From 56cc64d1cffc156691f2724311a684bfe1f6d413 Mon Sep 17 00:00:00 2001
From: Uwe Steinmann <steinm@debian.org>
Date: Tue, 7 May 2024 13:51:21 +0200
Subject: [PATCH] add new conversion service from text/html to text/plain

---
 inc/inc.ClassConversionMgr.php               |  1 +
 inc/inc.ClassConversionServiceHtmlToText.php | 51 ++++++++++++++++++++
 inc/inc.ConversionInit.php                   |  2 +
 3 files changed, 54 insertions(+)
 create mode 100644 inc/inc.ClassConversionServiceHtmlToText.php

diff --git a/inc/inc.ClassConversionMgr.php b/inc/inc.ClassConversionMgr.php
index b06430543..28e717c80 100644
--- a/inc/inc.ClassConversionMgr.php
+++ b/inc/inc.ClassConversionMgr.php
@@ -16,6 +16,7 @@ require_once("inc/inc.ClassConversionServiceImageToImage.php");
 require_once("inc/inc.ClassConversionServiceImageToText.php");
 require_once("inc/inc.ClassConversionServicePdfToImage.php");
 require_once("inc/inc.ClassConversionServiceTextToText.php");
+require_once("inc/inc.ClassConversionServiceHtmlToText.php");
 require_once("inc/inc.ClassConversionServiceTextToImage.php");
 
 /**
diff --git a/inc/inc.ClassConversionServiceHtmlToText.php b/inc/inc.ClassConversionServiceHtmlToText.php
new file mode 100644
index 000000000..c9342e564
--- /dev/null
+++ b/inc/inc.ClassConversionServiceHtmlToText.php
@@ -0,0 +1,51 @@
+<?php
+/**
+ * Implementation of conversion service class
+ *
+ * @category   DMS
+ * @package    SeedDMS
+ * @license    GPL 2
+ * @version    @version@
+ * @author     Uwe Steinmann <uwe@steinmann.cx>
+ * @copyright  Copyright (C) 2021 Uwe Steinmann
+ * @version    Release: @package_version@
+ */
+
+require_once("inc/inc.ClassConversionServiceBase.php");
+
+/**
+ * Implementation of conversion service class for text to text
+ *
+ * @category   DMS
+ * @package    SeedDMS
+ * @author     Uwe Steinmann <uwe@steinmann.cx>
+ * @copyright  Copyright (C) 2021 Uwe Steinmann
+ * @version    Release: @package_version@
+ */
+class SeedDMS_ConversionServiceHtmlToText extends SeedDMS_ConversionServiceBase {
+	public function __construct($from, $to) {
+		parent::__construct();
+		$this->from = $from;
+		$this->to = $to;
+	}
+
+	public function getInfo() {
+		return "Strip tags from document contents";
+	}
+
+	public function convert($infile, $target = null, $params = array()) {
+		$d = new DOMDocument;
+		$d->loadHTMLFile($infile);
+		$body = $d->getElementsByTagName('body')->item(0);
+		$str = '';
+		foreach($body->childNodes as $childNode) {
+			$str .= $d->saveHTML($childNode);
+		}
+		if($target) {
+			file_put_contents($target, strip_tags($str));
+			return true;
+		} else
+			return strip_tags($str);
+	}
+}
+
diff --git a/inc/inc.ConversionInit.php b/inc/inc.ConversionInit.php
index 46c034609..67014251a 100644
--- a/inc/inc.ConversionInit.php
+++ b/inc/inc.ConversionInit.php
@@ -45,6 +45,8 @@ $conversionmgr->addService(new SeedDMS_ConversionServiceTextToText('text/plain',
 $conversionmgr->addService(new SeedDMS_ConversionServiceTextToText('text/markdown', 'text/plain'))->setLogger($logger);
 $conversionmgr->addService(new SeedDMS_ConversionServiceTextToText('text/x-rst', 'text/plain'))->setLogger($logger);
 
+$conversionmgr->addService(new SeedDMS_ConversionServiceHtmlToText('text/html', 'text/plain'))->setLogger($logger);
+
 if(isset($GLOBALS['SEEDDMS_HOOKS']['initConversion'])) {
 	foreach($GLOBALS['SEEDDMS_HOOKS']['initConversion'] as $hookObj) {
 		if (method_exists($hookObj, 'getConversionServices')) {