Merge pull request #251 from ampproject/add/14-minify-html-transformer

src/Dom/Document.php

@@ -134,7 +134,7 @@ final class Document extends DOMDocument
 
     // Regex patterns used for securing and restoring the doctype node.
     const HTML_SECURE_DOCTYPE_IF_NOT_FIRST_PATTERN = '/(^[^<]*(?>\s*<!--[^>]*>\s*)+<)(!)(doctype)(\s+[^>]+?)(>)/i';
-    const HTML_RESTORE_DOCTYPE_PATTERN             = '/(^[^<]*(?>\s*<!--[^>]*>\s*)+<)'
+    const HTML_RESTORE_DOCTYPE_PATTERN             = '/(^[^<]*(?>\s*<!--[^>]*>\s*)*<)'
                                                      . '(!--amp-)(doctype)(\s+[^>]+?)(-->)/i';
 
     // Regex pattern used for removing Internet Explorer conditional comments.

src/Optimizer/Configuration/MinifyHtmlConfiguration.php

@@ -0,0 +1,141 @@
+<?php
+
+namespace AmpProject\Optimizer\Configuration;
+
+use AmpProject\Optimizer\Exception\InvalidConfigurationValue;
+use InvalidArgumentException;
+
+/**
+ * Configuration for the MinifyHtml transformer.
+ *
+ * @property bool   $minify                Whether minification is enabled.
+ * @property bool   $minifyAmpScript       Whether amp-script minification is enabled.
+ * @property bool   $minifyJSON            Whether JSON data minification is enabled.
+ * @property bool   $collapseWhitespace    Whether collapsing whitespace is enabled.
+ * @property bool   $removeComments        Whether comments should be removed.
+ * @property bool   $canCollapseWhitespace Whether whitespace can be collapsed.
+ * @property bool   $inBody                Whether the node is in the body.
+ * @property string $commentIgnorePattern  Regex pattern of comments to keep.
+ *
+ * @package ampproject/amp-toolbox
+ */
+final class MinifyHtmlConfiguration extends BaseTransformerConfiguration
+{
+
+    /**
+     * Whether minification is enabled.
+     *
+     * @var bool
+     */
+    const MINIFY = 'minify';
+
+    /**
+     * Whether amp-script minification is enabled.
+     *
+     * @var bool
+     */
+    const MINIFY_AMP_SCRIPT = 'minifyAmpScript';
+
+    /**
+     * Whether JSON data minification is enabled.
+     *
+     * @var bool
+     */
+    const MINIFY_JSON = 'minifyJSON';
+
+    /**
+     * Whether collapsing whitespace is enabled.
+     *
+     * @var bool
+     */
+    const COLLAPSE_WHITESPACE = 'collapseWhitespace';
+
+    /**
+     * Whether comments should be removed.
+     *
+     * @var bool
+     */
+    const REMOVE_COMMENTS = 'removeComments';
+
+    /**
+     * Regular expression pattern of comments to keep.
+     *
+     * @var string
+     */
+    const COMMENT_IGNORE_PATTERN = 'commentIgnorePattern';
+
+    /**
+     * Get the associative array of allowed keys and their respective default values.
+     *
+     * The array index is the key and the array value is the key's default value.
+     *
+     * @return array Associative array of allowed keys and their respective default values.
+     */
+    protected function getAllowedKeys()
+    {
+        return [
+            self::MINIFY                  => true,
+            self::MINIFY_AMP_SCRIPT       => false,
+            self::MINIFY_JSON             => true,
+            self::COLLAPSE_WHITESPACE     => true,
+            self::REMOVE_COMMENTS         => true,
+            self::COMMENT_IGNORE_PATTERN  => '',
+        ];
+    }
+
+    /**
+     * Validate an individual configuration entry.
+     *
+     * @param string $key   Key of the configuration entry to validate.
+     * @param mixed  $value Value of the configuration entry to validate.
+     * @return mixed Validated value.
+     */
+    protected function validate($key, $value)
+    {
+        switch ($key) {
+            case self::MINIFY:
+            case self::MINIFY_JSON:
+            case self::COLLAPSE_WHITESPACE:
+            case self::REMOVE_COMMENTS:
+                if (! is_bool($value)) {
+                    throw InvalidConfigurationValue::forInvalidSubValueType(
+                        self::class,
+                        $key,
+                        'boolean',
+                        is_object($value) ? get_class($value) : gettype($value)
+                    );
+                }
+                break;
+
+            case self::MINIFY_AMP_SCRIPT:
+                if (! is_bool($value)) {
+                    throw InvalidConfigurationValue::forInvalidSubValueType(
+                        self::class,
+                        $key,
+                        'boolean',
+                        is_object($value) ? get_class($value) : gettype($value)
+                    );
+                }
+
+                if ($value) {
+                    $message = 'Script minification feature is not implemented yet.';
+                    throw new InvalidArgumentException($message);
+                }
+                break;
+
+            case self::COMMENT_IGNORE_PATTERN:
+                if (! is_string($value)) {
+                    throw InvalidConfigurationValue::forInvalidSubValueType(
+                        self::class,
+                        self::COMMENT_IGNORE_PATTERN,
+                        'string',
+                        is_object($value) ? get_class($value) : gettype($value)
+                    );
+                }
+                $value = trim($value);
+                break;
+        }
+
+        return $value;
+    }
+}

src/Optimizer/DefaultConfiguration.php

@@ -3,6 +3,7 @@
 namespace AmpProject\Optimizer;
 
 use AmpProject\Optimizer\Configuration\AmpRuntimeCssConfiguration;
+use AmpProject\Optimizer\Configuration\MinifyHtmlConfiguration;
 use AmpProject\Optimizer\Configuration\OptimizeAmpBindConfiguration;
 use AmpProject\Optimizer\Configuration\OptimizeHeroImagesConfiguration;
 use AmpProject\Optimizer\Configuration\PreloadHeroImageConfiguration;
@@ -12,6 +13,7 @@
 use AmpProject\Optimizer\Exception\UnknownConfigurationClass;
 use AmpProject\Optimizer\Exception\UnknownConfigurationKey;
 use AmpProject\Optimizer\Transformer\AmpRuntimeCss;
+use AmpProject\Optimizer\Transformer\MinifyHtml;
 use AmpProject\Optimizer\Transformer\OptimizeAmpBind;
 use AmpProject\Optimizer\Transformer\OptimizeHeroImages;
 use AmpProject\Optimizer\Transformer\PreloadHeroImage;
@@ -49,6 +51,7 @@ class DefaultConfiguration implements Configuration
         PreloadHeroImage::class      => PreloadHeroImageConfiguration::class,
         RewriteAmpUrls::class        => RewriteAmpUrlsConfiguration::class,
         TransformedIdentifier::class => TransformedIdentifierConfiguration::class,
+        MinifyHtml::class            => MinifyHtmlConfiguration::class,
     ];
 
     /**

src/Optimizer/Error/InvalidJson.php

@@ -0,0 +1,37 @@
+<?php
+
+namespace AmpProject\Optimizer\Error;
+
+use AmpProject\Optimizer\Error;
+
+/**
+ * Optimizer error object for invalid JSON data.
+ *
+ * @package ampproject/amp-toolbox
+ */
+final class InvalidJson implements Error
+{
+    use ErrorProperties;
+
+    /**
+     * Instantiate an InvalidJson object after decoding JSON data.
+     *
+     * @return self
+     */
+    public static function fromLastErrorMsgAfterDecoding()
+    {
+        $errorMsg = 'Error decoding JSON: ' . json_last_error_msg();
+        return new self($errorMsg);
+    }
+
+    /**
+     * Instantiate an InvalidJson object after encoding JSON data.
+     *
+     * @return self
+     */
+    public static function fromLastErrorMsgAfterEncoding()
+    {
+        $errorMsg = 'Error encoding JSON: ' . json_last_error_msg();
+        return new self($errorMsg);
+    }
+}