Merge pull request #75 from square/tsutton/feat-sparse

Add --sparse-paths for sparse checkout feature

Author: timsutton

.rubocop.yml

@@ -24,3 +24,9 @@ Metrics/MethodLength:
 Metrics/BlockLength:
   Exclude:
     - 'spec/**/*'
+
+Metrics/CyclomaticComplexity:
+  Max: 15
+
+Metrics/PerceivedComplexity:
+  Max: 15

README.md

@@ -14,13 +14,13 @@ Why fastclone?
 Doing lots of repeated checkouts on a specific machine?
 
 | Repository | 1st Fastclone | 2nd Fastclone | git clone | cp -R |
-| -----------|---------------|---------------|-----------|-------|
-| angular.js |    8s         |     3s        | 6s        | 0.5s  |
-| bootstrap  |    26s        |     3s        | 11s       | 0.2s  |
-| gradle     |    25s        |     9s        | 19s       | 6.2s  |
-| linux      |    4m 53s     |     1m 6s     | 3m 51s    | 29s   |
-| react.js   |    18s        |     3s        | 8s        | 0.5s  |
-| tensorflow |    19s        |     4s        | 8s        | 1.5s  |
+| ---------- | ------------- | ------------- | --------- | ----- |
+| angular.js | 8s            | 3s            | 6s        | 0.5s  |
+| bootstrap  | 26s           | 3s            | 11s       | 0.2s  |
+| gradle     | 25s           | 9s            | 19s       | 6.2s  |
+| linux      | 4m 53s        | 1m 6s         | 3m 51s    | 29s   |
+| react.js   | 18s           | 3s            | 8s        | 0.5s  |
+| tensorflow | 19s           | 4s            | 8s        | 1.5s  |
 
 Above times captured using `time` without verbose mode.
 
@@ -51,6 +51,8 @@ Usage
         --lock-timeout N             Timeout in seconds to acquire a lock on any reference repo.
                 Default is 0 which waits indefinitely.
         --pre-clone-hook command     An optional command that should be invoked before cloning mirror repo
+        --sparse-paths PATHS         Comma-separated list of paths for sparse checkout.
+                                     Enables sparse checkout mode using git sparse-checkout.
 
 Change the default `REFERENCE_REPO_DIR` environment variable if necessary.
 
@@ -66,6 +68,11 @@ The hook is invoked with given arguments:
 1. path to the repo mirror location
 1. attempt number, 0-indexed
 
+Sparse checkout support
+-----------------------
+
+In passing `--sparse-paths`, git-fastclone will instead perform a sparse checkout, where the passed list of paths will be set up as patterns. This can be useful if you're interested only in a subset of paths in the repository.
+
 How to test?
 ------------
 Manual testing:

lib/git-fastclone.rb

@@ -85,7 +85,7 @@ class Runner
 
     attr_accessor :reference_dir, :prefetch_submodules, :reference_updated, :reference_mutex,
                   :options, :abs_clone_path, :using_local_repo, :verbose, :print_git_errors, :color,
-                  :flock_timeout_secs
+                  :flock_timeout_secs, :sparse_paths
 
     def initialize
       # Prefetch reference repos for submodules we've seen before
@@ -115,6 +115,8 @@ def initialize
       self.color = false
 
       self.flock_timeout_secs = 0
+
+      self.sparse_paths = nil
     end
 
     def run
@@ -172,6 +174,12 @@ def parse_options
                 'No-op when a file is missing') do |script_file|
           options[:pre_clone_hook] = script_file
         end
+
+        opts.on('--sparse-paths PATHS',
+                'Comma-separated list of paths for sparse checkout.',
+                'Enables sparse checkout mode using git sparse-checkout.') do |paths|
+          self.sparse_paths = paths.split(',').map(&:strip)
+        end
       end.parse!
     end
 
@@ -199,6 +207,16 @@ def parse_inputs
         raise msg
       end
 
+      # Validate that --branch is specified when using --sparse-paths
+      if sparse_paths && !options[:branch]
+        msg = "Error: --branch is required when using --sparse-paths\n" \
+              "Sparse checkouts need an explicit branch/revision to checkout.\n" \
+              'Usage: git-fastclone --sparse-paths <paths> --branch <branch> <url>'
+        raise msg.red if color
+
+        raise msg
+      end
+
       self.reference_dir = ENV['REFERENCE_REPO_DIR'] || DEFAULT_REFERENCE_REPO_DIR
       FileUtils.mkdir_p(reference_dir)
 
@@ -234,13 +252,23 @@ def clone(url, rev, src_dir, config)
         clear_clone_dest_if_needed(attempt_number, clone_dest)
 
         clone_commands = ['git', 'clone', verbose ? '--verbose' : '--quiet']
-        clone_commands << '--reference' << mirror.to_s << url.to_s << clone_dest
+        # For sparse checkouts, clone directly from the local mirror and skip the actual checkout process
+        # For normal clones, use --reference and clone from the remote URL
+        if sparse_paths
+          clone_commands.push('--no-checkout')
+          clone_commands << mirror.to_s << clone_dest
+        else
+          clone_commands << '--reference' << mirror.to_s << url.to_s << clone_dest
+        end
         clone_commands << '--config' << config.to_s unless config.nil?
         fail_on_error(*clone_commands, quiet: !verbose, print_on_failure: print_git_errors)
+
+        # Configure sparse checkout if enabled
+        perform_sparse_checkout(clone_dest, rev) if sparse_paths
       end
 
-      # Only checkout if we're changing branches to a non-default branch
-      if rev
+      # Only checkout if we're changing branches to a non-default branch (for non-sparse clones)
+      if !sparse_paths && rev
         fail_on_error('git', 'checkout', '--quiet', rev.to_s, quiet: !verbose,
                                                               print_on_failure: print_git_errors,
                                                               chdir: File.join(abs_clone_path, src_dir))
@@ -258,6 +286,22 @@ def clone(url, rev, src_dir, config)
       end
     end
 
+    def perform_sparse_checkout(clone_dest, rev)
+      puts 'Configuring sparse checkout...' if verbose
+
+      # Initialize sparse checkout with cone mode
+      fail_on_error('git', 'sparse-checkout', 'init', '--cone',
+                    quiet: !verbose, print_on_failure: print_git_errors, chdir: clone_dest)
+
+      # Set the sparse paths
+      fail_on_error('git', 'sparse-checkout', 'set', *sparse_paths,
+                    quiet: !verbose, print_on_failure: print_git_errors, chdir: clone_dest)
+
+      # Checkout the specified branch/revision
+      fail_on_error('git', 'checkout', '--quiet', rev.to_s,
+                    quiet: !verbose, print_on_failure: print_git_errors, chdir: clone_dest)
+    end
+
     def update_submodules(pwd, url)
       return unless File.exist?(File.join(abs_clone_path, pwd, '.gitmodules'))
 

lib/git-fastclone/version.rb

@@ -2,5 +2,5 @@
 
 # Version string for git-fastclone
 module GitFastCloneVersion
-  VERSION = '1.5.1'
+  VERSION = '1.6.0'
 end