<?php
/**
 * Tokenizes CSS code.
 *
 * @author    Greg Sherwood <gsherwood@squiz.net>
 * @copyright 2006-2015 Squiz Pty Ltd (ABN 77 084 670 600)
 * @license   https://github.com/squizlabs/PHP_CodeSniffer/blob/master/licence.txt BSD Licence
 */

namespace PHP_CodeSniffer\Tokenizers;

use PHP_CodeSniffer\Config;
use PHP_CodeSniffer\Exceptions\TokenizerException;
use PHP_CodeSniffer\Util;

class CSS extends PHP {


	/**
	 * Initialise the tokenizer.
	 *
	 * Pre-checks the content to see if it looks minified.
	 *
	 * @param string $content The content to tokenize,
	 * @param \PHP_CodeSniffer\Config $config The config data for the run.
	 * @param string $eolChar The EOL char used in the content.
	 *
	 * @return void
	 * @throws \PHP_CodeSniffer\Exceptions\TokenizerException If the file appears to be minified.
	 */
	public function __construct( $content, Config $config, $eolChar = '\n' ) {
		if ( $this->isMinifiedContent( $content, $eolChar ) === true ) {
			throw new TokenizerException( 'File appears to be minified and cannot be processed' );
		}

		parent::__construct( $content, $config, $eolChar );

	}//end __construct()


	/**
	 * Creates an array of tokens when given some CSS code.
	 *
	 * Uses the PHP tokenizer to do all the tricky work
	 *
	 * @param string $string The string to tokenize.
	 *
	 * @return array
	 */
	public function tokenize( $string ) {
		if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
			echo "\t*** START CSS TOKENIZING 1ST PASS ***" . PHP_EOL;
		}

		// If the content doesn't have an EOL char on the end, add one so
		// the open and close tags we add are parsed correctly.
		$eolAdded = false;
		if ( substr( $string, ( strlen( $this->eolChar ) * - 1 ) ) !== $this->eolChar ) {
			$string   .= $this->eolChar;
			$eolAdded = true;
		}

		$string = str_replace( '<?php', '^PHPCS_CSS_T_OPEN_TAG^', $string );
		$string = str_replace( '?>', '^PHPCS_CSS_T_CLOSE_TAG^', $string );
		$tokens = parent::tokenize( '<?php ' . $string . '?>' );

		$finalTokens    = [];
		$finalTokens[0] = [
			'code'    => T_OPEN_TAG,
			'type'    => 'T_OPEN_TAG',
			'content' => '',
		];

		$newStackPtr      = 1;
		$numTokens        = count( $tokens );
		$multiLineComment = false;
		for ( $stackPtr = 1; $stackPtr < $numTokens; $stackPtr ++ ) {
			$token = $tokens[ $stackPtr ];

			// CSS files don't have lists, breaks etc, so convert these to
			// standard strings early so they can be converted into T_STYLE
			// tokens and joined with other strings if needed.
			if ( $token['code'] === T_BREAK
			     || $token['code'] === T_LIST
			     || $token['code'] === T_DEFAULT
			     || $token['code'] === T_SWITCH
			     || $token['code'] === T_FOR
			     || $token['code'] === T_FOREACH
			     || $token['code'] === T_WHILE
			     || $token['code'] === T_DEC
			     || $token['code'] === T_NEW
			) {
				$token['type'] = 'T_STRING';
				$token['code'] = T_STRING;
			}

			$token['content'] = str_replace( '^PHPCS_CSS_T_OPEN_TAG^', '<?php', $token['content'] );
			$token['content'] = str_replace( '^PHPCS_CSS_T_CLOSE_TAG^', '?>', $token['content'] );

			if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
				$type    = $token['type'];
				$content = Util\Common::prepareForOutput( $token['content'] );
				echo "\tProcess token $stackPtr: $type => $content" . PHP_EOL;
			}

			if ( $token['code'] === T_BITWISE_XOR
			     && $tokens[ ( $stackPtr + 1 ) ]['content'] === 'PHPCS_CSS_T_OPEN_TAG'
			) {
				$content = '<?php';
				for ( $stackPtr += 3; $stackPtr < $numTokens; $stackPtr ++ ) {
					if ( $tokens[ $stackPtr ]['code'] === T_BITWISE_XOR
					     && $tokens[ ( $stackPtr + 1 ) ]['content'] === 'PHPCS_CSS_T_CLOSE_TAG'
					) {
						// Add the end tag and ignore the * we put at the end.
						$content  .= '?>';
						$stackPtr += 2;
						break;
					} else {
						$content .= $tokens[ $stackPtr ]['content'];
					}
				}

				if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
					echo "\t\t=> Found embedded PHP code: ";
					$cleanContent = Util\Common::prepareForOutput( $content );
					echo $cleanContent . PHP_EOL;
				}

				$finalTokens[ $newStackPtr ] = [
					'type'    => 'T_EMBEDDED_PHP',
					'code'    => T_EMBEDDED_PHP,
					'content' => $content,
				];

				$newStackPtr ++;
				continue;
			}//end if

			if ( $token['code'] === T_GOTO_LABEL ) {
				// Convert these back to T_STRING followed by T_COLON so we can
				// more easily process style definitions.
				$finalTokens[ $newStackPtr ] = [
					'type'    => 'T_STRING',
					'code'    => T_STRING,
					'content' => substr( $token['content'], 0, - 1 ),
				];
				$newStackPtr ++;
				$finalTokens[ $newStackPtr ] = [
					'type'    => 'T_COLON',
					'code'    => T_COLON,
					'content' => ':',
				];
				$newStackPtr ++;
				continue;
			}

			if ( $token['code'] === T_FUNCTION ) {
				// There are no functions in CSS, so convert this to a string.
				$finalTokens[ $newStackPtr ] = [
					'type'    => 'T_STRING',
					'code'    => T_STRING,
					'content' => $token['content'],
				];

				$newStackPtr ++;
				continue;
			}

			if ( $token['code'] === T_COMMENT
			     && substr( $token['content'], 0, 2 ) === '/*'
			) {
				// Multi-line comment. Record it so we can ignore other
				// comment tags until we get out of this one.
				$multiLineComment = true;
			}

			if ( $token['code'] === T_COMMENT
			     && $multiLineComment === false
			     && ( substr( $token['content'], 0, 2 ) === '//'
			          || $token['content'][0] === '#' )
			) {
				$content = ltrim( $token['content'], '#/' );

				// Guard against PHP7+ syntax errors by stripping
				// leading zeros so the content doesn't look like an invalid int.
				$leadingZero = false;
				if ( $content[0] === '0' ) {
					$content     = '1' . $content;
					$leadingZero = true;
				}

				$commentTokens = parent::tokenize( '<?php ' . $content . '?>' );

				// The first and last tokens are the open/close tags.
				array_shift( $commentTokens );
				array_pop( $commentTokens );

				if ( $leadingZero === true ) {
					$commentTokens[0]['content'] = substr( $commentTokens[0]['content'], 1 );
					$content                     = substr( $content, 1 );
				}

				if ( $token['content'][0] === '#' ) {
					// The # character is not a comment in CSS files, so
					// determine what it means in this context.
					$firstContent = $commentTokens[0]['content'];

					// If the first content is just a number, it is probably a
					// colour like 8FB7DB, which PHP splits into 8 and FB7DB.
					if ( ( $commentTokens[0]['code'] === T_LNUMBER
					       || $commentTokens[0]['code'] === T_DNUMBER )
					     && $commentTokens[1]['code'] === T_STRING
					) {
						$firstContent .= $commentTokens[1]['content'];
						array_shift( $commentTokens );
					}

					// If the first content looks like a colour and not a class
					// definition, join the tokens together.
					if ( preg_match( '/^[ABCDEF0-9]+$/i', $firstContent ) === 1
					     && $commentTokens[1]['content'] !== '-'
					) {
						array_shift( $commentTokens );
						// Work out what we trimmed off above and remember to re-add it.
						$trimmed                     = substr( $token['content'], 0, ( strlen( $token['content'] ) - strlen( $content ) ) );
						$finalTokens[ $newStackPtr ] = [
							'type'    => 'T_COLOUR',
							'code'    => T_COLOUR,
							'content' => $trimmed . $firstContent,
						];
					} else {
						$finalTokens[ $newStackPtr ] = [
							'type'    => 'T_HASH',
							'code'    => T_HASH,
							'content' => '#',
						];
					}
				} else {
					$finalTokens[ $newStackPtr ] = [
						'type'    => 'T_STRING',
						'code'    => T_STRING,
						'content' => '//',
					];
				}//end if

				$newStackPtr ++;

				array_splice( $tokens, $stackPtr, 1, $commentTokens );
				$numTokens = count( $tokens );
				$stackPtr --;
				continue;
			}//end if

			if ( $token['code'] === T_COMMENT
			     && substr( $token['content'], - 2 ) === '*/'
			) {
				// Multi-line comment is done.
				$multiLineComment = false;
			}

			$finalTokens[ $newStackPtr ] = $token;
			$newStackPtr ++;
		}//end for

		if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
			echo "\t*** END CSS TOKENIZING 1ST PASS ***" . PHP_EOL;
			echo "\t*** START CSS TOKENIZING 2ND PASS ***" . PHP_EOL;
		}

		// A flag to indicate if we are inside a style definition,
		// which is defined using curly braces.
		$inStyleDef = false;

		// A flag to indicate if an At-rule like "@media" is used, which will result
		// in nested curly brackets.
		$asperandStart = false;

		$numTokens = count( $finalTokens );
		for ( $stackPtr = 0; $stackPtr < $numTokens; $stackPtr ++ ) {
			$token = $finalTokens[ $stackPtr ];

			if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
				$type    = $token['type'];
				$content = Util\Common::prepareForOutput( $token['content'] );
				echo "\tProcess token $stackPtr: $type => $content" . PHP_EOL;
			}

			switch ( $token['code'] ) {
				case T_OPEN_CURLY_BRACKET:
					// Opening curly brackets for an At-rule do not start a style
					// definition. We also reset the asperand flag here because the next
					// opening curly bracket could be indeed the start of a style
					// definition.
					if ( $asperandStart === true ) {
						if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
							if ( $inStyleDef === true ) {
								echo "\t\t* style definition closed *" . PHP_EOL;
							}

							if ( $asperandStart === true ) {
								echo "\t\t* at-rule definition closed *" . PHP_EOL;
							}
						}

						$inStyleDef    = false;
						$asperandStart = false;
					} else {
						$inStyleDef = true;
						if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
							echo "\t\t* style definition opened *" . PHP_EOL;
						}
					}
					break;
				case T_CLOSE_CURLY_BRACKET:
					if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
						if ( $inStyleDef === true ) {
							echo "\t\t* style definition closed *" . PHP_EOL;
						}

						if ( $asperandStart === true ) {
							echo "\t\t* at-rule definition closed *" . PHP_EOL;
						}
					}

					$inStyleDef    = false;
					$asperandStart = false;
					break;
				case T_MINUS:
					// Minus signs are often used instead of spaces inside
					// class names, IDs and styles.
					if ( $finalTokens[ ( $stackPtr + 1 ) ]['code'] === T_STRING ) {
						if ( $finalTokens[ ( $stackPtr - 1 ) ]['code'] === T_STRING ) {
							$newContent = $finalTokens[ ( $stackPtr - 1 ) ]['content'] . '-' . $finalTokens[ ( $stackPtr + 1 ) ]['content'];

							if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
								echo "\t\t* token is a string joiner; ignoring this and previous token" . PHP_EOL;
								$old = Util\Common::prepareForOutput( $finalTokens[ ( $stackPtr + 1 ) ]['content'] );
								$new = Util\Common::prepareForOutput( $newContent );
								echo "\t\t=> token " . ( $stackPtr + 1 ) . " content changed from \"$old\" to \"$new\"" . PHP_EOL;
							}

							$finalTokens[ ( $stackPtr + 1 ) ]['content'] = $newContent;
							unset( $finalTokens[ $stackPtr ] );
							unset( $finalTokens[ ( $stackPtr - 1 ) ] );
						} else {
							$newContent = '-' . $finalTokens[ ( $stackPtr + 1 ) ]['content'];

							$finalTokens[ ( $stackPtr + 1 ) ]['content'] = $newContent;
							unset( $finalTokens[ $stackPtr ] );
						}
					} else if ( $finalTokens[ ( $stackPtr + 1 ) ]['code'] === T_LNUMBER ) {
						// They can also be used to provide negative numbers.
						if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
							echo "\t\t* token is part of a negative number; adding content to next token and ignoring *" . PHP_EOL;
							$content = Util\Common::prepareForOutput( $finalTokens[ ( $stackPtr + 1 ) ]['content'] );
							echo "\t\t=> token " . ( $stackPtr + 1 ) . " content changed from \"$content\" to \"-$content\"" . PHP_EOL;
						}

						$finalTokens[ ( $stackPtr + 1 ) ]['content'] = '-' . $finalTokens[ ( $stackPtr + 1 ) ]['content'];
						unset( $finalTokens[ $stackPtr ] );
					}//end if
					break;
				case T_COLON:
					// Only interested in colons that are defining styles.
					if ( $inStyleDef === false ) {
						break;
					}

					for ( $x = ( $stackPtr - 1 ); $x >= 0; $x -- ) {
						if ( isset( Util\Tokens::$emptyTokens[ $finalTokens[ $x ]['code'] ] ) === false ) {
							break;
						}
					}

					if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
						$type = $finalTokens[ $x ]['type'];
						echo "\t\t=> token $x changed from $type to T_STYLE" . PHP_EOL;
					}

					$finalTokens[ $x ]['type'] = 'T_STYLE';
					$finalTokens[ $x ]['code'] = T_STYLE;
					break;
				case T_STRING:
					if ( strtolower( $token['content'] ) === 'url' ) {
						// Find the next content.
						for ( $x = ( $stackPtr + 1 ); $x < $numTokens; $x ++ ) {
							if ( isset( Util\Tokens::$emptyTokens[ $finalTokens[ $x ]['code'] ] ) === false ) {
								break;
							}
						}

						// Needs to be in the format "url(" for it to be a URL.
						if ( $finalTokens[ $x ]['code'] !== T_OPEN_PARENTHESIS ) {
							continue 2;
						}

						// Make sure the content isn't empty.
						for ( $y = ( $x + 1 ); $y < $numTokens; $y ++ ) {
							if ( isset( Util\Tokens::$emptyTokens[ $finalTokens[ $y ]['code'] ] ) === false ) {
								break;
							}
						}

						if ( $finalTokens[ $y ]['code'] === T_CLOSE_PARENTHESIS ) {
							continue 2;
						}

						if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
							for ( $i = ( $stackPtr + 1 ); $i <= $y; $i ++ ) {
								$type    = $finalTokens[ $i ]['type'];
								$content = Util\Common::prepareForOutput( $finalTokens[ $i ]['content'] );
								echo "\tProcess token $i: $type => $content" . PHP_EOL;
							}

							echo "\t\t* token starts a URL *" . PHP_EOL;
						}

						// Join all the content together inside the url() statement.
						$newContent = '';
						for ( $i = ( $x + 2 ); $i < $numTokens; $i ++ ) {
							if ( $finalTokens[ $i ]['code'] === T_CLOSE_PARENTHESIS ) {
								break;
							}

							$newContent .= $finalTokens[ $i ]['content'];
							if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
								$content = Util\Common::prepareForOutput( $finalTokens[ $i ]['content'] );
								echo "\t\t=> token $i added to URL string and ignored: $content" . PHP_EOL;
							}

							unset( $finalTokens[ $i ] );
						}

						$stackPtr = $i;

						// If the content inside the "url()" is in double quotes
						// there will only be one token and so we don't have to do
						// anything except change its type. If it is not empty,
						// we need to do some token merging.
						$finalTokens[ ( $x + 1 ) ]['type'] = 'T_URL';
						$finalTokens[ ( $x + 1 ) ]['code'] = T_URL;

						if ( $newContent !== '' ) {
							$finalTokens[ ( $x + 1 ) ]['content'] .= $newContent;
							if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
								$content = Util\Common::prepareForOutput( $finalTokens[ ( $x + 1 ) ]['content'] );
								echo "\t\t=> token content changed to: $content" . PHP_EOL;
							}
						}
					} else if ( $finalTokens[ $stackPtr ]['content'][0] === '-'
					            && $finalTokens[ ( $stackPtr + 1 ) ]['code'] === T_STRING
					) {
						if ( isset( $finalTokens[ ( $stackPtr - 1 ) ] ) === true
						     && $finalTokens[ ( $stackPtr - 1 ) ]['code'] === T_STRING
						) {
							$newContent = $finalTokens[ ( $stackPtr - 1 ) ]['content'] . $finalTokens[ $stackPtr ]['content'] . $finalTokens[ ( $stackPtr + 1 ) ]['content'];

							if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
								echo "\t\t* token is a string joiner; ignoring this and previous token" . PHP_EOL;
								$old = Util\Common::prepareForOutput( $finalTokens[ ( $stackPtr + 1 ) ]['content'] );
								$new = Util\Common::prepareForOutput( $newContent );
								echo "\t\t=> token " . ( $stackPtr + 1 ) . " content changed from \"$old\" to \"$new\"" . PHP_EOL;
							}

							$finalTokens[ ( $stackPtr + 1 ) ]['content'] = $newContent;
							unset( $finalTokens[ $stackPtr ] );
							unset( $finalTokens[ ( $stackPtr - 1 ) ] );
						} else {
							$newContent = $finalTokens[ $stackPtr ]['content'] . $finalTokens[ ( $stackPtr + 1 ) ]['content'];

							$finalTokens[ ( $stackPtr + 1 ) ]['content'] = $newContent;
							unset( $finalTokens[ $stackPtr ] );
						}
					}//end if
					break;
				case T_ASPERAND:
					$asperandStart = true;
					if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
						echo "\t\t* at-rule definition opened *" . PHP_EOL;
					}
					break;
				default:
					// Nothing special to be done with this token.
					break;
			}//end switch
		}//end for

		// Reset the array keys to avoid gaps.
		$finalTokens = array_values( $finalTokens );
		$numTokens   = count( $finalTokens );

		// Blank out the content of the end tag.
		$finalTokens[ ( $numTokens - 1 ) ]['content'] = '';

		if ( $eolAdded === true ) {
			// Strip off the extra EOL char we added for tokenizing.
			$finalTokens[ ( $numTokens - 2 ) ]['content'] = substr(
				$finalTokens[ ( $numTokens - 2 ) ]['content'],
				0,
				( strlen( $this->eolChar ) * - 1 )
			);

			if ( $finalTokens[ ( $numTokens - 2 ) ]['content'] === '' ) {
				unset( $finalTokens[ ( $numTokens - 2 ) ] );
				$finalTokens = array_values( $finalTokens );
				$numTokens   = count( $finalTokens );
			}
		}

		if ( PHP_CODESNIFFER_VERBOSITY > 1 ) {
			echo "\t*** END CSS TOKENIZING 2ND PASS ***" . PHP_EOL;
		}

		return $finalTokens;

	}//end tokenize()


	/**
	 * Performs additional processing after main tokenizing.
	 *
	 * @return void
	 */
	public function processAdditional() {
		/*
			We override this method because we don't want the PHP version to
			run during CSS processing because it is wasted processing time.
		*/

	}//end processAdditional()


}//end class