Skip to content

CHANGELOG.md

Latest commit

 

History

History
238 lines (146 loc) · 10.6 KB

CHANGELOG.md

File metadata and controls

238 lines (146 loc) · 10.6 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.7.3] - 2026-04-07

Added

  • Smart Data File Handling: Data files (.csv, .json, .jsonl) now support nuanced inclusion strategies:
    • --truncate-data-files / --no-truncate-data-files: master toggle (config: truncate_data_files, default true).
    • --data-file-threshold-lines <num>: only truncate data files exceeding this many lines (config: data_file_threshold_lines, default 50). Small datasets are now included in full.
    • --data-file-max-per-dir <num>: when a directory contains more same-extension data files than this, only the first N are kept and the rest are summarised in an Omitted Data File Groups section. Set to 0 to disable. (config: data_file_max_per_dir, default 5.) This prevents per-record dataset layouts (50+ near-identical JSON files) from drowning out the rest of the prompt.

Fixed

  • CXML output: file content containing the literal string </document_content> no longer breaks the surrounding XML envelope; the closing tag is now neutralized inside content blocks.
  • Directory pruning: skipped directories (e.g. node_modules, .git) are now pruned at enqueue time during file discovery, avoiding unnecessary work in large monorepos.
  • GitHub fetcher: errors and rate-limits are now surfaced as warnings instead of being silently swallowed. Added support for GITHUB_TOKEN / GH_TOKEN environment variables to raise the API rate limit.
  • PDF extraction: pages are now joined with double newlines, so the last word of one page no longer runs into the first word of the next.
  • YouTube transcripts: now compatible with both the legacy YouTubeTranscriptApi.get_transcript() and the newer instance-based .fetch() API.
  • Snapshot diff: unchanged files are no longer re-hashed on every diff. The current-state index now reuses the snapshot's sha256 when both size and mtime match, drastically speeding up diffs on large repos.
  • Token-count fallback: snapshot diffs now report 0 tokens when tiktoken is unavailable, matching the rest of the tool instead of returning a misleading whitespace approximation.

[0.7.2] - 2025-08-11

Added

  • File Size Limits: Introduced flags to control the maximum size of individual files included in the prompt:
    • --file-max-lines <num>: Truncates file content after num lines.
    • --file-max-bytes <num>: Truncates file content after num bytes.
    • This is applied to regular files before compression is attempted, ensuring massive files don't overwhelm the context window. Files truncated by these limits are marked with a note in the final prompt.

[0.7.1] - 2025-08-08

Added

  • Snapshot Command: codetoprompt snapshot <PATH> --output <snapshot.json> to save a JSON snapshot of a local project (no git required).
    • --output is required for this command.
  • Diff Command: codetoprompt diff <PATH> --snapshot <snapshot.json> to show a unified diff between the current project state and a previous snapshot.
    • By default, the diff is copied to the clipboard when no --output is provided (if pyperclip is available; on Linux ensure xclip or wl-copy).
    • Providing --output <file> writes the diff to the given path instead of copying to the clipboard.
  • Configurable Snapshot Thresholds: Set via codetoprompt config:
    • snapshot_max_bytes (default 3 MB)
    • snapshot_max_lines (default 100,000)

Notes

  • Diff supports --use-snapshot-filters to reuse include/exclude and .gitignore behavior from the snapshot; you can override with flags.

[0.7.0] - 2025-07-31

Improved

  • Documentation: Overhauled README.md for improved clarity, professionalism, and visual appeal.
    • Replaced the introductory slogan with a more descriptive summary of the tool's purpose.
    • Added a prominent "Key Features" section for a quick overview of capabilities.
    • Updated the example outputs for the analyse command and the interactive mode (-i) to better reflect current functionality and showcase features.
    • Added more badges for CI status, Python versions, and license.
    • Restructured the feature sections and command-line reference for better readability.

[0.6.9] - 2025-07-19

Fixed

  • File Filtering: The --include and --exclude flags now correctly handle recursive folder patterns, behaving like .gitignore (e.g., --exclude "my_folder" or --exclude "my_folder/**" will exclude all contents).

[0.6.8] - 2025-07-15

Fixed

  • Single File Input: The CLI no longer errors when given a path to a single file (e.g., ctp my_file.py). It now correctly processes the specified file.
  • Jupyter Notebook Processing: Resolved an issue where processing .ipynb files could cause the tool to hang by including a necessary dependency (ipython).

[0.6.7] - 2025-07-12

Added

  • Jupyter Notebook Support: Automatically processes .ipynb files by extracting Python code from all cells and including it in the prompt. This requires the nbformat and nbconvert packages.

Fixed

  • Tokenization Errors: Fixed a critical bug where the tool would crash if a file's content included text that matched a special tiktoken token (e.g., <|endoftext|>). The tool now safely encodes such text.

[0.6.6] - 2025-07-11

Added

  • JavaScript Support in Web Scarping: Added Javascript Support that will help to scrape JS Enabled Website.

[0.6.5] - 2025-07-07

Added

  • Remote URL Processing: codetoprompt can now process content directly from the web, in addition to local directories.

    • GitHub Repositories: Pass a GitHub URL to get a prompt of the entire codebase.
    • Web Pages: Fetch and extract the main text from any website or documentation page.
    • YouTube Videos: Automatically get the full transcript from a video URL.
    • ArXiv Papers & PDFs: Extract text from ArXiv abstract pages or direct PDF links.

  • Documentation: Updated documentation.

[0.6.4] - 2025-07-05

Updated

  • Documentation: Updated documentation.

[0.6.3] - 2025-07-03

Updated

  • Lazy Loading Implemented for Interactive Mode: Implemented Lazy Loading for Interative Mode for Optimizing Interactive Mode on Huge Codebases.

[0.6.2] - 2025-07-01

Updated

  • Library Modularization: Refactored CLI functionality into modular components for improved readability and maintainability.

[0.6.1] - 2025-06-21

Added

  • Interactive Mode: Introduced a new --interactive (-i) flag that launches a file selection UI, allowing users to manually choose which files to include in the prompt. This provides fine-grained control over the context.

[0.6.0] - 2025-06-16

Updated

  • Documentation: Improved and updated documentation.

[0.5.9] - 2025-06-16

Fixed

  • Code Compressor Bugs: Resolved issues with unsupported languages in the compression process.

Updated

  • Dataset Detection: Now reads the first 5 lines of dataset files to generate more effective prompts while reducing token usage.

[0.5.7] - 2025-06-16

Updated

  • Code Compressor: Refactored into modular components.

Added

  • CLI Alias: You can now invoke the CLI using ctp as a shorthand for codetoprompt.
  • Version Flag: Added --version and -v flags to display the current version (e.g., codetoprompt --version or ctp -v).

[0.5.5] - 2025-06-15

Updated

  • Documentation: Added details for the compression feature and the --markdown (-m) and --cxml (-c) output modes.

[0.5.0] - 2025-06-15

Added

  • Code Compression: Introduced a --compress flag that uses tree-sitter to parse supported code files (Python, JS, TS, Java, C/C++, Rust) into a structural summary. This significantly reduces the token count by omitting implementation details while preserving classes, functions, and signatures.
  • Configurable Compression: The codetoprompt config wizard now supports setting compression as a default.
  • Fallback for Unsupported Files: Files in unsupported languages (e.g., Markdown) are included in full.

[0.4.2] - 2025-06-15

Added

  • Flexible Output Formats: Generate prompts in different formats using new CLI flags:
    • --markdown (-m): Outputs file contents in fenced Markdown code blocks with language hints.
    • --cxml (-c): Outputs file contents in a Claude-friendly XML structure.

Changed

  • Configuration Options: The interactive wizard (codetoprompt config) and config file now support setting a default output format (default, markdown, or cxml).

[0.4.1] - 2025-06-14

Added

  • Codebase Analysis: New codetoprompt analyze command for in-depth project statistics.
  • Enhanced Prompt Summary: Includes top files and file types by token count.

Fixed

  • Config Command Validation: The config command now prevents invalid flag usage.
  • UI Improvements: Panels now correctly size to their content.

[0.4.0] - 2025-06-13

Added

  • Interactive Configuration Wizard: Accessed via codetoprompt config.
  • Config Management: View and reset config with codetoprompt config --show and codetoprompt config --reset.
  • Persistent Configuration File: Stored at ~/.config/codetoprompt/config.toml.

Changed

  • CLI Default Behavior: Running codetoprompt with no arguments now shows the help menu.

[0.3.0] - 2025-06-12

Added

  • Project Structure Tree: A visual tree of the project structure is now included in the prompt.

Changed

  • CLI Output Improvements: Enhanced progress bar and more user feedback.

Fixed

  • Multiple bugs related to tree generation, depth handling, and relative paths.
  • Clipboard and output file handling issues.

[0.2.0] - 2025-06-12

Changed

  • File Processing: More robust handling of special tokens, problematic files, and file read errors.
  • CLI Enhancements: Added directory validation and improved argument parsing.

Fixed

  • Packaging & CI: Corrected pyproject.toml and GitHub Actions workflows.
  • Multiple bugs in token counting and file handling logic.

[0.1.0] - 2025-06-11

Added

  • Initial release.
  • Core functionality for converting a codebase into a single prompt.
  • File filtering, token counting, and a command-line interface.