lucarecord 0.2.23 → 0.2.28

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a408c56d2c0f47238bc8c17e4135867e1be31b9897f940370a9bc8e87f8d014e
-  data.tar.gz: 66fead2c3441f8b019f8d805fe564dbf9a9121abec424ea5d5482277d7bcc3ad
+  metadata.gz: c52d38b86c9670e3358ef9e94553902a3a511f626689566a4dc0b1d5e480c041
+  data.tar.gz: 064b974bbeeaa1dbb8e644532c5e2bee67cdd5d697e1cdde99087204698aa665
 SHA512:
-  metadata.gz: a6f3d2e4773bc6c29c657a7fbe5864893efa39ec739ae1df4eece68fa82907bf99feb72f5a43888b59e39bc9f9a93f023fa4c0243e2429d3c04ba2c2f09d2e20
-  data.tar.gz: 7fd98c339f4b81430dc961b009233e6de65228f6df83d361b81f4040857045740f0d2d18363e29bd2579be6b3d19a5a09cca37f9e39cfc1a8703212c32ba4e78
+  metadata.gz: 20ece81a5471e985bdc6b46aa341b3d859ad050b28a5976b0b20f15a44971ac0111b16a58db1053c91da73d91f8c37dc26e6bd825b4604a3c17ced65239a19b2
+  data.tar.gz: fc4af325295583e739fb972f3d96470f0a84d74ea4a4b330a16178d88de0c4700f64b28f36166539904c3b589c7540ec4486e36e74d6fd0d1e908da22e4cc685

data/CHANGELOG.md

@@ -1,3 +1,29 @@
+## LucaRecord 0.2.28
+
+* implement LucaSupport::Range, handle #by_month enumeration between several months.
+* @record_type = 'raw' is deprecated in favor of overriding LucaRecord::IO.load_data
+* change code search from exact match to prefix match
+
+## LucaRecord 0.2.27
+
+* Fix: update_digest
+
+## 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.

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

@@ -17,10 +17,11 @@ module LucaRecord
       set_driver
     end
 
-    # Search word with n-gram.
+    # 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
 
@@ -34,6 +35,12 @@ module LucaRecord
       end
     end
 
+    # Search with unique code.
+    #
+    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.
     #
@@ -49,7 +56,6 @@ module LucaRecord
       [obj[main_key], options.compact]
     end
 
-    #
     # Load CSV with config options
     #
     def load_csv(path)
@@ -58,7 +64,6 @@ module LucaRecord
       end
     end
 
-    #
     # load dictionary data
     #
     def self.load(file = @filename)
@@ -72,7 +77,6 @@ module LucaRecord
       end
     end
 
-    #
     # generate dictionary from TSV file. Minimum assumption is as bellows:
     # 1st row is converted symbol.
     #
@@ -101,7 +105,7 @@ module LucaRecord
         puts 'No error detected.'
         nil
       else
-        "Key #{errors.join(', ')} has nil #{target_key}."
+        puts "Key #{errors.join(', ')} has nil #{target_key}."
         errors.count
       end
     end
@@ -109,13 +113,19 @@ module LucaRecord
     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)
@@ -124,7 +134,7 @@ module LucaRecord
 
     def max_score_code(str)
       res = @definitions.map do |k, v|
-        [v, 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

@@ -64,11 +64,7 @@ module LucaRecord # :nodoc:
 
         LucaSupport::Code.encode_term(start_year, start_month, end_year, end_month).each do |subdir|
           open_records(basedir, subdir, nil, code) do |f, path|
-            if @record_type == 'raw'
-              yield f, path
-            else
-              yield load_data(f, path)
-            end
+            yield load_data(f, path)
           end
         end
       end
@@ -80,11 +76,7 @@ module LucaRecord # :nodoc:
 
         subdir = year.to_s + LucaSupport::Code.encode_month(month)
         open_records(basedir, subdir, LucaSupport::Code.encode_date(day), code) do |f, path|
-          if @record_type == 'raw'
-            yield f, path
-          else
-            yield load_data(f, path), path
-          end
+          yield load_data(f, path), path
         end
       end
 
@@ -243,7 +235,7 @@ 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 +244,17 @@ 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 +305,14 @@ 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 +350,11 @@ 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 'json'
+            # TODO: implement JSON parse
+          end
         else
           LucaSupport::Code.decimalize(YAML.load(io.read)).tap { |obj| validate_keys(obj) }
         end
@@ -362,17 +372,20 @@ 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
-      # false when file may have one
+      # True when file doesn't have record on code.
+      # False when file may have one.
+      # If filename doesn't record codes, always return false,
+      # so later check is required. This is partial optimization.
+      #
       def skip_on_unmatch_code(subpath, code = nil)
         # p filename.split('-')[1..-1]
         filename = subpath.split('/').last
         return false if code.nil? || filename.length <= 4
 
-        !filename.split('-')[1..-1].include?(code)
+        filename.split('-')[1..-1].select { |fragment| /^#{code}/.match(fragment) }.empty?
       end
 
       # AUTO INCREMENT

data/lib/luca_record/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module LucaRecord
-  VERSION = '0.2.23'
+  VERSION = '0.2.28'
 end

data/lib/luca_support/code.rb

@@ -5,8 +5,9 @@ require 'securerandom'
 require 'digest/sha1'
 require 'luca_support/config'
 
-# implement Luca IDs convention
-module LucaSupport
+module LucaSupport # :nodoc:
+  # implement Luca IDs convention
+  #
   module Code
     module_function
 
@@ -54,21 +55,24 @@ 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.
@@ -114,9 +118,24 @@ 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)
-      v_a = to_ngram(a, n)
-      v_b = to_ngram(b, n)
+      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
@@ -142,7 +161,7 @@ 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) }
@@ -170,7 +189,6 @@ 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:
     #
@@ -183,18 +201,23 @@ 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/range.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module LucaSupport # :nodoc:
+  # Partial range operation
+  #
+  module Range
+    def self.included(klass) # :nodoc:
+      klass.extend ClassMethods
+    end
+
+    def by_month(step = nil, from: nil, to: nil)
+      return enum_for(:by_month, step, from: from, to: to) unless block_given?
+
+      from ||= @start_date
+      to ||= @end_date
+      self.class.term_by_month(from, to, step || 1).each do |date|
+        @cursor_start = date
+        @cursor_end = step.nil? ? date : [date.next_month(step - 1), to].min
+        yield @cursor_start, @cursor_end
+      end
+    end
+
+    module ClassMethods
+      def term_by_month(start_date, end_date, step = 1)
+        Enumerator.new do |yielder|
+          each_month = start_date
+          while each_month <= end_date
+            yielder << each_month
+            each_month = each_month.next_month(step)
+          end
+        end
+      end
+    end
+  end
+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: 0.2.23
+  version: 0.2.28
 platform: ruby
 authors:
 - Chuma Takahiro
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-11-25 00:00:00.000000000 Z
+date: 2021-03-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mail
@@ -86,6 +86,7 @@ files:
 - lib/luca_support/code.rb
 - lib/luca_support/config.rb
 - lib/luca_support/mail.rb
+- lib/luca_support/range.rb
 - lib/luca_support/view.rb
 homepage: https://github.com/chumaltd/luca/tree/master/lucarecord
 licenses:
@@ -109,7 +110,7 @@ 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