/**
* Compile a string into a set of tokens.
*
* @param Mustache $mustache Mustache instance invoking compilation.
* @param string $string
* @param null|string $templateName Template to use in the case of a
* partial; can be null.
* @param array $scopedPragmas List of pragmas currently in scope; should
* only be used internally.
* @return array
* @throws Exception
*/
public function compile(Mustache $mustache, $string, $templateName = null, $scopedPragmas = [])
{
if (!is_string($string)) {
throw new Exception\InvalidTemplateException();
}
$len = strlen($string);
$delimStartLen = strlen($this->patterns[self::DS]);
$delimEndLen = strlen($this->patterns[self::DE]);
$state = self::STATE_CONTENT;
$pragmas = $mustache->getPragmas();
$tokens = [];
$content = '';
$tagData = '';
$rawTagData = '';
$sigilPattern = '/^[' . preg_quote('%=<>!&^#$') . ']/';
for ($i = 0; $i < $len;) {
switch ($state) {
case self::STATE_CONTENT:
$content .= $string[$i];
if (substr($content, -$delimStartLen) !== $this->patterns[self::DS]) {
// No start delimiter found in content yet; continue to next character.
$i += 1;
break;
}
// Create token for content
$tokens[] = $this->parseViaPragmas([self::TOKEN_CONTENT, substr($content, 0, -$delimStartLen)], $pragmas, $scopedPragmas);
$content = '';
// Switch to tag state
$state = self::STATE_TAG;
$tagData = '';
$i += 1;
break;
case self::STATE_TAG:
$tagData .= $string[$i];
if (substr($tagData, -$delimEndLen) !== $this->patterns[self::DE]) {
// Have not reached end of tag delimiter
$i += 1;
break;
}
// End of tag reached; start processing.
$tagData = substr($tagData, 0, -$delimEndLen);
// Whitespace check; whitespace not allowed before sigils
$rawTagData = $tagData;
$tagData = trim($tagData);
if (preg_match($sigilPattern, $tagData) && in_array($rawTagData[0], $this->whitespaceCharacters, true)) {
// Begins with a whitespace followed by sigil
throw new Exception\InvalidTemplateException('Whitespace is not allowed before sigils (#, ^, $, etc.)');
}
// Evaluate what kind of token we have
switch ($tagData[0]) {
case '#':
// Section start
$tagData = ltrim($tagData, '#');
$section = trim($tagData);
$sectionData = '';
$tokenType = self::TOKEN_SECTION;
$state = self::STATE_SECTION;
$i += 1;
break;
case '$':
// Placeholder start
//
// Placeholders are captured as a set of tokens. They are
// essentially a type of section.
//
// In a child template, any placeholders defined are then
// replaced with the tokens they contain.
$tagData = ltrim($tagData, '$');
$section = trim($tagData);
$sectionData = '';
$tokenType = self::TOKEN_PLACEHOLDER;
$state = self::STATE_SECTION;
$i += 1;
break;
case '^':
// Inverted section start
$tagData = ltrim($tagData, '^');
$section = trim($tagData);
$sectionData = '';
$tokenType = self::TOKEN_SECTION_INVERT;
$state = self::STATE_SECTION;
$i += 1;
break;
case '{':
// Raw value start (triple mustaches)
// Check that next character is a mustache; if
// not, we're basically still in the tag.
if ($i + 1 >= $len) {
// We've already reached the end of the string
$tagData .= $this->patterns[self::DE];
$i += 1;
break;
}
if ('}' !== $string[$i + 1]) {
// We don't have triple mustaches yet
$tagData .= $this->patterns[self::DE];
$i += 1;
break;
}
// Advance position by one
$i += 1;
// Create token
$tokens[] = $this->parseViaPragmas([self::TOKEN_VARIABLE_RAW, ltrim($tagData, '{')], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
$i += 1;
break;
case '&':
// Raw value start
$tagData = ltrim($tagData, '&');
$tagData = trim($tagData);
// Create token
$tokens[] = $this->parseViaPragmas([self::TOKEN_VARIABLE_RAW, $tagData], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
$i += 1;
break;
case '!':
// Comment
// Create token
$tokens[] = $this->parseViaPragmas([self::TOKEN_COMMENT, ltrim($tagData, '!')], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
$i += 1;
break;
case '>':
// Partial
// Trim the value of whitespace
$tagData = ltrim($tagData, '>');
$partial = trim($tagData);
// Create token
$tokens[] = $this->parseViaPragmas([self::TOKEN_PARTIAL, ['partial' => $partial]], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
$i += 1;
break;
case '<':
// Template inheritance
//
// Indicates that the content provides placeholders for
// the inherited template. The parent template is parsed,
// and then the content is parsed for placeholders. Any
// placeholders found are then used to replace placeholders
// of the same name in the parent template. The content
// is then replaced with the parent tokens.
//
// For purposes of first-pass lexing, it's a type of section.
$tagData = ltrim($tagData, '<');
$section = trim($tagData);
$sectionData = '';
$tokenType = self::TOKEN_CHILD;
$state = self::STATE_SECTION;
$i += 1;
break;
case '=':
// Delimiter set
if (!preg_match('/^=(\\S+)\\s+(\\S+)=$/', $tagData, $matches)) {
throw new Exception\InvalidDelimiterException('Did not find delimiters!');
}
$this->patterns[self::DS] = $delimStart = $matches[1];
$this->patterns[self::DE] = $delimEnd = $matches[2];
// Create token
$tokens[] = $this->parseViaPragmas([self::TOKEN_DELIM_SET, ['delim_start' => $delimStart, 'delim_end' => $delimEnd]], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
$i += 1;
break;
case '%':
// Pragmas
$data = ltrim($tagData, '%');
if (!preg_match('/^(?P<pragma>' . $this->patterns['pragma'] . ')( (?P<options>.*))?$/', $data, $matches)) {
throw new Exception\InvalidPragmaNameException();
}
$pragma = $matches['pragma'];
$options = [];
if (!empty($matches['options'])) {
$pairs = explode(' ', $matches['options']);
foreach ($pairs as $pair) {
if (!strstr($pair, '=')) {
$options[$pair] = null;
continue;
}
list($key, $value) = explode('=', $pair, 2);
$options[$key] = $value;
}
}
$tokens[] = $this->parseViaPragmas([self::TOKEN_PRAGMA, ['pragma' => $pragma, 'options' => $options]], $pragmas, $scopedPragmas);
$scopedPragmas[] = $pragma;
$state = self::STATE_CONTENT;
$i += 1;
break;
default:
// We have a simple variable replacement
// First, create the token, passing it to pragmas; this allows pragmas
// to filter the variable name if required.
$token = $this->parseViaPragmas([self::TOKEN_VARIABLE, $tagData], $pragmas, $scopedPragmas);
// Now filter the tag data (index 1 of the token) to ensure it is valid.
if (!preg_match($this->patterns[self::VARNAME], $token[1])) {
throw new Exception\InvalidVariableNameException(sprintf('Invalid variable name provided (%s)', $token[1]));
}
// Add the token to the list, and continue.
$tokens[] = $token;
$state = self::STATE_CONTENT;
$i += 1;
break;
}
break;
case self::STATE_SECTION:
$sectionData .= $string[$i];
$ds = $this->patterns[self::DS];
$de = $this->patterns[self::DE];
$pattern = $this->implodePregQuote('\\s*', [$ds, '/', $section, $de], '/');
if (preg_match('/' . $pattern . '$/', $sectionData)) {
// we have a match. Now, let's make sure we're balanced
$pattern = '/((' . $this->implodePregQuote('\\s*', [$ds . '#', $section, $de], '/') . ')|(' . $this->implodePregQuote('\\s*', [$ds . '/', $section, $de], '/') . '))/';
preg_match_all($pattern, $sectionData, $matches);
$open = 0;
$closed = 0;
foreach ($matches[3] as $match) {
if ('' === $match) {
$open += 1;
} else {
$closed += 1;
}
}
if ($closed > $open) {
// We're balanced if we have 1 more end tag then start tags
$endTag = end($matches[3]);
$sectionData = substr($sectionData, 0, strlen($sectionData) - strlen($endTag));
// compile sections later
$tokens[] = $this->parseViaPragmas([$tokenType, ['name' => $section, 'template' => $sectionData]], $pragmas, $scopedPragmas);
$state = self::STATE_CONTENT;
}
}
// Increment pointer
$i += 1;
break;
default:
throw new Exception\InvalidStateException(sprintf('Invalid state invoked ("%s")?', var_export($state, 1)));
}
}
// Create last token
switch ($state) {
case self::STATE_CONTENT:
// Un-collected content
$tokens[] = $this->parseViaPragmas([self::TOKEN_CONTENT, $content], $pragmas, $scopedPragmas);
break;
case self::STATE_TAG:
// Un-closed content
throw new Exception\UnbalancedTagException();
case self::STATE_SECTION:
// Un-closed section
throw new Exception\UnbalancedSectionException('Unbalanced section, placeholder, or inheritance in template');
default:
// Do nothing.
}
// Tokenize any partials, sections, placeholders, or child templates
// discovered, strip whitespaces as necessary
$replaceKeys = [];
foreach ($tokens as $key => $token) {
$type = $token[0];
switch ($type) {
case self::TOKEN_PARTIAL:
// If the partial name is the same as the template name
// provided, we should skip processing (prevents
// recursion).
if ($token[1]['partial'] === $templateName) {
break;
}
// First, reset the delimiters
$delimStart = $this->patterns[self::DS];
$delimEnd = $this->patterns[self::DE];
$this->patterns[self::DS] = self::DEFAULT_DELIM_START;
$this->patterns[self::DE] = self::DEFAULT_DELIM_END;
// Tokenize the partial
$partial = $token[1]['partial'];
$partialTokens = $mustache->tokenize($partial);
// Restore the delimiters
$this->patterns[self::DS] = $delimStart;
$this->patterns[self::DE] = $delimEnd;
$token[1]['tokens'] = $partialTokens;
$tokens[$key] = $token;
break;
case self::TOKEN_CHILD:
// We need to compile the content (compile($token[1]['template'])).
// Once done, we determine what placeholders were in the content.
$delimStart = $this->patterns[self::DS];
$delimEnd = $this->patterns[self::DE];
$child = $this->compile($mustache, $token[1]['template'], $templateName);
// Reset delimiters to retain scope
$this->patterns[self::DS] = $delimStart;
$this->patterns[self::DE] = $delimEnd;
// Get placeholders from child
$placeholders = [];
foreach ($child as $childToken) {
$childType = $childToken[0];
if ($childType !== self::TOKEN_PLACEHOLDER) {
continue;
}
$placeholders[$childToken[1]['name']] = $childToken[1]['content'];
}
// Now, tokenize the parent
$this->nestingLevel += 1;
$this->placeholders[$this->nestingLevel] = $placeholders;
$parent = $mustache->tokenize($token[1]['name'], false);
unset($this->placeholders[$this->nestingLevel]);
$this->nestingLevel -= 1;
// At this point, we hint that we need to remove the
// previous token, and inject the tokens of the parent in
// sequence.
$replaceKeys[$key] = $parent;
break;
case self::TOKEN_PLACEHOLDER:
if ($this->nestingLevel > 0) {
$placeholders = [];
foreach ($this->placeholders as $childPlaceholders) {
// 1 is deepest, and thus has precedence
$placeholders = array_merge($childPlaceholders, $placeholders);
}
$placeholder = $token[1]['name'];
if (isset($placeholders[$placeholder])) {
$token[1]['content'] = $placeholders[$placeholder];
$tokens[$key] = $token;
// Break out of the case; we had a replacement
break;
}
}
// intentionally fall through in the case of no nesting level
// intentionally fall through in the case of no nesting level
case self::TOKEN_SECTION:
case self::TOKEN_SECTION_INVERT:
$delimStart = $this->patterns[self::DS];
$delimEnd = $this->patterns[self::DE];
$token[1]['content'] = $this->compile($mustache, $token[1]['template'], $templateName, $scopedPragmas);
$tokens[$key] = $token;
// Reset delimiters to retain scope
$this->patterns[self::DS] = $delimStart;
$this->patterns[self::DE] = $delimEnd;
// Clean whitespace
if (!$this->disableStripWhitespace()) {
$this->stripWhitespace($tokens, $key);
}
break;
case self::TOKEN_DELIM_SET:
if (!$this->disableStripWhitespace()) {
$this->stripWhitespace($tokens, $key);
}
break;
default:
// do nothing
}
}
if (count($replaceKeys)) {
$tokens = $this->replaceTokens($tokens, $replaceKeys);
}
return $tokens;
}