Release v0.1.4: Add hardcoded patterns from json_repair Python library

This release integrates battle-tested cleanup patterns ported from the
Python json_repair library into JsonRemedy's Layer 3 processing pipeline.

## New Features

### Hardcoded Patterns Module (lib/json_remedy/layer3/hardcoded_patterns.ex)
- Smart quotes normalization: Converts ""curly"", «guillemets», ‹angles› to standard quotes
- Doubled quotes repair: Fixes ""value"" → "value" while preserving empty strings
- Number format normalization: Removes thousands separators (1,234,567 → 1234567)
- Unicode/hex escape sequences: Converts \uXXXX and \xXX (opt-in via config)

### Feature Flags
- :enable_hardcoded_patterns (default: true) - Master switch for all patterns
- :enable_escape_normalization (default: false) - Opt-in for escape sequences

### Examples & Documentation
- New examples/hardcoded_patterns_examples.exs with 8 comprehensive examples
- README.md updated with dedicated hardcoded patterns section
- Full attribution to source library: https://github.com/mangiucugna/json_repair

## Testing & Quality

- 47 new tests (100% pass rate)
- Total test suite: 499 tests, 0 failures
- Dialyzer: 0 type warnings
- Credo: 0 issues in new code
- Full UTF-8 international character support

## Technical Details

- Context-aware processing preserves commas in string values
- Smart quotes use Unicode byte sequences to avoid syntax conflicts
- Regex-based optimizations with minimal overhead
- Clean integration as Layer 3 pre-processing step

## Files Changed

Modified:
- .gitignore (added json_repair_python/)
- CHANGELOG.md (comprehensive v0.1.4 entry)
- README.md (version + hardcoded patterns section)
- lib/json_remedy/layer3/syntax_normalization.ex (integration)
- mix.exs (version 0.1.4)

New:
- lib/json_remedy/layer3/hardcoded_patterns.ex (250 lines)
- test/unit/layer3_hardcoded_patterns_test.exs (310 lines, 47 tests)
- examples/hardcoded_patterns_examples.exs (318 lines, 8 examples)

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

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

Author: 2 people

.gitignore

@@ -10,3 +10,4 @@ erl_crash.dump
 .elixir_ls/
 *.tar
 repomix*.xml
+json_repair_python/

CHANGELOG.md

@@ -7,6 +7,44 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.1.4] - 2025-10-07
+
+### Added
+- **Hardcoded Patterns Module**: New `JsonRemedy.Layer3.HardcodedPatterns` module with battle-tested cleanup patterns ported from Python's `json_repair` library
+  - **Smart quotes normalization**: Converts curly quotes (`""`), guillemets (`«»`), and angle quotes (`‹›`) to standard JSON quotes
+  - **Doubled quotes repair**: Fixes `""value""` → `"value"` while preserving empty strings `""`
+  - **Number format normalization**: Removes thousands separators from numbers: `1,234,567` → `1234567`
+  - **Unicode escape sequences**: Converts `\u263a` → `☺` (opt-in via `:enable_escape_normalization`)
+  - **Hex escape sequences**: Converts `\x41` → `A` (opt-in via `:enable_escape_normalization`)
+- **Comprehensive examples**: New `examples/hardcoded_patterns_examples.exs` with 8 detailed examples demonstrating:
+  - Smart quotes with international text (French, Japanese, German)
+  - Doubled quotes edge cases
+  - Number format cleaning while preserving string content
+  - Unicode/hex escape handling
+  - Combined patterns for real-world LLM output
+  - Full Layer 3 pipeline integration
+- **Feature flags**: Configurable pattern processing with safe defaults
+  - `:enable_hardcoded_patterns` (default: `true`)
+  - `:enable_escape_normalization` (default: `false` for safety)
+
+### Enhanced
+- **Layer 3 integration**: Hardcoded patterns run as pre-processing step before main syntax normalization
+- **Context-aware processing**: Number format normalization preserves commas in string values
+- **International support**: Full UTF-8 support with smart quotes from multiple languages
+- **Documentation**: README updated with dedicated hardcoded patterns section and attribution to source library
+
+### Technical Details
+- **Test coverage**: 47 new tests for hardcoded patterns (100% pass rate)
+- **Source attribution**: Patterns ported from [json_repair](https://github.com/mangiucugna/json_repair) by Stefano Baccianella
+- **Architecture**: Cleanly organized as Layer 3 subsection with proper separation of concerns
+- **Type safety**: Full Dialyzer compliance with zero warnings
+- **Performance**: Regex-based optimizations with minimal overhead
+
+### Performance
+- **Test suite**: 499 total tests, 0 failures (added 47 new tests)
+- **Zero regressions**: All existing functionality preserved
+- **Efficient processing**: Smart quotes and number normalization use optimized regex patterns
+
 ## [0.1.3] - 2025-07-05
 
 ### Fixed
@@ -117,6 +155,9 @@ This is a **100% rewrite** - all previous code has been replaced with the new la
 - Minimal memory overhead (< 8KB for repairs)
 - All operations pass performance thresholds
 
-[Unreleased]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.1...HEAD
+[Unreleased]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.4...HEAD
+[0.1.4]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.3...v0.1.4
+[0.1.3]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.2...v0.1.3
+[0.1.2]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.1...v0.1.2
 [0.1.1]: https://github.com/nshkrdotcom/json_remedy/compare/v0.1.0...v0.1.1
 [0.1.0]: https://github.com/nshkrdotcom/json_remedy/releases/tag/v0.1.0

README.md

@@ -79,6 +79,17 @@ Standard JSON parsers fail completely on these inputs. JsonRemedy fixes them int
 - **Unescaped quotes**: `"text "quoted" text"` → proper escaping
 - **Trailing backslashes**: Streaming artifact cleanup
 
+#### 🔧 **Hardcoded Patterns** *(ported from [json_repair](https://github.com/mangiucugna/json_repair) Python library)*
+Layer 3 includes battle-tested cleanup patterns for edge cases commonly found in LLM output:
+
+- **Smart quotes**: `"curly"`, `«guillemets»`, `‹angles›` → `"standard"`
+- **Doubled quotes**: `""value""` → `"value"` (preserves empty strings `""`)
+- **Number formats**: `1,234,567` → `1234567` (removes thousands separators)
+- **Unicode escapes**: `\u263a` → `☺` (opt-in via `:enable_escape_normalization`)
+- **Hex escapes**: `\x41` → `A` (opt-in via `:enable_escape_normalization`)
+
+These patterns run as a pre-processing step and can be controlled via feature flags. See `examples/hardcoded_patterns_examples.exs` for demonstrations.
+
 ### 🚀 **Fast Path Validation (Layer 4)**
 - **Jason.decode optimization**: Valid JSON uses battle-tested parser
 - **Performance monitoring**: Automatic fallback for complex repairs
@@ -128,7 +139,7 @@ Add JsonRemedy to your `mix.exs`:
 ```elixir
 def deps do
   [
-    {:json_remedy, "~> 0.1.3"}
+    {:json_remedy, "~> 0.1.4"}
   ]
 end
 ```
@@ -228,11 +239,23 @@ mix run examples/basic_usage.exs
 ```
 Learn the fundamentals with step-by-step examples:
 - Fixing unquoted keys
-- Normalizing quote styles  
+- Normalizing quote styles
 - Handling boolean/null variants
 - Repairing structural issues
 - Processing LLM outputs
 
+### 🔧 **Hardcoded Patterns Examples** ✨ *NEW*
+```bash
+mix run examples/hardcoded_patterns_examples.exs
+```
+Demonstrates advanced cleanup patterns ported from Python's `json_repair` library:
+- **Smart quotes normalization**: Curly quotes, guillemets, angle quotes
+- **Doubled quotes repair**: `""value""` → `"value"`
+- **Number format cleaning**: `1,234,567` → `1234567`
+- **Unicode/hex escapes**: `\u263a` → `☺`, `\x41` → `A`
+- **International text**: UTF-8 support with smart quotes
+- **Combined patterns**: Real-world LLM output examples
+
 ### 🌍 **Real-World Scenarios**
 ```bash
 mix run examples/real_world_scenarios.exs