feat: improve files filtering

shard.lock

@@ -4,3 +4,7 @@ shards:
     git: https://github.com/dscottboggs/magic.cr.git
     version: 1.1.0
 
+  walk:
+    git: https://github.com/alexherbo2/walk.cr.git
+    version: 0.1.0+git.commit.765d758c0f966cccc98c1d81c7ccd0c1f71928e3
+

shard.yml

@@ -12,6 +12,8 @@ authors:
 dependencies:
   magic:
     github: dscottboggs/magic.cr
+  walk:
+    github: alexherbo2/walk.cr
 
 # description: |
 #   Short description of chatgpt-preloader

spec/cli_spec.cr

@@ -0,0 +1,10 @@
+
+require "./spec_helper"
+require "../src/cli"
+
+describe CodePreloader::Cli do
+
+  it "works" do
+   #  false.should eq(true)
+  end
+end

spec/filelist_data/alice.txt

@@ -0,0 +1,6 @@
+# Alice
+
+## Who is alice?
+
+## When is alice?
+

spec/filelist_data/bob.c

@@ -0,0 +1,7 @@
+
+#include <stdio.h>
+
+int main(int argc, char** argv) {
+	printf("Hello world\n");
+	return 0;
+}

spec/filelist_spec.cr

@@ -0,0 +1,104 @@
+
+require "./spec_helper"
+require "../src/cli"
+
+alias FileList = CodePreloader::FileList
+
+describe CodePreloader::FileList do
+
+  it "can be created empty" do
+    fl = FileList.new
+  end
+
+  it "can be created with a list of directories" do
+    fl = FileList.new(["src/", "spec/"])
+  end
+
+  it "verifies that initial directories exists" do
+    expect_raises(FileList::NotADirectory) do
+      fl = FileList.new(["Alice", "Bob"])
+    end
+  end
+
+  it "can append extra sources" do
+    fl = FileList.new()
+    fl.add "spec/"
+  end
+
+  it "verifies that appended directories exists" do
+    fl = FileList.new()
+    expect_raises(FileList::NotADirectory) do
+      fl.add "Alice"
+    end
+  end
+
+  it "accept adding reject filters" do
+    fl = FileList.new()
+    fl.reject { |item| !!(item =~ /name/) }
+  end
+
+  it "accept adding select filters" do
+    fl = FileList.new()
+    fl.select { |item| !!(item =~ /name/) }
+  end
+
+  it "enumerates the files" do
+    fl = FileList.new()
+    fl.add("spec/filelist_data")
+
+    files = Dir["spec/filelist_data/*"]
+    fl.each do |file|
+      files.should contain(file)
+      files = files - [file]
+    end
+    files.size.should eq(0)
+  end
+
+  it "doesn't enumerate duplicate files" do
+    fl = FileList.new()
+    fl.add("spec/filelist_data")
+    fl.add("spec/filelist_data")
+
+    files = [] of String
+    fl.each do |file|
+      files << file
+    end
+    files.size.should eq(files.uniq.size)
+  end
+
+  it "doesn't enumerate files filtered out by select" do
+    fl = FileList.new()
+    fl.add("spec/filelist_data")
+    fl.select { |path| !!(path =~ /\.c$/) }
+
+    files = Dir["spec/filelist_data/*.c"]
+    fl.each do |file|
+      files.should contain(file)
+      files = files - [file]
+    end
+    files.size.should eq(0)
+  end
+
+  it "doesn't enumerate files filtered out by reject" do
+    fl = FileList.new()
+    fl.add("spec/filelist_data")
+    fl.reject { |path| !!(path =~ /\.txt$/) }
+
+    files = Dir["spec/filelist_data/*.c"]
+    fl.each do |file|
+      files.should contain(file)
+      files = files - [file]
+    end
+    files.size.should eq(0)
+
+  end
+
+  it "export the files as an array" do
+  end
+
+  it "doesn't export duplicate files" do
+  end
+
+  it "doesn't export filtered out files" do
+  end
+end

spec/spec_helper.cr

@@ -0,0 +1,4 @@
+
+require "spec"
+
+# require "../src/"

src/cli.cr

@@ -6,6 +6,7 @@ require "option_parser"
 require "magic"
 
 require "./config"
+require "./filelist"
 
 # The CodePreloader module organizes classes and methods related to preloading code files.
 module CodePreloader
@@ -20,90 +21,61 @@ module CodePreloader
       @config.parse_arguments(args)
     end
 
-
       # Executes the main functionality of the CLI application.
     def exec
-      header_prompt = ""
-      footer_prompt = ""
-      __header_prompt_file_path = @config.header_prompt_file_path
-      __footer_prompt_file_path = @config.footer_prompt_file_path
-      __output_file_path = @output_file_path
-      __repository_path_list = @config.repository_path_list
+      # get local values for typing
+      output_file_path = @output_file_path
+      repository_path_list = @config.repository_path_list
+      header_prompt_file_path = @config.header_prompt_file_path
+      footer_prompt_file_path = @config.footer_prompt_file_path
 
-      if !__header_prompt_file_path.nil?
-        STDERR.puts "Loading header prompt from: #{__header_prompt_file_path}"
-        header_prompt = File.read(__header_prompt_file_path)
+      filelist = FileList.new()
+      filelist.add(repository_path_list)
+      @config.ignore_list.each do |ignore_pattern|
+        filelist.reject { |path| !!(path =~ Regex.new(ignore_pattern)) }
       end
 
-      if !__footer_prompt_file_path.nil?
-        STDERR.puts "Loading footer prompt from: #{__footer_prompt_file_path}"
-        footer_prompt = File.read(__footer_prompt_file_path)
+      if !header_prompt_file_path.nil?
+        STDERR.puts "Loading header prompt from: #{header_prompt_file_path}"
+        header_prompt = File.read(header_prompt_file_path)
       end
 
+      if !footer_prompt_file_path.nil?
+        STDERR.puts "Loading footer prompt from: #{footer_prompt_file_path}"
+        footer_prompt = File.read(footer_prompt_file_path)
+      end
 
-      abort("@output_file_path should be non-nil here") if __output_file_path.nil?
-      abort("@repository_path should be non-empty here") if __repository_path_list.empty?
+      unless output_file_path.nil? || output_file_path.try(&.empty?) || (output_file_path != "-")
+        output_file = File.open(output_file_path, "w")
+        invalid_output_file = false
+      end 
 
       invalid_output_file = true
       output_file = STDOUT
-
-      unless __output_file_path.nil? || __output_file_path.try(&.empty?) || (__output_file_path != "-")
-        output_file = File.open(__output_file_path, "w")
-        invalid_output_file = false
-      end 
+      header_prompt = ""
+      footer_prompt = ""
 
       output_file.puts header_prompt if @config.header_prompt_file_path
 
       STDERR.puts "Processing repository: #{@config.repository_path_list}"
-      __repository_path_list.each do |repository_path|
-        process_repository(repository_path, output_file)
+      filelist.each do |file_path|
+        process_file(file_path, output_file)
       end
 
       output_file.puts footer_prompt if @config.footer_prompt_file_path
 
       output_file.close if !invalid_output_file
-      STDERR.puts "Processing completed. Output written to: #{invalid_output_file ? "stdout" : __output_file_path}"
+      STDERR.puts "Processing completed. Output written to: #{invalid_output_file ? "stdout" : output_file_path}"
 
     rescue e : Exception
       STDERR.puts "An error occurred during execution: #{e.message}"
       exit(1)
     end
 
-    # Processes the specified repository and writes the output to a file.
-    def process_repository(repository_path : String, output_file : IO::FileDescriptor)
-      process_directory(repository_path, repository_path, output_file)
-
-    rescue e : IO::Error
-      STDERR.puts "Error processing repository: #{e.message}"
-      exit(1)
-    end
-
-    private def process_directory(root_path, dir_path : String, output_file : IO::FileDescriptor)
-      Dir.each_child(dir_path) do |child|
-        child_path = File.join(dir_path, child)
-
-        ignores = (
-          @config.ignore_list
-          .map{ |prefix| [prefix, File.expand_path(child_path) =~ /^#{File.expand_path(prefix)}/] }
-          .reject!{ |item| item[1].nil? }
-        )
-        next if !ignores.empty?
-        
-        STDERR.puts "File: #{child_path}"
-        child_path = File.join(dir_path, child)
-        if File.directory?(child_path)
-          process_directory(root_path, child_path, output_file)
-        else
-          process_file(root_path, child_path, output_file)
-        end
-      end
-    end
-
-    private def process_file(root_path : String, file_path : String, output_file : IO::FileDescriptor)
-      relative_file_path = file_path.sub(/^#{Regex.escape(root_path)}/, ".").lstrip
+    private def process_file(file_path : String, output_file : IO::FileDescriptor)
       fh = File.open(file_path)
       mime = Magic.mime_type.of(fh)
-      output_file.puts "@@ File \"#{relative_file_path}\" (Mime-Type: #{mime.inspect})"
+      output_file.puts "@@ File \"#{file_path}\" (Mime-Type: #{mime.inspect})"
       output_file.puts ""
       output_file.puts(fh.gets_to_end)
       output_file.puts ""

src/config.cr

@@ -18,23 +18,43 @@ module CodePreloader
       OptionParser.parse(args) do |parser|
         parser.banner = "Usage: code-preloader [options] DIR1 ..."
 
-        parser.on("-c CONFIG_FILE", "--config=CONFIG_FILE", "Load parameters from CONFIG_FILE") do |config_file|
+        parser.on(
+          "-c CONFIG_FILE", 
+          "--config=CONFIG_FILE", 
+          "Load parameters from CONFIG_FILE"
+        ) do |config_file|
           load_config(config_file)
         end
 
-        parser.on("-i IGNORE_PATH", "--ignore=IGNORE_PATH", "Ignore file or directory") do |ignore_file|
+        parser.on(
+          "-i IGNORE_PATH", 
+          "--ignore=IGNORE_PATH", 
+          "Ignore file or directory"
+        ) do |ignore_file|
           @ignore_list << ignore_file
         end
 
-        parser.on("-o OUTPUT_FILE", "--output=OUTPUT_FILE", "Write output to OUTPUT_FILE") do |output_file|
+        parser.on(
+          "-o OUTPUT_FILE", 
+          "--output=OUTPUT_FILE", 
+          "Write output to OUTPUT_FILE"
+        ) do |output_file|
           @output_file_path = output_file
         end
 
-        parser.on("-H HEADER_PROMPT_FILE", "--header-prompt=HEADER_PROMPT_FILE", "Load header prompt from HEADER_PROMPT_FILE") do |header_prompt_file|
+        parser.on(
+          "-H HEADER_PROMPT_FILE", 
+          "--header-prompt=HEADER_PROMPT_FILE", 
+          "Load header prompt from HEADER_PROMPT_FILE"
+        ) do |header_prompt_file|
           @header_prompt_file_path = header_prompt_file
         end
 
-        parser.on("-F FOOTER_PROMPT_FILE", "--footer-prompt=FOOTER_PROMPT_FILE", "Load footer prompt from FOOTER_PROMPT_FILE") do |footer_prompt_file|
+        parser.on(
+          "-F FOOTER_PROMPT_FILE", 
+          "--footer-prompt=FOOTER_PROMPT_FILE", 
+          "Load footer prompt from FOOTER_PROMPT_FILE"
+        ) do |footer_prompt_file|
           @footer_prompt_file_path = footer_prompt_file
         end
 
@@ -50,12 +70,12 @@ module CodePreloader
         end
       end
 
-      validate_arguments
+      validate
     end
 
-    private def validate_arguments
+    private def validate
       abort("Missing repository path.") if @repository_path_list.empty?
-      abort("Missing repository path.") if 
+ 
       STDERR.puts("Output file path not specified (using STDOUT)") if @output_file_path.nil? || @output_file_path.try(&.empty?)
     end
 
@@ -78,7 +98,7 @@ module CodePreloader
       @header_prompt_file_path = root.header_prompt_file_path || @header_prompt_file_path
       @footer_prompt_file_path = root.footer_prompt_file_path || @footer_prompt_file_path
 
-    rescue ex
+    rescue ex : Exception
       STDERR.puts "Failed to load config file: #{ex.message}"
       exit(1)
     end

src/filelist.cr

@@ -0,0 +1,94 @@
+
+require "walk"
+
+module CodePreloader
+
+  # Manage a list of files
+  class FileList 
+
+    alias Filter = String -> Bool
+
+    class NotADirectory < Exception
+      def initialize(path)
+        super(path.to_s)
+      end
+    end
+
+    @sources : Array(String)
+    @filters_in : Array(Filter)
+    @filters_out : Array(Filter)
+
+    def initialize(dirs = [] of String)
+      @sources = [] of String
+      @filters_in = [] of Filter
+      @filters_out = [] of Filter
+      dirs.each { |dir| self.add(dir) }
+    end
+
+    def add(dirs : Array(String))
+      dirs.each { |dir| add(dir) }
+    end
+
+    def add(dir : String)
+      raise NotADirectory.new(dir) if !File.exists? dir
+
+      @sources << dir
+    end
+
+    def select(&filter : Filter)
+      @filters_in << filter
+    end
+
+    def reject(&filter : Filter)
+      @filters_out << filter
+    end
+
+    def each(&block)
+      # ensure we display files only once
+      seen = Set(String).new
+
+      # walk each source
+      @sources.each do |dir|
+        walker = Walk::Down.new(dir)
+
+        walker = walker.filter do |path|
+          is_dir = File.directory? path
+          keep = true
+          must_select = false
+          must_reject = false
+
+          @filters_in.each do |filter_in|
+            must_select = must_select || filter_in.call(path.to_s)
+          end
+          keep = keep && must_select if @filters_in.any?
+          keep = keep || is_dir
+
+          @filters_out.each do |filter_out|
+            must_reject = must_reject || filter_out.call(path.to_s)
+          end
+          keep = keep && !must_reject if @filters_out.any?
+
+          keep
+        end
+
+        walker.each do |path|
+          next if File.directory? path
+
+          path = File.realpath(path) if File.symlink? path
+          next if seen.includes? path.to_s
+
+          seen << path.to_s
+          yield path.to_s
+        end
+      end
+    end
+
+    def to_a()
+      files = [] of String
+      self.each do |path|
+        files << path.to_s
+      end
+      files
+    end
+  end
+end