From 567c52d80aafdd940e095bc8f3708ec4aa9a5097 Mon Sep 17 00:00:00 2001
From: Gabriel Arazas <foodogsquared@foodogsquared.one>
Date: Sat, 11 Mar 2023 16:34:12 +0800
Subject: [PATCH] Add SWHID include processor

---
 lib/asciidoctor/foodogsquared-extensions.rb   |  2 +
 .../swhid-include-processor/README.adoc       | 43 ++++++++++++++++
 .../swhid-include-processor/extension.rb      | 51 +++++++++++++++++++
 3 files changed, 96 insertions(+)
 create mode 100644 lib/asciidoctor/swhid-include-processor/README.adoc
 create mode 100644 lib/asciidoctor/swhid-include-processor/extension.rb

diff --git a/lib/asciidoctor/foodogsquared-extensions.rb b/lib/asciidoctor/foodogsquared-extensions.rb
index adb965a..af6de49 100644
--- a/lib/asciidoctor/foodogsquared-extensions.rb
+++ b/lib/asciidoctor/foodogsquared-extensions.rb
@@ -4,6 +4,7 @@ require 'asciidoctor'
 require 'asciidoctor/extensions'
 require_relative 'man-inline-macro/extension'
 require_relative 'swhid-inline-macro/extension'
+require_relative 'swhid-include-processor/extension'
 require_relative 'github-link-inline-macro/extension'
 require_relative 'github-raw-content-include-processor/extension'
 require_relative 'gitlab-link-inline-macro/extension'
@@ -13,6 +14,7 @@ Asciidoctor::Extensions.register do
   inline_macro ManInlineMacro
 
   inline_macro SWHInlineMacro
+  include_processor SWHIDIncludeProcessor
 
   inline_macro GitHubLinkInlineMacro
   include_processor GitHubRawIncludeProcessor
diff --git a/lib/asciidoctor/swhid-include-processor/README.adoc b/lib/asciidoctor/swhid-include-processor/README.adoc
new file mode 100644
index 0000000..6c8ed46
--- /dev/null
+++ b/lib/asciidoctor/swhid-include-processor/README.adoc
@@ -0,0 +1,43 @@
+= SWHID include processor
+:toc:
+
+
+This is an include processor extension for easily fetching SWHIDs, only with the `cnt` schema type.
+
+
+== Synopsis
+
+[source, asciidoc]
+----
+\include::$SWHID[]
+----
+
+Where `$SWHID` is a link:https://docs.softwareheritage.org/devel/swh-model/persistent-identifiers.html[SWHID].
+This could accept SWHIDs with qualifiers.
+
+Take note this include processor will only give the raw content with the `cnt` schema type.
+Anything else will be skipped and log a warning instead.
+
+
+
+== Extra notes
+
+[source, asciidoc]
+----
+= doctitle
+:swhid-gpl3: swh:1:cnt:94a9ed024d3859793618152ea559a168bbcbb5e2
+
+\include::{swhid-nixpkgs}[]
+----
+
+This include processor also respects the safe mode setting.
+This means in order to permit including by SWHID, you have to permit link:https://docs.asciidoctor.org/asciidoc/latest/directives/include-uri/[includes by URIs].
+
+Lastly, this include processor uses the Software Heritage API which includes a limitation.
+
+
+== Example usage
+
+- SWHID with a bare core identifier: `include::swh:1:cnt:94a9ed024d3859793618152ea559a168bbcbb5e2[]`.
+
+- SWHID with full contextual information: `include::swh:1:cnt:4c6ad635164b25b9bc2ebe17d2c3b7c0835f6035;origin=https://github.com/NixOS/nixpkgs;visit=swh:1:snp:6ea7d28dfd4789609e0be2b64179fc9c12931beb;anchor=swh:1:rev:7f5639fa3b68054ca0b062866dc62b22c3f11505;path=/README.md`.
diff --git a/lib/asciidoctor/swhid-include-processor/extension.rb b/lib/asciidoctor/swhid-include-processor/extension.rb
new file mode 100644
index 0000000..77d9a38
--- /dev/null
+++ b/lib/asciidoctor/swhid-include-processor/extension.rb
@@ -0,0 +1,51 @@
+# frozen_string_literal: true
+
+require 'json'
+require 'open-uri'
+require 'uri'
+
+class SWHIDIncludeProcessor < Asciidoctor::Extensions::IncludeProcessor
+  def handles?(target)
+    target.start_with? 'swh:'
+  end
+
+  def process(doc, reader, target, attributes)
+    swhid = target
+    swhid_core_identifier = swhid.split(';').at(0)
+    swhid_object_type = (swhid_core_identifier.split ':').at 2
+
+    unless (doc.safe <= Asciidoctor::SafeMode::SERVER) && (doc.attr? 'allow-uri-read')
+      raise %('swh:' include cannot be used in safe mode level > SERVER and without attribute 'allow-uri-read')
+    end
+
+    # We're already going to throw out anything that is not content object type
+    # just to make the later pipelines easier to construct.
+    if swhid_object_type != 'cnt'
+      warn %(SWHID '#{swhid_core_identifier}' is not of 'cnt' type; ignoring)
+      return reader
+    end
+
+    version = '1'
+
+    content = begin
+      uri = URI.parse %(https://archive.softwareheritage.org/api/#{version}/resolve/#{target}/)
+
+      headers = {}
+      headers['Authorization'] = "Bearer #{ENV['SWH_API_BEARER_TOKEN']}" if ENV['SWH_BEARER_TOKEN']
+      headers['Accept'] = 'application/json'
+
+      metadata = OpenURI.open_uri(uri, headers) { |f| JSON.parse(f.read) }
+      object_hash = metadata['object_id']
+
+      uri = URI.parse %(https://archive.softwareheritage.org/api/#{version}/content/sha1_git:#{object_hash}/raw/)
+      OpenURI.open_uri(uri, headers, &:read)
+    rescue OpenURI::HTTPError => e
+      warning = %(error while getting '#{swhid_core_identifier}': #{e})
+      warn warning
+      warning
+    end
+
+    reader.push_include content, target, target, 1, attributes
+    reader
+  end
+end