Skip to content

Commit 399173f

Browse files
cappuccappuc
committed
added support for tags with raw body
1 parent 6645801 commit 399173f

File tree

4 files changed

+79
-38
lines changed

4 files changed

+79
-38
lines changed

‎src/Parse/Lexer.php‎

Lines changed: 60 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -3,6 +3,7 @@
33
namespace Keepsuit\Liquid\Parse;
44

55
use Keepsuit\Liquid\Exceptions\SyntaxException;
6+
use Keepsuit\Liquid\TagBlock;
67
use RuntimeException;
78

89
class Lexer
@@ -36,6 +37,11 @@ class Lexer
3637

3738
protected int $position;
3839

40+
/**
41+
* @var string[]
42+
*/
43+
protected array $rawBodyTags;
44+
3945
public function __construct(
4046
protected ParseContext $parseContext,
4147
) {}
@@ -53,6 +59,14 @@ public function tokenize(string $source): TokenStream
5359
$this->state = LexerState::Data;
5460
$this->tokens = [];
5561

62+
$this->rawBodyTags = array_keys(array_filter($this->parseContext->environment->tagRegistry->all(), function ($tag) {
63+
if (! is_subclass_of($tag, TagBlock::class)) {
64+
return false;
65+
}
66+
67+
return $tag::hasRawBody();
68+
}));
69+
5670
$this->parseContext->lineNumber = 1;
5771

5872
preg_match_all(LexerOptions::tokenStartRegex(), $this->source, $matches, PREG_OFFSET_CAPTURE);
@@ -159,18 +173,41 @@ protected function lexVariable(): void
159173
*/
160174
protected function lexBlock(): void
161175
{
162-
if (preg_match(LexerOptions::blockEndRegex(), $this->source, $matches, offset: $this->cursor) === 1) {
163-
$this->pushToken(TokenType::BlockEnd);
164-
$this->moveCursor($matches[0]);
165-
$this->popState();
176+
$tag = null;
166177

167-
// trim?
168-
if (trim($matches[0])[0] === LexerOptions::WhitespaceTrim->value) {
169-
preg_match('/\s+/A', $this->source, $matches, offset: $this->cursor);
170-
$this->moveCursor($matches[0] ?? '');
178+
// Parse the full expression inside {% ... %}
179+
while (preg_match(LexerOptions::blockEndRegex(), $this->source, $matches, offset: $this->cursor) !== 1) {
180+
$this->lexExpression();
181+
182+
$lastToken = $this->tokens[array_key_last($this->tokens)];
183+
184+
if ($tag === null && $lastToken->type === TokenType::Identifier) {
185+
$tag = $lastToken;
171186
}
187+
}
188+
189+
// Move the cursor to the end of the block
190+
$this->moveCursor($matches[0]);
191+
192+
// trim?
193+
if (trim($matches[0])[0] === LexerOptions::WhitespaceTrim->value) {
194+
preg_match('/\s+/A', $this->source, $matches, offset: $this->cursor);
195+
$this->moveCursor($matches[0] ?? '');
196+
}
197+
198+
// If the last token is a block start, we remove the node
199+
$lastToken = $this->tokens[array_key_last($this->tokens)];
200+
if ($lastToken->type === TokenType::BlockStart) {
201+
array_pop($this->tokens);
172202
} else {
173-
$this->lexExpression();
203+
$this->pushToken(TokenType::BlockEnd);
204+
}
205+
206+
$this->popState();
207+
208+
// If the tag is a raw body tag, we need to lex the body as raw data instead of liquid blocks
209+
if ($tag !== null && in_array($tag->data, $this->rawBodyTags, true)) {
210+
$this->laxRawBodyTag($tag->data);
174211
}
175212
}
176213

@@ -227,6 +264,19 @@ protected function ensureStreamNotEnded(): void
227264
}
228265
}
229266

267+
private function laxRawBodyTag(string $tag): void
268+
{
269+
if (preg_match(LexerOptions::blockRawBodyTagDataRegex($tag), $this->source, $matches, flags: PREG_OFFSET_CAPTURE, offset: $this->cursor) !== 1) {
270+
throw SyntaxException::tagBlockNeverClosed($tag);
271+
}
272+
273+
$rawBody = substr($this->source, $this->cursor, $matches[0][1] - $this->cursor);
274+
275+
$this->moveCursor($rawBody);
276+
277+
$this->pushToken(TokenType::RawData, $rawBody);
278+
}
279+
230280
protected function lexRawData(): void
231281
{
232282
if (preg_match(LexerOptions::blockRawDataRegex(), $this->source, $matches, flags: PREG_OFFSET_CAPTURE, offset: $this->cursor) !== 1) {
@@ -265,24 +315,7 @@ protected function lexInlineComment(): void
265315

266316
$text = substr($this->source, $this->cursor, $matches[0][1] - $this->cursor);
267317

268-
$this->moveCursor($text.$matches[0][0]);
269-
270-
if ($matches[1][0] === "\n") {
271-
return;
272-
}
273-
274-
$lastToken = $this->tokens[count($this->tokens) - 1] ?? null;
275-
276-
if ($lastToken?->type === TokenType::BlockStart) {
277-
array_pop($this->tokens);
278-
} else {
279-
$this->pushToken(TokenType::BlockEnd);
280-
}
281-
282-
if ($matches[1][0] === LexerOptions::WhitespaceTrim->value) {
283-
preg_match('/\s+/A', $this->source, $matches2, offset: $this->cursor);
284-
$this->moveCursor($matches2[0] ?? '');
285-
}
318+
$this->moveCursor($text);
286319
}
287320

288321
protected function pushToken(TokenType $type, string $value = ''): void

‎src/Parse/LexerOptions.php‎

Lines changed: 13 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -33,44 +33,47 @@ public static function tokenStartRegex(): string
3333
return $regex;
3434
}
3535

36-
public static function commentBlockRegex(): string
36+
public static function variableEndRegex(): string
3737
{
3838
static $regex;
3939

4040
if ($regex === null) {
4141
$regex = sprintf(
42-
"{\s*comment\s*(?:%s|%s')}Asx",
43-
preg_quote(LexerOptions::WhitespaceTrim->value.LexerOptions::TagBlockEnd->value),
44-
preg_quote(LexerOptions::TagBlockEnd->value),
42+
'{\s*(?:%s|%s)}Ax',
43+
preg_quote(LexerOptions::WhitespaceTrim->value.LexerOptions::TagVariableEnd->value),
44+
preg_quote(LexerOptions::TagVariableEnd->value),
4545
);
4646
}
4747

4848
return $regex;
4949
}
5050

51-
public static function variableEndRegex(): string
51+
public static function blockEndRegex(): string
5252
{
5353
static $regex;
5454

5555
if ($regex === null) {
5656
$regex = sprintf(
5757
'{\s*(?:%s|%s)}Ax',
58-
preg_quote(LexerOptions::WhitespaceTrim->value.LexerOptions::TagVariableEnd->value),
59-
preg_quote(LexerOptions::TagVariableEnd->value),
58+
preg_quote(LexerOptions::WhitespaceTrim->value.LexerOptions::TagBlockEnd->value),
59+
preg_quote(LexerOptions::TagBlockEnd->value),
6060
);
6161
}
6262

6363
return $regex;
6464
}
6565

66-
public static function blockEndRegex(): string
66+
public static function blockRawBodyTagDataRegex(string $tag): string
6767
{
6868
static $regex;
6969

7070
if ($regex === null) {
7171
$regex = sprintf(
72-
'{\s*(?:%s|%s)}Ax',
73-
preg_quote(LexerOptions::WhitespaceTrim->value.LexerOptions::TagBlockEnd->value),
72+
'{%s(%s)?\s*end%s\s*(%s)?%s}sx',
73+
preg_quote(LexerOptions::TagBlockStart->value),
74+
LexerOptions::WhitespaceTrim->value,
75+
preg_quote($tag),
76+
LexerOptions::WhitespaceTrim->value,
7477
preg_quote(LexerOptions::TagBlockEnd->value),
7578
);
7679
}

‎src/TagBlock.php‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -20,4 +20,9 @@ public function parseTreeVisitorChildren(): array
2020
{
2121
return [];
2222
}
23+
24+
public static function hasRawBody(): bool
25+
{
26+
return false;
27+
}
2328
}

‎tests/Integration/Tags/InlineCommentTagTest.php‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -51,5 +51,5 @@
5151
});
5252

5353
test('inline comment does not support nested tags', function () {
54-
assertMatchSyntaxError('Liquid syntax error (line 1): Unexpected token type: %}', "{%- # {% echo 'hello world' %} -%}");
54+
assertTemplateResult(' -%}', "{%- # {% echo 'hello world' %} -%}");
5555
});

0 commit comments

Comments
 (0)