pHash 1.0.2 → 1.1.0

data/lib/phash.rb

@@ -1,13 +1,23 @@
 require 'ffi'
 
 module Phash
-  class HashData
+  class Data
     attr_reader :data, :length
     def initialize(data, length = nil)
       @data, @length = data, length
     end
   end
 
+  class HashData < Data
+    def similarity(other, *args)
+      Phash.send("#{self.class.hash_type}_similarity", self, other, *args)
+    end
+
+    def self.hash_type
+      @hash_type ||= self.name.split('::').last.sub(/Hash$/, '').downcase
+    end
+  end
+
   class FileHash
     attr_reader :path
 
@@ -28,8 +38,18 @@ module Phash
       @phash ||= compute_phash
     end
 
-    def %(other)
-      similarity(other)
+    def compute_phash
+      Phash.send("#{self.class.hash_type}_hash", @path)
+    end
+
+    # Similarity with other phash
+    def similarity(other, *args)
+      phash.similarity(other.phash, *args)
+    end
+    alias_method :%, :similarity
+
+    def self.hash_type
+      @hash_type ||= self.name.split('::').last.downcase
     end
   end
 

data/lib/phash/audio.rb

@@ -49,23 +49,38 @@ module Phash
   #
   attach_function :ph_audio_distance_ber, [:pointer, :int, :pointer, :int, :float, :int, :pointer], :pointer, :blocking => true
 
-  class AudioHash < HashData; end
   class << self
-    # Read audio file specified by path and optional length using <tt>ph_readaudio</tt> and get its hash using <tt>ph_audiohash</tt>
-    def audio_hash(path, length = 0)
-      sample_rate = 8000
+    DEFAULT_SAMPLE_RATE = 8000
+
+    # Read audio file specified by path and optional length using <tt>ph_readaudio</tt>
+    def audio_data(path, length = 0, sample_rate = nil)
+      sample_rate ||= DEFAULT_SAMPLE_RATE
       audio_data_length_p = FFI::MemoryPointer.new :int
       if audio_data = ph_readaudio(path.to_s, sample_rate, 1, nil, audio_data_length_p, length.to_f)
         audio_data_length = audio_data_length_p.get_int(0)
         audio_data_length_p.free
 
-        hash_data_length_p = FFI::MemoryPointer.new :int
-        if hash_data = ph_audiohash(audio_data, audio_data_length, sample_rate, hash_data_length_p)
-          hash_data_length = hash_data_length_p.get_int(0)
-          hash_data_length_p.free
+        Data.new(audio_data, audio_data_length)
+      end
+    end
+
+    # Get hash of audio data using <tt>ph_audiohash</tt>
+    def audio_data_hash(audio_data, sample_rate = nil)
+      sample_rate ||= DEFAULT_SAMPLE_RATE
+      hash_data_length_p = FFI::MemoryPointer.new :int
+      if hash_data = ph_audiohash(audio_data.data, audio_data.length, sample_rate, hash_data_length_p)
+        hash_data_length = hash_data_length_p.get_int(0)
+        hash_data_length_p.free
+
+        AudioHash.new(hash_data, hash_data_length)
+      end
+    end
 
-          AudioHash.new(hash_data, hash_data_length)
-        end
+    # Use <tt>audio_data</tt> and <tt>audio_data_hash</tt> to compute hash for file at path, specify max length in seconds to read
+    def audio_hash(path, length = 0, sample_rate = nil)
+      sample_rate ||= DEFAULT_SAMPLE_RATE
+      if audio_data = audio_data(path, length, sample_rate)
+        audio_data_hash(audio_data, sample_rate)
       end
     end
 
@@ -92,6 +107,10 @@ module Phash
     end
   end
 
+  # Class to store audio hash and compare to other
+  class AudioHash < HashData
+  end
+
   # Class to store audio file hash and compare to other
   class Audio < FileHash
     attr_reader :length
@@ -101,13 +120,6 @@ module Phash
       @path, @length = path, length
     end
 
-    # Similarity with other audio
-    def similarity(other, *args)
-      Phash.audio_similarity(phash, other.phash, *args)
-    end
-
-  private
-
     def compute_phash
       Phash.audio_hash(@path, @length)
     end

data/lib/phash/image.rb

@@ -25,16 +25,16 @@ module Phash
         hash = hash_p.get_uint64(0)
         hash_p.free
 
-        hash
+        ImageHash.new(hash)
       end
     end
 
     # Get distance between two image hashes using <tt>ph_hamming_distance</tt>
     def image_hamming_distance(hash_a, hash_b)
-      hash_a.is_a?(Integer) or raise ArgumentError.new('hash_a is not an Integer')
-      hash_b.is_a?(Integer) or raise ArgumentError.new('hash_b is not an Integer')
+      hash_a.is_a?(ImageHash) or raise ArgumentError.new('hash_a is not an ImageHash')
+      hash_b.is_a?(ImageHash) or raise ArgumentError.new('hash_b is not an ImageHash')
 
-      ph_hamming_distance(hash_a, hash_b)
+      ph_hamming_distance(hash_a.data, hash_b.data)
     end
 
     # Get similarity from hamming_distance
@@ -43,17 +43,11 @@ module Phash
     end
   end
 
+  # Class to store image hash and compare to other
+  class ImageHash < HashData
+  end
+
   # Class to store image file hash and compare to other
   class Image < FileHash
-    # Similarity with other image
-    def similarity(other)
-      Phash.image_similarity(phash, other.phash)
-    end
-
-  private
-
-    def compute_phash
-      Phash.image_hash(@path)
-    end
   end
 end

data/lib/phash/text.rb

@@ -35,7 +35,6 @@ module Phash
   #
   attach_function :ph_compare_text_hashes, [:pointer, :int, :pointer, :int, :pointer], :pointer, :blocking => true
 
-  class TextHash < HashData; end
   class << self
     # Get text file hash using <tt>ph_texthash</tt>
     def text_hash(path)
@@ -66,7 +65,6 @@ module Phash
 
     def text_similarity(hash_a, hash_b)
       matches = text_hash_matches(hash_a, hash_b)
-      # p [hash_a.length, hash_b.length, matches.length]
       matched_a = Array.new(hash_a.length)
       matched_b = Array.new(hash_b.length)
       matches.each do |match|
@@ -83,17 +81,11 @@ module Phash
     end
   end
 
+  # Class to store text hash and compare to other
+  class TextHash < HashData
+  end
+
   # Class to store text file hash and compare to other
   class Text < FileHash
-    # Distance from other file, for now bit useless thing
-    def similarity(other)
-      Phash.text_similarity(phash, other.phash)
-    end
-
-  private
-
-    def compute_phash
-      Phash.text_hash(@path)
-    end
   end
 end

data/lib/phash/video.rb

@@ -13,7 +13,6 @@ module Phash
   #
   attach_function :ph_dct_videohash_dist, [:pointer, :int, :pointer, :int, :int], :double, :blocking => true
 
-  class VideoHash < HashData; end
   class << self
     # Get video hash using <tt>ph_dct_videohash</tt>
     def video_hash(path)
@@ -38,17 +37,11 @@ module Phash
     alias_method :video_similarity, :video_dct_distance
   end
 
+  # Class to store video hash and compare to other
+  class VideoHash < HashData
+  end
+
   # Class to store video file hash and compare to other
   class Video < FileHash
-    # Similarity with other video
-    def similarity(other, *args)
-      Phash.video_similarity(phash, other.phash, *args)
-    end
-
-  private
-
-    def compute_phash
-      Phash.video_hash(@path)
-    end
   end
 end

data/pHash.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name        = 'pHash'
-  s.version     = '1.0.2'
+  s.version     = '1.1.0'
   s.summary     = %q{Use pHash with ruby}
   s.homepage    = "http://github.com/toy/#{s.name}"
   s.authors     = ['Ivan Kuchin']

metadata

@@ -5,9 +5,9 @@ version: !ruby/object:Gem::Version
   prerelease: 
   segments: 
   - 1
+  - 1
   - 0
-  - 2
-  version: 1.0.2
+  version: 1.1.0
 platform: ruby
 authors: 
 - Ivan Kuchin
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-12-26 00:00:00 Z
+date: 2011-12-27 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rspec