md_metadata_extractor 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 50071436090d012df8b8310ee9c0063c25486c8792856d9921f20870b7416cc1
-  data.tar.gz: 6a1c992cc84bd4a96d6a6adf77ea17f268eef3c2d8a7df88bb0c687a7246810b
+  metadata.gz: a9f0b91416a5b9ae2f53c23a870285a0a2a9866558cbf84b4203c413084b32b4
+  data.tar.gz: dde346391a2ad7a160a8c710b16aa68b625f55e76e084e90e426623389dfa9d7
 SHA512:
-  metadata.gz: ee81b949e9e842fc947290ac662059424c52f95c3ec53d81f0e35fc4cb7a84baacda45dae4384768a18df1c3a8bb285e8b3b3eea425fee99c2ef7aa73fa2711d
-  data.tar.gz: 1b504b93576c7bed3f3787b722b33b304a13f54aebbf0a93dc631837dc4a0588e72f62b39f8e9021938a92c1dd0248c95eb361b3f7e195245e167c84689df10b
+  metadata.gz: 14c93f64ddbd586b8517910a6b8dbf0b34a6e6a9033c37425cb36066d9e7be5fd75627a5e8f4d31cef87adbfec07cb97c73013aeef67dbcd67e1c8c68adc84f8
+  data.tar.gz: 69be39484b1196899e19bdfa805cdaa5c7de644ecd90acad95b1baeedb0c660867acd0f680b54cc3ee53e73782e1de33e4c23be381a9439dd29cd9ed6eb35957

data/lib/md_metadata_extractor.rb

@@ -2,40 +2,52 @@ require "pp"
 module MdMetadata
   class Extractor
     attr_accessor :mode
-    def initialize
+    def initialize(mode = :hugo)
       # Define input data
       @input = ""
-      @result = {}
-      @mode = :hugo
+      @result = { :metadata => {} , :content => ""}
+      @mode = mode
+    end
+    def metadata_regex
+        /#{Regexp.quote(delimiter)}(.*)#{Regexp.quote(delimiter)}/m
+    end
+    def content_regex
+        /#{Regexp.quote(delimiter)}.*#{Regexp.quote(delimiter)}(.*)\z/m
+    end
+    def delimiter
+      case @mode
+      when :hugo
+        "+++"
+      when :jekyll
+        "---"
+      end
     end
 
-    ##
-    # Loads an input string for processing, returns a hash of values
-    # based on said metadata
-    def load_md(input)
-      @input = input
-      @metadata = ""
+    def separator
       case @mode
       when :hugo
-        @metadata = /\+\+\+(.*)\+\+\+/m.match(@input).to_s.delete("+++")
+        "="
       when :jekyll
-        @metadata = /\-\-\-(.*)\-\-\-/m.match(@input).to_s.delete("---")
+        ":"
       end
+    end
+    
+    ##
+    # Loads an input string for processing, returns a hash of values
+    # based on said metadata
+    def extract(input)
+      raise "Malformed input file" unless metadata_regex.match?(input)
+      @metadata = metadata_regex.match(input).to_s.delete(delimiter)
       @metadata.each_line do |line|
-        case @mode
-        when :hugo
-          meta = line.split("=")
-        when :jekyll
-          meta = line.split(":")
-        end
+        meta = line.split(separator)
         unless meta[1].nil?
           meta[0].strip!
-          meta[1].strip!.gsub! '"', ''
-          @result[meta[0]] = meta[1]
+          meta[1].gsub!( '"', ' ').strip!
+          @result[:metadata][meta[0]] = meta[1]
         end
       end
+      @result[:content] = content_regex.match(input)[-1].to_s.strip!
       @result
     end
-
   end
 end

data/test/md_metadata_extractor_test.rb

@@ -8,8 +8,10 @@ class TestExtractor < Minitest::Test
         ---
         title : "My Title"
         date : "2014-11-17 19:38:31 +0000 UTC"
+
         description : "here goes the post description"
         keywords : "here goes the metadata"
+
         ---
         Lorem ipsum dolor sit amet, consectetur adipiscing elit.
     MD
@@ -22,18 +24,38 @@ class TestExtractor < Minitest::Test
         +++ 
         Lorem ipsum dolor sit amet, consectetur adipiscing elit.
     MD
+    @testCaseBroken = <<-MD
+        +-+
+        title = "My Title"
+        date  "2014-11-17 19:38:31 +0000 UTC" =
+        description = "here goes the post description"
+        keywords = "here goes the metadata"
+        +++ 
+        Lorem ipsum dolor sit amet, consectetur adipiscing elit.
+    MD
   end
-  def test_load_md
-    output = @importer.load_md @testCaseHugo
+  def test_extract
+    output = @importer.extract @testCaseHugo
     assert_instance_of Hash, output
   end
   def test_extracts_title_from_hugo_metadata
-    output = @importer.load_md @testCaseHugo
-    assert_equal "My Title", output["title"]
+    output = @importer.extract @testCaseHugo
+    assert_equal "My Title", output[:metadata]["title"]
   end
   def test_extracts_title_from_jekyll_metadata
     @importer.mode = :jekyll
-    output = @importer.load_md @testCaseJekyll
-    assert_equal "My Title", output["title"]
-  end 
+    output = @importer.extract @testCaseJekyll
+    assert_equal "My Title", output[:metadata]["title"]
+  end
+  def test_extracts_content_from_hugo_md
+    output = @importer.extract @testCaseHugo
+    assert_equal "Lorem ipsum dolor sit amet, consectetur adipiscing elit.", output[:content]
+  end
+
+  
+  def test_raises_exception_on_malformed_file
+    assert_raises RuntimeError do
+      output = @importer.extract @testCaseBroken
+    end
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: md_metadata_extractor
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Juan Francisco Giménez Silva