From 8cf133e74aea1653e10a130dbc7e1a85b49d129d Mon Sep 17 00:00:00 2001
From: Mat Moore <matmoore@users.noreply.github.com>
Date: Sun, 26 May 2024 16:41:54 +0100
Subject: [PATCH] Add mechanism to mark and prune subtrees

This enables you to trim down the vault after parsing the content, e.g.
you can filter out anything without a specific header, or that isn't
linked to by another page
---
 CHANGELOG.md                      | 11 +++++++++
 lib/obsidian/parser/page.rb       | 24 ++++++++++++++++++++
 spec/obsidian/parser/page_spec.rb | 37 +++++++++++++++++++++++++++++++
 3 files changed, 72 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 40533e9..4e7c6c0 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,35 +1,46 @@
 ## [Unreleased]
+
 - Added support for wikilinks that embed files. These are rendered as images or links in the HTML content.
 - Added method `#page.generate_html` to replace `#page.content.generate_html`, and removed the `MarkdownDocument` class. `#page.content` is now a callable that returns the markdown content.
+- Added `#mark_referenced` and `#referenced?` to page objects
+- Added `#prune!` method to page objects, to remove non-referenced pages
 
 ## [0.7.0] - 2023-08-03
+
 - Fix wikilinks pointing to slugs with spaces not rendering properly.
 - Links created from wikilinks now include a leading slash
 
 ## [0.6.1] - 2023-08-03
+
 - Prevent `HtmlRenderer` state being shared across documents
 
 ## [0.6.0] - 2023-08-03
+
 - Replace Kramdown with Markly
 - Enabled support for Github Flavored Markdown tables and tasklists
 - Rename `MarkdownContent` -> `MarkdownDocument`, `ObsidianFlavoredMarkdown` -> `MarkdownParser`
 
 ## [0.5.4] - 2023-08-02
+
 - Fix page getting clobbered when wikilinks point to non-existent pages.
 - Expand `[[foo/index]]` wiklinks to `[foo](foo)`.
 
 ## [0.5.3] - 2023-08-01
+
 - Support non-fully qualified titles when parsing wikilink syntax.
 - Autolink raw URLs.
 
 ## [0.5.2] - 2023-07-30
+
 - Fix handling of `index.md` at the root level.
 
 ## [0.5.0] - 2023-07-30
+
 - Fix ordering of `Page#children` so that index pages come first.
 - Fix handling of `index.md` documents so that the slug reflects the directory path.
 
 ## [0.4.0] - 2023-07-30
+
 - Unify `Note` and `Index` classes into `Page`. This is a breaking API change. `Parser#notes is replaced by Parse#pages`. Call `Page#is_index?`to distinguish between directory derived pages and documents.
 - Remove `Parser#table_of_contents` and `Parser#walk_tree`.
 - Add `Page#find_in_tree` to recursively search for a page with a matching slug.
diff --git a/lib/obsidian/parser/page.rb b/lib/obsidian/parser/page.rb
index 7560554..9606440 100644
--- a/lib/obsidian/parser/page.rb
+++ b/lib/obsidian/parser/page.rb
@@ -32,6 +32,7 @@ def initialize(title:, slug:, last_modified: nil, content: nil, parent: nil, con
       @content_type = content_type
       @media_root = media_root
       @source_path = source_path
+      @referenced = false
     end
 
     def is_index?
@@ -163,6 +164,29 @@ def generate_html(markdown_parser: MarkdownParser.new)
       markdown_parser.parse(content.call, root: root, media_root: media_root).to_html
     end
 
+    def referenced?
+      @referenced
+    end
+
+    # Mark the tree containing this page as being "referenced"
+    # i.e. reachable through links
+    def mark_referenced
+      @referenced = true
+      parent&.mark_referenced
+    end
+
+    # Remove any child paths that are unreferenced,
+    # i.e. not reachable through links
+    def prune!
+      @children = @children.delete_if do |k, v|
+        !v.referenced?
+      end
+
+      @children.values.each do |page|
+        page.prune!
+      end
+    end
+
     attr_reader :title
     attr_reader :slug
     attr_reader :last_modified
diff --git a/spec/obsidian/parser/page_spec.rb b/spec/obsidian/parser/page_spec.rb
index 815afbe..af46892 100644
--- a/spec/obsidian/parser/page_spec.rb
+++ b/spec/obsidian/parser/page_spec.rb
@@ -99,4 +99,41 @@
       expect(root.find_in_tree("foo/index")).to eq(page.parent)
     end
   end
+
+  describe "#referenced?" do
+    it "is false by default" do
+      expect(root.referenced?).to eq(false)
+    end
+
+    it "is true after #mark_referenced is called" do
+      root.mark_referenced
+      expect(root.referenced?).to eq(true)
+    end
+
+    it "is true after #mark_referenced is called on a child node" do
+      page = root.add_page("foo/bar")
+      page.mark_referenced
+      expect(page.referenced?).to eq(true)
+      expect(page.parent.referenced?).to eq(true)
+      expect(root.referenced?).to eq(true)
+    end
+  end
+
+  describe "#prune!" do
+    it "does not delete referenced pages" do
+      root.add_page("foo/bar").mark_referenced
+      root.prune!
+      expect(root.find_in_tree("foo/bar")).not_to be_nil
+    end
+
+    it "deletes unreferenced pages" do
+      root.add_page("foo/bar").mark_referenced
+      page = root.add_page("foo/baz")
+      expect(page.referenced?).to eq(false)
+      root.prune!
+      expect(root.find_in_tree("foo")).not_to be_nil
+      expect(root.find_in_tree("foo/bar")).not_to be_nil
+      expect(root.find_in_tree("foo/baz")).to be_nil
+    end
+  end
 end