phash-rb 0.2.0 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 995d2669bfb1ed93482da8a528db5da15a4a01095ab4b04798a7a8bd2ef2d207
-  data.tar.gz: 6abdafea8b495b031378b22b36bba58a1b69bcef263fa1f8981fd2dd3c774828
+  metadata.gz: cb8320eac1ac54f8b57ce98e88cefc376c7921e57e3874e40e4bd046544077ad
+  data.tar.gz: 62a21503a180cf5c42f5681d661f059032fe37e9947a5037d6dcb53727bb1a8a
 SHA512:
-  metadata.gz: 7416ab673a2218b225a138e4934d7ca117af2617256d8ffeb8919ad54a23b18025b81795d1f1f3e592532f823bf30db3cca127acc22abd6fb0c4dd07f7ed9552
-  data.tar.gz: 47830995008c37150316b48c08504747ece664d652fae7116752d1ec996fd53f444b612c70ff3a447e97e2e36a3946c62edc6d87eafc8564bf9b4636f2f878e4
+  metadata.gz: 923e41cf1d048b71ca224f47c6809315ce071348e85f916f508c4f1eca9b8a31c2508c9f8a47fd916cb3729b2d0a86010c0f69c67930f874338785ac8bc3ebc0
+  data.tar.gz: af270246aaafcd1671a40206fd7d93b7e154a550ad85720769a517227d31ae4e42c4e92f4e845126dfdebcc7cdd23e434db42f10369c3802e977b3efc46bbe65

data/CHANGELOG.md

@@ -2,6 +2,18 @@
 
 All notable changes to this project will be documented in this file.
 
+### 0.3.1
+
+- Fix fingerprinting for 4-band images (RGBA with alpha channel) on vips 8.15.1+
+
+### 0.3.0
+
+- Optimize fingerprint calculation for large images (~24x faster for images >500x500)
+- Cache DCT matrix for faster batch processing
+- Lower required Ruby version from 3.0 to 2.6
+- Improve CLI: support multiple files, `--compare`, `--help`, `--version`
+- Add `Phash.distance(fp1, fp2)` class method for computing Hamming distance
+
 ### 0.2.0
 
 Allow passing Vips::Image directly to fingerprint function

data/README.md

@@ -8,7 +8,7 @@ Phashion replacement without native extension (however it currently relies on li
 ## Requirements
 
 - libvips (see requirements for [ruby-vips](https://github.com/libvips/ruby-vips))
-- Ruby 3.0.0 or later
+- Ruby 2.6.0 or later
 
 ## Installation
 
@@ -52,11 +52,34 @@ require 'phash'
 Phash::Image.new(filename1).fingerprint # 3714852948054213970
 ```
 
+You can also compute the Hamming distance between two fingerprints directly:
+
+```ruby
+fp1 = Phash.fingerprint(filename1)
+fp2 = Phash.fingerprint(filename2)
+Phash.distance(fp1, fp2) # 30
+```
+
 Fingerprint is also available in a command `phash`:
 
 ```bash
 $ phash test/fixtures/test.jpg
 3714852948054213970
+
+$ phash image1.jpg image2.jpg image3.jpg
+3714852948054213970
+5378904591010983442
+1234567890123456789
+
+$ phash --compare image1.jpg image2.jpg
+image1.jpg: 3714852948054213970
+image2.jpg: 5378904591010983442
+Hamming distance: 30
+Result: different images
+
+$ phash --help
+Usage: phash [options] <file> [file...]
+...
 ```
 
 Additionally you can pass `Vips::Image` directly to fingerprint function:

data/exe/phash

@@ -2,12 +2,62 @@
 
 require_relative "../lib/phash"
 
-filename = ARGV[0]
+def print_help
+  puts "Usage: phash [options] <file> [file...]"
+  puts ""
+  puts "Commands:"
+  puts "  phash <file>                    Print fingerprint for file"
+  puts "  phash <file1> <file2> ...       Print fingerprints for multiple files"
+  puts "  phash --compare <file1> <file2> Compare two images and print distance"
+  puts ""
+  puts "Options:"
+  puts "  -h, --help                      Show this help message"
+  puts "  -v, --version                   Show version"
+  puts "  -c, --compare                   Compare two images"
+end
+
+def print_version
+  puts "phash-rb #{Phash::VERSION}"
+end
 
-begin
+def compute_fingerprint(filename)
   puts Phash.fingerprint(filename)
 rescue => e
-  puts "Failed to compute pHash for #{filename}."
-  puts e
+  warn "Failed to compute pHash for #{filename}: #{e.message}"
+  exit 1
+end
+
+def compare_images(file1, file2)
+  fp1 = Phash.fingerprint(file1)
+  fp2 = Phash.fingerprint(file2)
+  distance = Phash.distance(fp1, fp2)
+
+  puts "#{file1}: #{fp1}"
+  puts "#{file2}: #{fp2}"
+  puts "Hamming distance: #{distance}"
+  puts distance < 10 ? "Result: likely duplicates" : "Result: different images"
+rescue => e
+  warn "Failed to compare images: #{e.message}"
   exit 1
 end
+
+if ARGV.empty? || ARGV.include?("-h") || ARGV.include?("--help")
+  print_help
+  exit 0
+end
+
+if ARGV.include?("-v") || ARGV.include?("--version")
+  print_version
+  exit 0
+end
+
+if ARGV.include?("-c") || ARGV.include?("--compare")
+  args = ARGV.reject { |a| a == "-c" || a == "--compare" }
+  if args.size != 2
+    warn "Error: --compare requires exactly 2 files"
+    exit 1
+  end
+  compare_images(args[0], args[1])
+else
+  ARGV.each { |file| compute_fingerprint(file) }
+end

data/lib/phash/image.rb

@@ -13,7 +13,7 @@ module Phash
     end
 
     def distance_from(other_image)
-      (fingerprint ^ other_image.fingerprint).to_s(2).count('1')
+      Phash.distance(fingerprint, other_image.fingerprint)
     end
   end
 end

data/lib/phash/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Phash
-  VERSION = "0.2.0"
+  VERSION = "0.3.1"
 end

data/lib/phash.rb

@@ -10,8 +10,14 @@ module Phash
   CIMG_V = CIMG_PI / 2 / 32
   CIMG_SCALE = 2**8 + 1
 
+  def self.distance(fingerprint1, fingerprint2)
+    (fingerprint1 ^ fingerprint2).to_s(2).count('1')
+  end
+
   def self.fingerprint(path_or_img)
     img = path_or_img.is_a?(Vips::Image) ? path_or_img : Vips::Image.new_from_file(path_or_img)
+    # Strip alpha channel if present (4-band images)
+    img = img[0..2] if img.bands == 4
     #Y = (66*R + 129*G + 25*B + 128)/256 + 16
     img = img * [66.0 / 256, 129.0 / 256, 25.0 / 256]
     r, g, b = img.bandsplit
@@ -35,19 +41,36 @@ module Phash
   private
 
   def self.ph_dct_matrix
-    v = 1 / Math.sqrt(32)
-    c1 = Math.sqrt(2.0 / 32)
+    @dct_matrix ||= begin
+      v = 1 / Math.sqrt(32)
+      c1 = Math.sqrt(2.0 / 32)
 
-    Matrix.build(32, 32) do |y, x|
-      (y < 1 ? v : c1 * Math.cos(CIMG_V * y * (2*x + 1))).round(6)
+      Matrix.build(32, 32) do |y, x|
+        (y < 1 ? v : c1 * Math.cos(CIMG_V * y * (2*x + 1))).round(6)
+      end
     end
   end
 
+  def self.large_image?(img)
+    img.width * img.height > 250_000
+  end
+
   def self.sample(img)
-    w, h = img.width / 32.0, img.height / 32.0
-    src = img.to_a
-    Matrix.build(32, 32) do |y, x|
-      src[y*h][x*w][0] / CIMG_SCALE
+    w, h = img.width, img.height
+    w_step = w / 32.0
+    h_step = h / 32.0
+
+    if large_image?(img)
+      # Large images: use getpoint to avoid copying entire image to Ruby array
+      Matrix.build(32, 32) do |y, x|
+        img.getpoint((x * w_step).to_i, (y * h_step).to_i)[0] / CIMG_SCALE
+      end
+    else
+      # Small images: to_a is faster than 1024 getpoint calls
+      src = img.to_a
+      Matrix.build(32, 32) do |y, x|
+        src[(y * h_step).to_i][(x * w_step).to_i][0] / CIMG_SCALE
+      end
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: phash-rb
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.3.1
 platform: ruby
 authors:
 - Tomasz Ratajczak
@@ -70,14 +70,14 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: 3.0.0
+      version: 2.6.0
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.7
+rubygems_version: 4.0.2
 specification_version: 4
 summary: Ruby implementation of pHash library
 test_files: []