API encoding tuning (#2120)

Use only minimal XML->Unicode encoding for articles title.
Follow-up of https://github.com/FreshRSS/FreshRSS/pull/2093

3 files changed, 18 insertions, 13 deletions

diff --git a/app/Models/Feed.php b/app/Models/Feed.php
index a5ef33d6b..acf3bd981 100644
--- a/app/Models/Feed.php
+++ b/app/Models/Feed.php
@@ -424,7 +424,7 @@ class FreshRSS_Feed extends Minz_Model {
 			$author_names = '';
 			if (is_array($authors)) {
 				foreach ($authors as $author) {
-					$author_names .= escapeToUnicodeAlternative(strip_tags($author->name == '' ? $author->email : $author->name)) . '; ';
+					$author_names .= escapeToUnicodeAlternative(strip_tags($author->name == '' ? $author->email : $author->name), true) . '; ';
 				}
 			}
 			$author_names = substr($author_names, 0, -2);
diff --git a/lib/lib_rss.php b/lib/lib_rss.php
index 52e4408d2..c445874c8 100644
--- a/lib/lib_rss.php
+++ b/lib/lib_rss.php
@@ -102,16 +102,21 @@ function safe_ascii($text) {
 	return filter_var($text, FILTER_DEFAULT, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_STRIP_HIGH);
 }
 
-function escapeToUnicodeAlternative($text) {
+function escapeToUnicodeAlternative($text, $extended = true) {
 	$text = htmlspecialchars_decode($text, ENT_QUOTES);
+
+	//Problematic characters
+	$problem = array('&', '<', '>');
+	//Use their fullwidth Unicode form instead:
+	$replace = array('＆', '＜', '＞');
+
 	// https://raw.githubusercontent.com/mihaip/google-reader-api/master/wiki/StreamId.wiki
-	return trim(str_replace(
-			//Problematic characters
-			array("'", '"', '^', '<', '>', '?', '&', '\\', '/', ',', ';'),
-			//Use their fullwidth Unicode form instead:
-			array("’", '＂', '＾', '＜', '＞', '？', '＆', '＼', '／', '，', '；'),
-			$text
-		));
+	if ($extended) {
+		$problem += array("'", '"', '^', '?', '\\', '/', ',', ';');
+		$replace += array("’", '＂', '＾', '？', '＼', '／', '，', '；');
+	}
+
+	return trim(str_replace($problem, $replace, $text));
 }
 
 /**
diff --git a/p/api/greader.php b/p/api/greader.php
index 7c5c54951..7cd312f2c 100644
--- a/p/api/greader.php
+++ b/p/api/greader.php
@@ -300,7 +300,7 @@ function subscriptionList() {
 	foreach ($res as $line) {
 		$subscriptions[] = array(
 			'id' => 'feed/' . $line['id'],
-			'title' => escapeToUnicodeAlternative($line['name']),
+			'title' => escapeToUnicodeAlternative($line['name'], true),
 			'categories' => array(
 				array(
 					'id' => 'user/-/label/' . htmlspecialchars_decode($line['c_name'], ENT_QUOTES),
@@ -506,7 +506,7 @@ function entriesToArray($entries) {
 			'crawlTimeMsec' => substr($entry->id(), 0, -3),
 			'timestampUsec' => '' . $entry->id(),	//EasyRSS
 			'published' => $entry->date(true),
-			'title' => escapeToUnicodeAlternative($entry->title()),
+			'title' => escapeToUnicodeAlternative($entry->title(), false),
 			'summary' => array('content' => $entry->content()),
 			'alternate' => array(
 				array('href' => htmlspecialchars_decode($entry->link(), ENT_QUOTES)),
@@ -517,14 +517,14 @@ function entriesToArray($entries) {
 			),
 			'origin' => array(
 				'streamId' => 'feed/' . $f_id,
-				'title' => escapeToUnicodeAlternative($f_name),	//EasyRSS
+				'title' => escapeToUnicodeAlternative($f_name, true),	//EasyRSS
 				//'htmlUrl' => $line['f_website'],
 			),
 		);
 		$author = $entry->authors(true);
 		$author = trim($author, '; ');
 		if ($author != '') {
-			$item['author'] = escapeToUnicodeAlternative($author);
+			$item['author'] = escapeToUnicodeAlternative($author, false);
 		}
 		if ($entry->isRead()) {
 			$item['categories'][] = 'user/-/state/com.google/read';