jeeves-git-commit 3.0.0 → 3.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9dfc1f28648d8afaf7035592aff4dc5cc2bacda8bb2ae760caa04df038fc2359
-  data.tar.gz: 8d9f768afab1305f0ded8f14192af519f2d0c45cd9bcb7062c876ea113599488
+  metadata.gz: c677877fa585dc14ffd287c273bd0a79c82b4592c0f84c5e532c2171b576d061
+  data.tar.gz: fcfcd962506dc36115358ae8e5b20afe7273e1a8f531101d4b9c92442bc24b4c
 SHA512:
-  metadata.gz: f7bb5569cd7f8b204660bebc3c69f2cf25c96b120b3556a924d6f8bbfe16319a6acaa37550e98d9130966eefb420f1367e4942b1fe4325eb7c1970697c00f4dc
-  data.tar.gz: c1556ab3fa0eac2a91fd73579f4061a1968a6eae4134b875cff5f165ffb2ce0de5f9aa092106708909acfbccdc379f6d1d86d6c03222ba433ed99684370ac650
+  metadata.gz: 0773d6fa8d0504583fbca079ffcb0a418c0de29b09a8ae799b5ac42b0e917fbba3d7f06b2b1a78ab69ae1255a5f36aa118dccdcc06fab41edb10ad452a35aacf
+  data.tar.gz: 05a7466354d8ff50a486d4c3bf4487a68a746fdce0c64209133ccb2f52e30a85a4ddf5637f016393293d2fc40e3137c685ab915559f383f1bb6f017f03b964cb

data/README.md

@@ -95,6 +95,11 @@ Git and generation failures stop the command with a nonzero exit status.
 Jeeves checks that staged changes and HEAD still match what it reviewed before
 committing. See [architecture and Git behavior](docs/architecture.md).
 
+Local generation defaults to a 65,536-token context. Oversized diffs are
+automatically shortened, with a warning and omission notices for the model.
+Only the model input is shortened; Git commits the complete staged changes.
+See [input limits](docs/configuration.md#output-checks-and-large-diffs).
+
 ## Develop
 
 ```sh

data/lib/jeeves/diff.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+
+module Jeeves
+  class Diff
+    NOTICE = "[Jeeves: diff shortened. Excerpts may omit files, hunks, or parts of lines. " \
+             "Hunk ranges are from the original diff; omitted content is unknown.]\n"
+    OMITTED = "\n[... content omitted ...]\n"
+
+    def initialize(text)
+      @text = text
+    end
+
+    def fit(limit)
+      return @text if @text.bytesize <= limit
+
+      budget = limit - NOTICE.bytesize
+      if budget < OMITTED.bytesize + 8
+        raise Error, 'Too little room for a shortened diff. Shorten the prompt or increase ' \
+                     'GIT_COMMIT_LOCAL_CONTEXT and GIT_COMMIT_MAX_DIFF_BYTES.'
+      end
+      compact = without_context
+      return NOTICE + compact if compact.bytesize <= budget
+
+      files = compact.split(/(?=^diff --(?:git |cc |combined ))/)
+      NOTICE + fit_parts(files, budget) { |file, size| fit_file(file, size) }
+    end
+
+    private
+
+    def without_context
+      context_prefix = nil
+      @text.each_line.reject do |line|
+        context_prefix = nil if line.start_with?('diff --')
+        match = line.match(/\A(@{2,}) /)
+        context_prefix = ' ' * (match[1].length - 1) if match
+        context_prefix && line.start_with?(context_prefix)
+      end.join
+    end
+
+    def fit_file(file, budget)
+      return file if file.bytesize <= budget
+
+      header = header_for(file)
+      body = file.delete_prefix(header)
+      hunks = body.split(/(?=^@@)/)
+      header + fit_parts(hunks, budget - header.bytesize) do |hunk, size|
+        hunk_header = header_for(hunk)
+        hunk_header + excerpt(hunk.delete_prefix(hunk_header), size - hunk_header.bytesize)
+      end
+    end
+
+    def header_for(part)
+      lines = part.lines
+      return lines.first.to_s if lines.first&.start_with?('@@')
+
+      lines.take_while { |line| !line.start_with?('@@') }.join
+    end
+
+    def fit_parts(parts, budget)
+      return parts.join if parts.sum(&:bytesize) <= budget
+
+      # Keep file metadata and hunk headers before sharing space for changes.
+      minimums = parts.map { |part| [part.bytesize, header_for(part).bytesize + OMITTED.bytesize + 8].min }
+      return excerpt(parts.join, budget) if minimums.sum > budget
+
+      sizes = minimums.dup
+      remaining = budget - sizes.sum
+      order = parts.each_index.sort_by { |index| parts[index].bytesize - sizes[index] }
+      order.each_with_index do |index, position|
+        extra = [parts[index].bytesize - sizes[index], remaining / (order.length - position)].min
+        sizes[index] += extra
+        remaining -= extra
+      end
+      parts.each_with_index.map { |part, index| yield part, sizes[index] }.join
+    end
+
+    def excerpt(text, budget)
+      return text if text.bytesize <= budget
+
+      available = budget - OMITTED.bytesize
+      head_size = (available + 1) / 2
+      tail_size = available / 2
+      head = text.byteslice(0, head_size).scrub('')
+      tail = text.byteslice(-tail_size, tail_size).scrub('')
+      head = head[0..head.rindex("\n")] if head.include?("\n")
+      tail = tail[(tail.index("\n") + 1)..] if tail.include?("\n")
+      head + OMITTED + tail
+    end
+  end
+end

data/lib/jeeves/prompt.rb

@@ -10,7 +10,7 @@ module Jeeves
       @errors = errors
     end
 
-    def render(diff)
+    def render(diff, settings:)
       path = @repository && File.join(@repository, '.jeeves_prompt')
       unless path && File.file?(path)
         path = @global_path
@@ -19,7 +19,21 @@ module Jeeves
       template = File.read(path, encoding: 'UTF-8')
       raise Error, "Prompt must contain {{DIFF}}: #{path}" unless template.include?('{{DIFF}}')
 
-      template.gsub('{{DIFF}}') { diff }
+      limit = settings.max_diff_bytes
+      if settings.prompt_budget
+        overhead = template.gsub('{{DIFF}}', '').bytesize
+        available = (settings.prompt_budget - overhead) / template.scan('{{DIFF}}').length
+        if available < 1
+          raise Error, 'The prompt leaves no room for a diff. Shorten the prompt or increase GIT_COMMIT_LOCAL_CONTEXT.'
+        end
+        limit = [limit, available].min
+      end
+      shortened = Diff.new(diff).fit(limit)
+      if shortened != diff
+        @errors.puts "Warning: diff shortened from #{diff.bytesize} to #{shortened.bytesize} bytes. " \
+                     'Some content is omitted; the commit message may miss changes.'
+      end
+      template.gsub('{{DIFF}}') { shortened }
     end
 
     private

data/lib/jeeves/providers.rb

@@ -80,9 +80,7 @@ module Jeeves
       end
 
       def generate(prompt)
-        # One byte per token is a conservative bound without a model-specific tokenizer.
-        budget = @settings.context - 1000 - 256
-        if prompt.bytesize > budget
+        if prompt.bytesize > @settings.prompt_budget
           raise Error, 'Prompt exceeds the conservative local context budget. Split the changes or increase GIT_COMMIT_LOCAL_CONTEXT.'
         end
         body = {

data/lib/jeeves/settings.rb

@@ -16,7 +16,7 @@ module Jeeves
                                   end
       raise Error, 'The configured model must not be empty' if @model.strip.empty?
 
-      @context = integer('GIT_COMMIT_LOCAL_CONTEXT', 32_768) if @provider == 'ollama'
+      @context = integer('GIT_COMMIT_LOCAL_CONTEXT', 65_536) if @provider == 'ollama'
       @max_diff_bytes = integer('GIT_COMMIT_MAX_DIFF_BYTES', 65_536)
       @message_format = env.fetch('GIT_COMMIT_MESSAGE_FORMAT', 'conventional')
       return if %w[conventional plain].include?(@message_format)
@@ -27,9 +27,11 @@ module Jeeves
     def check_diff!(diff)
       raise Error, 'Diff must be valid UTF-8 text.' unless diff.valid_encoding?
       raise Error, 'No diff content provided.' if diff.strip.empty?
-      return if diff.bytesize <= max_diff_bytes
+    end
 
-      raise Error, "Diff is too large (#{diff.bytesize} bytes; limit #{max_diff_bytes}). Split the changes or increase GIT_COMMIT_MAX_DIFF_BYTES."
+    def prompt_budget
+      # One byte per token, with room for output and chat framing.
+      context - 1000 - 256 if context
     end
 
     private

data/lib/jeeves/version.rb

@@ -1,3 +1,3 @@
 module Jeeves
-  VERSION = '3.0.0'
+  VERSION = '3.1.0'
 end

data/lib/jeeves.rb

@@ -10,6 +10,7 @@ module Jeeves
 end
 
 require 'jeeves/settings'
+require 'jeeves/diff'
 require 'jeeves/prompt'
 require 'jeeves/message'
 require 'jeeves/providers'
@@ -38,7 +39,7 @@ module Jeeves
       provider = settings.provider == 'ollama' ? Providers::Ollama.new(settings) : Providers::OpenRouter.new(settings)
       piped = !@input.tty? && !@input.closed?
       if piped
-        diff = (@input.read(settings.max_diff_bytes + 1) || +'').force_encoding(Encoding::UTF_8)
+        diff = (@input.read || +'').force_encoding(Encoding::UTF_8)
       else
         raise Error, 'Not inside a Git working tree.' unless @git.root
 
@@ -47,7 +48,8 @@ module Jeeves
         diff = options[:all] && options[:dry_run] ? @git.preview_all : @git.diff
       end
       settings.check_diff!(diff)
-      prompt = Prompt.new(home: @env.fetch('HOME') { Dir.home }, repository: @git.root, errors: @errors).render(diff)
+      prompt = Prompt.new(home: @env.fetch('HOME') { Dir.home }, repository: @git.root, errors: @errors)
+                     .render(diff, settings: settings)
       @errors.puts "Using #{settings.provider}: #{settings.model}" unless piped
       message = Message.prepare(provider.generate(prompt), format: settings.message_format)
       return show(message) if piped || options[:dry_run]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jeeves-git-commit
 version: !ruby/object:Gem::Version
-  version: 3.0.0
+  version: 3.1.0
 platform: ruby
 authors:
 - Justin Bishop
@@ -79,6 +79,7 @@ files:
 - bin/jeeves
 - config/prompt
 - lib/jeeves.rb
+- lib/jeeves/diff.rb
 - lib/jeeves/git_repository.rb
 - lib/jeeves/message.rb
 - lib/jeeves/prompt.rb