|
| 1 | +# CLAUDE.md |
| 2 | + |
| 3 | +This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository. |
| 4 | + |
| 5 | +## Project Overview |
| 6 | + |
| 7 | +**Nette NEON** is a parser/encoder library for the NEON format (Nette Object Notation), a human-readable structured data format similar to YAML but with unique features like entities. It's used primarily for configuration files in the Nette Framework. |
| 8 | + |
| 9 | +- **Repository**: https://github.com/nette/neon |
| 10 | +- **Homepage**: https://neon.nette.org |
| 11 | +- **PHP Requirements**: 8.2 - 8.5 |
| 12 | +- **Current Version**: 3.5-dev |
| 13 | + |
| 14 | +## Essential Commands |
| 15 | + |
| 16 | +### Testing |
| 17 | +```bash |
| 18 | +# Run all tests |
| 19 | +composer run tester |
| 20 | + |
| 21 | +# Run specific test file |
| 22 | +vendor/bin/tester tests/Neon/Decoder.phpt -s |
| 23 | + |
| 24 | +# Run tests in specific directory |
| 25 | +vendor/bin/tester tests/Neon/ -s |
| 26 | +``` |
| 27 | + |
| 28 | +### Static Analysis |
| 29 | +```bash |
| 30 | +# Run PHPStan (level 6) |
| 31 | +composer run phpstan |
| 32 | +``` |
| 33 | + |
| 34 | +### Validation |
| 35 | +```bash |
| 36 | +# Lint NEON files |
| 37 | +vendor/bin/neon-lint <path> |
| 38 | +``` |
| 39 | + |
| 40 | +## Architecture |
| 41 | + |
| 42 | +### Parser Pipeline |
| 43 | + |
| 44 | +The library follows a classic lexer → parser → AST → value pipeline: |
| 45 | + |
| 46 | +1. **Lexer** (`src/Neon/Lexer.php`): Tokenizes NEON input using regex patterns |
| 47 | + - Produces tokens with type, text, and position information |
| 48 | + - Token types: `String`, `Literal`, `Char`, `Comment`, `Newline`, `Whitespace`, `End` |
| 49 | + |
| 50 | +2. **TokenStream** (`src/Neon/TokenStream.php`): Manages token buffer |
| 51 | + - Handles whitespace/comment skipping |
| 52 | + - Provides token consumption and seeking |
| 53 | + |
| 54 | +3. **Parser** (`src/Neon/Parser.php`): Builds Abstract Syntax Tree |
| 55 | + - Converts token stream to Node tree |
| 56 | + - Handles block (indentation-based) and inline (bracket-based) notation |
| 57 | + - Detects mixed tabs/spaces errors |
| 58 | + |
| 59 | +4. **Node Hierarchy** (`src/Neon/Node/`): AST representation |
| 60 | + - All nodes extend abstract `Node` class with `toValue()` and `toString()` methods |
| 61 | + - Node types: |
| 62 | + - `LiteralNode` - numbers, booleans, null, dates |
| 63 | + - `StringNode` - quoted and unquoted strings |
| 64 | + - `EntityNode` - NEON entities like `Column(type: int)` |
| 65 | + - `EntityChainNode` - chained entities |
| 66 | + - `BlockArrayNode` / `InlineArrayNode` - arrays in different notations |
| 67 | + - `ArrayItemNode` - individual array items with optional key |
| 68 | + - All nodes track position (`start`, `end` as `Position` objects) |
| 69 | + |
| 70 | +5. **Decoder** (`src/Neon/Decoder.php`): Public API for parsing |
| 71 | + - `parseToNode()` - returns AST |
| 72 | + - `decode()` - returns PHP value (calls `toValue()` on root node) |
| 73 | + |
| 74 | +6. **Encoder** (`src/Neon/Encoder.php`): Converts PHP values to NEON |
| 75 | + - `valueToNode()` - creates Node tree from PHP value |
| 76 | + - `encode()` - returns NEON string (calls `toString()` on nodes) |
| 77 | + - Supports block and inline rendering modes |
| 78 | + |
| 79 | +7. **Traverser** (`src/Neon/Traverser.php`): AST manipulation utility |
| 80 | + - Visitor pattern for node processing |
| 81 | + - Supports `enter` and `leave` callbacks |
| 82 | + - Return constants: `RemoveNode`, `StopTraversal`, `DontTraverseChildren` |
| 83 | + |
| 84 | +### Key Design Patterns |
| 85 | + |
| 86 | +- **Position Tracking**: All tokens and nodes include `Position` (line, column, offset) for precise error reporting |
| 87 | +- **Readonly Classes**: `Token` and `Position` are readonly for immutability |
| 88 | +- **Final by Default**: All concrete classes are `final` unless designed for extension |
| 89 | +- **Entity Pattern**: `Entity` class extends `stdClass` with `value` and `attributes` properties |
| 90 | +- **Type Safety**: All parameters, properties, and return types are strictly typed |
| 91 | + |
| 92 | +## Code Style |
| 93 | + |
| 94 | +### PHP Standards |
| 95 | +- Every file must start with `declare(strict_types=1)` |
| 96 | +- Use tabs for indentation |
| 97 | +- Use single quotes for strings (except when escaping needed) |
| 98 | +- PHP 8.2+ features are standard (readonly, union types, named parameters) |
| 99 | + |
| 100 | +### Naming Conventions |
| 101 | +- Classes: PascalCase (e.g., `BlockArrayNode`) |
| 102 | +- Methods: camelCase (e.g., `toValue()`, `toString()`) |
| 103 | +- Constants: PascalCase (e.g., `Token::String`, `Traverser::RemoveNode`) |
| 104 | +- Namespace: `Nette\Neon` |
| 105 | + |
| 106 | +### Documentation |
| 107 | +- Minimalist phpDoc - only document non-obvious information |
| 108 | +- Don't duplicate type information already in signatures |
| 109 | +- Use `@internal` for classes not meant for public use |
| 110 | +- Focus on explaining "why" and edge cases, not "what" (which is in the signature) |
| 111 | + |
| 112 | +### Return Type Formatting |
| 113 | +```php |
| 114 | +public function example( |
| 115 | + string $param, |
| 116 | + array $options, |
| 117 | +): ReturnType |
| 118 | +{ |
| 119 | + // method body on new line after opening brace |
| 120 | +} |
| 121 | +``` |
| 122 | + |
| 123 | +## Testing Patterns |
| 124 | + |
| 125 | +Tests use **Nette Tester** with `.phpt` extension: |
| 126 | + |
| 127 | +```php |
| 128 | +<?php |
| 129 | + |
| 130 | +declare(strict_types=1); |
| 131 | + |
| 132 | +use Tester\Assert; |
| 133 | +use Nette\Neon\Neon; |
| 134 | + |
| 135 | +require __DIR__ . '/../bootstrap.php'; |
| 136 | + |
| 137 | +test('description of what is being tested', function () { |
| 138 | + $result = Neon::decode('hello: world'); |
| 139 | + Assert::same(['hello' => 'world'], $result); |
| 140 | +}); |
| 141 | + |
| 142 | +test('handles edge case properly', function () { |
| 143 | + Assert::exception( |
| 144 | + fn() => Neon::decode('invalid [[['), |
| 145 | + Nette\Neon\Exception::class, |
| 146 | + 'Unexpected %a%', // %a% = any text wildcard |
| 147 | + ); |
| 148 | +}); |
| 149 | +``` |
| 150 | + |
| 151 | +**Key Points**: |
| 152 | +- Use `test()` function for each test case (provided by bootstrap) |
| 153 | +- First parameter is clear description (no comments needed) |
| 154 | +- Use `Assert::same()` for exact comparisons |
| 155 | +- Use `Assert::exception()` for exception testing |
| 156 | +- Message patterns support `%a%` (any text) placeholders |
| 157 | + |
| 158 | +## NEON Syntax Rules |
| 159 | + |
| 160 | +Understanding these rules is critical when working with the parser: |
| 161 | + |
| 162 | +### Block vs Inline Notation |
| 163 | +- **Block notation**: Items on separate lines with consistent indentation |
| 164 | +- **Inline notation**: Enclosed in brackets `[]` or braces `{}`, indentation irrelevant |
| 165 | +- **Critical limitation**: Block notation CANNOT be used inside inline notation |
| 166 | + |
| 167 | +```neon |
| 168 | +# This works |
| 169 | +pets: [Cat, Dog] |
| 170 | +
|
| 171 | +# This does NOT work |
| 172 | +item: [ |
| 173 | + pets: |
| 174 | + - Cat # INVALID! |
| 175 | + - Dog |
| 176 | +] |
| 177 | +``` |
| 178 | + |
| 179 | +### Indentation Rules |
| 180 | +- Both tabs and spaces allowed for indentation |
| 181 | +- **Cannot mix** tabs and spaces (parser will report error) |
| 182 | +- Indentation level determines structure hierarchy |
| 183 | +- Space after `:` in key-value pairs is **required** |
| 184 | + |
| 185 | +### Mappings (Associative Arrays) |
| 186 | +- Format: `key: value` (space after `:` required) |
| 187 | +- Alternative: `key=value` (both block and inline) |
| 188 | +- Inline: `{key: value, key2: value2}` or multiline with `{}` |
| 189 | + |
| 190 | +### Sequences (Indexed Arrays) |
| 191 | +- Block: Lines starting with `- ` (hyphen + space) |
| 192 | +- Inline: `[item1, item2]` |
| 193 | +- Hyphens cannot be used in inline notation |
| 194 | + |
| 195 | +### String Quoting |
| 196 | +Must quote strings containing: `# " ' , : = - [ ] { } ( )` |
| 197 | +- Single quotes: Escape quotes by doubling `''` |
| 198 | +- Double quotes: JSON escape sequences + `\_` for non-breaking space |
| 199 | +- Unquoted: Simple strings without special characters |
| 200 | + |
| 201 | +### Multiline Strings |
| 202 | +```neon |
| 203 | +''' |
| 204 | + content here |
| 205 | + ''' |
| 206 | +``` |
| 207 | +- Triple quotes on separate lines |
| 208 | +- Indentation of first line ignored for all lines |
| 209 | +- Use `"""` for escape sequences |
| 210 | + |
| 211 | +### Comments |
| 212 | +- Start with `#` |
| 213 | +- All characters to the right are ignored |
| 214 | +```neon |
| 215 | +# Full line comment |
| 216 | +key: value # Inline comment |
| 217 | +``` |
| 218 | + |
| 219 | +### Mixed Arrays |
| 220 | +PHP uses same structure for mappings and sequences, so they can be merged: |
| 221 | +```neon |
| 222 | +- Cat # indexed item 0 |
| 223 | +street: 742 Evergreen Terrace # key 'street' |
| 224 | +- Goldfish # indexed item 1 |
| 225 | +``` |
| 226 | + |
| 227 | +Becomes: |
| 228 | +```php |
| 229 | +['Cat', 'street' => '742 Evergreen Terrace', 'Goldfish'] |
| 230 | +``` |
| 231 | + |
| 232 | +## Common Patterns |
| 233 | + |
| 234 | +### Parsing NEON |
| 235 | +```php |
| 236 | +// To PHP value |
| 237 | +$value = Nette\Neon\Neon::decode($neonString); |
| 238 | + |
| 239 | +// From file (removes BOM automatically) |
| 240 | +$value = Nette\Neon\Neon::decodeFile('config.neon'); |
| 241 | + |
| 242 | +// To AST (for manipulation) |
| 243 | +$decoder = new Nette\Neon\Decoder; |
| 244 | +$node = $decoder->parseToNode($neonString); |
| 245 | +``` |
| 246 | + |
| 247 | +### Error Handling |
| 248 | +All methods throw `Nette\Neon\Exception` on error with position information: |
| 249 | + |
| 250 | +```php |
| 251 | +try { |
| 252 | + $value = Neon::decode($neonString); |
| 253 | +} catch (Nette\Neon\Exception $e) { |
| 254 | + // Exception includes line/column position |
| 255 | + echo $e->getMessage(); |
| 256 | +} |
| 257 | +``` |
| 258 | + |
| 259 | +### Encoding to NEON |
| 260 | +```php |
| 261 | +// Inline mode (single line) |
| 262 | +$neon = Nette\Neon\Neon::encode($value); |
| 263 | + |
| 264 | +// Block mode (multiline) |
| 265 | +$neon = Nette\Neon\Neon::encode($value, blockMode: true); |
| 266 | + |
| 267 | +// Custom indentation (default is tab) |
| 268 | +$neon = Nette\Neon\Neon::encode($value, blockMode: true, indentation: ' '); |
| 269 | +``` |
| 270 | + |
| 271 | +### AST Traversal |
| 272 | +```php |
| 273 | +$traverser = new Nette\Neon\Traverser; |
| 274 | +$newNode = $traverser->traverse($node, |
| 275 | + enter: function ($node) { |
| 276 | + // Process node before children |
| 277 | + if ($node instanceof StringNode) { |
| 278 | + return new StringNode($node->value . ' modified', $node->start, $node->end); |
| 279 | + } |
| 280 | + }, |
| 281 | + leave: function ($node) { |
| 282 | + // Process node after children |
| 283 | + } |
| 284 | +); |
| 285 | +``` |
| 286 | + |
| 287 | +## Special Features |
| 288 | + |
| 289 | +### JSON Compatibility |
| 290 | +**JSON is a subset of NEON** - any valid JSON can be parsed as NEON: |
| 291 | +```php |
| 292 | +$value = Neon::decode('{"key": "value"}'); // Valid NEON |
| 293 | +``` |
| 294 | + |
| 295 | +### Entities |
| 296 | +NEON entities are function-like structures parsed into `Entity` objects: |
| 297 | + |
| 298 | +```neon |
| 299 | +Column(type: int, nulls: yes) |
| 300 | +``` |
| 301 | + |
| 302 | +Becomes: |
| 303 | +```php |
| 304 | +new Nette\Neon\Entity('Column', ['type' => 'int', 'nulls' => true]) |
| 305 | +``` |
| 306 | + |
| 307 | +Chained entities use `Entity::Chain`: |
| 308 | +```neon |
| 309 | +Column(type: int) Field(id: 1) |
| 310 | +``` |
| 311 | + |
| 312 | +Inside entity parentheses, inline notation rules apply (multiline allowed, commas optional): |
| 313 | +```neon |
| 314 | +Column( |
| 315 | + type: int |
| 316 | + nulls: yes |
| 317 | +) |
| 318 | +``` |
| 319 | + |
| 320 | +### Literals (Booleans and Nulls) |
| 321 | + |
| 322 | +**Booleans** - case insensitive variants: |
| 323 | +- `true` / `TRUE` / `True` |
| 324 | +- `false` / `FALSE` / `False` |
| 325 | +- `yes` / `YES` / `Yes` |
| 326 | +- `no` / `NO` / `No` |
| 327 | + |
| 328 | +**Nulls** - case insensitive variants: |
| 329 | +- `null` / `NULL` / `Null` |
| 330 | +- Empty value (just `key:` without value) |
| 331 | + |
| 332 | +### Date Parsing |
| 333 | +NEON automatically converts date strings to `DateTimeImmutable`: |
| 334 | +```neon |
| 335 | +- 2016-06-03 # date |
| 336 | +- 2016-06-03 19:00:00 # date & time |
| 337 | +- 2016-06-03 19:00:00.1234 # date & microtime |
| 338 | +- 2016-06-03 19:00:00 +0200 # date & time & timezone |
| 339 | +- 2016-06-03 19:00:00 +02:00 # date & time & timezone |
| 340 | +``` |
| 341 | + |
| 342 | +### Number Formats |
| 343 | +Supports binary, octal, hexadecimal, and scientific notation: |
| 344 | +```neon |
| 345 | +- 12 # integer |
| 346 | +- 12.3 # float |
| 347 | +- +1.2e-34 # scientific notation |
| 348 | +- 0b11010 # binary |
| 349 | +- 0o666 # octal |
| 350 | +- 0x7A # hexadecimal |
| 351 | +``` |
| 352 | + |
| 353 | +## Important Notes |
| 354 | + |
| 355 | +- **This is a parser library** - focus on correct format handling, not business logic |
| 356 | +- **Position tracking is critical** - all parsing includes line/column info for error messages |
| 357 | +- **Minimal dependencies** - only requires `ext-json` |
| 358 | +- **Strict types everywhere** - proper type declarations are essential |
| 359 | +- **Final by default** - classes are `final` unless designed for extension |
| 360 | +- **Format preservation** - upcoming feature; be careful with changes that affect formatting |
0 commit comments