 {
-    let tags: Vec<Tag> = Vec::deserialize(d)?;
+    Author::deserialize(d).map(|author| AUTHORS.intern(author))
+}
 
-    Ok(tags.into_iter().map(|tag| TAGS.intern(tag)).collect())
+fn tags_as_static<'de, D>(d: D) -> Result<Vec<Arc<Tag>>, D::Error>
+where
+    D: Deserializer<'de>,
+{
+    Vec::deserialize(d)?
+        .into_iter()
+        .map(|tag| TAGS.intern(tag))
+        .map(Result::Ok)
+        .collect()
 }
 
 impl<'de> Deserialize<'de> for Color {