v0.5.1

Code quality fixes for MCP module

• Hoist get_pk_constraint out of per-column loop in schema introspection
• Use sets for O(1) membership tests in collection remove operations
• Remove dead code in mcp_serve CLI command
• Align setup.py with pyproject.toml (python_requires, classifiers, extras)
• Test cleanup: shared fixtures, unused imports

v0.5.0: MCP Server, Major Simplification

What's New

MCP Server (ebk mcp-serve) — AI integration via Model Context Protocol with 3 tools:

• get_schema — introspect library database structure
• execute_sql — read-only SQL with 3-layer defense (regex, ?mode=ro, authorizer)
• update_books — batch update metadata, tags, authors, subjects, merge duplicates

What's Removed (~8,500 lines)

Module	Lines	Reason
ebk/ai/	~3,300	Replaced by MCP server
ebk/repl/	~2,300	Unused
ebk/vfs/	~2,800	Redundant with REST API
ebk/skills/	~180	Replaced by MCP server

Also removed: LLM config, ebk enrich command, VFS web router, stale references.

Install / Upgrade

pip install --upgrade ebk
# With MCP support:
pip install --upgrade ebk[mcp]

Full Changelog

v0.4.4...v0.5.0

v0.4.0 - Views DSL

Major Feature: Views DSL

A SICP-inspired domain-specific language for defining composable, non-destructive views over your library.

Core Concepts

• Primitives: all, none, filter, ids, view references
• Combination: union, intersect, difference
• Abstraction: Named views become new primitives
• Closure: Combining views yields a view

CLI Commands

# Create views
ebk view create favorites --favorite
ebk view create programming --subject programming
ebk view create top-rated --rating-gte 4

# List and show views
ebk view list
ebk view show favorites

# Curate membership
ebk view add my-collection 42
ebk view remove my-collection 99

# Override metadata (non-destructive)
ebk view set my-view 42 --title "Better Title"
ebk view set my-view 42 --description "My notes"

# Import/export YAML
ebk view export my-view -o my-view.yaml
ebk view import my-view.yaml
ebk view edit my-view  # Opens in $EDITOR

Export Integration

ebk export html ~/library ~/output.html --view favorites
ebk export json ~/library ~/backup.json --view programming
ebk export csv ~/library ~/books.csv --view top-rated
ebk export opds ~/library ~/catalog.xml --view recent

Built-in Views

• all - All books
• favorites - Books marked as favorites
• reading - Currently reading
• completed - Finished books
• unread - Not yet started
• recent - Recently added
• top-rated - Rating >= 4

YAML Definition Example

name: cs-fundamentals
description: Core CS texts, curated

select:
  intersect:
    - {filter: {subject: programming}}
    - {filter: {favorite: true}}

transform:
  override:
    42:
      title: "Structure and Interpretation of Computer Programs"
      description: "The wizard book. Start here."

order:
  by: title

v0.3.8

Changes

• Fix VFS commands: use config default library and fix type error
  • VFS commands now accept -L/--library option instead of positional argument
  • VFS commands use config default library when not specified
  • Fixed TypeError where LibraryShell received Library object instead of Path

v0.3.7: CLI List Filters

New Features

CLI List Filters

Added new filter options to ebk list command:

• --favorite / -f: Filter by favorite status
• --status / -S: Filter by reading status (reading, completed, unread)

Examples

# List favorites
ebk list --favorite
ebk list -f

# Filter by reading status
ebk list --status reading      # Currently reading
ebk list --status completed    # Finished books
ebk list --status unread       # Not started

# Combine filters
ebk list --favorite --status reading
ebk list -f -a "Knuth"         # Favorite books by Knuth

Installation

pip install --upgrade ebk

v0.3.6: Web Interface Fixes & Reading Status Filters

Bug Fixes

• Fixed JavaScript syntax error that prevented books from loading in web interface
  • Changed quote escaping from \' to ' HTML entities in onclick handlers

New Features

Reading Status Filter

• Added reading_status parameter to /api/books endpoint
• Filter books by: reading, completed, unread
• Sidebar filters now work for "Currently Reading" and "Completed"

Extended Stats API

• /api/stats now returns:
  • favorites_count: Number of favorited books
  • reading_count: Books currently being read
  • completed_count: Books marked as completed
• Sidebar counts now display accurate numbers

Tests

• Added 29 new tests for reading status filter, stats API, and JavaScript validation
• Total test suite: 989 tests passing

Installation

pip install --upgrade ebk

v0.3.5: Extended Import/Export Support

New Features

Import Types

• URL Import: Import ebooks directly from URLs (ebk import url)
• OPDS Import: Import from OPDS catalog feeds (ebk import opds)
• ISBN Import: Create book entries via Google Books/Open Library lookup (ebk import isbn)

Export Types

• OPDS Export: Export library to OPDS catalog format (ebk export opds)

Web Interface (ebk serve)

• 6-tab import modal: File, URL, Folder, Calibre, OPDS, ISBN
• All import types accessible from the web UI

CLI/REST API Parity

All import types are now available in both the CLI and REST API.

New Files

• ebk/calibre_import.py: Calibre library import helper
• ebk/exports/opds_export.py: OPDS catalog export module

Testing

• 54 comprehensive import tests
• 71 comprehensive export tests
• 125 new tests total

Installation

```bash
pip install --upgrade ebk
```

v0.3.4: OPDS Server, Reading Queue, Annotation Extraction

New Features

OPDS 1.2 Catalog Server

E-reader apps can now connect to your library! Works with Foliate, KOReader, Moon+ Reader, and more.

• Browse by author, subject, language
• Full search functionality with pagination
• Cover images and direct file downloads
• Start with: ebk serve
• Connect at: http://localhost:9000/opds/

Reading Queue

Manage your "up next" reading list:

ebk queue add 42              # Add book to queue
ebk queue add 42 -p 1         # Add to top of queue
ebk queue list                # Show queue
ebk queue next                # Show next book
ebk queue move 42 3           # Reorder
ebk queue remove 42           # Remove from queue
ebk queue clear               # Clear entire queue

Annotation Extraction

Extract highlights and notes from your PDFs:

ebk note extract 42           # Extract annotations from book files
ebk note export 42            # Export as markdown
ebk note export 42 -f json    # Export as JSON

Improvements

• Web server filters (favorite, rating) now work correctly with pagination
• API returns proper total counts for filtered queries
• All CLI commands support config default library path
• Better error handling throughout

Testing

• 91 new OPDS tests with 100% coverage
• 839 total tests passing

v0.3.3: Comprehensive test improvements

This release focuses on significantly improving test coverage and quality through TDD best practices.

Test Coverage Improvements

• search_parser.py: 44% → 98% coverage (+54 points)
• vfs/resolver.py: 55% → 99% coverage (+44 points)
• repl/shell.py: 56% → 64% coverage (+8 points)

New Test Suites

• test_search_parser.py (83 tests, 1,026 lines)

  • Field-specific searches (title:Python, author:Knuth)
  • Boolean operators (AND, OR, NOT)
  • Comparison operators (rating:>=4, rating:3-5)
  • SQL generation and SQL injection protection
  • Edge cases and error handling

• test_vfs_resolver.py (47 tests, 695 lines)

  • Path resolution (absolute, relative, special paths)
  • Parent directory navigation
  • Symlink resolution and tab completion
  • Edge cases and error handling

• test_repl.py (35 additional tests)

  • Pipeline execution and chaining
  • Output redirection
  • Text processing commands
  • Error handling and validation

Test Quality Improvements

Refactored 12 brittle tests to focus on behavior, not implementation:

• Removed __repr__ format tests → replaced with inspection tests
• Refactored hash tests to focus on deduplication behavior
• Refactored unique ID tests to focus on uniqueness guarantees
• Refactored sort name tests to focus on ordering behavior
• Replaced magic number assertions with calculated expectations
• Replaced exact title assertions with ordering logic

Test Design Principles

All new and refactored tests follow TDD best practices:

• Given-When-Then structure for clarity
• Test behavior, not implementation - resilient to refactoring
• Independent and isolated - can run in any order
• Clear failure messages - easy debugging
• Living documentation - tests explain what the system does

Results

• Total tests: 662 passing (no regressions)
• Coverage: 75%+ maintained on refactored modules
• Test quality: All tests survive refactoring
• New tests added: 165 comprehensive tests
• Tests refactored: 12 tests improved for resilience

Benefits

• Can now change hash algorithms without breaking tests
• Can modify ID formats without test failures
• Can improve sort algorithms freely
• Tests serve as executable specifications
• Fearless refactoring enabled

Installation

pip install ebk==0.3.3

Documentation

https://queelius.github.io/ebk/

v0.3.2: Project cleanup, test improvements, and new features

This release focuses on code organization, improved test coverage, and new features:

Project Cleanup

• Reorganized documentation to /docs/development/
• Moved utility scripts to /scripts/
• Archived legacy code and obsolete requirements files
• Updated .gitignore for better artifact management

Test Improvements

• Improved library_db.py coverage: 56% → 82%
• Improved vfs/nodes/books.py coverage: 42% → 82%
• Added 18 new tests across multiple modules
• Total: 497 passing tests

New Features

• REPL shell implementation with find, grep, text utilities
• Tag service with hierarchical navigation
• VFS nodes for authors, books, subjects, tags
• Similarity detection system
• Database migrations support

Bug Fixes

• Fixed MkDocs broken links
• Cleaned up strict mode build warnings

Documentation

• All design docs now in /docs/development/
• New implementation summaries for tags, piping, shell
• Clean mkdocs build in strict mode
• Documentation site: https://queelius.github.io/ebk/