xlg 0.4.2 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 487577ad7b07e1220696b70de189c3c098a39547c84dc8db0204b5420c4ae0af
-  data.tar.gz: 8e58a4ba07793fd06036bbeb3c03069938dab6fe5f3eb0aa21dc38a7d94912bf
+  metadata.gz: 33bc5a5ea33c1b5b7a68bb4d7a7061c03323eb1be97cbb68c2b156361ec896d0
+  data.tar.gz: 5fd055f3b0c8bdd53d83400e4b1818e97ac84380abc11bcc894f8b477f3e5bb4
 SHA512:
-  metadata.gz: 54c70a935c0b50d11d63e70f9d0203172afd17022f8bd058a2ddb8cc8e66de14524a561881d8070ca301d93aede144305c909c9a0fb9d389628bc69ee1dfeeff
-  data.tar.gz: da330328bb724784d21f0d1e59322f4eec091fbe57c48fc55ab03cee9238470867c8b321a2b833894a96cf69c80b830f528bfa800188b44b7744aee66516ca21
+  metadata.gz: dedc5f0004b795af9b98782e350c9e7dc1ccaee8bdd094950dad05b2bab945c2e12d73833bf53506ae24fab55bb9b23a1934424e604d416921417f1a39441dbd
+  data.tar.gz: 5097f1343ad1a3bb517bd95267034598b120ca74da92e896b1b1338b78c03e2b2a0099ece782811a079bccf331675454c9a1b92adf7db18d103b6b97c9595cad

data/README.md

@@ -5,6 +5,7 @@
 ## Features
 
 - Search text in Excel files (.xlsx, .xls, .xlsm)
+- **Regular expression support** for advanced pattern matching
 - Support for single file, multiple files, and directory-based searches
 - Case-insensitive keyword matching
 - Grep-like output format: `filename:sheet:cell_ref:matched_text`
@@ -41,19 +42,49 @@ xlg 'keyword' file1.xlsx file2.xlsx
 xlg 'keyword' /path/to/directory/
 ```
 
+### Regular Expression Support
+
+`xlg` supports Ruby-style regular expressions for advanced pattern matching:
+
+```bash
+# Search for date patterns (YYYY-MM-DD format)
+xlg '\d{4}-\d{2}-\d{2}' data.xlsx
+
+# Search for email addresses
+xlg '[a-zA-Z0-9]+@[a-zA-Z0-9]+\.[a-zA-Z]+' contacts.xlsx
+
+# Search for strings starting with capital letters
+xlg '^[A-Z].*' documents.xlsx
+
+# Search for "test" OR "demo"
+xlg 'test|demo' samples.xlsx
+
+# Case-insensitive regex using flags
+xlg '/hello/i' greetings.xlsx
+
+# Search for phone numbers
+xlg '\d{3}-\d{3}-\d{4}' phonebook.xlsx
+```
+
 ### Command Line Options
 
 ```
-xlg KEYWORD FILE.xlsx                    # Single file search
-xlg KEYWORD FILE1.xlsx FILE2.xlsx       # Multiple file search
-xlg KEYWORD /path/to/directory/          # Directory search
+xlg PATTERN FILE.xlsx                    # Single file search
+xlg PATTERN FILE1.xlsx FILE2.xlsx       # Multiple file search
+xlg PATTERN /path/to/directory/          # Directory search
 
 Options:
   -h, --help  Show help message
 ```
 
+Where `PATTERN` can be:
+- A simple text string (e.g., `'hello'`)
+- A regular expression (e.g., `'\d+'` for numbers)
+- A regex with flags (e.g., `'/pattern/i'` for case-insensitive)
+
 ### Examples
 
+#### Text Search
 ```bash
 # Search for "test" in sample.xlsx
 xlg 'test' sample.xlsx
@@ -65,6 +96,24 @@ xlg 'データ' file1.xlsx file2.xlsx
 xlg 'キーワード' /home/user/documents/
 ```
 
+#### Regular Expression Search
+```bash
+# Find all dates in YYYY-MM-DD format
+xlg '\d{4}-\d{2}-\d{2}' reports.xlsx
+
+# Find email addresses
+xlg '\w+@\w+\.\w+' contacts.xlsx
+
+# Find cells starting with "Total"
+xlg '^Total' financial.xlsx
+
+# Find prices in dollar format
+xlg '\$\d+\.\d{2}' prices.xlsx
+
+# Case-insensitive search
+xlg '/error/i' logs.xlsx
+```
+
 ## Output Format
 
 The output follows a grep-like format:
@@ -97,7 +146,7 @@ The tool consists of four main components:
 
 - `ExcelGrep` (lib/excel_grep.rb): Core search functionality for single files
 - `MultiFileSearcher` (lib/multi_file_searcher.rb): Handles multiple files and directories
-- `CellMatcher` (lib/cell_matcher.rb): Performs case-insensitive text matching
+- `CellMatcher` (lib/cell_matcher.rb): Performs text and regular expression matching
 - `OutputFormatter` (lib/output_formatter.rb): Formats output in grep-like style
 
 ## Development
@@ -105,7 +154,13 @@ The tool consists of four main components:
 ### Running Tests
 
 ```bash
-ruby test/test_*.rb
+# Run individual test files
+RUBYLIB=lib bundle exec ruby test/test_cell_matcher.rb
+RUBYLIB=lib bundle exec ruby test/test_regex_integration.rb
+RUBYLIB=lib bundle exec ruby test/test_edge_cases.rb
+
+# Or run all tests
+RUBYLIB=lib bundle exec ruby test/test_*.rb
 ```
 
 ### Project Structure

data/bin/xlg

@@ -5,14 +5,15 @@ require 'multi_file_searcher'
 
 def show_help
   puts "使用方法:"
-  puts "  xlg KEYWORD FILE.xlsx                    # 単一ファイル"
-  puts "  xlg KEYWORD FILE1.xlsx FILE2.xlsx       # 複数ファイル"
-  puts "  xlg KEYWORD /path/to/directory/          # ディレクトリ内全検索"
+  puts "  xlg PATTERN FILE.xlsx                    # 単一ファイル"
+  puts "  xlg PATTERN FILE1.xlsx FILE2.xlsx       # 複数ファイル"
+  puts "  xlg PATTERN /path/to/directory/          # ディレクトリ内全検索"
   puts ""
   puts "Excel ファイル内の文字列を検索し、grep形式で出力します。"
+  puts "検索パターンには通常の文字列または正規表現を指定できます。"
   puts ""
   puts "引数:"
-  puts "  KEYWORD         検索するキーワード"
+  puts "  PATTERN         検索パターン（文字列または正規表現）"
   puts "  FILE.xlsx       検索対象のExcelファイル（複数指定可能）"
   puts "  /path/to/dir/   検索対象のディレクトリ"
   puts ""
@@ -22,9 +23,15 @@ def show_help
   puts "  .xlsm (マクロ有効Excel)"
   puts ""
   puts "例:"
-  puts "  xlg 'test' sample.xlsx"
-  puts "  xlg 'データ' file1.xlsx file2.xlsx"
-  puts "  xlg 'キーワード' /home/user/documents/"
+  puts "  xlg 'test' sample.xlsx                  # 通常の文字列検索"
+  puts "  xlg 'データ' file1.xlsx file2.xlsx      # 複数ファイル検索"
+  puts "  xlg 'キーワード' /home/user/documents/   # ディレクトリ検索"
+  puts ""
+  puts "正規表現の例:"
+  puts "  xlg '\\d{4}-\\d{2}-\\d{2}' data.xlsx      # 日付形式 (YYYY-MM-DD)"
+  puts "  xlg '^[A-Z].*' data.xlsx               # 大文字で始まる文字列"
+  puts "  xlg 'test|demo' data.xlsx              # testまたはdemoを含む"
+  puts "  xlg '/pattern/i' data.xlsx             # 大文字小文字を区別しない"
   puts ""
   puts "オプション:"
   puts "  -h, --help  このヘルプを表示"
@@ -38,12 +45,12 @@ def main
 
   if ARGV.length < 2
     $stderr.puts "エラー: 引数の数が正しくありません"
-    $stderr.puts "使用方法: xlg KEYWORD FILE.xlsx [FILE2.xlsx ...]"
+    $stderr.puts "使用方法: xlg PATTERN FILE.xlsx [FILE2.xlsx ...]"
     $stderr.puts "詳細は 'xlg --help' を参照してください"
     exit 1
   end
 
-  keyword = ARGV[0]
+  pattern = ARGV[0]
   paths = ARGV[1..-1]
 
   begin
@@ -52,16 +59,16 @@ def main
       path = paths[0]
       if File.directory?(path)
         # ディレクトリの場合はMultiFileSearcherを使用
-        searcher = MultiFileSearcher.new(keyword, [path])
+        searcher = MultiFileSearcher.new(pattern, [path])
         success = searcher.search
       else
         # 単一ファイルの場合はExcelGrepを使用
-        grep = ExcelGrep.new(keyword, path)
+        grep = ExcelGrep.new(pattern, path)
         success = grep.search
       end
     else
       # 複数パスの場合はMultiFileSearcherを使用
-      searcher = MultiFileSearcher.new(keyword, paths)
+      searcher = MultiFileSearcher.new(pattern, paths)
       success = searcher.search
     end
     
@@ -69,6 +76,10 @@ def main
   rescue ArgumentError => e
     $stderr.puts "エラー: #{e.message}"
     exit 1
+  rescue RegexpError => e
+    $stderr.puts "正規表現エラー: 不正な正規表現パターンです - #{e.message}"
+    $stderr.puts "正規表現の構文については 'xlg --help' を参照してください"
+    exit 1
   rescue => e
     $stderr.puts "予期しないエラーが発生しました: #{e.message}"
     exit 1

data/lib/cell_matcher.rb

@@ -1,22 +1,93 @@
 class CellMatcher
-  def match?(cell_value, keyword)
-    return false if cell_value.nil? || keyword.nil? || keyword.empty?
+  def initialize(pattern)
+    raise ArgumentError, "パターンが空です" if pattern.nil? || pattern.empty?
+    @pattern = pattern
+    @compiled_pattern = compile_pattern(pattern)
+  end
+
+  def match?(cell_value)
+    return false if cell_value.nil? || @pattern.nil? || @pattern.empty?
     
     cell_str = cell_value.to_s
     return false if cell_str.empty?
     
-    cell_str.downcase.include?(keyword.downcase)
+    if is_regex?(@pattern)
+      if @compiled_pattern.is_a?(Regexp)
+        !(@compiled_pattern =~ cell_str).nil?
+      else
+        # Fallback to literal matching if regex compilation failed
+        cell_str.downcase.include?(@pattern.downcase)
+      end
+    else
+      cell_str.downcase.include?(@pattern.downcase)
+    end
   end
 
-  def extract_match(cell_value, keyword)
-    return nil unless match?(cell_value, keyword)
+  def extract_match(cell_value)
+    return nil unless match?(cell_value)
     
     cell_str = cell_value.to_s
-    keyword_lower = keyword.downcase
     
-    start_index = cell_str.downcase.index(keyword_lower)
-    return nil if start_index.nil?
-    
-    cell_str[start_index, keyword.length]
+    if is_regex?(@pattern)
+      if @compiled_pattern.is_a?(Regexp)
+        match_result = @compiled_pattern.match(cell_str)
+        match_result ? match_result[0] : nil
+      else
+        # Fallback to literal matching if regex compilation failed
+        pattern_lower = @pattern.downcase
+        start_index = cell_str.downcase.index(pattern_lower)
+        return nil if start_index.nil?
+        
+        cell_str[start_index, @pattern.length]
+      end
+    else
+      pattern_lower = @pattern.downcase
+      start_index = cell_str.downcase.index(pattern_lower)
+      return nil if start_index.nil?
+      
+      cell_str[start_index, @pattern.length]
+    end
+  end
+
+  private
+
+  def is_regex?(pattern)
+    # Check if pattern looks like a regex (contains regex metacharacters)
+    pattern.match?(/[.*+?^${}()|\\]/) || 
+    pattern.include?('[') || pattern.include?(']') ||
+    pattern.start_with?('/') && pattern.end_with?('/') ||
+    pattern.start_with?('/') && pattern.match?(/\/[gimxo]*$/)
+  end
+
+  def compile_pattern(pattern)
+    begin
+      if is_regex?(pattern)
+        # Handle /pattern/flags format
+        if pattern.start_with?('/') && pattern.match?(/\/([gimxo]*)$/)
+          # Extract pattern and flags
+          match = pattern.match(/^\/(.*)\/([gimxo]*)$/)
+          if match
+            regex_pattern = match[1]
+            flags = match[2]
+            
+            options = 0
+            options |= Regexp::IGNORECASE if flags.include?('i')
+            options |= Regexp::MULTILINE if flags.include?('m')
+            options |= Regexp::EXTENDED if flags.include?('x')
+            
+            return Regexp.new(regex_pattern, options)
+          end
+        end
+        
+        # Try to compile as-is for patterns with metacharacters
+        Regexp.new(pattern, Regexp::IGNORECASE)
+      else
+        # For plain text, return the pattern as-is
+        pattern
+      end
+    rescue RegexpError => e
+      # If regex compilation fails, treat as plain text
+      pattern
+    end
   end
 end

data/lib/excel_grep.rb

@@ -3,15 +3,15 @@ require 'cell_matcher'
 require 'output_formatter'
 
 class ExcelGrep
-  attr_reader :keyword, :file_path
+  attr_reader :pattern, :file_path
 
-  def initialize(keyword, file_path)
-    raise ArgumentError, "キーワードが空です" if keyword.nil? || keyword.empty?
+  def initialize(pattern, file_path)
+    raise ArgumentError, "検索パターンが空です" if pattern.nil? || pattern.empty?
     raise ArgumentError, "ファイルパスが空です" if file_path.nil? || file_path.empty?
     
-    @keyword = keyword
+    @pattern = pattern
     @file_path = file_path
-    @matcher = CellMatcher.new
+    @matcher = CellMatcher.new(pattern)
     @formatter = OutputFormatter.new
   end
 
@@ -45,9 +45,10 @@ class ExcelGrep
             next if cell.nil? || cell.value.nil?
             
             cell_value = cell.value.to_s
-            if @matcher.match?(cell_value, @keyword)
+            if @matcher.match?(cell_value)
               cell_ref = RubyXL::Reference.ind2ref(row_index, col_index)
-              puts @formatter.format(file_name, sheet_name, cell_ref, cell_value)
+              match_text = @matcher.extract_match(cell_value) || cell_value
+              puts @formatter.format(file_name, sheet_name, cell_ref, match_text)
             end
           end
         end

data/lib/multi_file_searcher.rb

@@ -1,13 +1,13 @@
 require 'excel_grep'
 
 class MultiFileSearcher
-  attr_reader :keyword, :paths
+  attr_reader :pattern, :paths
 
-  def initialize(keyword, paths)
-    raise ArgumentError, "キーワードが空です" if keyword.nil? || keyword.empty?
+  def initialize(pattern, paths)
+    raise ArgumentError, "検索パターンが空です" if pattern.nil? || pattern.empty?
     raise ArgumentError, "パスが空です" if paths.nil? || paths.empty?
     
-    @keyword = keyword
+    @pattern = pattern
     @paths = paths
   end
 
@@ -17,7 +17,7 @@ class MultiFileSearcher
     
     expanded_files.each do |file_path|
       begin
-        grep = ExcelGrep.new(@keyword, file_path)
+        grep = ExcelGrep.new(@pattern, file_path)
         if grep.search
           success_count += 1
         end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: xlg
 version: !ruby/object:Gem::Version
-  version: 0.4.2
+  version: 1.0.0
 platform: ruby
 authors:
 - Kazto TAKAHASHI
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-08-08 00:00:00.000000000 Z
+date: 2025-08-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rubyXL