From 05ae1b0d2684cea4eda664c5ea1a995cb9f0c4b9 Mon Sep 17 00:00:00 2001
From: Alexandre Alapetite <alexandre@alapetite.fr>
Date: Thu, 9 Feb 2023 13:57:20 +0100
Subject: XML+XPath (#5076)

* XML+XPath
#fix https://github.com/FreshRSS/FreshRSS/issues/5075
Implementation allowing to take an XML document as input using an XML parser (instead of an HTML parser for HTML+XPath)

* Remove noise from another PR

* Better MIME for XML

* And add glob *.xml for cache cleaning

* Minor syntax

* Add glob json for clean cache
---
 app/Models/Feed.php | 29 ++++++++++++++++++++++++-----
 1 file changed, 24 insertions(+), 5 deletions(-)

(limited to 'app/Models/Feed.php')

diff --git a/app/Models/Feed.php b/app/Models/Feed.php
index f7ff76768..7c46199a5 100644
--- a/app/Models/Feed.php
+++ b/app/Models/Feed.php
@@ -17,6 +17,11 @@ class FreshRSS_Feed extends Minz_Model {
 	 * @var int
 	 */
 	const KIND_HTML_XPATH = 10;
+	/**
+	 * Normal XML with XPath scraping
+	 * @var int
+	 */
+	const KIND_XML_XPATH = 15;
 	/**
 	 * Normal JSON with XPath scraping
 	 * @var int
@@ -586,7 +591,7 @@ class FreshRSS_Feed extends Minz_Model {
 	/**
 	 * @return SimplePie|null
 	 */
-	public function loadHtmlXpath(bool $loadDetails = false, bool $noCache = false) {
+	public function loadHtmlXpath() {
 		if ($this->url == '') {
 			return null;
 		}
@@ -614,8 +619,9 @@ class FreshRSS_Feed extends Minz_Model {
 			return null;
 		}
 
-		$cachePath = FreshRSS_Feed::cacheFilename($feedSourceUrl, $this->attributes(), FreshRSS_Feed::KIND_HTML_XPATH);
-		$html = httpGet($feedSourceUrl, $cachePath, 'html', $this->attributes());
+		$cachePath = FreshRSS_Feed::cacheFilename($feedSourceUrl, $this->attributes(), $this->kind());
+		$html = httpGet($feedSourceUrl, $cachePath,
+			$this->kind() === FreshRSS_Feed::KIND_XML_XPATH ? 'xml' : 'html', $this->attributes());
 		if (strlen($html) <= 0) {
 			return null;
 		}
@@ -630,7 +636,18 @@ class FreshRSS_Feed extends Minz_Model {
 			$doc = new DOMDocument();
 			$doc->recover = true;
 			$doc->strictErrorChecking = false;
-			$doc->loadHTML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
+
+			switch ($this->kind()) {
+				case FreshRSS_Feed::KIND_HTML_XPATH:
+					$doc->loadHTML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
+					break;
+				case FreshRSS_Feed::KIND_XML_XPATH:
+					$doc->loadXML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
+					break;
+				default:
+					return null;
+			}
+
 			$xpath = new DOMXPath($doc);
 			$view->rss_title = $xPathFeedTitle == '' ? $this->name() :
 				htmlspecialchars(@$xpath->evaluate('normalize-space(' . $xPathFeedTitle . ')'), ENT_COMPAT, 'UTF-8');
@@ -776,8 +793,10 @@ class FreshRSS_Feed extends Minz_Model {
 	public static function cacheFilename(string $url, array $attributes, int $kind = FreshRSS_Feed::KIND_RSS): string {
 		$simplePie = customSimplePie($attributes);
 		$filename = $simplePie->get_cache_filename($url);
-		if ($kind == FreshRSS_Feed::KIND_HTML_XPATH) {
+		if ($kind === FreshRSS_Feed::KIND_HTML_XPATH) {
 			return CACHE_PATH . '/' . $filename . '.html';
+		} elseif ($kind === FreshRSS_Feed::KIND_XML_XPATH) {
+			return CACHE_PATH . '/' . $filename . '.xml';
 		} else {
 			return CACHE_PATH . '/' . $filename . '.spc';
 		}
-- 
cgit v1.2.3