From af2ce29d01b98448ff5fccda2112540137547ea0 Mon Sep 17 00:00:00 2001
From: Joakim Soderlund <joakim.soderlund@gmail.com>
Date: Fri, 22 May 2020 21:19:01 +0200
Subject: [PATCH] Add ineffective deduplication experiment

---
 Cargo.toml            |  2 +-
 src/archive/parser.rs | 47 ++++++++++++++++++++++++++++++++++++++++---
 src/archive/story.rs  | 39 +++++------------------------------
 3 files changed, 50 insertions(+), 38 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 03b94a0..4cabbcc 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -35,7 +35,7 @@ version = "*"
 
 [dependencies.serde]
 version = "*"
-features = ["derive"]
+features = ["derive", "rc"]
 
 [dependencies.serde_json]
 version = "*"
diff --git a/src/archive/parser.rs b/src/archive/parser.rs
index cad15ad..cc14185 100644
--- a/src/archive/parser.rs
+++ b/src/archive/parser.rs
@@ -1,7 +1,9 @@
 //! Index parser.
 
+use std::collections::HashSet;
 use std::io::BufRead;
 use std::sync::mpsc::{channel, Receiver};
+use std::sync::Arc;
 use std::thread::spawn;
 
 use rayon::prelude::*;
@@ -9,7 +11,7 @@ use serde::de::Error;
 use serde_json::error::Result;
 use serde_json::from_str;
 
-use super::story::Story;
+use super::story::{Author, Story, Tag};
 
 const TRIM: &[char] = &['"', ',', ' ', '\t', '\n', '\r'];
 
@@ -46,9 +48,48 @@ pub fn parse(reader: impl BufRead) -> Result<Vec<Story>> {
         return Err(Error::custom("Invalid file structure"));
     }
 
-    rx.recv().map_err(|e| match e {
+    let result = rx.recv().map_err(|e| match e {
         _ => Error::custom("Missing parser result"),
-    })?
+    })?;
+
+    Ok(dedup(result?))
+}
+
+fn dedup(mut stories: Vec<Story>) -> Vec<Story> {
+    let mut authors: HashSet<Arc<Author>> = HashSet::new();
+    let mut tags: HashSet<Arc<Tag>> = HashSet::new();
+
+    for story in stories.iter_mut() {
+        if let Some(author) = authors.get(&story.author) {
+            story.author = author.clone();
+        } else {
+            authors.insert(story.author.clone());
+        }
+    }
+
+    for story in stories.iter_mut() {
+        let unseen = story
+            .tags
+            .iter()
+            .filter(|tag| !tags.contains(*tag))
+            .map(|tag| tag.clone())
+            .collect::<Vec<_>>();
+
+        tags.extend(unseen);
+
+        story.tags = story
+            .tags
+            .iter()
+            .filter_map(|tag| tags.get(tag))
+            .map(|tag| tag.clone())
+            .collect();
+    }
+
+    for tag in tags.iter() {
+        println!("{}: {}", tag.name, Arc::strong_count(tag));
+    }
+
+    stories
 }
 
 fn spawn_parser(stream: Receiver<String>) -> Receiver<Result<Vec<Story>>> {
diff --git a/src/archive/story.rs b/src/archive/story.rs
index 092a738..c6e133b 100644
--- a/src/archive/story.rs
+++ b/src/archive/story.rs
@@ -1,7 +1,6 @@
 //! Story meta.
 
-use std::collections::HashSet;
-use std::sync::Mutex;
+use std::sync::Arc;
 
 use chrono::prelude::*;
 
@@ -9,16 +8,10 @@ use serde::de::Error;
 use serde::{Deserialize, Deserializer};
 use serde_json::Value;
 
-use lazy_static::lazy_static;
-
-lazy_static! {
-    static ref TAGS: Mutex<HashSet<&'static Tag>> = Mutex::new(HashSet::new());
-}
-
 #[derive(Clone, Debug, Deserialize)]
 pub struct Story {
     pub archive: Archive,
-    pub author: Author,
+    pub author: Arc<Author>,
     pub chapters: Vec<Chapter>,
     pub color: Option<Color>,
     pub completion_status: CompletionStatus,
@@ -43,8 +36,7 @@ pub struct Story {
     pub short_description: String,
     pub status: Status,
     pub submitted: bool,
-    #[serde(deserialize_with = "interned_tag")]
-    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 +52,7 @@ pub struct Archive {
     pub path: String,
 }
 
-#[derive(Clone, Debug, Deserialize)]
+#[derive(Clone, Debug, Hash, PartialEq, Eq, Deserialize)]
 pub struct Author {
     pub avatar: Option<Avatar>,
     pub bio_html: Option<String>,
@@ -74,7 +66,7 @@ pub struct Author {
     pub url: String,
 }
 
-#[derive(Clone, Debug, Deserialize)]
+#[derive(Clone, Debug, Hash, PartialEq, Eq, Deserialize)]
 pub struct Avatar {
     #[serde(rename = "16")]
     pub x16: Option<String>,
@@ -200,27 +192,6 @@ where
     }
 }
 
-fn interned_tag<'de, D>(d: D) -> Result<Vec<&'static Tag>, D::Error>
-where
-    D: Deserializer<'de>,
-{
-    let tags = Vec::<Tag>::deserialize(d)?;
-    let mut store = TAGS.lock().unwrap();
-
-    Ok(tags
-        .into_iter()
-        .map(|tag| match store.get(&tag) {
-            Some(tag) => tag,
-            None => {
-                let boxed: Box<Tag> = Box::new(tag);
-                let leaked: &'static Tag = Box::leak(boxed);
-                store.insert(leaked);
-                leaked
-            }
-        })
-        .collect())
-}
-
 impl<'de> Deserialize<'de> for Color {
     fn deserialize<D>(d: D) -> Result<Color, D::Error>
     where