…sses

…dia type detection

… functions

… base64 data

Replace outdated gemini-1.5-flash-latest with gemini-2.5-flash in multimodal content tests to use the latest available model version.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

This workflow automatically creates releases every 2 weeks if:
- All tests pass (linting, type checking, core tests)
- There are changes since the last release
- Final verification tests pass after version bump

Features:
- Smart version bumping based on commit messages
- Comprehensive test coverage before release
- Auto-generated changelog from commit history
- Manual trigger option for testing
- Detailed logging and failure notifications
- Skips release if no meaningful changes

The workflow runs every Monday at 9 AM UTC and only proceeds
if all safety checks pass.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

Fixed boolean default value and secrets access pattern in workflow configuration.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

- Added dry_run input for testing workflow without pushing
- Made LLM tests non-blocking with better error handling
- Added dry run summary output for testing
- Improved robustness of test execution

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

## 🚀 Automated Bi-weekly Releases

This PR adds a GitHub Actions workflow that automatically creates
releases every 2 weeks with comprehensive safety checks.

## ✨ Features

### 🛡️ Safety First
- **All tests must pass** before any release (linting, type checking,
core tests)
- **Smart version bumping** based on conventional commit messages
- **Final verification** tests run after version bump
- **Skips empty releases** - won't create releases with no meaningful
changes

### 🔧 Smart Automation
- **Cron schedule**: Every Monday at 9 AM UTC (bi-weekly)
- **Manual trigger**: Can be run manually for testing
- **Dry run mode**: Test the workflow without actually pushing changes
- **Non-blocking LLM tests**: Runs provider tests if API keys available

### 📊 Version Management
- **Patch bump**: For bug fixes and chores (`fix:`, `chore:`)
- **Minor bump**: For new features (`feat:`) or breaking changes
(`BREAKING`)
- **Auto-generated changelog**: Based on commit history since last
release

### 🔍 Testing Strategy
- Core tests (no API calls required)
- Linting with Ruff
- Type checking with PyRight  
- Optional LLM provider tests (if secrets available)
- Post-bump verification tests

## 🎯 Workflow Logic

```mermaid
graph TD
    A[Scheduled Trigger] --> B[Install Dependencies]
    B --> C[Run Linting]
    C --> D[Run Type Checking] 
    D --> E[Run Core Tests]
    E --> F[Run LLM Tests*]
    F --> G[Check for Changes]
    G --> H{Has Changes?}
    H -->|No| I[Skip Release]
    H -->|Yes| J[Determine Version Bump]
    J --> K[Update Version]
    K --> L[Update Lockfile]
    L --> M[Final Test Run]
    M --> N[Generate Changelog]
    N --> O[Create Commit & Tag]
    O --> P{Dry Run?}
    P -->|Yes| Q[Show Summary]
    P -->|No| R[Push to GitHub]
    R --> S[Create GitHub Release]
```

## 🧪 Testing

You can test this workflow with:

```bash
# Dry run (no changes pushed)
gh workflow run scheduled-release.yml -f dry_run=true

# Skip LLM tests for faster testing  
gh workflow run scheduled-release.yml -f skip_tests=true -f dry_run=true
```

## 📋 Checklist

- [x] Comprehensive test coverage
- [x] Smart version bumping logic
- [x] Error handling and logging
- [x] Dry run mode for testing
- [x] Auto-generated changelogs
- [x] Non-blocking LLM tests
- [x] Skip logic for empty releases
- [x] Detailed workflow summaries

## 🔗 Configuration

The workflow will need these repository secrets for full functionality:
- `OPENAI_API_KEY` (optional - for LLM tests)
- `ANTHROPIC_API_KEY` (optional - for LLM tests)
- `PYPI_TOKEN` (optional - for automatic PyPI publishing)

## 📈 Benefits

1. **Consistent releases** - No more manual version management
2. **Quality assurance** - All tests must pass before release
3. **Zero surprises** - Only releases when there are actual changes
4. **Full transparency** - Detailed logs and summaries
5. **Easy testing** - Dry run mode for validation

---

🤖 Generated with [Claude Code](https://claude.ai/code)
<!-- ELLIPSIS_HIDDEN -->

----

> [!IMPORTANT]
> Adds a GitHub Actions workflow for automated bi-weekly releases with
comprehensive safety checks and smart version management.
> 
>   - **Workflow**:
> - Adds `scheduled-release.yml` for automated bi-weekly releases via
GitHub Actions.
>     - Triggers every 2 weeks on Monday at 9 AM UTC or manually.
>     - Includes dry run mode and optional LLM tests.
>   - **Safety Checks**:
> - Requires all tests (linting, type checking, core tests) to pass
before release.
>     - Skips release if no meaningful changes are detected.
>   - **Version Management**:
> - Smart version bumping based on commit messages (`patch` for `fix:`,
`chore:`, `minor` for `feat:`, `BREAKING`).
>     - Auto-generates changelog from commit history.
>   - **Testing**:
>     - Core tests run without API calls.
>     - Optional LLM tests if API keys are available.
>     - Final verification tests after version bump.
>   - **Misc**:
>     - Generates detailed workflow summaries and logs.
>     - Provides error notifications on failure.
> 
> <sup>This description was created by </sup>[<img alt="Ellipsis"
src="https://img.shields.io/badge/Ellipsis-blue?color=175173">](https://www.ellipsis.dev?ref=567-labs%2Finstructor&utm_source=github&utm_medium=referral)<sup>
for da24f61. You can
[customize](https://app.ellipsis.dev/567-labs/settings/summaries) this
summary. It will automatically update as commits are pushed.</sup>

<!-- ELLIPSIS_HIDDEN -->

…andardized errors

## Improvements Made:
- **Configurable timeout**: All `from_gs_url()` methods now accept optional `timeout` parameter (default: 30s)
- **Standardized error messages**: Unified error format across Image, Audio, and PDF classes
- **Enhanced documentation**: Added comprehensive GCS examples and usage notes for all media types

## Changes:
- `Image.from_gs_url()`: Added timeout parameter, standardized error message
- `Audio.from_gs_url()`: Added timeout parameter (error message was already correct)
- `PDF.from_gs_url()`: Added timeout parameter (error message was already correct)
- Updated multimodal.md with GCS examples for all three media types
- Added method tables showing GCS support across providers

## Testing:
- All existing tests pass (48/49, 1 skipped)
- GCS functionality verified for all media types
- Timeout parameter tested and working correctly

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

…1788)

## 🚀 **Enhanced Google Cloud Storage (GCS) Support for Multimodal
Classes**

This PR adds comprehensive Google Cloud Storage support to the
instructor library's multimodal functionality with significant
enhancements over the basic implementation.

### ✨ **Key Features**

#### 🔧 **Configurable Timeout Support**
- All `from_gs_url()` methods now accept optional `timeout` parameter
- Default: 30 seconds (configurable per request)
- **Usage**: `Image.from_gs_url("gs://bucket/image.jpg", timeout=60)`

#### 📝 **Standardized Error Messages**
- Unified error format across all media types
- Clear, user-friendly messages: "Failed to access GCS {type} (must be
publicly readable)"
- Consistent error handling patterns

#### 📚 **Complete Documentation**
- Added comprehensive GCS examples for **all three media types** (Image,
Audio, PDF)
- Updated method support tables showing GCS availability across
providers
- Added usage notes about public accessibility requirements
- Copy-paste ready examples for all providers

### 🎯 **What's Implemented**

#### **Image Class**
- ✅ `Image.from_gs_url(url, timeout=30)` - Load images from GCS
- ✅ `Image.from_url()` - Automatically handles GCS URLs
- ✅ `Image.autodetect()` - Recognizes GCS URLs
- ✅ Provider support: OpenAI, Anthropic, Google GenAI

#### **Audio Class**  
- ✅ `Audio.from_gs_url(url, timeout=30)` - Load audio from GCS
- ✅ `Audio.from_url()` - Automatically handles GCS URLs
- ✅ `Audio.autodetect()` - Recognizes GCS URLs
- ✅ Provider support: OpenAI, Google GenAI
- ❓ Anthropic: Correctly shows "not supported" (API limitation)

#### **PDF Class**
- ✅ `PDF.from_gs_url(url, timeout=30)` - Load PDFs from GCS  
- ✅ `PDF.from_url()` - Automatically handles GCS URLs
- ✅ `PDF.autodetect()` - Recognizes GCS URLs
- ✅ Provider support: OpenAI, Anthropic, Google GenAI, Mistral

#### **Enhanced Message Processing**
- ✅ `autodetect_images=True` now supports all media types including GCS
URLs
- ✅ New `autodetect_media()` function for unified media detection
- ✅ PDF support added to message conversion pipeline

### 📊 **Testing Status**

- ✅ **48/49 tests passing** (1 unrelated skip)
- ✅ **All GCS functionality verified** for Image, Audio, PDF
- ✅ **Timeout parameter tested** and working correctly  
- ✅ **Pre-commit hooks passed** (linting & formatting)
- ✅ **Backward compatibility maintained**

### 🔐 **Security & Reliability**

- **Public Access Only**: Only supports publicly accessible GCS objects
- **Proper URL Validation**: Validates GCS URL format before processing
- **Timeout Protection**: Configurable timeouts prevent hanging requests
- **Error Handling**: Comprehensive error messages for debugging

### 📖 **Usage Examples**

#### **Basic GCS Usage**
```python
from instructor.processing.multimodal import Image, Audio, PDF

# Load from GCS with default timeout (30s)
image = Image.from_gs_url("gs://my-bucket/image.jpg")
audio = Audio.from_gs_url("gs://my-bucket/audio.wav")  
pdf = PDF.from_gs_url("gs://my-bucket/document.pdf")

# Load with custom timeout
image = Image.from_gs_url("gs://my-bucket/large-image.jpg", timeout=60)
```

#### **Automatic Detection**
```python
import instructor

client = instructor.from_provider("openai/gpt-4.1-mini")

# Automatic GCS detection with autodetect_images=True
response = client.chat.completions.create(
    response_model=MyModel,
    autodetect_images=True,  # Now detects GCS URLs automatically
    messages=[{
        "role": "user", 
        "content": ["Analyze this:", "gs://my-bucket/document.pdf"]
    }]
)
```

### 🔄 **Relationship to Existing Work**

This PR enhances and supersedes the basic GCS implementation in PR #1763
by adding:
- Configurable timeouts (addresses hanging request concerns)
- Standardized error messages (eliminates inconsistencies)  
- Complete documentation coverage (all media types documented)
- Enhanced testing coverage

### 🛠 **Technical Details**

- **URL Conversion**: `gs://bucket/path` →
`https://storage.googleapis.com/bucket/path`
- **Content Encoding**: All content is base64-encoded for API
compatibility
- **MIME Type Detection**: Proper Content-Type header validation
- **Provider Compatibility**: Works seamlessly across all supported
providers

### 📋 **Migration Path**

This implementation is fully backward compatible. Existing code will
continue to work unchanged, and users can gradually adopt the new
timeout parameters and enhanced features.

---

**Ready for review!** This implementation provides production-ready GCS
support with comprehensive documentation, testing, and error handling.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>
<!-- ELLIPSIS_HIDDEN -->

----

> [!IMPORTANT]
> Enhances Google Cloud Storage support for Image, Audio, and PDF
classes with new methods, error handling, and documentation updates in
the instructor library.
> 
>   - **Behavior**:
> - Adds `from_gs_url()` method to `Image`, `Audio`, and `PDF` classes
in `multimodal.py` for loading from GCS with optional `timeout`
parameter (default 30s).
> - Enhances `autodetect()` and `autodetect_safely()` methods to
recognize GCS URLs.
> - Implements `autodetect_media()` function for unified media
detection.
>   - **Error Handling**:
> - Standardizes error messages for GCS access failures across media
types.
>     - Validates GCS URL format and ensures public accessibility.
>   - **Documentation**:
> - Updates `multimodal.md` with GCS usage examples for Image, Audio,
and PDF classes.
>     - Adds notes on public accessibility requirements for GCS URLs.
>   - **Testing**:
> - Adds tests for `from_gs_url()` methods and `autodetect_media()` in
`test_multimodal.py`.
>     - Verifies GCS functionality and timeout parameter handling.
>   - **Misc**:
>     - Maintains backward compatibility with existing code.
> - Updates method support tables in documentation to reflect GCS
support.
> 
> <sup>This description was created by </sup>[<img alt="Ellipsis"
src="https://img.shields.io/badge/Ellipsis-blue?color=175173">](https://www.ellipsis.dev?ref=567-labs%2Finstructor&utm_source=github&utm_medium=referral)<sup>
for d3f8046. You can
[customize](https://app.ellipsis.dev/567-labs/settings/summaries) this
summary. It will automatically update as commits are pushed.</sup>

<!-- ELLIPSIS_HIDDEN -->

…tions imports

This commit addresses a breaking change introduced in v1.11.0 where the
backward compatibility module `instructor/exceptions.py` was removed,
causing imports like `from instructor.exceptions import InstructorRetryException`
to fail.

Changes:
- Restore instructor/exceptions.py with proper re-exports from instructor.core.exceptions
- Add deprecation warnings to guide users to the new import path
- Bump version to 1.11.2 (patch version for backwards compatibility fix)

The deprecated import path will continue to work but will show a
deprecation warning directing users to import from instructor.core instead.

Fixes: Breaking change from v1.11.0 removal of backwards compatibility files

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

…tions imports (#1789)

## Summary
- Restore backwards compatibility for `instructor.exceptions` imports
that broke in v1.11.0
- Add deprecation warnings to guide users to new import path
`instructor.core`
- Bump version to 1.11.2 (patch version for backwards compatibility fix)

## Problem
Version 1.11.0 introduced a breaking change by removing the backwards
compatibility module `instructor/exceptions.py`. This caused imports
like `from instructor.exceptions import InstructorRetryException` to
fail, breaking downstream applications without warning.

## Solution
- Recreated `instructor/exceptions.py` with proper re-exports from
`instructor.core.exceptions`
- Added deprecation warnings to guide users to the recommended import
path
- Used proper semantic versioning (patch bump) for a backwards
compatibility fix

## Test plan
- [x] Test that old import path works: `from instructor.exceptions
import InstructorRetryException`
- [x] Test that new import path still works: `from instructor.core
import InstructorRetryException`
- [x] Verify deprecation warning is shown when using old import path
- [x] Run linting and type checking
- [x] Version bump to 1.11.2

🤖 Generated with [Claude Code](https://claude.ai/code)
<!-- ELLIPSIS_HIDDEN -->

----

> [!IMPORTANT]
> Restores backwards compatibility for `instructor.exceptions` imports
with deprecation warnings and bumps version to 1.11.2.
> 
>   - **Behavior**:
> - Reintroduces `instructor/exceptions.py` to restore backwards
compatibility for imports like `from instructor.exceptions import
InstructorRetryException`.
> - Adds deprecation warning in `instructor/exceptions.py` to guide
users to `instructor.core`.
>   - **Versioning**:
> - Bumps version to 1.11.2 in `pyproject.toml` and `uv.lock` for
backwards compatibility fix.
> 
> <sup>This description was created by </sup>[<img alt="Ellipsis"
src="https://img.shields.io/badge/Ellipsis-blue?color=175173">](https://www.ellipsis.dev?ref=567-labs%2Finstructor&utm_source=github&utm_medium=referral)<sup>
for ff9d88f. You can
[customize](https://app.ellipsis.dev/567-labs/settings/summaries) this
summary. It will automatically update as commits are pushed.</sup>

<!-- ELLIPSIS_HIDDEN -->