rvpacker-txt 1.11.0 → 1.12.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0be6dfc912a9d15dd4be7c37ecee0c316d650ca10f89eb99af19d5baf38addf2
-  data.tar.gz: a0ed09eac3d11c36db6d9014e4c0bc6ffa2408ac8c4a78f55fdfbb08285b759a
+  metadata.gz: dc893f5cf25ad608e961d2b85408d1acaedaed3e508eaf74688bc0814f4a740c
+  data.tar.gz: 8c157c1d2446add716a05615e1f1d10dd758bcbb30feba4938a2420b350e892d
 SHA512:
-  metadata.gz: bfc00222eff1dc676c73e62bc9048a89a4ea01fdce63eda1e7373a8d4f5a4d9975418111844704127cae24f2d01848d8486ba544f8f2de409346801f9563325c
-  data.tar.gz: 473d0707b8359f54dd49ed665addf7cbfedf4a80ccadfaf6f7d997db9e4f691ae05a4882187a5dad3c8a6dc37231c11efb6e997a573bd89a70bc3f366f6e8695
+  metadata.gz: 3dd75b346fa21bcbc16b15c7e081b7fbd72a97796afac4343fc43c656b9d84ff504cdfc51962a771e4a54a9a17c29134a0e23fc3e158cab245c8226935c23035
+  data.tar.gz: 5fe5adf14cc73bd8681c61e16186a4652e22223581265784b207553ecc592252b574b15881f0f6964074c90fdd45b58a455ccb10469932248477ef79a91c07e5

data/lib/classes.rb

@@ -75,43 +75,6 @@ class Rect
     end
 end
 
-# Fuck using an array with set, that's just straight dumb and not efficient
-class IndexSet
-    def initialize
-        @hash = Hash.new
-    end
-
-    def add(item)
-        return if @hash.include?(item)
-        @hash[item] = @hash.size
-        @hash
-    end
-
-    def include?(item)
-        @hash.include?(item)
-    end
-
-    def each(&block)
-        @hash.each_key(&block)
-    end
-
-    def to_a
-        @hash.dup
-    end
-
-    def join(delimiter = '')
-        @hash.keys.join(delimiter)
-    end
-
-    def length
-        @hash.size
-    end
-
-    def empty?
-        @hash.empty?
-    end
-end
-
 class Hash
     def insert_at_index(index, key, value)
         return self[key] = value if index >= size

data/lib/extensions.rb

@@ -127,110 +127,79 @@ def self.shuffle_words(array)
     end
 end
 
-# @param [String] string A parsed scripts code string, containing raw Ruby code
-# @param [Symbol] mode Mode to extract quoted strings
-# @return [IndexSet<String>] Set of extracted strings
-def extract_quoted_strings(string, mode)
-    if mode == :read
-        result = IndexSet.new
-
-        skip_block = false
-        in_quotes = false
-        quote_type = nil
-        buffer = []
-
-        string.each_line do |line|
-            stripped = line.strip
-
-            next if stripped[0] == '#' ||
-                (!in_quotes && !stripped.match?(/["']/)) ||
-                stripped.start_with?(/(Win|Lose)|_Fanfare/) ||
-                stripped.match?(/eval\(/)
-
-            skip_block = true if stripped.start_with?('=begin')
-            skip_block = false if stripped.start_with?('=end')
-
-            next if skip_block
-
-            line.each_char do |char|
-                if %w[' "].include?(char)
-                    unless quote_type.nil? || char == quote_type
-                        buffer.push(char)
-                        next
-                    end
-
-                    quote_type = char
-                    in_quotes = !in_quotes
-                    result.add(buffer.join)
-                    buffer.clear
-                    next
-                end
-
-                next unless in_quotes
+def escaped?(line, index)
+    backslash_count = 0
 
-                if char == "\r"
-                    next
-                elsif char == "\n"
-                    buffer.push('\#')
-                    next
-                end
-
-                buffer.push(char)
-            end
-        end
-
-        result
-    else
-        strings_array = []
-        indices_array = []
-
-        skip_block = false
-        in_quotes = false
-        quote_type = nil
-        buffer = []
+    (0..index).reverse_each do |i|
+        break if line[i] != '\\'
+        backslash_count += 1
+    end
 
-        current_string_index = 0
-        string.each_line do |line|
-            stripped = line.strip
+    backslash_count.even?
+end
 
-            if stripped[0] == '#' || stripped.start_with?(/(Win|Lose)|_Fanfare/)
-                current_string_index += line.length
+# @param [String] ruby_code
+def extract_strings(ruby_code, mode = false)
+    strings = mode ? [] : Set.new
+    indices = []
+    inside_string = false
+    inside_multiline_comment = false
+    string_start_index = 0
+    current_quote_type = ''
+
+    global_index = 0
+    ruby_code.each_line do |line|
+        stripped = line.strip
+
+        unless inside_string
+            if stripped[0] == '#'
+                global_index += line.length
                 next
             end
 
-            skip_block = true if stripped.start_with?('=begin')
-            skip_block = false if stripped.start_with?('=end')
-
-            if skip_block
-                current_string_index += line.length
-                next
+            if stripped.start_with?('=begin')
+                inside_multiline_comment = true
+            elsif stripped.start_with?('=end')
+                inside_multiline_comment = false
             end
+        end
 
-            buffer.push('\#') if in_quotes
-
-            line.each_char.each_with_index do |char, index|
-                if %w[' "].include?(char)
-                    unless quote_type.nil? || char == quote_type
-                        buffer.push(char)
-                        next
-                    end
+        if inside_multiline_comment
+            global_index += line.length
+            next
+        end
 
-                    quote_type = char
-                    in_quotes = !in_quotes
+        i = 0
+        while i < line.length
+            char = line[i]
 
-                    strings_array.push(buffer.join)
-                    indices_array.push(current_string_index + index)
+            if !inside_string && char == '#'
+                break
+            end
 
-                    buffer.clear
-                    next
+            if !inside_string && ['"', "'"].include?(char)
+                inside_string = true
+                string_start_index = global_index + i
+                current_quote_type = char
+            elsif inside_string && char == current_quote_type && escaped?(line, i - 1)
+                extracted_string = ruby_code[string_start_index + 1...global_index + i].gsub(/\r?\n/, '\#')
+
+                if mode
+                    strings << extracted_string
+                    indices << string_start_index + 1
+                else
+                    strings.add(extracted_string)
                 end
 
-                buffer.push(char) if in_quotes
+                inside_string = false
+                current_quote_type = ''
             end
 
-            current_string_index += line.length
+            i += 1
         end
 
-        [strings_array, indices_array]
+        global_index += line.length
     end
-end
+
+    mode ? [strings, indices] : strings.to_a
+end

data/lib/read.rb

@@ -92,8 +92,8 @@ def self.read_map(maps_files_paths, output_path, romanize, logging, game_type, p
 
     maps_object_map = Hash[maps_files_paths.map { |f| [File.basename(f), Marshal.load(File.binread(f))] }]
 
-    maps_lines = IndexSet.new
-    names_lines = IndexSet.new
+    maps_lines = Set.new
+    names_lines = Set.new
 
     maps_translation_map = nil
     names_translation_map = nil
@@ -161,9 +161,10 @@ def self.read_map(maps_files_paths, output_path, romanize, logging, game_type, p
 
                                 maps_lines.add(parsed)
                             end
+
+                            line.clear
                         end
 
-                        line.clear
                         in_sequence = false
                     end
 
@@ -262,7 +263,7 @@ def self.read_other(other_files_paths, output_path, romanize, logging, game_type
             next
         end
 
-        other_lines = IndexSet.new
+        other_lines = Set.new
         other_translation_map = nil
 
         if processing_mode == :append
@@ -339,9 +340,10 @@ def self.read_other(other_files_paths, output_path, romanize, logging, game_type
 
                                     other_lines.add(parsed)
                                 end
+
+                                line.clear
                             end
 
-                            line.clear
                             in_sequence = false
                         end
 
@@ -449,7 +451,7 @@ def self.read_system(system_file_path, ini_file_path, output_path, romanize, log
 
     system_object = Marshal.load(File.binread(system_file_path))
 
-    system_lines = IndexSet.new
+    system_lines = Set.new
     system_translation_map = nil
 
     if processing_mode == :append
@@ -575,7 +577,7 @@ def self.read_scripts(scripts_file_path, output_path, romanize, logging, process
 
     script_entries = Marshal.load(File.binread(scripts_file_path))
 
-    scripts_lines = IndexSet.new
+    scripts_lines = Set.new
     scripts_translation_map = nil
 
     if processing_mode == :append
@@ -593,8 +595,6 @@ def self.read_scripts(scripts_file_path, output_path, romanize, logging, process
     # This code was fun before `that` game used Windows-1252 degree symbol
     script_entries.each do |script|
         code = Zlib::Inflate.inflate(script[2]).force_encoding('UTF-8')
-        # we're fucking cloning because of encoding issue
-        codes_content.push(code.clone)
 
         # I figured how String#encode works - now everything is good
         unless code.valid_encoding?
@@ -610,62 +610,35 @@ def self.read_scripts(scripts_file_path, output_path, romanize, logging, process
             end
         end
 
-        extract_quoted_strings(code, :read).each do |string|
-            # Removes the U+3000 Japanese typographical space to check if string, when stripped, is truly empty
-            string = string.strip.delete('　')
+        codes_content.push(code)
+    end
+
+    extracted = extract_strings(codes_content.join)
 
-            next if string.empty?
+    extracted.each do |string|
+        # Removes the U+3000 Japanese typographical space to check if string, when stripped, is truly empty
+        string = string.gsub('　', ' ').strip
 
-            # Maybe this mess will remove something that mustn't be removed, but it needs to be tested
-            next if string.start_with?(/([#!?$@]|(\.\/)?(Graphics|Data|Audio|CG|Movies|Save)\/)/) ||
-                string.match?(/^[^\p{L}]+$/) ||
-                string.match?(/^\d+$/) ||
-                string.match?(/%.*(\d|\+|\*)d\]?:?$/) ||
-                string.match?(/^\[((ON|OFF)|P[EMRT])\]$/) ||
-                string.match?(/^\[\]$/) ||
-                string.match?(/^(.)\1{2,}$/) ||
-                string.match?(/^(false|true)$/) ||
-                string.match?(/^[wr]b$/) ||
-                string.match?(/^(?=.*\d)[A-Za-z0-9-]+$/) ||
-                string.match?(/^[a-z\-()\/ +'&]*$/) ||
-                string.match?(/^[A-Za-z]+[+-]$/) ||
-                string.match?(STRING_IS_ONLY_SYMBOLS_RE) ||
-                string.match?(/^Tile.*[A-Z]$/) ||
-                string.match?(/^[a-zA-Z][a-z]+([A-Z][a-z]*)+$/) ||
-                string.match?(/^Cancel Action$|^Invert$|^End$|^Individual$|^Missed File$|^Bitmap$|^Audio$/) ||
-                string.match?(/^(?=.*%d)(?=.*%m)(?=.*%Y).*$/) ||
-                string.match?(/^\\\\ALPHAC/) ||
-                string.match?(/^[A-Z]{,3}-[A-Z][A-Za-z]+/) ||
-                string.match?(/\.(mp3|ogg|jpg|png|ini|txt)$/i) ||
-                string.match?(/\/(\d.*)?$/) ||
-                string.match?(/FILE$/) ||
-                string.match?(/#\{/) ||
-                string.match?(/(?<!\\)\\(?![\\G#])/) ||
-                string.match?(/\+?=?=/) ||
-                string.match?(/[}{_<>]/) ||
-                string.match?(/r[vx]data/) ||
-                string.match?(/No such file or directory/) ||
-                string.match?(/level \*\*/) ||
-                string.match?(/Courier New|Comic Sans|Lucida|Verdana|Tahoma|Arial|Times New Roman/) ||
-                string.match?(/Player start location/) ||
-                string.match?(/Common event call has exceeded/) ||
-                string.match?(/se-/) ||
-                string.match?(/Start Pos/) ||
-                string.match?(/An error has occurred/) ||
-                string.match?(/Define it first/) ||
-                string.match?(/Process Skill/) ||
-                string.match?(/Wpn Only/) ||
-                string.match?(/Don't Wait/) ||
-                string.match?(/Clear image/) ||
-                string.match?(/Can Collapse/)
+        next if string.empty?
 
-            string = romanize_string(string) if romanize
+        next if string.match?(/(Graphics|Data|Audio|Movies|System)\/.*\/?/) ||
+            string.match?(/r[xv]data2?$/) ||
+            string.match?(STRING_IS_ONLY_SYMBOLS_RE) ||
+            string.match?(/@window/) ||
+            string.match?(/\$game/) ||
+            string.match?(/_/) ||
+            string.match?(/^\\e/) ||
+            string.match?(/.*\(/) ||
+            string.match?(/^([d\d\p{P}+-]*|[d\p{P}+-]*)$/) ||
+            string.match?(/ALPHAC/) ||
+            string.match?(/^(Actor<id>|ExtraDropItem|EquipLearnSkill|GameOver|Iconset|Window|true|false|MActor%d|w[rb]|\\f|\\n|\[[A-Z]*\])$/)
 
-            scripts_translation_map.insert_at_index(scripts_lines.length, string, '') if processing_mode == :append &&
-                !scripts_translation_map.include?(string)
+        string = romanize_string(string) if romanize
 
-            scripts_lines.add(string)
-        end
+        scripts_translation_map.insert_at_index(scripts_lines.length, string, '') if processing_mode == :append &&
+            !scripts_translation_map.include?(string)
+
+        scripts_lines.add(string)
     end
 
     puts "Parsed #{scripts_filename}" if logging

data/lib/write.rb

@@ -164,7 +164,7 @@ def self.write_map(original_files_paths, maps_path, output_path, shuffle_level,
                                 end
 
                                 if split_length > line_length
-                                    list[item_indices.last].parameters[0] = split[line_length..].join("\n")
+                                    list[item_indices.last].parameters[0] = split[line_length - 1..].join("\n")
                                 end
                             end
                         end
@@ -335,7 +335,7 @@ def self.write_other(original_files_paths, other_path, output_path, shuffle_leve
                                     end
 
                                     if split_length > line_length
-                                        list[item_indices.last].parameters[0] = split[line_length..].join("\n")
+                                        list[item_indices.last].parameters[0] = split[line_length - 1..].join("\n")
                                     end
                                 end
                             end
@@ -536,15 +536,10 @@ def self.write_scripts(scripts_file_path, other_path, output_path, romanize, log
 
     scripts_translation_map = Hash[scripts_original_text.zip(scripts_translated_text)]
 
-    # Shuffle can possibly break the game in scripts, so no shuffling
-    codes = []
-
-    # This code was fun before `that` game used Windows-1252 degree symbol
     script_entries.each do |script|
         code = Zlib::Inflate.inflate(script[2])
         code.force_encoding('UTF-8')
 
-        # I figured how String#encode works - now everything is good
         unless code.valid_encoding?
             [Encoding::UTF_8, Encoding::WINDOWS_1252, Encoding::SHIFT_JIS].each do |encoding|
                 encoded = code.encode(code.encoding, encoding)
@@ -559,23 +554,21 @@ def self.write_scripts(scripts_file_path, other_path, output_path, romanize, log
         end
 
         # this shit finally works and requires NO further changes
-        string_array, index_array = extract_quoted_strings(code, :write)
+        string_array, index_array = extract_strings(code, true)
 
         string_array.zip(index_array).reverse_each do |string, index|
-            string = string.strip.delete('　')
+            string = string.gsub('　', '').strip
             next if string.empty? || !scripts_translation_map.include?(string)
 
             string = romanize_string(string) if romanize
 
             gotten = scripts_translation_map[string]
-            code[index - string.length, string.length] = gotten unless gotten.nil? || gotten.empty?
+            code[index, string.length] = gotten unless gotten.nil? || gotten.empty?
         end
 
-        codes.push(code)
         script[2] = Zlib::Deflate.deflate(code, Zlib::BEST_COMPRESSION)
     end
 
-    # File.binwrite(File.join(output_path, 'scripts_plain.txt'), codes.join("\n")) - debug line
     File.binwrite(File.join(output_path, scripts_basename), Marshal.dump(script_entries))
     puts "Written #{scripts_basename}" if logging
 end

data/rvpacker-txt.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |spec|
     spec.name = 'rvpacker-txt'
-    spec.version = '1.11.0'
+    spec.version = '1.12.0'
     spec.authors = ['Howard Jeng', 'Andrew Kesterson', 'Solistra', 'Darkness9724', 'savannstm']
     spec.email = ['savannstm@gmail.com']
     spec.summary = 'Reads RPG Maker XP/VX/VXAce game text to .txt files and writes them to their initial form.'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rvpacker-txt
 version: !ruby/object:Gem::Version
-  version: 1.11.0
+  version: 1.12.0
 platform: ruby
 authors:
 - Howard Jeng
@@ -12,7 +12,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-07-28 00:00:00.000000000 Z
+date: 2024-08-02 00:00:00.000000000 Z
 dependencies: []
 description:
 email: