lucarecord 0.2.22 → 0.2.27

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 217630d4010c2341954f05600982cf641470b244440849c7816873ed8d10e8d2
-  data.tar.gz: f535546634069491d64b3e0359fc9a40f10059cd08d216f8a947c28d717d1c6c
+  metadata.gz: 3e81c1ccc95c688bdfa2da3af37635324f1f744f1976b0dd3654514e2a1254b4
+  data.tar.gz: 2ed07a97b4bca099f1eb7004476960c3c2a8da03d06ee7f4db6e48226294f431
 SHA512:
-  metadata.gz: 9e69a01066562a854ebe78128ead7c7d87be1227cfbce7453edf5510fa03f7b2acfbea5b334837b1c80e49e71a8a858d2d273277769e096baf29e4c90977816b
-  data.tar.gz: c7caf90c504940b0533441f19cde460fbb680f69937412659ac04670f009b74ed84c5a2aa120f2554a7e4cfab8d4eba62309cd5c0ace1e057fc09400fdc05cee
+  metadata.gz: fcdd0ddae31a118c6532c872361535d89336fa712ba80395a10d5dda7a597c6300214ac2c1cdbab6782bd861d46e9da1e32c48e3fae182c33d63e78b0fcf7424
+  data.tar.gz: 4ec6114ea0f761e7ca4f743fce6c2594cac704c0bb6a32082ba35cbc5b9c6e639ce421d13ccf33a46a5359a4030974ff85447133391641beb83d2efaab3e1a91

data/CHANGELOG.md

@@ -1,3 +1,23 @@
+## LucaRecord 0.2.26
+
+* Support #dig / #search for TSV dictionary
+* Fix: shorten n-gram split factor on search word length < specified factor
+
+## LucaRecord 0.2.25
+
+* Implement `dir_digest()` for data validation.
+* support defunct without effective history record
+
+## LucaRecord 0.2.24
+
+* Digit delimiter for `delimit_num` can be customized through `thousands_separator` and `decimal_separator` in config.yml.
+* Const `CONFIG` and `PJDIR` is defined at `LucaRecord::Base`.
+* add `LucaSupport::Code.keys_stringify()`
+
+## LucaRecord 0.2.23
+
+* Enhance Dictionary, supporting extensible options.
+
 ## LucaRecord 0.2.22
 
 * add `LucaSupport::View.nushell()`, render nushell table directly.

data/lib/luca_record/base.rb

@@ -6,6 +6,8 @@ require 'luca_support'
 
 module LucaRecord
   class Base
+    CONFIG = LucaSupport::CONFIG
+    PJDIR = LucaSupport::PJDIR
     include LucaRecord::IO
     include LucaSupport::View
   end

data/lib/luca_record/dict.rb

@@ -6,7 +6,6 @@ require 'yaml'
 require 'pathname'
 require 'luca_support'
 
-#
 # Low level API
 #
 module LucaRecord
@@ -14,69 +13,49 @@ module LucaRecord
     include LucaSupport::Code
 
     def initialize(file = @filename)
-      #@path = file
       @path = self.class.dict_path(file)
       set_driver
     end
 
-    def search(word, default_word = nil)
-      res = max_score_code(word.gsub(/[[:space:]]/, ''))
-      if res[1] > 0.4
-        res[0]
+    # Search code with n-gram word.
+    # If dictionary has Hash or Array, it returns [label, options].
+    #
+    def search(word, default_word = nil, main_key: 'label', options: nil)
+      definitions_lazyload
+      res, score = max_score_code(word.gsub(/[[:space:]]/, ''))
+      return default_word if score < 0.4
+
+      case res
+      when Hash
+        hash2multiassign(res, main_key, options: options)
+      when Array
+        res.map { |item| hash2multiassign(item, main_key, options: options) }
       else
-        default_word
+        res
       end
     end
 
+    # Search with unique code.
     #
-    # Column number settings for CSV/TSV convert
-    #
-    # :label
-    #   for double entry data
-    # :counter_label
-    #   must be specified with label
-    # :debit_label
-    #   for double entry data
-    # * debit_value
-    # :credit_label
-    #   for double entry data
-    # * credit_value
-    # :note
-    #   can be the same column as another label
-    #
-    # :encoding
-    #   file encoding
+    def dig(*args)
+      @data.dig(*args)
+    end
+
+    # Separate main item from other options.
+    # If options specified as Array of string, it works as safe list filter.
     #
-    def csv_config
-      {}.tap do |config|
-        if @config.dig('label')
-          config[:label] = @config['label'].to_i
-          if @config.dig('counter_label')
-            config[:counter_label] = @config['counter_label']
-            config[:type] = 'single'
-          end
-        elsif @config.dig('debit_label')
-          config[:debit_label] = @config['debit_label'].to_i
-          if @config.dig('credit_label')
-            config[:credit_label] = @config['credit_label'].to_i
-            config[:type] = 'double'
-          end
+    def hash2multiassign(obj, main_key = 'label', options: nil)
+      options = {}.tap do |opt|
+        obj.map do |k, v|
+          next if k == main_key
+          next if !options.nil? && !options.include?(k)
+
+          opt[k.to_sym] = v
         end
-        config[:type] ||= 'invalid'
-        config[:debit_value] = @config['debit_value'].to_i if @config.dig('debit_value')
-        config[:credit_value] = @config['credit_value'].to_i if @config.dig('credit_value')
-        config[:note] = @config['note'] if @config.dig('note')
-        config[:encoding] = @config['encoding'] if @config.dig('encoding')
-
-        config[:year] = @config['year'] if @config.dig('year')
-        config[:month] = @config['month'] if @config.dig('month')
-        config[:day] = @config['day'] if @config.dig('day')
-        config[:default_debit] = @config['default_debit'] if @config.dig('default_debit')
-        config[:default_credit] = @config['default_credit'] if @config.dig('default_credit')
       end
+      [obj[main_key], options.compact]
     end
 
-    #
     # Load CSV with config options
     #
     def load_csv(path)
@@ -85,7 +64,6 @@ module LucaRecord
       end
     end
 
-    #
     # load dictionary data
     #
     def self.load(file = @filename)
@@ -99,7 +77,6 @@ module LucaRecord
       end
     end
 
-    #
     # generate dictionary from TSV file. Minimum assumption is as bellows:
     # 1st row is converted symbol.
     #
@@ -122,16 +99,33 @@ module LucaRecord
       end
     end
 
+    def self.validate(filename, target_key = :label)
+      errors = load(filename).map { |k, v| v[target_key].nil? ? k : nil }.compact
+      if errors.empty?
+        puts 'No error detected.'
+        nil
+      else
+        puts "Key #{errors.join(', ')} has nil #{target_key}."
+        errors.count
+      end
+    end
+
     private
 
     def set_driver
-      input = self.class.load(@path)
-      @config = input['config']
-      @definitions = input['definitions']
+      @data = self.class.load(@path)
+      @config = @data['config']
+      @definitions = @data['definitions']
+    end
+
+    # Build Reverse dictionary for TSV data
+    #
+    def definitions_lazyload
+      @definitions ||= @data.each_with_object({}) { |(k, entry), h| h[entry[:label]] = k if entry[:label] }
     end
 
     def self.dict_path(filename)
-      Pathname(LucaSupport::Config::Pjdir) / 'dict' / filename
+      Pathname(LucaSupport::PJDIR) / 'dict' / filename
     end
 
     def self.reverse(dict)
@@ -140,7 +134,7 @@ module LucaRecord
 
     def max_score_code(str)
       res = @definitions.map do |k, v|
-        [v, LucaSupport.match_score(str, k, 3)]
+        [v, match_score(str, k, 2)]
       end
       res.max { |x, y| x[1] <=> y[1] }
     end

data/lib/luca_record/io.rb

@@ -243,7 +243,7 @@ module LucaRecord # :nodoc:
       end
 
       # test if having required dirs/files under exec path
-      def valid_project?(path = LucaSupport::Config::Pjdir)
+      def valid_project?(path = LucaSupport::PJDIR)
         project_dir = Pathname(path)
         FileTest.file?((project_dir + 'config.yml').to_s) and FileTest.directory?( (project_dir + 'data').to_s)
       end
@@ -252,6 +252,17 @@ module LucaRecord # :nodoc:
         LucaSupport::Code.encode_txid(new_record_no(basedir, date_obj))
       end
 
+      # Calculate md5sum under specific month directory.
+      #
+      def dir_digest(year, month, basedir = @dirname)
+        subdir = year.to_s + LucaSupport::Code.encode_month(month)
+        digest = String.new
+        open_records(basedir, subdir).each do |f, path|
+          digest = update_digest(digest, f.read, path[1])
+        end
+        digest
+      end
+
       private
 
       # define new transaction ID & write data at once
@@ -302,6 +313,14 @@ module LucaRecord # :nodoc:
         end
       end
 
+      # Calculate md5sum with original digest, file content and filename(optional).
+      #
+      def update_digest(digest, str, filename = nil)
+        str = filename.nil? ? str : filename + str
+        content = Digest::MD5.new.update(str).hexdigest
+        Digest::MD5.new.update(digest + content).hexdigest
+      end
+
       # git object like structure
       #
       def open_hashed(basedir, id, mode = 'r')
@@ -339,12 +358,14 @@ module LucaRecord # :nodoc:
       # If specific decode is needed, override this method in each class.
       #
       def load_data(io, path = nil)
-        case @record_type
-        when 'raw'
-          # TODO: raw may be unneeded in favor of override
-          io
-        when 'json'
-        # TODO: implement JSON parse
+        if @record_type
+          case @record_type
+          when 'raw'
+            # TODO: raw may be unneeded in favor of override
+            io
+          when 'json'
+            # TODO: implement JSON parse
+          end
         else
           LucaSupport::Code.decimalize(YAML.load(io.read)).tap { |obj| validate_keys(obj) }
         end
@@ -362,7 +383,7 @@ module LucaRecord # :nodoc:
 
       # TODO: replace with data_dir method
       def abs_path(base_dir)
-        Pathname(LucaSupport::Config::Pjdir) / 'data' / base_dir
+        Pathname(LucaSupport::PJDIR) / 'data' / base_dir
       end
 
       # true when file doesn't have record on code

data/lib/luca_record/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module LucaRecord
-  VERSION = '0.2.22'
+  VERSION = '0.2.27'
 end

data/lib/luca_support.rb

@@ -6,15 +6,4 @@ module LucaSupport
   autoload :Config, 'luca_support/config'
   autoload :Mail, 'luca_support/mail'
   autoload :View, 'luca_support/view'
-
-  def self.match_score(a, b, n = 2)
-    v_a = to_ngram(a, n)
-    v_b = to_ngram(b, n)
-
-    v_a.map { |item| v_b.include?(item) ? 1 : 0 }.sum / v_a.length.to_f
-  end
-
-  def self.to_ngram(str, n = 2)
-    str.each_char.each_cons(n).map(&:join)
-  end
 end

data/lib/luca_support/code.rb

@@ -54,21 +54,24 @@ module LucaSupport
     def delimit_num(num, decimal: nil, delimiter: nil)
       return nil if num.nil?
 
-      decimal ||= LucaSupport::Config::DECIMAL_NUM
-      str = case num
-            when BigDecimal
-              if decimal == 0
-                num.floor.to_s.reverse.gsub!(/(\d{3})(?=\d)/, '\1 ').reverse!
-              else
-                fragments = num.floor(decimal).to_s('F').split('.')
-                fragments[0].reverse!.gsub!(/(\d{3})(?=\d)/, '\1 ').reverse!
-                fragments[1].gsub!(/(\d{3})(?=\d)/, '\1 ')
-                fragments.join('.')
-              end
-            else
-              num.to_s.reverse.gsub!(/(\d{3})(?=\d)/, '\1 ').reverse!
-            end
-      delimiter.nil? ? str : str.gsub!(/\s/, delimiter)
+      decimal ||= LucaSupport::CONFIG['decimal_num']
+      delimiter ||= LucaSupport::CONFIG['thousands_separator']
+      case num
+      when BigDecimal
+        if decimal == 0
+          num.floor.to_s.reverse!.gsub(/(\d{3})(?=\d)/, '\1 ').reverse!
+            .gsub(/\s/, delimiter)
+        else
+          fragments = num.floor(decimal).to_s('F').split('.')
+          fragments[0].reverse!.gsub!(/(\d{3})(?=\d)/, '\1 ')
+          fragments[0].reverse!.gsub!(/\s/, delimiter)
+          fragments[1].gsub!(/(\d{3})(?=\d)/, '\1 ')
+          fragments.join(LucaSupport::CONFIG['decimal_separator'])
+        end
+      else
+        num.to_s.reverse.gsub(/(\d{3})(?=\d)/, '\1 ').reverse!
+          .gsub(/\s/, delimiter)
+      end
     end
 
     # encode directory name from year and month.
@@ -101,7 +104,7 @@ module LucaSupport
       (start_year..end_year).to_a.map do |y|
         g1 = y == start_year ? encode_month(start_month) : encode_month(1)
         g2 = y == end_year ? encode_month(end_month) : encode_month(12)
-        "#{y}[#{g1}-#{g2}]"
+        g1 == g2 ? "#{y}#{g1}" : "#{y}[#{g1}-#{g2}]"
       end
     end
 
@@ -114,6 +117,32 @@ module LucaSupport
       Digest::SHA1.hexdigest(SecureRandom.uuid)
     end
 
+    # Convert Hash keys to string recursively.
+    # Required for YAML compatibility.
+    #
+    def keys_stringify(dat)
+      case dat
+      when Array
+        dat.map { |d| keys_stringify(d) }
+      when Hash
+        dat.map { |k, v| [k.to_s, keys_stringify(v)] }.to_h
+      else
+        dat
+      end
+    end
+
+    def match_score(a, b, n = 2)
+      split_factor = [a.length, b.length, n].min
+      v_a = to_ngram(a, split_factor)
+      v_b = to_ngram(b, split_factor)
+
+      v_a.map { |item| v_b.include?(item) ? 1 : 0 }.sum / v_a.length.to_f
+    end
+
+    def to_ngram(str, n = 2)
+      str.each_char.each_cons(n).map(&:join)
+    end
+
     def decimalize(obj)
       case obj.class.name
       when 'Array'
@@ -131,7 +160,7 @@ module LucaSupport
       end
     end
 
-    def readable(obj, len = LucaSupport::Config::DECIMAL_NUM)
+    def readable(obj, len = LucaSupport::CONFIG['decimal_num'])
       case obj
       when Array
         obj.map { |i| readable(i) }
@@ -159,7 +188,6 @@ module LucaSupport
       end
     end
 
-    #
     # return current value with effective/defunct on target @date
     # For multiple attribues, return hash on other than 'val'. Examples are as bellows:
     #
@@ -172,18 +200,23 @@ module LucaSupport
     #     point: 1000
     #   => { 'effective' => 2020-1-1, 'rank' => 5, 'point' => 1000 }
     #
+    #   - defunct: 2020-1-1
+    #     val: 3000
+    #   => nil
+    #
     def take_current(dat, item)
-      target = dat.dig(item)
-      if target.is_a?(Array) && target.map(&:keys).flatten.include?('effective')
-        latest = target
-                 .filter { |a| Date.parse(a.dig('effective').to_s) < @date }
-                 .max { |a, b| Date.parse(a.dig('effective').to_s) <=> Date.parse(b.dig('effective').to_s) }
-        return nil if !latest.dig('defunct').nil? && Date.parse(latest.dig('defunct').to_s) < @date
-
-        latest.dig('val') || latest
-      else
-        target
-      end
+      target = dat&.dig(item)
+      return target unless target.is_a?(Array)
+
+      keys = target.map(&:keys).flatten
+      return target if !keys.include?('effective') && !keys.include?('defunct')
+
+      latest = target
+                 .reject { |a| a['defunct'] && Date.parse(a['defunct'].to_s) < @date  }
+                 .filter { |a| a['effective'] && Date.parse(a['effective'].to_s) < @date }
+                 .max { |a, b| Date.parse(a['effective'].to_s) <=> Date.parse(b['effective'].to_s) }
+
+      latest&.dig('val') || latest
     end
 
     def has_status?(dat, status)

data/lib/luca_support/config.rb

@@ -8,19 +8,15 @@ require 'yaml'
 module LucaSupport
   PJDIR = ENV['LUCA_TEST_DIR'] || Dir.pwd.freeze
   CONFIG = begin
-             YAML.load_file(Pathname(PJDIR) / 'config.yml', **{})
+             {
+               'decimal_separator' => '.',
+               'thousands_separator' => ','
+             }.merge(YAML.load_file(Pathname(PJDIR) / 'config.yml'))
            rescue Errno::ENOENT
-             {}
+             {
+               'decimal_separator' => '.',
+               'thousands_separator' => ','
+             }
            end
-
-  module Config
-    # Project top directory.
-    Pjdir = ENV['LUCA_TEST_DIR'] || Dir.pwd.freeze
-    if File.exist?(Pathname(Pjdir) / 'config.yml')
-      # DECIMAL_NUM = YAML.load_file(Pathname(Pjdir) / 'config.yml', **{})['decimal_number']
-      COUNTRY = YAML.load_file(Pathname(Pjdir) / 'config.yml', **{})['country']
-      DECIMAL_NUM ||= 0 if COUNTRY == 'jp'
-    end
-    DECIMAL_NUM ||= 2
-  end
+  CONFIG['decimal_num'] ||= CONFIG['country'] == 'jp' ? 0 : 2
 end

data/lib/luca_support/view.rb

@@ -34,9 +34,13 @@ module LucaSupport
       out
     end
 
+    # Search existing file and return path under:
+    # 1. 'templates/' in Project directory that data resides
+    # 2. 'templates/' in Library directory that calls LucaSupport::View#search_template
+    #
     def search_template(file, dir = 'templates')
       # TODO: load config
-      [@pjdir, lib_path].each do |base|
+      [LucaSupport::PJDIR, lib_path].each do |base|
         path = (Pathname(base) / dir / file)
         return path.to_path if path.file?
       end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: lucarecord
 version: !ruby/object:Gem::Version
-  version: 0.2.22
+  version: 0.2.27
 platform: ruby
 authors:
 - Chuma Takahiro
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-11-22 00:00:00.000000000 Z
+date: 2021-03-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mail
@@ -109,7 +109,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.2
+rubygems_version: 3.2.3
 signing_key: 
 specification_version: 4
 summary: ERP File operation framework