Rewrite ignore patterns to support globs

Author: john-shaffer

composer.json

@@ -30,7 +30,9 @@
         "ext-simplexml": "*",
         "lib-libxml": "*",
         "yahnis-elsts/plugin-update-checker": "^5.6",
-        "masterminds/html5": "^2.9"
+        "masterminds/html5": "^2.9",
+        "phlak/splat": "^5.0",
+        "symfony/finder": "=6.0.3"
     },
     "require-dev": {
         "thecodingmachine/phpstan-strict-rules": "*",

composer.lock

@@ -22,7 +22,7 @@
               let rel = baseNameOf path;
               in rel == "composer.json" || rel == "composer.lock";
           };
-          vendorHash = "sha256-4DTiW/rykJGZAnWCRSqlAaBtDthXKkZshPNe3yYCfA4=";
+          vendorHash = "sha256-dIGRVRFG7Rc1HhN0gCiYm1DgHsZAmc1KP490A26Gj2A=";
         });
         wp2static = runCommand "wp2static" {} ''
           export PLUGIN_DIR="$TMPDIR/${name}"

flake.nix

@@ -297,7 +297,7 @@ public static function optionSpecs() : array {
                 'filenamesToIgnore',
                 '1',
                 'Directory and File Names to Ignore',
-                'Directories and files with these names will be ignored.',
+                'Directories and files with these names will be ignored. Glob syntax is supported via <a href="https://github.com/PHLAK/Splat">Splat</a>.',
                 implode(
                     "\n",
                     [

src/CoreOptions.php

@@ -2,41 +2,50 @@
 
 namespace WP2Static;
 
+use PHLAK\Splat\Anchors;
+use PHLAK\Splat\Pattern;
 use WP2Static\CoreOptions;
+use WP2Static\FileIgnorePattern;
 use WP2Static\SiteInfo;
 
 class FileFiltering {
 
     /**
-     * @var array<string>
-     * File and directory names to ignore
+     * @var array<FileIgnorePattern>
+     * Files and directories to ignore
      */
-    private $filenames_to_ignore;
-
-    /**
-     * @var array<string>
-     * File extensions to ignore
-     */
-    private $file_extensions_to_ignore;
+    private $patterns_to_ignore;
 
     public function __construct() {
+        $this->patterns_to_ignore = [];
+
         $filenames_to_ignore = CoreOptions::getLineDelimitedBlobValue( 'filenamesToIgnore' );
 
-        $this->filenames_to_ignore =
+        $filenames_to_ignore =
             apply_filters(
                 'wp2static_filenames_to_ignore',
                 $filenames_to_ignore
             );
 
+        foreach ( $filenames_to_ignore as $filename ) {
+            $this->patterns_to_ignore[] = new FileIgnorePattern( $filename );
+        }
+
         $file_extensions_to_ignore = CoreOptions::getLineDelimitedBlobValue(
             'fileExtensionsToIgnore'
         );
 
-        $this->file_extensions_to_ignore =
+        $file_extensions_to_ignore =
             apply_filters(
                 'wp2static_file_extensions_to_ignore',
                 $file_extensions_to_ignore
             );
+
+        foreach ( $file_extensions_to_ignore as $extension ) {
+            $this->patterns_to_ignore[] = new FileIgnorePattern(
+                "**$extension"
+            );
+        }
     }
 
     /**
@@ -49,6 +58,8 @@ public function __construct() {
     public function crawlableFiles(
         string $directory,
     ) : \Iterator {
+        $abs_base_dir = ( new \SplFileInfo( $directory ) )->getPathname();
+
         $dir_iter = new \RecursiveDirectoryIterator(
             $directory,
             \RecursiveDirectoryIterator::SKIP_DOTS,
@@ -59,32 +70,14 @@ public function crawlableFiles(
         // blocked directories.
         $filter_iter = new \RecursiveCallbackFilterIterator(
             $dir_iter,
-            function ( $current, $key, $iterator ) {
-                $filename = $current->getFilename();
-
+            function ( $current, $key, $iterator ) use ( $abs_base_dir ) {
                 // Filter out both directories and files
-                foreach ( $this->filenames_to_ignore as $filename_to_ignore ) {
-                    if ( $filename === $filename_to_ignore ) {
+                foreach ( $this->patterns_to_ignore as $pattern ) {
+                    if ( $pattern->matches( $abs_base_dir, $current ) ) {
                         return false;
                     }
                 }
 
-                // Filter only files
-                if ( $current->isFile() ) {
-                    /*
-                      Prepare the file extension list for regex:
-                      - Add prepending (escaped) \ for a literal . at the start of
-                        the file extension
-                      - Add $ at the end to match end of string
-                      - Add i modifier for case insensitivity
-                    */
-                    foreach ( $this->file_extensions_to_ignore as $extension ) {
-                        if ( preg_match( "/\\{$extension}$/i", $filename ) ) {
-                            return false;
-                        }
-                    }
-                }
-
                 return true;
             }
         );
@@ -118,33 +111,14 @@ public function getListOfLocalFilesByDir(
     }
 
     /**
-     * Ensure a given filepath has an allowed filename and extension.
-     *
-     * @param string $file_name
-     * @return bool  True if the given file does not have a disallowed filename
-     *               or extension.
+     * @param string $path
+     * @return bool  True if the given path does not match an ignore pattern
      */
     public function pathLooksCrawlable(
-        string $file_name,
+        string $path,
     ) : bool {
-        $filename_matches = 0;
-
-        str_ireplace( $this->filenames_to_ignore, '', $file_name, $filename_matches );
-
-        // If we found matches we don't need to go any further
-        if ( $filename_matches ) {
-            return false;
-        }
-
-        /*
-          Prepare the file extension list for regex:
-          - Add prepending (escaped) \ for a literal . at the start of
-            the file extension
-          - Add $ at the end to match end of string
-          - Add i modifier for case insensitivity
-        */
-        foreach ( $this->file_extensions_to_ignore as $extension ) {
-            if ( preg_match( "/\\{$extension}$/i", $file_name ) ) {
+        foreach ( $this->patterns_to_ignore as $pattern ) {
+            if ( $pattern->matchesPath( $path ) ) {
                 return false;
             }
         }

src/FileFiltering.php

@@ -0,0 +1,81 @@
+<?php
+
+namespace WP2Static;
+
+use PHLAK\Splat\Pattern;
+use WP2Static\Utils;
+use WP2Static\WsLog;
+
+class FileIgnorePattern {
+    /**
+     * @var bool
+     * Match only directories and not files.
+     */
+    private $only_directories;
+
+    /**
+     * @var string
+     * Regex tested against file and directory paths.
+     */
+    private $regex;
+
+    /**
+     * Following gitignore rules:
+     * - Patterns ending in / match only directories.
+     * - Patterns with a / at the start or middle match
+     *   from the site root only.
+     *   - Other patterns can match starting from any subdirectory.
+     *
+     * Splat handles the rest of the logic of turning glob
+     * patterns into regexes.
+     */
+    public function __construct(
+        string $pattern,
+    ) {
+        if ( substr( $pattern, -1 ) === '/' ) {
+            $this->only_directories = true;
+            $pattern = substr( $pattern, 0, -1 );
+        } else {
+            $this->only_directories = false;
+        }
+
+        if ( strpos( $pattern, '/' ) === false ) {
+            $pattern = '**/' . $pattern;
+        } else if ( substr( $pattern, 0, 1 ) !== '/' ) {
+            $pattern = '/' . $pattern;
+        }
+
+        $regex = Pattern::make( $pattern )->toRegex( Pattern::BOTH_ANCHORS );
+        // Make it case-insensitive
+        $this->regex = $regex . 'i';
+
+        //WsLog::l("Created regex $this->regex for pattern $pattern");
+    }
+
+    public function matches(
+         string $abs_base_dir,
+         \SplFileInfo $file,
+    ) : bool {
+        if ( $this->only_directories && !$file->isDir() ) {
+            return false;
+        }
+
+        $path = $file->getPathname();
+        $path = Utils::str_replace_first( $abs_base_dir, '', $path );
+
+        return self::matchesPath( $path );
+    }
+
+    public function matchesPath(
+        string $path,
+   ) : bool {
+       if ( preg_match( $this->regex, $path ) ) {
+           //WsLog::l(
+           //    "Ignoring $path with regex $this->regex"
+           //);
+           return true;
+       }
+
+       return false;
+   }
+}