file-digests 0.0.18 → 0.0.19

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d69e75a52c05cbc2caf912491be9fbaaadb5136a2dfc723920a9010d3e4c2592
-  data.tar.gz: f90deae82f2581d301d1cb7ba7c730e684b24f5c9582d6deb2581b9f2e6fa557
+  metadata.gz: 7aeebbfa688dc871d736ca71e740b3bdd4804701d29e4b960da16d1f75bdb04f
+  data.tar.gz: 136b2f908e41f56ba2ef243130d4a8a13b156f49b1c179a37225d3d19b0eff1c
 SHA512:
-  metadata.gz: 5ad20e936d21d42f56ed20250728eb0b7c2b9a034877e764f528c6e297853bb7724d6780a56b7c0f11a3df975c1d51fed04852ddb3f3ff9848f4c546167902e8
-  data.tar.gz: 27f86166310f420ac5858fa86c0233af666bbb2af40f53501a9f8e02205a3753151db5cfd6777558bbf0040e2472f9bc296d9382afebb8dd45f65fb981fc173b
+  metadata.gz: 9dd356eb305cd87874c3c8bc84905dec4388172e14db4b6a0d78004cf9bf183dc6c508537aa6f2806cc8d0f7eaeec68ccb6d05ea7cc5cf18db95fe86db9f4a15
+  data.tar.gz: 8a2d67681bc07b46fa4e93c280c8355eb639168e7bc1f244d3dae6cfd57e58fde6ded1a4d25aa2f63364bc639a55038936bedf990cc0608b267e2facd60bf556

data/lib/file-digests.rb

@@ -5,7 +5,7 @@ require 'fileutils'
 require 'pathname'
 require 'sqlite3'
 
-module FileDigests
+class FileDigests
 
   def self.perform_check
     options = {
@@ -13,271 +13,262 @@ module FileDigests
       quiet: (ENV["QUIET"] == "true"),
       test_only: (ENV["TEST_ONLY"] == "true")
     }
-    checker = Checker.new ARGV[0], ARGV[1], options
-    checker.perform_check
+    file_digests = self.new ARGV[0], ARGV[1], options
+    file_digests.perform_check
   end
 
-  class DigestDatabase
-    def initialize path, options = {}
-      @options = options
-
-      @db = SQLite3::Database.new path.to_s
-      @db.results_as_hash = true
-
-      execute 'PRAGMA journal_mode = "WAL"'
-      execute 'PRAGMA synchronous = "NORMAL"'
-      execute 'PRAGMA locking_mode = "EXCLUSIVE"'
-      execute 'PRAGMA cache_size = "5000"'
-
-      unless execute("SELECT name FROM sqlite_master WHERE type='table' AND name = 'digests'").length == 1
-        execute 'PRAGMA encoding = "UTF-8"'
-        execute "CREATE TABLE digests (
-          id INTEGER PRIMARY KEY,
-          filename TEXT,
-          mtime TEXT,
-          digest TEXT,
-          digest_check_time TEXT)"
-        execute "CREATE UNIQUE INDEX digests_filename ON digests(filename)"
-      end
+  def initialize files_path, digest_database_path, options = {}
+    @options = options
+    @files_path = cleanup_path(files_path || ".")
+    @prefix_to_remove = @files_path.to_s + '/'
 
-      @missing_files = Hash[@db.prepare("SELECT filename, digest FROM digests").execute!]
-      @new_files = {}
+    raise "Files path must be a readable directory" unless (File.directory?(@files_path) && File.readable?(@files_path))
 
-      prepare_method :insert, "INSERT INTO digests (filename, mtime, digest, digest_check_time) VALUES (?, ?, ?, datetime('now'))"
-      prepare_method :find_by_filename, "SELECT id, mtime, digest FROM digests WHERE filename = ?"
-      prepare_method :touch_digest_check_time, "UPDATE digests SET digest_check_time = datetime('now') WHERE id = ?"
-      prepare_method :update_mtime_and_digest, "UPDATE digests SET mtime = ?, digest = ?, digest_check_time = datetime('now') WHERE id = ?"
-      prepare_method :update_mtime, "UPDATE digests SET mtime = ?, digest_check_time = datetime('now') WHERE id = ?"
-      prepare_method :delete_by_filename, "DELETE FROM digests WHERE filename = ?"
+    @digest_database_path = if digest_database_path
+      cleanup_path(digest_database_path)
+    else
+      @files_path + '.file-digests.sqlite'
     end
 
-    def insert_or_update file_path, mtime, digest, counters
-      result = find_by_filename file_path
-
-      if found = result.next_hash
-        raise "Multiple records found" if result.next
-
-        @missing_files.delete(file_path)
-
-        if found['digest'] == digest
-          counters[:good] += 1
-          # puts "GOOD: #{file_path}" unless @options[:quiet]
-          unless @options[:test_only]
-            if found['mtime'] == mtime
-              touch_digest_check_time found['id']
-            else
-              update_mtime mtime, found['id']
-            end
-          end
-        else
-          if found['mtime'] == mtime # Digest is different and mtime is the same
-            counters[:likely_damaged] += 1
-            STDERR.puts "LIKELY DAMAGED: #{file_path}"
-          else
-            counters[:updated] += 1
-            puts "UPDATED: #{file_path}" unless @options[:quiet]
-            unless @options[:test_only]
-              update_mtime_and_digest mtime, digest, found['id']
-            end
-          end
-        end
-      else
-        counters[:new] += 1
-        puts "NEW: #{file_path}" unless @options[:quiet]
-        unless @options[:test_only]
-          @new_files[file_path] = digest
-          insert file_path, mtime, digest
-        end
-      end
+    if File.directory?(@digest_database_path)
+      @digest_database_path = @digest_database_path + '.file-digests.sqlite'
     end
 
-    def track_renames counters
-      @missing_files.delete_if do |filename, digest|
-        if @new_files.value?(digest)
-          counters[:renamed] += 1
-          unless @options[:test_only]
-            delete_by_filename filename
-          end
-          true
-        end
-      end
-      counters[:missing] = @missing_files.length
+    if @files_path == @digest_database_path.dirname
+      @skip_file_digests_sqlite = true
     end
 
-    def any_missing_files?
-      @missing_files.length > 0
-    end
+    ensure_dir_exists @digest_database_path.dirname
 
-    def print_missing_files
-      puts "\nMISSING FILES:"
-      @missing_files.sort.to_h.each do |filename, digest|
-        puts filename
-      end
+    # Please do not use this flag, support for sha512 is here for backward compatibility, and one day it will be removed.
+    if File.exist?(@digest_database_path.dirname + '.file-digests.sha512')
+      @use_sha512 = true
     end
 
-    def remove_missing_files
-      @db.transaction do
-        @missing_files.each do |filename, digest|
-          delete_by_filename filename
-        end
-      end
-    end
-
-    private
+    initialize_database @digest_database_path
 
-    def execute *args, &block
-      @db.execute *args, &block
-    end
-
-    def prepare_method name, query
-      variable = "@#{name}"
-      instance_variable_set(variable, @db.prepare(query))
-      define_singleton_method name do |*args, &block|
-        instance_variable_get(variable).execute(*args, &block)
-      end
-    end
+    @counters = {good: 0, updated: 0, new: 0, missing: 0, renamed: 0, likely_damaged: 0, exceptions: 0}
   end
 
-  class Checker
-    def initialize files_path, digest_database_path, options = {}
-      @options = options
-      @files_path = cleanup_path(files_path || ".")
-      @prefix_to_remove = @files_path.to_s + '/'
-
-      raise "Files path must be a readable directory" unless (File.directory?(@files_path) && File.readable?(@files_path))
+  def initialize_database path
+    @db = SQLite3::Database.new path.to_s
+    @db.results_as_hash = true
+
+    execute 'PRAGMA journal_mode = "WAL"'
+    execute 'PRAGMA synchronous = "NORMAL"'
+    execute 'PRAGMA locking_mode = "EXCLUSIVE"'
+    execute 'PRAGMA cache_size = "5000"'
+
+    unless execute("SELECT name FROM sqlite_master WHERE type='table' AND name = 'digests'").length == 1
+      execute 'PRAGMA encoding = "UTF-8"'
+      execute "CREATE TABLE digests (
+        id INTEGER PRIMARY KEY,
+        filename TEXT,
+        mtime TEXT,
+        digest TEXT,
+        digest_check_time TEXT)"
+      execute "CREATE UNIQUE INDEX digests_filename ON digests(filename)"
+    end
 
-      @digest_database_path = if digest_database_path
-        cleanup_path(digest_database_path)
-      else
-        @files_path + '.file-digests.sqlite'
-      end
+    @missing_files = Hash[@db.prepare("SELECT filename, digest FROM digests").execute!]
+    @new_files = {}
 
-      if File.directory?(@digest_database_path)
-        @digest_database_path = @digest_database_path + '.file-digests.sqlite'
-      end
+    prepare_method :insert, "INSERT INTO digests (filename, mtime, digest, digest_check_time) VALUES (?, ?, ?, datetime('now'))"
+    prepare_method :find_by_filename, "SELECT id, mtime, digest FROM digests WHERE filename = ?"
+    prepare_method :touch_digest_check_time, "UPDATE digests SET digest_check_time = datetime('now') WHERE id = ?"
+    prepare_method :update_mtime_and_digest, "UPDATE digests SET mtime = ?, digest = ?, digest_check_time = datetime('now') WHERE id = ?"
+    prepare_method :update_mtime, "UPDATE digests SET mtime = ?, digest_check_time = datetime('now') WHERE id = ?"
+    prepare_method :delete_by_filename, "DELETE FROM digests WHERE filename = ?"
+  end
 
-      if @files_path == @digest_database_path.dirname
-        @skip_file_digests_sqlite = true
+  def perform_check
+    measure_time do
+      walk_files do |filename|
+        process_file filename
       end
+    end
 
-      ensure_dir_exists @digest_database_path.dirname
+    track_renames
 
-      # Please do not use this flag, support for sha512 is here for backward compatibility, and one day it will be removed.
-      if File.exist?(@digest_database_path.dirname + '.file-digests.sha512')
-        @use_sha512 = true
+    if any_missing_files?
+      print_missing_files
+      if !@options[:test_only] && (@options[:auto] || confirm("Remove missing files from the database"))
+        remove_missing_files
       end
+    end
 
-      @digest_database = DigestDatabase.new @digest_database_path, @options
-      @counters = {good: 0, updated: 0, new: 0, missing: 0, renamed: 0, likely_damaged: 0, exceptions: 0}
+    if @counters[:likely_damaged] > 0 || @counters[:exceptions] > 0
+      STDERR.puts "ERRORS WERE OCCURRED"
     end
 
-    def perform_check
-      measure_time do
-        walk_files do |filename|
-          process_file filename
-        end
-      end
+    puts @counters.inspect
+  end
 
-      @digest_database.track_renames @counters
+  private
 
-      if @digest_database.any_missing_files?
-        @digest_database.print_missing_files
-        if !@options[:test_only] && (@options[:auto] || confirm("Remove missing files from the database"))
-          @digest_database.remove_missing_files
-        end
-      end
+  def process_file filename
+    return if File.symlink? filename
 
-      if @counters[:likely_damaged] > 0 || @counters[:exceptions] > 0
-        STDERR.puts "ERRORS WERE OCCURRED"
-      end
+    stat = File.stat filename
 
-      puts @counters.inspect
+    return if stat.blockdev?
+    return if stat.chardev?
+    return if stat.directory?
+    return if stat.pipe?
+    unless stat.readable?
+      raise "File is not readable"
     end
+    return if stat.socket?
+
+    if @skip_file_digests_sqlite
+      basename = File.basename(filename)
+      return if basename == '.file-digests.sha512'
+      return if basename == '.file-digests.sqlite'
+      return if basename == '.file-digests.sqlite-wal'
+      return if basename == '.file-digests.sqlite-shm'
+    end
+
+    insert_or_update(
+      filename.delete_prefix(@prefix_to_remove).encode('utf-8', universal_newline: true).unicode_normalize(:nfkc),
+      stat.mtime.utc.strftime('%Y-%m-%d %H:%M:%S'),
+      get_file_digest(filename)
+      )
+  rescue => exception
+    @counters[:exceptions] += 1
+    STDERR.puts "EXCEPTION: #{filename.encode('utf-8', universal_newline: true)}: #{exception.message}"
+  end
 
-    private
+  def patch_path_string path
+    Gem.win_platform? ? path.gsub(/\\/, '/') : path
+  end
 
+  def cleanup_path path
+    Pathname.new(patch_path_string(path)).cleanpath
+  end
 
-    def confirm text
-      if STDIN.tty? && STDOUT.tty?
-        puts "#{text} (y/n)?"
-        STDIN.gets.strip.downcase == "y"
+  def ensure_dir_exists path
+    if File.exist?(path)
+      unless File.directory?(path)
+        raise "#{path} is not a directory"
       end
+    else
+      FileUtils.mkdir_p path
     end
+  end
 
-    def process_file filename
-      return if File.symlink? filename
-
-      stat = File.stat filename
+  def walk_files
+    Dir.glob(@files_path + '**' + '*', File::FNM_DOTMATCH) do |filename|
+      yield filename
+    end
+  end
 
-      return if stat.blockdev?
-      return if stat.chardev?
-      return if stat.directory?
-      return if stat.pipe?
-      unless stat.readable?
-        raise "File is not readable"
+  def get_file_digest filename
+    File.open(filename, 'rb') do |io|
+      digest = (@use_sha512 ? Digest::SHA512 : Digest::SHA256).new
+      buffer = ""
+      while io.read(40960, buffer)
+        digest.update(buffer)
       end
-      return if stat.socket?
-
-      if @skip_file_digests_sqlite
-        basename = File.basename(filename)
-        return if basename == '.file-digests.sha512'
-        return if basename == '.file-digests.sqlite'
-        return if basename == '.file-digests.sqlite-wal'
-        return if basename == '.file-digests.sqlite-shm'
-      end
-
-      @digest_database.insert_or_update(
-        filename.delete_prefix(@prefix_to_remove).encode('utf-8', universal_newline: true).unicode_normalize(:nfkc),
-        stat.mtime.utc.strftime('%Y-%m-%d %H:%M:%S'),
-        get_file_digest(filename),
-        @counters
-        )
-    rescue => exception
-      @counters[:exceptions] += 1
-      STDERR.puts "EXCEPTION: #{filename.encode('utf-8', universal_newline: true)}: #{exception.message}"
+      return digest.hexdigest
     end
+  end
 
-    def patch_path_string path
-      Gem.win_platform? ? path.gsub(/\\/, '/') : path
+  def confirm text
+    if STDIN.tty? && STDOUT.tty?
+      puts "#{text} (y/n)?"
+      STDIN.gets.strip.downcase == "y"
     end
+  end
 
-    def cleanup_path path
-      Pathname.new(patch_path_string(path)).cleanpath
-    end
+  def measure_time
+    start = Process.clock_gettime(Process::CLOCK_MONOTONIC)
+    yield
+    elapsed = (Process.clock_gettime(Process::CLOCK_MONOTONIC) - start).to_i
+    puts "Elapsed time: #{elapsed / 3600}h #{(elapsed % 3600) / 60}m #{elapsed % 60}s" unless @options[:quiet]
+  end
+
+  def insert_or_update file_path, mtime, digest
+    result = find_by_filename file_path
 
-    def ensure_dir_exists path
-      if File.exist?(path)
-        unless File.directory?(path)
-          raise "#{path} is not a directory"
+    if found = result.next_hash
+      raise "Multiple records found" if result.next
+
+      @missing_files.delete(file_path)
+
+      if found['digest'] == digest
+        @counters[:good] += 1
+        # puts "GOOD: #{file_path}" unless @options[:quiet]
+        unless @options[:test_only]
+          if found['mtime'] == mtime
+            touch_digest_check_time found['id']
+          else
+            update_mtime mtime, found['id']
+          end
         end
       else
-        FileUtils.mkdir_p path
+        if found['mtime'] == mtime # Digest is different and mtime is the same
+          @counters[:likely_damaged] += 1
+          STDERR.puts "LIKELY DAMAGED: #{file_path}"
+        else
+          @counters[:updated] += 1
+          puts "UPDATED: #{file_path}" unless @options[:quiet]
+          unless @options[:test_only]
+            update_mtime_and_digest mtime, digest, found['id']
+          end
+        end
       end
-    end
-
-    def walk_files
-      Dir.glob(@files_path + '**' + '*', File::FNM_DOTMATCH) do |filename|
-        yield filename
+    else
+      @counters[:new] += 1
+      puts "NEW: #{file_path}" unless @options[:quiet]
+      unless @options[:test_only]
+        @new_files[file_path] = digest
+        insert file_path, mtime, digest
       end
     end
+  end
 
-    def get_file_digest filename
-      File.open(filename, 'rb') do |io|
-        digest = (@use_sha512 ? Digest::SHA512 : Digest::SHA256).new
-        buffer = ""
-        while io.read(40960, buffer)
-          digest.update(buffer)
+  def track_renames
+    @missing_files.delete_if do |filename, digest|
+      if @new_files.value?(digest)
+        @counters[:renamed] += 1
+        unless @options[:test_only]
+          delete_by_filename filename
         end
-        return digest.hexdigest
+        true
       end
     end
+    @counters[:missing] = @missing_files.length
+  end
 
-    def measure_time
-      start = Process.clock_gettime(Process::CLOCK_MONOTONIC)
-      yield
-      elapsed = (Process.clock_gettime(Process::CLOCK_MONOTONIC) - start).to_i
-      puts "Elapsed time: #{elapsed / 3600}h #{(elapsed % 3600) / 60}m #{elapsed % 60}s" unless @options[:quiet]
+  def any_missing_files?
+    @missing_files.length > 0
+  end
+
+  def print_missing_files
+    puts "\nMISSING FILES:"
+    @missing_files.sort.to_h.each do |filename, digest|
+      puts filename
+    end
+  end
+
+  def remove_missing_files
+    @db.transaction do
+      @missing_files.each do |filename, digest|
+        delete_by_filename filename
+      end
     end
+  end
 
+  def execute *args, &block
+    @db.execute *args, &block
   end
+
+  def prepare_method name, query
+    variable = "@#{name}"
+    instance_variable_set(variable, @db.prepare(query))
+    define_singleton_method name do |*args, &block|
+      instance_variable_get(variable).execute(*args, &block)
+    end
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: file-digests
 version: !ruby/object:Gem::Version
-  version: 0.0.18
+  version: 0.0.19
 platform: ruby
 authors:
 - Stanislav Senotrusov