Add option to set page.date automatically with the file creation date

.gitignore

@@ -2,6 +2,7 @@
 _site/
 Gemfile.lock
 spec/fixtures/_posts/1992-09-11-last-modified-at.md
+spec/fixtures/_posts/date-no.md
 spec/fixtures/.jekyll-metadata
 spec/fixtures/.jekyll-cache
 spec/dev/out.txt

README.md

@@ -82,3 +82,18 @@ To format such a time, you'll need to rely on Liquid's `date` filter:
 ```
 
 (It's generally [more performant to use the `page.last_modified_at` version](https://github.com/gjtorikian/jekyll-last-modified-at/issues/24#issuecomment-55431108) of this plugin.)
+
+## `page.date`
+
+Additionally, this plugin automatically sets a default `date` value on every page based on when the file was **first** commited in git. To disable this, set `set-page-date` to `false` in your config yaml:
+
+ ```yml
+plugins:
+  - jekyll-last-modified-at
+
+last-modified-at:
+    set-page-date: false
+```
+
+If a post's date is already set via [the filename](https://jekyllrb.com/docs/posts/#creating-posts) or a page's date is set in its [frontmatter](https://jekyllrb.com/docs/variables/#page-variables), those values will override the value provided by this plugin. If a git date isn't available, `ctime` is used.
+

lib/jekyll-last-modified-at/determinator.rb

@@ -4,41 +4,63 @@ module Jekyll
   module LastModifiedAt
     class Determinator
       @repo_cache = {}
-      @path_cache = {}
+      @last_mod_cache = {}
+      @first_mod_cache = {}
       class << self
         # attr_accessor so we can flush externally
         attr_accessor :repo_cache
-        attr_accessor :path_cache
+        attr_accessor :last_mod_cache
+        attr_accessor :first_mod_cache
       end
 
       attr_reader :site_source, :page_path, :use_git_cache
       attr_accessor :format
 
-      def initialize(site_source, page_path, format = nil, use_git_cache = true) # rubocop:disable Style/OptionalBooleanParameter
+      def initialize(site_source, page_path, format = nil, use_git_cache = true, first_time = false) # rubocop:disable Style/OptionalBooleanParameter
         @site_source   = site_source
         @page_path     = page_path
         @format        = format || '%d-%b-%y'
         @use_git_cache = use_git_cache
+        @first_time    = first_time
       end
 
       def git
         return self.class.repo_cache[site_source] unless self.class.repo_cache[site_source].nil?
 
         self.class.repo_cache[site_source] = Git.new(site_source)
-        self.class.repo_cache[site_source]
       end
 
       def formatted_last_modified_date
         last_modified_at_time.strftime(@format)
       end
 
+      def formatted_first_modified_date
+        first_modified_at_time.strftime(@format)
+      end
+
+      def first_modified_at_time
+        return self.class.first_mod_cache[page_path] unless self.class.first_mod_cache[page_path].nil?
+
+        raise Errno::ENOENT, "#{absolute_path_to_article} does not exist!" unless File.exist? absolute_path_to_article
+
+        self.class.first_mod_cache[page_path] = Time.at(first_modified_at_unix.to_i)
+      end
+
+      def first_modified_at_unix
+        if git.git_repo?
+          first_commit_date = git.first_commit_date(relative_path_from_git_dir, use_git_cache)
+          first_commit_date.nil? || first_commit_date.empty? ? ctime(absolute_path_to_article) : first_commit_date
+        else
+          ctime(absolute_path_to_article)
+        end
+      end
+
       def last_modified_at_time
-        return self.class.path_cache[page_path] unless self.class.path_cache[page_path].nil?
+        return self.class.last_mod_cache[page_path] unless self.class.last_mod_cache[page_path].nil?
 
         raise Errno::ENOENT, "#{absolute_path_to_article} does not exist!" unless File.exist? absolute_path_to_article
 
-        self.class.path_cache[page_path] = Time.at(last_modified_at_unix.to_i)
-        self.class.path_cache[page_path]
+        self.class.last_mod_cache[page_path] = Time.at(last_modified_at_unix.to_i)
       end
 
       def last_modified_at_unix
@@ -51,12 +73,36 @@ def last_modified_at_unix
         end
       end
 
+      def to_i
+        if @first_time
+          @to_i ||= first_modified_at_unix.to_i
+        else
+          @to_i ||= last_modified_at_unix.to_i
+        end
+      end
+
       def to_s
-        @to_s ||= formatted_last_modified_date
+        if @first_time
+          @to_s ||= formatted_first_modified_date
+        else
+          @to_s ||= formatted_last_modified_date
+        end
       end
 
       def to_liquid
-        @to_liquid ||= last_modified_at_time
+        if @first_time
+          @to_liquid ||= first_modified_at_time
+        else
+          @to_liquid ||= last_modified_at_time
+        end
+      end
+
+      def to_time
+        to_liquid
+      end
+
+      def strftime(*args)
+        return to_liquid().strftime(*args)
       end
 
       private
@@ -77,6 +123,10 @@ def relative_path_from_git_dir
       def mtime(file)
         File.mtime(file).to_i.to_s
       end
+
+      def ctime(file)
+        File.ctime(file).to_i.to_s
+      end
     end
   end
 end

lib/jekyll-last-modified-at/git.rb

@@ -9,6 +9,7 @@ def initialize(site_source)
         @site_source = site_source
         @is_git_repo = nil
         @lcd_cache = {}
+        @lce_cache = {}
       end
 
       def top_level_directory
@@ -37,7 +38,7 @@ def git_repo?
 
       def last_commit_date(path, use_git_cache = false) # rubocop:disable Style/OptionalBooleanParameter
         if use_git_cache
-          build_lcd_cache if @lcd_cache.empty?
+          build_cache if @lcd_cache.empty?
           @lcd_cache[path]
         else
           Executor.sh(
@@ -54,16 +55,35 @@ def last_commit_date(path, use_git_cache = false) # rubocop:disable Style/Option
         end
       end
 
+      def first_commit_date(path, use_git_cache = false) # rubocop:disable Style/OptionalBooleanParameter
+        if use_git_cache
+          build_cache if @lce_cache.empty?
+          @lce_cache[path]
+        else
+          Executor.sh(
+            'git',
+            '--git-dir',
+            top_level_directory,
+            'log',
+            '--follow',
+            '--diff-filter=A',
+            '--format="%ct"',
+            '--',
+            path
+          )&.split("\n")&[-1]&[/\d+/]
+        end
+      end
+
       private
 
       # generates hash of `path => unix time stamp (string)`
-      def build_lcd_cache
+      def build_cache
         # example output:
         #
-        # %jekyll-last-modified-at:1621042992
+        # %these-files-modified-at:1621042992
         #
         # Dockerfile.production
-        # %jekyll-last-modified-at:1621041929
+        # %these-files-modified-at:1621041929
         #
         # assets/css/style.52513a5600efd4015668ccb9b702256e.css
         # assets/css/style.52513a5600efd4015668ccb9b702256e.css.gz
@@ -74,24 +94,21 @@ def build_lcd_cache
           'log',
           '--name-only',
           '--date=unix',
-          '--pretty=%%jekyll-last-modified-at:%ct'
+          '--pretty=%%these-files-modified-at:%ct'
         )
 
-        lcd = nil
+        timestamp = nil
         lines.split("\n").each do |line|
           next if line.empty?
 
-          if line.start_with?('%jekyll-last-modified-at:')
+          if line.start_with?('%these-files-modified-at:')
             # new record
-            lcd = line.split(':')[1]
+            timestamp = line.split(':')[1]
             next
           end
 
-          # we already have it
-          next if @lcd_cache[line]
-
-          # we don't have it
-          @lcd_cache[line] = lcd
+          @lcd_cache[line] = timestamp unless @lcd_cache.key?(line)
+          @lce_cache[line] = timestamp
         end
       end
     end

lib/jekyll-last-modified-at/hook.rb

@@ -6,21 +6,18 @@ module Hook
       def self.add_determinator_proc
         proc { |item|
           format = item.site.config.dig('last-modified-at', 'date-format')
-          use_git_cache = item.site.config.dig('last-modified-at', 'use-git-cache')
+          use_git_cache = item.site.config.dig('last-modified-at', 'use-git-cache') != false
           item.data['last_modified_at'] = Determinator.new(item.site.source, item.relative_path,
                                                            format, use_git_cache)
+          if item.site.config.dig('last-modified-at', 'set-page-date') != false
+            # The "date" field will be converted to a string first by Jekyll and it must be
+            # in the format given below: https://jekyllrb.com/docs/variables/#page-variables
+            item.data['date'] = Determinator.new(item.site.source, item.relative_path,
+                                                 '%Y-%m-%d %H:%M:%S %z', use_git_cache, true)
+          end
         }
       end
 
-      Jekyll::Hooks.register :site, :after_reset do |site|
-        use_git_cache = site.config.dig('last-modified-at', 'use-git-cache')
-        if use_git_cache
-          # flush the caches so we can detect commits while server is running
-          Determinator.repo_cache = {}
-          Determinator.path_cache = {}
-        end
-      end
-
       Jekyll::Hooks.register :posts, :post_init, &Hook.add_determinator_proc
       Jekyll::Hooks.register :pages, :post_init, &Hook.add_determinator_proc
       Jekyll::Hooks.register :documents, :post_init, &Hook.add_determinator_proc

spec/fixtures/_layouts/date_with_format.html

@@ -0,0 +1,10 @@
+<html>
+  <head>
+    <title>date-with-format</title>
+  </head>
+  <body>
+    <span class="last-modified-at">Article created on {% page.date | date: "%Y:%B:%A:%d" %}</span>
+
+    {{ content }}
+  </body>
+</html>

spec/fixtures/_posts/no-date.md

@@ -0,0 +1,6 @@
+---
+title: Testing Last Modified At
+layout: date_with_format
+---
+
+Yay.

spec/jekyll-last-modified-at/date_spec.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe 'Date Tag' do
+  context 'A committed post file' do
+    def setup(file, layout)
+      @post = setup_post(file)
+      do_render(@post, layout)
+    end
+
+    it 'has date' do
+      setup('no-date.md', 'date_with_format.html')
+      expect(@post.output).to match(/Article created on 09-Nov-22/)
+    end
+  end
+
+  context 'An uncommitted post file' do
+    before(:all) do
+      cheater_file = 'no-date.md'
+      uncommitted_file = 'date_no.md'
+      duplicate_post(cheater_file, uncommitted_file)
+      @post = setup_post(uncommitted_file)
+      do_render(@post, 'date_with_format.html')
+    end
+
+    it 'has date' do
+      expect(@post.output).to match(Regexp.new("Article created on #{Time.new.utc.strftime('%d-%b-%y')}"))
+    end
+  end
+end