Merge pull request #7799 from wallabag/export2md

Add Markdown export
2024-11-25 18:41:05 +00:00 · 2024-11-13 18:25:11 +01:00 · 2024-11-13 18:25:11 +01:00 · cdb44fe1f0
commit cdb44fe1f0
parent 2560826b41 2a382b15c1
9 changed files with 170 additions and 22 deletions
--- a/app/config/wallabag.yml
+++ b/app/config/wallabag.yml
@ -114,6 +114,10 @@ parameters:
            name: export_xml
            value: 1
            section: export
+        -
+            name: export_md
+            value: 1
+            section: export
        -
            name: import_with_redis
            value: 0
--- a/composer.json
+++ b/composer.json
@ -87,6 +87,7 @@
        "jms/serializer-bundle": "^5.4",
        "laminas/laminas-code": "^4.7.1",
        "lcobucci/jwt": "^4.3",
+        "league/html-to-markdown": "^5.1",
        "mgargano/simplehtmldom": "^1.5",
        "mnapoli/piwik-twig-extension": "^3.0",
        "monolog/monolog": "^2.9",
--- a/composer.lock
+++ b/composer.lock
@ -4,7 +4,7 @@
        "Read more about it at https://getcomposer.org/doc/01-basic-usage.md#installing-dependencies",
        "This file is @generated automatically"
    ],
-    "content-hash": "6407ed5fbd4b0973ed565c2a136d3a81",
+    "content-hash": "a45ce1bad60f024c66e17b6aca7ad88d",
    "packages": [
        {
            "name": "babdev/pagerfanta-bundle",
@ -5303,6 +5303,95 @@
            ],
            "time": "2023-01-02T13:28:00+00:00"
        },
+        {
+            "name": "league/html-to-markdown",
+            "version": "5.1.1",
+            "source": {
+                "type": "git",
+                "url": "https://github.com/thephpleague/html-to-markdown.git",
+                "reference": "0b4066eede55c48f38bcee4fb8f0aa85654390fd"
+            },
+            "dist": {
+                "type": "zip",
+                "url": "https://api.github.com/repos/thephpleague/html-to-markdown/zipball/0b4066eede55c48f38bcee4fb8f0aa85654390fd",
+                "reference": "0b4066eede55c48f38bcee4fb8f0aa85654390fd",
+                "shasum": ""
+            },
+            "require": {
+                "ext-dom": "*",
+                "ext-xml": "*",
+                "php": "^7.2.5 || ^8.0"
+            },
+            "require-dev": {
+                "mikehaertl/php-shellcommand": "^1.1.0",
+                "phpstan/phpstan": "^1.8.8",
+                "phpunit/phpunit": "^8.5 || ^9.2",
+                "scrutinizer/ocular": "^1.6",
+                "unleashedtech/php-coding-standard": "^2.7 || ^3.0",
+                "vimeo/psalm": "^4.22 || ^5.0"
+            },
+            "bin": [
+                "bin/html-to-markdown"
+            ],
+            "type": "library",
+            "extra": {
+                "branch-alias": {
+                    "dev-master": "5.2-dev"
+                }
+            },
+            "autoload": {
+                "psr-4": {
+                    "League\\HTMLToMarkdown\\": "src/"
+                }
+            },
+            "notification-url": "https://packagist.org/downloads/",
+            "license": [
+                "MIT"
+            ],
+            "authors": [
+                {
+                    "name": "Colin O'Dell",
+                    "email": "colinodell@gmail.com",
+                    "homepage": "https://www.colinodell.com",
+                    "role": "Lead Developer"
+                },
+                {
+                    "name": "Nick Cernis",
+                    "email": "nick@cern.is",
+                    "homepage": "http://modernnerd.net",
+                    "role": "Original Author"
+                }
+            ],
+            "description": "An HTML-to-markdown conversion helper for PHP",
+            "homepage": "https://github.com/thephpleague/html-to-markdown",
+            "keywords": [
+                "html",
+                "markdown"
+            ],
+            "support": {
+                "issues": "https://github.com/thephpleague/html-to-markdown/issues",
+                "source": "https://github.com/thephpleague/html-to-markdown/tree/5.1.1"
+            },
+            "funding": [
+                {
+                    "url": "https://www.colinodell.com/sponsor",
+                    "type": "custom"
+                },
+                {
+                    "url": "https://www.paypal.me/colinpodell/10.00",
+                    "type": "custom"
+                },
+                {
+                    "url": "https://github.com/colinodell",
+                    "type": "github"
+                },
+                {
+                    "url": "https://tidelift.com/funding/github/packagist/league/html-to-markdown",
+                    "type": "tidelift"
+                }
+            ],
+            "time": "2023-07-12T21:21:09+00:00"
+        },
        {
            "name": "masterminds/html5",
            "version": "2.9.0",
@ -20034,7 +20123,7 @@
        "ext-tokenizer": "*",
        "ext-xml": "*"
    },
-    "platform-dev": [],
+    "platform-dev": {},
    "platform-overrides": {
        "php": "7.4.29"
    },
--- a/migrations/Version20241112193044.php
+++ b/migrations/Version20241112193044.php
@ -0,0 +1,24 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Application\Migrations;
+
+use Doctrine\DBAL\Schema\Schema;
+use Wallabag\Doctrine\WallabagMigration;
+
+/**
+ * Added the internal setting to export articles in markdown.
+ */
+final class Version20241112193044 extends WallabagMigration
+{
+    public function up(Schema $schema): void
+    {
+        $this->addSql('INSERT INTO ' . $this->getTable('internal_setting') . " (name, value, section) VALUES ('export_md', '1', 'export');");
+    }
+
+    public function down(Schema $schema): void
+    {
+        $this->addSql('DELETE FROM' . $this->getTable('internal_setting') . " WHERE name = 'export_md';");
+    }
+}
--- a/src/Controller/ExportController.php
+++ b/src/Controller/ExportController.php
@ -20,7 +20,7 @@ class ExportController extends AbstractController
     * Gets one entry content.
     *
     * @Route("/export/{id}.{format}", name="export_entry", requirements={
-     *     "format": "epub|pdf|json|xml|txt|csv",
+     *     "format": "epub|pdf|json|xml|txt|csv|md",
     *     "id": "\d+"
     * })
     *
@ -54,7 +54,7 @@ class ExportController extends AbstractController
     * Export all entries for current user.
     *
     * @Route("/export/{category}.{format}", name="export_entries", requirements={
-     *     "format": "epub|pdf|json|xml|txt|csv",
+     *     "format": "epub|pdf|json|xml|txt|csv|md",
     *     "category": "all|unread|starred|archive|tag_entries|untagged|search|annotated|same_domain"
     * })
     *
--- a/src/Helper/EntriesExport.php
+++ b/src/Helper/EntriesExport.php
@ -5,6 +5,7 @@ namespace Wallabag\Helper;
 use Html2Text\Html2Text;
 use JMS\Serializer\SerializationContext;
 use JMS\Serializer\SerializerBuilder;
+use League\HTMLToMarkdown\HtmlConverter;
 use PHPePub\Core\EPub;
 use PHPePub\Core\Structure\OPF\DublinCore;
 use Symfony\Component\HttpFoundation\Response;
@ -129,10 +130,8 @@ class EntriesExport

    /**
     * Use PHPePub to dump a .epub file.
-     *
-     * @return Response
     */
-    private function produceEpub()
+    private function produceEpub(): Response
    {
        $user = $this->tokenStorage->getToken() ? $this->tokenStorage->getToken()->getUser() : null;
        \assert($user instanceof User);
@ -249,10 +248,8 @@ class EntriesExport

    /**
     * Use TCPDF to dump a .pdf file.
-     *
-     * @return Response
     */
-    private function producePdf()
+    private function producePdf(): Response
    {
        $user = $this->tokenStorage->getToken() ? $this->tokenStorage->getToken()->getUser() : null;
        \assert($user instanceof User);
@ -326,10 +323,8 @@ class EntriesExport

    /**
     * Inspired from CsvFileDumper.
-     *
-     * @return Response
     */
-    private function produceCsv()
+    private function produceCsv(): Response
    {
        $delimiter = ';';
        $enclosure = '"';
@ -372,10 +367,8 @@ class EntriesExport

    /**
     * Dump a JSON file.
-     *
-     * @return Response
     */
-    private function produceJson()
+    private function produceJson(): Response
    {
        return Response::create(
            $this->prepareSerializingContent('json'),
@ -390,10 +383,8 @@ class EntriesExport

    /**
     * Dump a XML file.
-     *
-     * @return Response
     */
-    private function produceXml()
+    private function produceXml(): Response
    {
        return Response::create(
            $this->prepareSerializingContent('xml'),
@ -408,10 +399,8 @@ class EntriesExport

    /**
     * Dump a TXT file.
-     *
-     * @return Response
     */
-    private function produceTxt()
+    private function produceTxt(): Response
    {
        $content = '';
        $bar = str_repeat('=', 100);
@ -432,6 +421,29 @@ class EntriesExport
        );
    }

+    /**
+     * Dump a Markdown file.
+     */
+    private function produceMd(): Response
+    {
+        $content = '';
+        $converter = new HtmlConverter();
+        $converter->getConfig()->setOption('strip_tags', true);
+        foreach ($this->entries as $entry) {
+            $content .= $converter->convert('<h1>' . $entry->getTitle() . '</h1>' . $entry->getContent());
+        }
+
+        return Response::create(
+            $content,
+            200,
+            [
+                'Content-type' => 'text/markdown',
+                'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.md"',
+                'Content-Transfer-Encoding' => 'UTF-8',
+            ]
+        );
+    }
+
    /**
     * Return a Serializer object for producing processes that need it (JSON & XML).
     *
--- a/templates/Entry/entries.html.twig
+++ b/templates/Entry/entries.html.twig
@ -105,6 +105,7 @@
            {% if craue_setting('export_csv') %}<li class="bold"><a class="waves-effect" href="{{ path('export_entries', {'category': current_route, 'format': 'csv', 'tag': current_tag, 'search_entry[term]': export_search_term, 'currentRoute': previous_route, 'entry': entry}) }}">CSV</a></li>{% endif %}
            {% if craue_setting('export_txt') %}<li class="bold"><a class="waves-effect" href="{{ path('export_entries', {'category': current_route, 'format': 'txt', 'tag': current_tag, 'search_entry[term]': export_search_term, 'currentRoute': previous_route, 'entry': entry}) }}">TXT</a></li>{% endif %}
            {% if craue_setting('export_xml') %}<li class="bold"><a class="waves-effect" href="{{ path('export_entries', {'category': current_route, 'format': 'xml', 'tag': current_tag, 'search_entry[term]': export_search_term, 'currentRoute': previous_route, 'entry': entry}) }}">XML</a></li>{% endif %}
+            {% if craue_setting('export_md') %}<li class="bold"><a class="waves-effect" href="{{ path('export_entries', {'category': current_route, 'format': 'md', 'tag': current_tag, 'search_entry[term]': export_search_term, 'currentRoute': previous_route, 'entry': entry}) }}">Markdown</a></li>{% endif %}
        </ul>
    </div>

--- a/templates/Entry/entry.html.twig
+++ b/templates/Entry/entry.html.twig
@ -242,6 +242,7 @@
                    {% if craue_setting('export_json') %}<li><a href="{{ path('export_entry', {'id': entry.id, 'format': 'json'}) }}" title="Generate JSON file">JSON</a></li>{% endif %}
                    {% if craue_setting('export_txt') %}<li><a href="{{ path('export_entry', {'id': entry.id, 'format': 'txt'}) }}" title="Generate TXT file">TXT</a></li>{% endif %}
                    {% if craue_setting('export_xml') %}<li><a href="{{ path('export_entry', {'id': entry.id, 'format': 'xml'}) }}" title="Generate XML file">XML</a></li>{% endif %}
+                    {% if craue_setting('export_md') %}<li><a href="{{ path('export_entry', {'id': entry.id, 'format': 'md'}) }}" title="Generate MD file">Markdown</a></li>{% endif %}
                </ul>
            </div>
        </li>
--- a/tests/Controller/ExportControllerTest.php
+++ b/tests/Controller/ExportControllerTest.php
@ -307,6 +307,22 @@ class ExportControllerTest extends WallabagTestCase
        $this->assertNotEmpty('updated_at', (string) $content->entry[0]->updated_at);
    }

+    public function testMdExport()
+    {
+        $this->logInAs('admin');
+        $client = $this->getTestClient();
+        $client->request('GET', '/export/all.md');
+
+        $this->assertSame(200, $client->getResponse()->getStatusCode());
+
+        $headers = $client->getResponse()->headers;
+        $content = $client->getResponse()->getContent();
+        $this->assertSame('text/markdown; charset=UTF-8', $headers->get('content-type'));
+        $this->assertSame('attachment; filename="All articles.md"', $headers->get('content-disposition'));
+        $this->assertSame('UTF-8', $headers->get('content-transfer-encoding'));
+        $this->assertStringContainsString('=================', $content);
+    }
+
    public function testJsonExportFromSameDomain()
    {
        $this->logInAs('admin');