[Data Liberation] Build markdown importer as phar (#2094)

Builds data-liberation-markdown.phar.gz (200KB) to enable downloading
the
Markdown importer only when needed instead of on every page load.

A part of:

* #2080
* #1894

 ## Testing instructions

Run `nx build playground-data-liberation-markdown`, confirm it finished
without errors. A smoke test of the built phar file is included in the
build command.

Author: adamziel

packages/playground/data-liberation-markdown/bin/build/DataLiberationBoxCompactor.php

@@ -0,0 +1,33 @@
+<?php
+
+use KevinGH\Box\Compactor\Compactor;
+
+class DataLiberationBoxCompactor implements Compactor
+{
+    /**
+     * {@inheritdoc}
+     */
+    public function compact(string $file, string $contents): string
+    {
+        if (!preg_match('/\.(php|json|lock)$/', $file)) {
+            return '';
+        }
+
+        if (
+            str_contains($file, 'platform_check.php') ||
+            str_contains($file, '/tests/') ||
+            str_contains($file, '/.git/') ||
+            str_contains($file, '/.github/') ||
+            str_contains($file, '/bin/')
+        ) {
+            return '';
+        }
+
+        if( str_contains($contents, 'Your Composer dependencies require ') ) {
+            return '';
+        }
+
+
+        return $contents;
+    }
+}

packages/playground/data-liberation-markdown/bin/build/box.php

@@ -0,0 +1,6 @@
+<?php
+
+$box_base_path = dirname(getenv('BOX_BASE_PATH'));
+require_once $box_base_path . '/../autoload.php';
+require_once __DIR__ . '/DataLiberationBoxCompactor.php';
+require_once $box_base_path . '/box';

packages/playground/data-liberation-markdown/bin/build/core.php

@@ -0,0 +1,2 @@
+<?php
+require_once __DIR__ . '/../../bootstrap.php';

packages/playground/data-liberation-markdown/bin/build/smoke-test.php

@@ -0,0 +1,23 @@
+<?php
+
+require_once __DIR__ . '/../../../data-liberation/dist/data-liberation-core.phar.gz';
+require_once __DIR__ . '/../../dist/data-liberation-markdown.phar';
+
+/**
+ * None of this will actually try to parse a file or import
+ * any data. We're just making sure the importer can
+ * be created without throwing an exception.
+ */
+$markdown_root = __DIR__ . '/markdown-test-data';
+$c = WP_Markdown_Importer::create_for_markdown_directory(
+    $markdown_root,
+    array(
+        'source_site_url' => 'file://' . $markdown_root,
+        'local_markdown_assets_root' => $markdown_root,
+        'local_markdown_assets_url_prefix' => '@site/',
+    ),
+    $import['cursor'] ?? null
+);
+
+echo 'Markdown importer created!';
+

packages/playground/data-liberation-markdown/bin/build/truncate-composer-checks.php

@@ -0,0 +1,46 @@
+<?php
+
+$file = $argv[1];
+$phar = new Phar($file);
+$phar->startBuffering();
+
+
+/**
+ * Box, includes an autoloader with a fixed name in every build.
+ * However, we want to load two .phar files built with Box, not
+ * one. Unfortunately this yields an error:
+ *
+ *     Cannot declare class ComposerAutoloaderInitHumbugBox451
+ *
+ * Therefore, we're giving all the HumbugBox classes a unique suffix.
+ */
+$autoloadSuffix = substr(md5(__FILE__), 0, 8);
+foreach (new RecursiveIteratorIterator($phar) as $file) {
+    if(!$file->isFile()) {
+        continue;
+    }
+    $relativePath = $file->getPathname();
+    $relativePath = str_replace('phar://', '', $relativePath);
+    $relativePath = str_replace($phar->getPath().'/', '', $relativePath);
+    $contents = $file->getContent();
+    $updated_contents = $contents;
+    foreach([
+        'InitHumbugBox',
+    ] as $class) {
+        $updated_contents = str_replace($class, $class . $autoloadSuffix, $updated_contents);
+    }
+    if($updated_contents !== $contents) {
+        $phar[$relativePath] = $updated_contents;
+    }
+}
+
+/**
+ * Box, very annoyingly, force-adds a platform_check.php file
+ * into the final built .phar archive. The vendor libraries
+ * do work with a PHP version lower than 8.1 enforced by that
+ * platform_check.php file, so let's just truncate it.
+ */
+$phar['vendor/composer/platform_check.php'] = '';
+$phar['.box/bin/check-requirements.php'] = '';
+$phar->stopBuffering();
+

packages/playground/data-liberation-markdown/box.json

@@ -0,0 +1,12 @@
+{
+	"$schema": "https://raw.githubusercontent.com/box-project/box/refs/heads/main/res/schema.json",
+	"main": "src/bootstrap.php",
+	"output": "dist/data-liberation-markdown.phar",
+	"force-autodiscovery": true,
+	"compactors": [
+		"KevinGH\\Box\\Compactor\\Php",
+		"DataLiberationBoxCompactor"
+	],
+	"annotations": false,
+	"directories": ["src/", "vendor/"]
+}

packages/playground/data-liberation-markdown/composer.json

@@ -1,4 +1,5 @@
 {
+	"name": "wordpress/data-liberation-markdown",
 	"name": "wordpress/data-liberation-markdown",
 	"prefer-stable": true,
 	"require": {

packages/playground/data-liberation-markdown/dist/.gitkeep

@@ -0,0 +1,26 @@
+#!/bin/bash
+
+# Builds the standalone dist/core-data-liberation.phar.gz file meant for
+# use in the importWxr Blueprint step.
+#
+# This is a temporary measure until we have a canonical way of distributing,
+# versioning, and using the Data Liberation modules and their dependencies.
+# Possible solutions might include composer packages, WordPress plugins, or
+# tree-shaken zip files with each module and its composer deps.
+
+set -e
+echo "Building data liberation plugin"
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+DATA_LIBERATION_DIR=$SCRIPT_DIR
+BUILD_DIR=$DATA_LIBERATION_DIR/bin/build
+DIST_DIR=$DATA_LIBERATION_DIR/dist
+
+rm $DIST_DIR/* > /dev/null 2>&1 || true
+export BOX_BASE_PATH=$(type -a box | grep -v 'alias' | awk '{print $3}')
+php $BUILD_DIR/box.php compile -d $DATA_LIBERATION_DIR -c $DATA_LIBERATION_DIR/box.json
+php -d 'phar.readonly=0' $BUILD_DIR/truncate-composer-checks.php $DIST_DIR/data-liberation-markdown.phar
+php $BUILD_DIR/smoke-test.php
+PHP=8.0 bun $DATA_LIBERATION_DIR/../../php-wasm/cli/src/main.ts $BUILD_DIR/smoke-test.php
+cd $DIST_DIR
+gzip data-liberation-markdown.phar
+ls -sgh $DIST_DIR