2.0.0

Version 2.0.0 - Enhanced Cookie Handling and Stability Improvements

This major release fundamentally changes how cookies are managed and addresses several critical issues with redirect handling and session persistence. The TLS client now handles cookies natively through the shared library rather than Python-side implementations.

Major Changes

🍪 Cookie Handling Improvements

• Native cookie management through shared library fixes redirect cookie persistence (#120)
• Added session.get_cookies() and session.add_cookies() methods
• Automatic domain/path validation using standard library rules
• Full support for secure/httpOnly flags and expiration
• Cookies now persist correctly through redirect chains

⚠ Breaking Changes

• Dropped requests dependency and related structures
• session.cookies is now read-only (CookieJar interface)
• CaseInsensitiveDict replaced with nocasedict implementation

🛠 Known Limitations

• Manual cookie modification via session.cookies is not supported (issues #25, #73)
• Cookie deletion/clearing methods are not yet implemented in the shared library
• Session cookies should be managed through add_cookies()/get_cookies() methods

🚀 Other Improvements

• Added proper type hints and request options typing

Upgrade Instructions

1. Replace any session.cookies manipulation with explicit cookie methods:

   # Get cookies for domain
   cookies = await session.get_cookies("https://example.com")
   
   # Add new cookies
   await session.add_cookies([
       {"name": "test", "value": "123", "domain": "example.com"}
   ], "https://example.com")

2. Remove any requests-specific code from your implementation

1.2

Release 1.2 - "We Copied Requests So You Don't Have To" 😴

I'm too sleepy to write proper release notes, but here’s the gist:

• We stopped reinventing the wheel 🛞
  Instead of copying requests internals (like CaseInsensitiveDict), we now just use them directly. Less code, fewer bugs, more import requests. Though slightly changed RequestsCookieJar class is still present in cookies.py...

• Tests exist now! 🎯
  Turns out, writing tests before midnight is a good idea. Who knew?

---

Breaking Changes?
Only if you were digging into our internal structures (which you shouldn’t have been doing anyway).

Why Upgrade?

• Fewer bugs 🐛
• More sleep for the maintainer 😴 (me)

1.1.1

This release introduces significant improvements in packaging, dependency management, and code organization while maintaining backward compatibility with the original tls-client implementation.

📦 Packaging Modernization

• ✅ Migrated to modern pyproject.toml-based packaging (PEP 621 compliant)
• 🧹 Removed legacy files: setup.py, __version__.py, requirements.txt
• 🛠️ Enhanced build process with setuptools>=65.0 and wheel
• 📁 Improved package data handling with explicit binary file inclusion

🔄 Dynamic Library Updates

• 🔄 New dynamic shared library fetcher that automatically:
  • 📡 Fetches latest 1.x releases from bogdanfinn/tls-client
  • 🗃️ Renames binaries consistently across platforms
  • 🔄 Keeps dependencies up-to-date without manual intervention

🧱 Code Improvements

• 📁 Restructured session module:
  • Moved to session/async_session.py for better organization
  • Added proper module initialization
• 🛠️ Fixed critical typo in exception class name: TLSClientExeption → TLSClientException
• 🧼 Cleaned up imports and type hints
• 🔄 Renamed settings.py → types.py for clarity

1.1

This release introduces a critical fix for handling binary responses, such as images or PDFs, ensuring they are properly decoded and accessible through response.content and response.text.

---

🔧 Summary of Changes

The Go-based TLS client now always returns byte-level responses (when isByteResponse=true) in base64 format with MIME type prefixes like:

data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAASw...

This allows the Python wrapper to:

• Safely decode base64 data
• Preserve binary content integrity
• Avoid UTF-8 decoding errors

The Python side was updated to:

• Handle malformed or missing base64 prefixes gracefully
• Decode response bodies safely

---

📦 What’s Fixed

1. Binary Response Corruption

Previously, binary data was treated as UTF-8 text, leading to corruption or exceptions when trying to read files like images.

Now, thanks to the Go-side change and updated Python logic:

response = await session.get("https://example.com/image.png")
with open("image.png", "wb") as f:
    f.write(response.content)  # This now works reliably

2. Text Decoding Improvements

Used errors='replace' instead of 'ignore' to avoid silent data loss:

response.text = decoded.decode(errors='replace')

---

💡 Technical Highlights

Component	Change
Python (response.py)	Safely decodes base64 responses and handles malformed ones
Python (sessions.py)	Sets "isByteResponse": True by default

---

🐛 Related Issues

This release resolves several long-standing issues related to binary responses:

• FlorianREGAZ/Python-Tls-Client#83
• FlorianREGAZ/Python-Tls-Client#119
• FlorianREGAZ/Python-Tls-Client#117
• FlorianREGAZ/Python-Tls-Client#83

---

🚀 Installation

pip install -U async_tls_client

---

📌 Contributors

• @3droj7 – PR
• @diprog – Small enhancements