Add support for intelligent recognition of LaTeX and custom internal hosts config

.github/workflows/release.yml

@@ -29,6 +29,7 @@ jobs:
           mkdir temp
           cp -r vendor temp/
           cp composer.json temp/
+          cp MarkdownParse.php temp/
           php -d phar.readonly=0 vendor/bin/phar-composer build temp vendor.phar
 
       - name: Package Files
@@ -37,7 +38,6 @@ jobs:
           mkdir MarkdownParse
           mv vendor.phar MarkdownParse/
           cp LICENSE.md MarkdownParse/
-          cp MarkdownParse.php MarkdownParse/
           cp Plugin.php MarkdownParse/
           cp README.md MarkdownParse/
           zip -r MarkdownParse.zip MarkdownParse

MarkdownParse.php

@@ -2,11 +2,7 @@
 
 namespace TypechoPlugin\MarkdownParse;
 
-if (file_exists(__DIR__ . '/vendor/autoload.php')) {
-    require_once __DIR__ . '/vendor/autoload.php';
-} elseif (file_exists(__DIR__ . 'vendor.phar')) {
-    require_once 'phar://' . __DIR__ . '/vendor.phar/vendor/autoload.php';
-}
+require_once __DIR__ . '/vendor/autoload.php';
 
 use League\CommonMark\Environment\Environment;
 use League\CommonMark\Extension\Autolink\AutolinkExtension;
@@ -38,6 +34,9 @@ class MarkdownParse
     // Flag to determine if LaTex support is needed
     private bool $isNeedLaTex = false;
 
+    // The internal hosts, supports regular expressions ("/(^|\.)example\.com$/"), multiple values can be separated by commas.
+    private string $internalHosts = '';
+
     // Singleton instance of MarkdownParse
     private static ?MarkdownParse $instance = null;
 
@@ -74,11 +73,66 @@ public static function getInstance(): MarkdownParse
      */
     public function parse(string $text, array $config = []): string
     {
+        list($text, $config) = $this->preParse($text, $config);
+
         $environment = new Environment(array_merge($this->getConfig(), $config));
 
         $this->addCommonMarkExtensions($environment);
 
-        return (new MarkdownConverter($environment))->convert($text)->getContent();
+        $htmlContent = (new MarkdownConverter($environment))->convert($text)->getContent();
+
+        list($htmlContent, $config) = $this->postParse($htmlContent, $config);
+
+        return $htmlContent;
+    }
+
+    /**
+     * Placeholder function for actions to be performed before parsing
+     *
+     * @param string $text The input text
+     * @param array $config Optional configuration for the parsing process
+     * @return array Result of actions before parsing
+     */
+    public function preParse(string $text, array $config = []): array
+    {
+        // Remove Table of Contents config if it is not enabled
+        if (!$this->isTocEnable) {
+            $config['table_of_contents']['placeholder'] = '';
+        }
+
+        // Set internal hosts for external link config
+        if (!empty($this->internalHosts)) {
+            $config['external_link']['internal_hosts'] = explode(',', $this->internalHosts);
+        }
+
+        // Check if LaTeX is needed by searching for $$ or $ in the text
+        if (!$this->isNeedLaTex) {
+            $this->isNeedLaTex = (bool) preg_match('/\${1,2}[^`]*?\${1,2}/m', $text);
+        }
+
+        // Replace double $$ at the beginning and end of the text with <div> tags
+        $count             = 0;
+        $text              = preg_replace('/(^\${2,})(.*)(\${2,}$)/ms', '<div>$1$2$3</div>', $text, -1, $count);
+        $this->isNeedLaTex = $this->isNeedLaTex || $count > 0;
+
+        return [$text, $config];
+    }
+
+    /**
+     * Placeholder function for actions to be performed after parsing
+     *
+     * @param string $htmlContent The parsed HTML content
+     * @param array $config Optional configuration for the parsing process
+     * @return array Result of actions after parsing
+     */
+    public function postParse(string $htmlContent, array $config = []): array
+    {
+        // If LaTeX is needed, remove <div> tags added during preParse
+        if ($this->isNeedLaTex) {
+            $htmlContent = str_replace(['<div>$$', '$$</div>'], '$$', $htmlContent);
+        }
+
+        return [$htmlContent, $config];
     }
 
     /**
@@ -96,7 +150,7 @@ public function getConfig(): array
                 'placeholder' => '[TOC]',
             ],
             'external_link' => [
-                'internal_hosts'     => ['foo.example.com', 'bar.example.com', '/(^|\.)google\.com$/'],
+                'internal_hosts'     => [],
                 'open_in_new_window' => true,
             ],
             'default_attributes' => [
@@ -113,11 +167,6 @@ public function getConfig(): array
             ]
         ];
 
-        // Remove Table of Contents config if it is not enabled
-        if (!$this->isTocEnable) {
-            $defaultConfig['table_of_contents']['placeholder'] = '';
-        }
-
         return $defaultConfig;
     }
 
@@ -203,4 +252,24 @@ public function setIsNeedLaTex(bool $isNeedLaTex): void
     {
         $this->isNeedLaTex = $isNeedLaTex;
     }
+
+    /**
+     * Get the internal hosts value
+     *
+     * @return string The internal hosts value
+     */
+    public function getInternalHosts(): string
+    {
+        return $this->internalHosts;
+    }
+
+    /**
+     * Set the internal hosts value
+     *
+     * @param string $internalHosts The internal hosts value
+     */
+    public function setInternalHosts(string $internalHosts): void
+    {
+        $this->internalHosts = $internalHosts;
+    }
 }

Plugin.php

@@ -2,6 +2,8 @@
 
 namespace TypechoPlugin\MarkdownParse;
 
+require_once 'phar://' . __DIR__ . '/vendor.phar/MarkdownParse.php';
+
 use Typecho\Plugin\PluginInterface;
 use Typecho\Widget\Helper\Form;
 use Widget\Options;
@@ -60,6 +62,9 @@ public static function config(Form $form)
         $elementCDNSource = new Form\Element\Radio('cdn_source', array_combine(array_keys(self::CDN_SOURCE_MERMAID), array_map('_t', array_keys(self::CDN_SOURCE_MERMAID))), self::CDN_SOURCE_DEFAULT);
         $form->addInput($elementCDNSource);
 
+        $elementInternalHosts = new Form\Element\Text('internal_hosts', null, parse_url(Options::alloc()->siteUrl(), PHP_URL_HOST), _t('设置内部链接'), _t('默认为本站点地址，支持正则表达式("/(^|\.)example\.com$/")，多个可用英文逗号分隔。外部链接解析策略：默认在新窗口中打开，并加上 "noopener noreferrer" 属性'));
+        $form->addInput($elementInternalHosts);
+
         $elementHelper = new Form\Element\Radio('show_help_info', [], self::RADIO_VALUE_DISABLE, _t('<a href="https://www.chengxiaobai.cn/php/markdown-parser-library.html/">点击查看更新信息</a>'), _t('<a href="https://www.chengxiaobai.cn/record/markdown-concise-grammar-manual.html/">点击查看语法手册</a>'));
         $form->addInput($elementHelper);
     }
@@ -71,9 +76,10 @@ public static function personalConfig(Form $form)
 
     public static function parse($text)
     {
-        $markdownParser = ParsedownExtension::getInstance();
+        $markdownParser = MarkdownParse::getInstance();
 
         $markdownParser->setIsTocEnable((bool)Options::alloc()->plugin('MarkdownParse')->is_available_toc);
+        $markdownParser->setInternalHosts((string)Options::alloc()->plugin('MarkdownParse')->internal_hosts);
 
         return $markdownParser->parse($text);
     }
@@ -83,12 +89,9 @@ public static function resourceLink()
         $configMermaid      = (int)Options::alloc()->plugin('MarkdownParse')->is_available_mermaid;
         $configLaTex        = (int)Options::alloc()->plugin('MarkdownParse')->is_available_mathjax;
         $configCDN          = (string)Options::alloc()->plugin('MarkdownParse')->cdn_source;
-        $markdownParser     = ParsedownExtension::getInstance();
+        $markdownParser     = MarkdownParse::getInstance();
         $isAvailableMermaid = $configMermaid === self::RADIO_VALUE_FORCE || ($markdownParser->getIsNeedMermaid() && $configMermaid === self::RADIO_VALUE_AUTO);
-
-        // @Todo performance for latex
-        // $isAvailableMathjax = $configLaTex === self::RADIO_VALUE_FORCE || ($markdownParser->getIsNeedLaTex() && $configLaTex === self::RADIO_VALUE_AUTO);
-        $isAvailableMathjax = $configLaTex === self::RADIO_VALUE_FORCE || $configLaTex === self::RADIO_VALUE_AUTO;
+        $isAvailableMathjax = $configLaTex   === self::RADIO_VALUE_FORCE || ($markdownParser->getIsNeedLaTex() && $configLaTex === self::RADIO_VALUE_AUTO);
 
         $resourceContent = '';
 

composer.json

@@ -1,5 +1,6 @@
 {
     "require": {
+        "php": "^8.0",
         "league/commonmark": "^2.4",
         "wnx/commonmark-mark-extension": "^1.2",
         "simonvomeyser/commonmark-ext-lazy-image": "^2.0",

tests/test.php

@@ -0,0 +1,11 @@
+<?php
+
+require_once 'phar://' . __DIR__ . '/../vendor.phar/MarkdownParse.php';
+
+use TypechoPlugin\MarkdownParse\MarkdownParse;
+
+$markdownParser = MarkdownParse::getInstance();
+
+$markdownParser->setIsTocEnable(true);
+
+echo $markdownParser->parse('Hello World!');