Merge branch 'ncmpcpp:master' into master

CHANGELOG.md

@@ -1,3 +1,7 @@
+# ncmpcpp-0.10.2 (2025-??-??)
+* Update lyrics fetchers.
+* Add support for hexadecimal HTML escape codes.
+
 # ncmpcpp-0.10.1 (2024-10-24)
 * Fix compilation with `libc++`.
 * Remove `autogen.sh` in favour of `autoreconf`.

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([ncmpcpp],[0.10.1])
+AC_INIT([ncmpcpp],[0.10.2_dev])
 AC_CONFIG_SRCDIR([configure.ac])
 AC_CONFIG_HEADERS(config.h)
 AM_INIT_AUTOMAKE([subdir-objects])

src/lyrics_fetcher.h

@@ -84,7 +84,7 @@ struct GeniusFetcher : public GoogleLyricsFetcher
 	virtual const char *name() const override { return "genius.com"; }
 
 protected:
-	virtual const char *regex() const override { return "<div.*?class=\"(?:lyrics|Lyrics__Container).*?>(.*?)</div>"; }
+	virtual const char *regex() const override { return "<div data-lyrics-container.*?>(.*?)</div>"; }
 };
 
 struct JahLyricsFetcher : public GoogleLyricsFetcher
@@ -116,7 +116,7 @@ struct ZeneszovegFetcher : public GoogleLyricsFetcher
 	virtual const char *name() const override { return "zeneszoveg.hu"; }
 
 protected:
-	virtual const char *regex() const override { return "<div id=\"tartalom_slide_content\"> (.*?)<style>"; }
+	virtual const char *regex() const override { return "<div class=\"lyrics-plain-text trans_original\">(.*?)</div>"; }
 };
 
 struct InternetLyricsFetcher : public GoogleLyricsFetcher

src/utility/html.cpp

@@ -20,16 +20,29 @@
 
 #include <algorithm>
 #include <boost/algorithm/string/replace.hpp>
+#include <cstdlib>
 #include "utility/html.h"
 
 std::string unescapeHtmlUtf8(const std::string &data)
 {
+	int base;
+	size_t offset;
 	std::string result;
 	for (size_t i = 0, j; i < data.length(); ++i)
 	{
 		if (data[i] == '&' && data[i+1] == '#' && (j = data.find(';', i)) != std::string::npos)
 		{
-			int n = atoi(&data.c_str()[i+2]);
+			if (data[i+2] == 'x')
+			{
+				offset = 3;
+				base = 16;
+			}
+			else
+			{
+				offset = 2;
+				base = 10;
+			}
+			int n = strtol(&data.c_str()[i+offset], nullptr, base);
 			if (n >= 0x800)
 			{
 				result += (0xe0 | ((n >> 12) & 0x0f));