1: <?php
  2: 
  3: /**
  4:  * This file is part of the Nette Framework (https://nette.org)
  5:  * Copyright (c) 2004 David Grudl (http://davidgrudl.com)
  6:  * @package Nette\Latte
  7:  */
  8: 
  9: 
 10: 
 11: /**
 12:  * Latte parser.
 13:  *
 14:  * @author     David Grudl
 15:  * @package Nette\Latte
 16:  */
 17: class NParser extends NObject
 18: {
 19:     /** @internal regular expression for single & double quoted PHP string */
 20:     const RE_STRING = '\'(?:\\\\.|[^\'\\\\])*\'|"(?:\\\\.|[^"\\\\])*"';
 21: 
 22:     /** @internal special HTML attribute prefix */
 23:     const N_PREFIX = 'n:';
 24: 
 25:     /** @var string default macro tag syntax */
 26:     public $defaultSyntax = 'latte';
 27: 
 28:     /** @var array */
 29:     public $syntaxes = array(
 30:         'latte' => array('\\{(?![\\s\'"{}])', '\\}'), // {...}
 31:         'double' => array('\\{\\{(?![\\s\'"{}])', '\\}\\}'), // {{...}}
 32:         'asp' => array('<%\s*', '\s*%>'), /* <%...%> */
 33:         'python' => array('\\{[{%]\s*', '\s*[%}]\\}'), // {% ... %} | {{ ... }}
 34:         'off' => array('[^\x00-\xFF]', ''),
 35:     );
 36: 
 37:     /** @var string */
 38:     private $macroRe;
 39: 
 40:     /** @var string source template */
 41:     private $input;
 42: 
 43:     /** @var NLatteToken[] */
 44:     private $output;
 45: 
 46:     /** @var int  position on source template */
 47:     private $offset;
 48: 
 49:     /** @var array */
 50:     private $context;
 51: 
 52:     /** @var string */
 53:     private $lastHtmlTag;
 54: 
 55:     /** @var string used by filter() */
 56:     private $syntaxEndTag;
 57: 
 58:     /** @var bool */
 59:     private $xmlMode;
 60: 
 61:     /** @internal states */
 62:     const CONTEXT_TEXT = 'text',
 63:         CONTEXT_CDATA = 'cdata',
 64:         CONTEXT_TAG = 'tag',
 65:         CONTEXT_ATTRIBUTE = 'attribute',
 66:         CONTEXT_NONE = 'none',
 67:         CONTEXT_COMMENT = 'comment';
 68: 
 69: 
 70:     /**
 71:      * Process all {macros} and <tags/>.
 72:      * @param  string
 73:      * @return array
 74:      */
 75:     public function parse($input)
 76:     {
 77:         if (substr($input, 0, 3) === "\xEF\xBB\xBF") { // BOM
 78:             $input = substr($input, 3);
 79:         }
 80:         if (!NStrings::checkEncoding($input)) {
 81:             throw new InvalidArgumentException('Template is not valid UTF-8 stream.');
 82:         }
 83:         $input = str_replace("\r\n", "\n", $input);
 84:         $this->input = $input;
 85:         $this->output = array();
 86:         $this->offset = 0;
 87: 
 88:         $this->setSyntax($this->defaultSyntax);
 89:         $this->setContext(self::CONTEXT_TEXT);
 90:         $this->lastHtmlTag = $this->syntaxEndTag = NULL;
 91: 
 92:         while ($this->offset < strlen($input)) {
 93:             $matches = $this->{"context".$this->context[0]}();
 94: 
 95:             if (!$matches) { // EOF
 96:                 break;
 97: 
 98:             } elseif (!empty($matches['comment'])) { // {* *}
 99:                 $this->addToken(NLatteToken::COMMENT, $matches[0]);
100: 
101:             } elseif (!empty($matches['macro'])) { // {macro}
102:                 $token = $this->addToken(NLatteToken::MACRO_TAG, $matches[0]);
103:                 list($token->name, $token->value, $token->modifiers) = $this->parseMacroTag($matches['macro']);
104:             }
105: 
106:             $this->filter();
107:         }
108: 
109:         if ($this->offset < strlen($input)) {
110:             $this->addToken(NLatteToken::TEXT, substr($this->input, $this->offset));
111:         }
112:         return $this->output;
113:     }
114: 
115: 
116:     /**
117:      * Handles CONTEXT_TEXT.
118:      */
119:     private function contextText()
120:     {
121:         $matches = $this->match('~
122:             (?:(?<=\n|^)[ \t]*)?<(?P<closing>/?)(?P<tag>[a-z0-9:]+)|  ##  begin of HTML tag <tag </tag - ignores <!DOCTYPE
123:             <(?P<htmlcomment>!--(?!>))|     ##  begin of HTML comment <!--, but not <!-->
124:             '.$this->macroRe.'              ##  macro tag
125:         ~xsi');
126: 
127:         if (!empty($matches['htmlcomment'])) { // <!--
128:             $this->addToken(NLatteToken::HTML_TAG_BEGIN, $matches[0]);
129:             $this->setContext(self::CONTEXT_COMMENT);
130: 
131:         } elseif (!empty($matches['tag'])) { // <tag or </tag
132:             $token = $this->addToken(NLatteToken::HTML_TAG_BEGIN, $matches[0]);
133:             $token->name = $matches['tag'];
134:             $token->closing = (bool) $matches['closing'];
135:             $this->lastHtmlTag = $matches['closing'] . strtolower($matches['tag']);
136:             $this->setContext(self::CONTEXT_TAG);
137:         }
138:         return $matches;
139:     }
140: 
141: 
142:     /**
143:      * Handles CONTEXT_CDATA.
144:      */
145:     private function contextCData()
146:     {
147:         $matches = $this->match('~
148:             </(?P<tag>'.$this->lastHtmlTag.')(?![a-z0-9:])| ##  end HTML tag </tag
149:             '.$this->macroRe.'              ##  macro tag
150:         ~xsi');
151: 
152:         if (!empty($matches['tag'])) { // </tag
153:             $token = $this->addToken(NLatteToken::HTML_TAG_BEGIN, $matches[0]);
154:             $token->name = $this->lastHtmlTag;
155:             $token->closing = TRUE;
156:             $this->lastHtmlTag = '/' . $this->lastHtmlTag;
157:             $this->setContext(self::CONTEXT_TAG);
158:         }
159:         return $matches;
160:     }
161: 
162: 
163:     /**
164:      * Handles CONTEXT_TAG.
165:      */
166:     private function contextTag()
167:     {
168:         $matches = $this->match('~
169:             (?P<end>\ ?/?>)([ \t]*\n)?|  ##  end of HTML tag
170:             '.$this->macroRe.'|          ##  macro tag
171:             \s*(?P<attr>[^\s/>={]+)(?:\s*=\s*(?P<value>["\']|[^\s/>{]+))? ## begin of HTML attribute
172:         ~xsi');
173: 
174:         if (!empty($matches['end'])) { // end of HTML tag />
175:             $this->addToken(NLatteToken::HTML_TAG_END, $matches[0]);
176:             $this->setContext(!$this->xmlMode && in_array($this->lastHtmlTag, array('script', 'style'), TRUE) ? self::CONTEXT_CDATA : self::CONTEXT_TEXT);
177: 
178:         } elseif (isset($matches['attr']) && $matches['attr'] !== '') { // HTML attribute
179:             $token = $this->addToken(NLatteToken::HTML_ATTRIBUTE, $matches[0]);
180:             $token->name = $matches['attr'];
181:             $token->value = isset($matches['value']) ? $matches['value'] : '';
182: 
183:             if ($token->value === '"' || $token->value === "'") { // attribute = "'
184:                 if (NStrings::startsWith($token->name, self::N_PREFIX)) {
185:                     $token->value = '';
186:                     if ($m = $this->match('~(.*?)' . $matches['value'] . '~xsi')) {
187:                         $token->value = $m[1];
188:                         $token->text .= $m[0];
189:                     }
190:                 } else {
191:                     $this->setContext(self::CONTEXT_ATTRIBUTE, $matches['value']);
192:                 }
193:             }
194:         }
195:         return $matches;
196:     }
197: 
198: 
199:     /**
200:      * Handles CONTEXT_ATTRIBUTE.
201:      */
202:     private function contextAttribute()
203:     {
204:         $matches = $this->match('~
205:             (?P<quote>'.$this->context[1].')|  ##  end of HTML attribute
206:             '.$this->macroRe.'                 ##  macro tag
207:         ~xsi');
208: 
209:         if (!empty($matches['quote'])) { // (attribute end) '"
210:             $this->addToken(NLatteToken::TEXT, $matches[0]);
211:             $this->setContext(self::CONTEXT_TAG);
212:         }
213:         return $matches;
214:     }
215: 
216: 
217:     /**
218:      * Handles CONTEXT_COMMENT.
219:      */
220:     private function contextComment()
221:     {
222:         $matches = $this->match('~
223:             (?P<htmlcomment>-->)|   ##  end of HTML comment
224:             '.$this->macroRe.'      ##  macro tag
225:         ~xsi');
226: 
227:         if (!empty($matches['htmlcomment'])) { // -->
228:             $this->addToken(NLatteToken::HTML_TAG_END, $matches[0]);
229:             $this->setContext(self::CONTEXT_TEXT);
230:         }
231:         return $matches;
232:     }
233: 
234: 
235:     /**
236:      * Handles CONTEXT_NONE.
237:      */
238:     private function contextNone()
239:     {
240:         $matches = $this->match('~
241:             '.$this->macroRe.'     ##  macro tag
242:         ~xsi');
243:         return $matches;
244:     }
245: 
246: 
247:     /**
248:      * Matches next token.
249:      * @param  string
250:      * @return array
251:      */
252:     private function match($re)
253:     {
254:         if ($matches = NStrings::match($this->input, $re, PREG_OFFSET_CAPTURE, $this->offset)) {
255:             $value = substr($this->input, $this->offset, $matches[0][1] - $this->offset);
256:             if ($value !== '') {
257:                 $this->addToken(NLatteToken::TEXT, $value);
258:             }
259:             $this->offset = $matches[0][1] + strlen($matches[0][0]);
260:             foreach ($matches as $k => $v) $matches[$k] = $v[0];
261:         }
262:         return $matches;
263:     }
264: 
265: 
266:     /**
267:      * @return self
268:      */
269:     public function setContext($context, $quote = NULL)
270:     {
271:         $this->context = array($context, $quote);
272:         return $this;
273:     }
274: 
275: 
276:     /**
277:      * Changes macro tag delimiters.
278:      * @param  string
279:      * @return self
280:      */
281:     public function setSyntax($type)
282:     {
283:         $type = ($tmp=$type) ? $tmp : $this->defaultSyntax;
284:         if (isset($this->syntaxes[$type])) {
285:             $this->setDelimiters($this->syntaxes[$type][0], $this->syntaxes[$type][1]);
286:         } else {
287:             throw new InvalidArgumentException("Unknown syntax '$type'");
288:         }
289:         return $this;
290:     }
291: 
292: 
293:     /**
294:      * Changes macro tag delimiters.
295:      * @param  string  left regular expression
296:      * @param  string  right regular expression
297:      * @return self
298:      */
299:     public function setDelimiters($left, $right)
300:     {
301:         $this->macroRe = '
302:             (?P<comment>' . $left . '\\*.*?\\*' . $right . '\n{0,2})|
303:             ' . $left . '
304:                 (?P<macro>(?:' . self::RE_STRING . '|\{
305:                         (?P<inner>' . self::RE_STRING . '|\{(?P>inner)\}|[^\'"{}])*+
306:                 \}|[^\'"{}])+?)
307:             ' . $right . '
308:             (?P<rmargin>[ \t]*(?=\n))?
309:         ';
310:         return $this;
311:     }
312: 
313: 
314:     /**
315:      * Parses macro tag to name, arguments a modifiers parts.
316:      * @param  string {name arguments | modifiers}
317:      * @return array
318:      */
319:     public function parseMacroTag($tag)
320:     {
321:         $match = NStrings::match($tag, '~^
322:             (
323:                 (?P<name>\?|/?[a-z]\w*+(?:[.:]\w+)*+(?!::|\(|\\\\))|   ## ?, name, /name, but not function( or class:: or namespace\
324:                 (?P<noescape>!?)(?P<shortname>/?[=\~#%^&_]?)      ## !expression, !=expression, ...
325:             )(?P<args>.*?)
326:             (?P<modifiers>\|[a-z](?:'.NParser::RE_STRING.'|[^\'"])*)?
327:         ()\z~isx');
328: 
329:         if (!$match) {
330:             return FALSE;
331:         }
332:         $modifiers = preg_replace('#\|noescape\s?(?=\||\z)#i', '', $match['modifiers'], -1, $noescape);
333:         if ($match['name'] === '') {
334:             $match['name'] = ($tmp=$match['shortname']) ? $tmp : '=';
335:             if (!$noescape && !$match['noescape'] && substr($match['shortname'], 0, 1) !== '/') {
336:                 $modifiers .= '|escape';
337:             }
338:         }
339:         return array($match['name'], trim($match['args']), $modifiers);
340:     }
341: 
342: 
343:     private function addToken($type, $text)
344:     {
345:         $this->output[] = $token = new NLatteToken;
346:         $token->type = $type;
347:         $token->text = $text;
348:         $token->line = substr_count($this->input, "\n", 0, max(1, $this->offset - 1)) + 1;
349:         return $token;
350:     }
351: 
352: 
353:     /**
354:      * Process low-level macros.
355:      */
356:     protected function filter()
357:     {
358:         $token = end($this->output);
359:         if ($token->type === NLatteToken::MACRO_TAG && $token->name === '/syntax') {
360:             $this->setSyntax($this->defaultSyntax);
361:             $token->type = NLatteToken::COMMENT;
362: 
363:         } elseif ($token->type === NLatteToken::MACRO_TAG && $token->name === 'syntax') {
364:             $this->setSyntax($token->value);
365:             $token->type = NLatteToken::COMMENT;
366: 
367:         } elseif ($token->type === NLatteToken::HTML_ATTRIBUTE && $token->name === 'n:syntax') {
368:             $this->setSyntax($token->value);
369:             $this->syntaxEndTag = '/' . $this->lastHtmlTag;
370:             $token->type = NLatteToken::COMMENT;
371: 
372:         } elseif ($token->type === NLatteToken::HTML_TAG_END && $this->lastHtmlTag === $this->syntaxEndTag) {
373:             $this->setSyntax($this->defaultSyntax);
374: 
375:         } elseif ($token->type === NLatteToken::MACRO_TAG && $token->name === 'contentType') {
376:             if (preg_match('#html|xml#', $token->value, $m)) {
377:                 $this->xmlMode = $m[0] === 'xml';
378:                 $this->setContext(self::CONTEXT_TEXT);
379:             } else {
380:                 $this->setContext(self::CONTEXT_NONE);
381:             }
382:         }
383:     }
384: 
385: }
386: