Fetch and return metadata

lib/manifique/agent.rb

@@ -3,6 +3,7 @@ require "manifique/web_client"
 
 module Manifique
   class Agent
+
     def initialize(options={})
       @options = options
 
@@ -15,7 +16,7 @@ module Manifique
 
     def fetch_metadata
       web_client = WebClient.new(url: @url)
-      web_client.fetch_web_manifest
+      web_client.fetch_metadata
     end
 
     private
@@ -27,5 +28,6 @@ module Manifique
     rescue URI::InvalidURIError
       false
     end
+
   end
 end

lib/manifique/metadata.rb

@@ -0,0 +1,15 @@
+require 'ostruct'
+
+module Manifique
+  class Metadata
+    attr_accessor :manifest
+
+    def initialize
+    end
+
+    def to_json
+      # TODO serialize into JSON
+    end
+
+  end
+end

lib/manifique/web_client.rb

@@ -1,21 +1,37 @@
+require 'ostruct'
 require 'faraday'
 require 'faraday_middleware'
 require "nokogiri"
-require 'nitlink/response'
-
+require 'manifique/metadata'
 require 'pry'
 
 module Manifique
   class WebClient
+
     def initialize(options={})
       @options = options
       @url = options[:url]
+      @metadata = Metadata.new
+    end
+
+    def fetch_metadata
+      fetch_website
+      manifest = fetch_web_manifest
+
+      if @metadata.manifest = manifest
+        return @metadata
+      else
+        #TODO assemble from HTML elements
+      end
+
+      @metadata
     end
 
     def fetch_website
       res = do_get_request @url
-      @links = parse_http_link_header(res)
       @html = Nokogiri::HTML(res.body)
+    rescue
+      false
     end
 
     def fetch_web_manifest
@@ -23,7 +39,7 @@ module Manifique
 
       unless manifest_url.match(/^https?\:\/\//)
         # Link is just the manifest path, not an absolute URL
-        manifest_url = @url + manifest_url
+        manifest_url = [@url.gsub(/\/$/, ''), manifest_url.gsub(/^\//, '')].join('/')
       end
 
       res = do_get_request manifest_url
@@ -39,22 +55,18 @@ module Manifique
         b.adapter :net_http
       end
       res = conn.get url
-      if res.status > 400
-        raise "Could not fetch #{url} successfully (#{res.status})"
-      else
+      if res.status < 400
         res
+      else
+        raise "Could not fetch #{url} successfully (#{res.status})"
       end
     end
 
-    def parse_http_link_header(response)
-      link_parser = Nitlink::Parser.new
-      link_parser.parse(response)
-    end
-
     def discover_web_manifest_url(html)
       html.at_css("link[rel=manifest]").attributes["href"].value
     rescue
       false
     end
+
   end
 end

spec/manifique/agent_spec.rb

@@ -1,6 +1,7 @@
 require "spec_helper"
 
 RSpec.describe Manifique::Agent do
+
   describe "options" do
     describe "URL validation" do
       context "with invalid URL" do
@@ -19,4 +20,9 @@ RSpec.describe Manifique::Agent do
       end
     end
   end
+
+  describe "#fetch_metadata" do
+
+  end
+
 end

spec/manifique/metadata_spec.rb

@@ -0,0 +1,7 @@
+require "spec_helper"
+require "manifique/metadata"
+
+RSpec.describe Manifique::Metadata do
+# RSpec.describe "Manifique::MetaData" do
+
+end

spec/manifique/web_client_spec.rb

@@ -2,6 +2,7 @@ require "spec_helper"
 require "manifique/web_client"
 
 RSpec.describe Manifique::WebClient do
+
   describe "#do_get_request" do
     before do
       stub_request(:get, "http://example.com/404").
@@ -49,7 +50,29 @@ RSpec.describe Manifique::WebClient do
     end
   end
 
+  describe "#fetch_website" do
+    let(:web_client) { Manifique::WebClient.new(url: "https://kosmos.social/") }
+
+    before do
+      index_html = File.read(File.join(__dir__, "..", "fixtures", "mastodon.html"));
+      stub_request(:get, "https://kosmos.social/").
+        to_return(body: index_html, status: 200, headers: {
+          "Content-Type": "text/html; charset=utf-8"
+        })
+
+      web_client.fetch_website
+    end
+
+    it "instantiates an HTML parser object" do
+      html = web_client.instance_variable_get("@html")
+
+      expect(html).to be_kind_of(Nokogiri::HTML::Document)
+    end
+  end
+
   describe "#fetch_web_manifest" do
+    let(:web_client) { Manifique::WebClient.new(url: "https://kosmos.social/") }
+
     context "link[rel=manifest] present" do
       before do
         index_html = File.read(File.join(__dir__, "..", "fixtures", "mastodon.html"));
@@ -62,12 +85,11 @@ RSpec.describe Manifique::WebClient do
           to_return(body: manifest, status: 200, headers: {
             "Content-Type": "application/json; charset=utf-8"
           })
+
+        web_client.fetch_website
       end
 
-      let(:web_client) { Manifique::WebClient.new(url: "https://kosmos.social") }
-
       subject do
-        web_client.fetch_website
         web_client.fetch_web_manifest
       end
 
@@ -84,12 +106,11 @@ RSpec.describe Manifique::WebClient do
           to_return(body: index_html, status: 200, headers: {
             "Content-Type": "text/html; charset=utf-8"
           })
+
+        web_client.fetch_website
       end
 
-      let(:web_client) { Manifique::WebClient.new(url: "https://kosmos.social") }
-
       subject do
-        web_client.fetch_website
         web_client.fetch_web_manifest
       end
 
@@ -98,4 +119,35 @@ RSpec.describe Manifique::WebClient do
       end
     end
   end
+
+  describe "#fetch_metadata" do
+    let(:web_client) { Manifique::WebClient.new(url: "https://kosmos.social/") }
+
+    context "web app manifest present" do
+      before do
+        index_html = File.read(File.join(__dir__, "..", "fixtures", "mastodon.html"));
+        stub_request(:get, "https://kosmos.social/").
+          to_return(body: index_html, status: 200, headers: {
+            "Content-Type": "text/html; charset=utf-8"
+          })
+        manifest = File.read(File.join(__dir__, "..", "fixtures", "mastodon-web-app-manifest.json"));
+        stub_request(:get, "https://kosmos.social/mastodon-web-app-manifest.json").
+          to_return(body: manifest, status: 200, headers: {
+            "Content-Type": "application/json; charset=utf-8"
+          })
+      end
+
+      subject { web_client.fetch_metadata }
+
+      it "returns a metadata object" do
+        expect(subject).to be_kind_of(Manifique::Metadata)
+      end
+
+      it "stores the web app manifest data" do
+        expect(subject.manifest).to be_kind_of(Hash)
+        expect(subject.manifest["name"]).to eq("kosmos.social")
+      end
+    end
+  end
+
 end

spec/manifique_spec.rb

@@ -1,6 +1,7 @@
 require "spec_helper"
 
 RSpec.describe Manifique do
+
   it "has a version number" do
     expect(Manifique::VERSION).not_to be nil
   end