Merge pull request #232 from hkalexling/rc/0.24.0

v0.24.0

Author: hkalexling

.dockerignore

@@ -1,2 +1,9 @@
 node_modules
 lib
+Dockerfile
+Dockerfile.arm32v7
+Dockerfile.arm64v8
+README.md
+.all-contributorsrc
+env.example
+.github/

README.md

@@ -51,7 +51,7 @@ The official docker images are available on [Dockerhub](https://hub.docker.com/r
 ### CLI
 
 ```
-  Mango - Manga Server and Web Reader. Version 0.23.0
+  Mango - Manga Server and Web Reader. Version 0.24.0
 
   Usage:
 
@@ -86,6 +86,10 @@ log_level: info
 upload_path: ~/mango/uploads
 plugin_path: ~/mango/plugins
 download_timeout_seconds: 30
+library_cache_path: ~/mango/library.yml.gz
+cache_enabled: false
+cache_size_mbs: 50
+cache_log_enabled: true
 disable_login: false
 default_username: ""
 auth_proxy_header_name: ""
@@ -97,12 +101,12 @@ mangadex:
   download_queue_db_path: ~/mango/queue.db
   chapter_rename_rule: '[Vol.{volume} ][Ch.{chapter} ]{title|id}'
   manga_rename_rule: '{title}'
-  subscription_update_interval_hours: 24
 ```
 
 - `scan_interval_minutes`, `thumbnail_generation_interval_hours` and `db_optimization_interval_hours` can be any non-negative integer. Setting them to `0` disables the periodic tasks
 - `log_level` can be `debug`, `info`, `warn`, `error`, `fatal` or `off`. Setting it to `off` disables the logging
 - You can disable authentication by setting `disable_login` to true. Note that `default_username` must be set to an existing username for this to work.
+- By setting `cache_enabled` to `true`, you can enable an experimental feature where Mango caches library metadata to improve page load time. You can further fine-tune the feature with `cache_size_mbs` and `cache_log_enabled`.
 
 ### Library Structure
 

shard.yml

@@ -1,5 +1,5 @@
 name: mango
-version: 0.23.0
+version: 0.24.0
 
 authors:
   - Alex Ling <hkalexling@gmail.com>

spec/util_spec.cr

@@ -61,3 +61,13 @@ describe "chapter_sort" do
     end.should eq ary
   end
 end
+
+describe "sanitize_filename" do
+  it "returns a random string for empty sanitized string" do
+    sanitize_filename("..").should_not eq sanitize_filename("..")
+  end
+  it "sanitizes correctly" do
+    sanitize_filename("..  \n\v.\rマンゴー/|*()<[1/2] 3.14 hello world ")
+      .should eq "マンゴー_()[1_2] 3.14 hello world"
+  end
+end

src/config.cr

@@ -11,6 +11,8 @@ class Config
   property session_secret : String = "mango-session-secret"
   property library_path : String = File.expand_path "~/mango/library",
     home: true
+  property library_cache_path = File.expand_path "~/mango/library.yml.gz",
+    home: true
   property db_path : String = File.expand_path "~/mango/mango.db", home: true
   property scan_interval_minutes : Int32 = 5
   property thumbnail_generation_interval_hours : Int32 = 24
@@ -20,6 +22,9 @@ class Config
   property plugin_path : String = File.expand_path "~/mango/plugins",
     home: true
   property download_timeout_seconds : Int32 = 30
+  property cache_enabled = false
+  property cache_size_mbs = 50
+  property cache_log_enabled = true
   property disable_login = false
   property default_username = ""
   property auth_proxy_header_name = ""

src/library/cache.cr

@@ -0,0 +1,188 @@
+require "digest"
+
+require "./entry"
+require "./types"
+
+# Base class for an entry in the LRU cache.
+# There are two ways to use it:
+#   1. Use it as it is by instantiating with the appropriate `SaveT` and
+#     `ReturnT`. Note that in this case, `SaveT` and `ReturnT` must be the
+#     same type. That is, the input value will be stored as it is without
+#     any transformation.
+#   2. You can also subclass it and provide custom implementations for
+#     `to_save_t` and `to_return_t`. This allows you to transform and store
+#     the input value to a different type. See `SortedEntriesCacheEntry` as
+#     an example.
+private class CacheEntry(SaveT, ReturnT)
+  getter key : String, atime : Time
+
+  @value : SaveT
+
+  def initialize(@key : String, value : ReturnT)
+    @atime = @ctime = Time.utc
+    @value = self.class.to_save_t value
+  end
+
+  def value
+    @atime = Time.utc
+    self.class.to_return_t @value
+  end
+
+  def self.to_save_t(value : ReturnT)
+    value
+  end
+
+  def self.to_return_t(value : SaveT)
+    value
+  end
+
+  def instance_size
+    instance_sizeof(CacheEntry(SaveT, ReturnT)) + # sizeof itself
+      instance_sizeof(String) + @key.bytesize +   # allocated memory for @key
+      @value.instance_size
+  end
+end
+
+class SortedEntriesCacheEntry < CacheEntry(Array(String), Array(Entry))
+  def self.to_save_t(value : Array(Entry))
+    value.map &.id
+  end
+
+  def self.to_return_t(value : Array(String))
+    ids_to_entries value
+  end
+
+  private def self.ids_to_entries(ids : Array(String))
+    e_map = Library.default.deep_entries.to_h { |entry| {entry.id, entry} }
+    entries = [] of Entry
+    begin
+      ids.each do |id|
+        entries << e_map[id]
+      end
+      return entries if ids.size == entries.size
+    rescue
+    end
+  end
+
+  def instance_size
+    instance_sizeof(SortedEntriesCacheEntry) +  # sizeof itself
+      instance_sizeof(String) + @key.bytesize + # allocated memory for @key
+      @value.size * (instance_sizeof(String) + sizeof(String)) +
+      @value.sum(&.bytesize) # elements in Array(String)
+  end
+
+  def self.gen_key(book_id : String, username : String,
+                   entries : Array(Entry), opt : SortOptions?)
+    entries_sig = Digest::SHA1.hexdigest (entries.map &.id).to_s
+    user_context = opt && opt.method == SortMethod::Progress ? username : ""
+    sig = Digest::SHA1.hexdigest (book_id + entries_sig + user_context +
+                                  (opt ? opt.to_tuple.to_s : "nil"))
+    "#{sig}:sorted_entries"
+  end
+end
+
+class String
+  def instance_size
+    instance_sizeof(String) + bytesize
+  end
+end
+
+struct Tuple(*T)
+  def instance_size
+    sizeof(T) + # total size of non-reference types
+      self.sum do |e|
+        next 0 unless e.is_a? Reference
+        if e.responds_to? :instance_size
+          e.instance_size
+        else
+          instance_sizeof(typeof(e))
+        end
+      end
+  end
+end
+
+alias CacheableType = Array(Entry) | String | Tuple(String, Int32)
+alias CacheEntryType = SortedEntriesCacheEntry |
+                       CacheEntry(String, String) |
+                       CacheEntry(Tuple(String, Int32), Tuple(String, Int32))
+
+def generate_cache_entry(key : String, value : CacheableType)
+  if value.is_a? Array(Entry)
+    SortedEntriesCacheEntry.new key, value
+  else
+    CacheEntry(typeof(value), typeof(value)).new key, value
+  end
+end
+
+# LRU Cache
+class LRUCache
+  @@limit : Int128 = Int128.new 0
+  @@should_log = true
+  # key => entry
+  @@cache = {} of String => CacheEntryType
+
+  def self.enabled
+    Config.current.cache_enabled
+  end
+
+  def self.init
+    cache_size = Config.current.cache_size_mbs
+    @@limit = Int128.new cache_size * 1024 * 1024 if enabled
+    @@should_log = Config.current.cache_log_enabled
+  end
+
+  def self.get(key : String)
+    return unless enabled
+    entry = @@cache[key]?
+    if @@should_log
+      Logger.debug "LRUCache #{entry.nil? ? "miss" : "hit"} #{key}"
+    end
+    return entry.value unless entry.nil?
+  end
+
+  def self.set(cache_entry : CacheEntryType)
+    return unless enabled
+    key = cache_entry.key
+    @@cache[key] = cache_entry
+    Logger.debug "LRUCache cached #{key}" if @@should_log
+    remove_least_recent_access
+  end
+
+  def self.invalidate(key : String)
+    return unless enabled
+    @@cache.delete key
+  end
+
+  def self.print
+    return unless @@should_log
+    sum = @@cache.sum { |_, entry| entry.instance_size }
+    Logger.debug "---- LRU Cache ----"
+    Logger.debug "Size: #{sum} Bytes"
+    Logger.debug "List:"
+    @@cache.each do |k, v|
+      Logger.debug "#{k} | #{v.atime} | #{v.instance_size}"
+    end
+    Logger.debug "-------------------"
+  end
+
+  private def self.is_cache_full
+    sum = @@cache.sum { |_, entry| entry.instance_size }
+    sum > @@limit
+  end
+
+  private def self.remove_least_recent_access
+    if @@should_log && is_cache_full
+      Logger.debug "Removing entries from LRUCache"
+    end
+    while is_cache_full && @@cache.size > 0
+      min_tuple = @@cache.min_by { |_, entry| entry.atime }
+      min_key = min_tuple[0]
+      min_entry = min_tuple[1]
+
+      Logger.debug "  \
+        Target: #{min_key}, \
+        Last Access Time: #{min_entry.atime}" if @@should_log
+      invalidate min_key
+    end
+  end
+end

src/library/entry.cr

@@ -1,6 +1,9 @@
 require "image_size"
+require "yaml"
 
 class Entry
+  include YAML::Serializable
+
   getter zip_path : String, book : Title, title : String,
     size : String, pages : Int32, id : String, encoded_path : String,
     encoded_title : String, mtime : Time, err_msg : String?
@@ -46,31 +49,23 @@ class Entry
     file.close
   end
 
-  def to_slim_json : String
+  def build_json(*, slim = false)
     JSON.build do |json|
       json.object do
         {% for str in ["zip_path", "title", "size", "id"] %}
         json.field {{str}}, @{{str.id}}
       {% end %}
         json.field "title_id", @book.id
         json.field "pages" { json.number @pages }
+        unless slim
+          json.field "display_name", @book.display_name @title
+          json.field "cover_url", cover_url
+          json.field "mtime" { json.number @mtime.to_unix }
+        end
       end
     end
   end
 
-  def to_json(json : JSON::Builder)
-    json.object do
-      {% for str in ["zip_path", "title", "size", "id"] %}
-        json.field {{str}}, @{{str.id}}
-      {% end %}
-      json.field "title_id", @book.id
-      json.field "display_name", @book.display_name @title
-      json.field "cover_url", cover_url
-      json.field "pages" { json.number @pages }
-      json.field "mtime" { json.number @mtime.to_unix }
-    end
-  end
-
   def display_name
     @book.display_name @title
   end
@@ -81,9 +76,17 @@ class Entry
 
   def cover_url
     return "#{Config.current.base_url}img/icon.png" if @err_msg
+
+    unless @book.entry_cover_url_cache
+      TitleInfo.new @book.dir do |info|
+        @book.entry_cover_url_cache = info.entry_cover_url
+      end
+    end
+    entry_cover_url = @book.entry_cover_url_cache
+
     url = "#{Config.current.base_url}api/cover/#{@book.id}/#{@id}"
-    TitleInfo.new @book.dir do |info|
-      info_url = info.entry_cover_url[@title]?
+    if entry_cover_url
+      info_url = entry_cover_url[@title]?
       unless info_url.nil? || info_url.empty?
         url = File.join Config.current.base_url, info_url
       end
@@ -170,6 +173,16 @@ class Entry
   # For backward backward compatibility with v0.1.0, we save entry titles
   #   instead of IDs in info.json
   def save_progress(username, page)
+    LRUCache.invalidate "#{@book.id}:#{username}:progress_sum"
+    @book.parents.each do |parent|
+      LRUCache.invalidate "#{parent.id}:#{username}:progress_sum"
+    end
+    [false, true].each do |ascend|
+      sorted_entries_cache_key = SortedEntriesCacheEntry.gen_key @book.id,
+        username, @book.entries, SortOptions.new(SortMethod::Progress, ascend)
+      LRUCache.invalidate sorted_entries_cache_key
+    end
+
     TitleInfo.new @book.dir do |info|
       if info.progress[username]?.nil?
         info.progress[username] = {@title => page}