From 6cd3e04abbbc7d9ab554dea079b9cb3a171adcc4 Mon Sep 17 00:00:00 2001
From: Rich Hodgkins <rhodgkins@gmail.com>
Date: Fri, 29 Apr 2022 02:25:51 +0100
Subject: [PATCH] Added encoding for toString() method (#482)

* Failing test case for stripped out entities

* Allow encoding to be specified for toString()
---
 src/xml_document.cc | 10 +++++++++-
 test/html_parser.js |  8 ++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/src/xml_document.cc b/src/xml_document.cc
index 8e4f8251..2ed3dd56 100644
--- a/src/xml_document.cc
+++ b/src/xml_document.cc
@@ -177,6 +177,7 @@ NAN_METHOD(XmlDocument::ToString)
     assert(document);
 
     int options = 0;
+    const char *encoding = "UTF-8";
 
     if (info.Length() > 0) {
     if (info[0]->IsBoolean()) {
@@ -186,6 +187,13 @@ NAN_METHOD(XmlDocument::ToString)
     } else if (info[0]->IsObject()) {
         v8::Local<v8::Object> obj = Nan::To<v8::Object>(info[0]).ToLocalChecked();
 
+        // choose encoding declaration
+        v8::Local<v8::Value> encodingOpt = Nan::Get(obj, Nan::New<v8::String>("encoding").ToLocalChecked()).ToLocalChecked();
+        if (encodingOpt->IsString()) {
+            std::string encoding_ = *Nan::Utf8String(encodingOpt);
+            encoding = encoding_.c_str();
+        }
+
         // drop the xml declaration
         if (Nan::Get(obj, Nan::New<v8::String>("declaration").ToLocalChecked()).ToLocalChecked()->IsFalse()) {
             options |= XML_SAVE_NO_DECL;
@@ -227,7 +235,7 @@ NAN_METHOD(XmlDocument::ToString)
     }
 
     xmlBuffer* buf = xmlBufferCreate();
-    xmlSaveCtxt* savectx = xmlSaveToBuffer(buf, "UTF-8", options);
+    xmlSaveCtxt* savectx = xmlSaveToBuffer(buf, encoding, options);
     xmlSaveTree(savectx, (xmlNode*)document->xml_obj);
     xmlSaveFlush(savectx);
     xmlSaveClose(savectx);
diff --git a/test/html_parser.js b/test/html_parser.js
index bde64eab..1bbd2bea 100644
--- a/test/html_parser.js
+++ b/test/html_parser.js
@@ -151,5 +151,13 @@ module.exports.toString = function(assert) {
     assert.ok(doc.toString({ type: 'xml' }).indexOf('<?xml') > -1);
     assert.ok(doc.toString({ type: 'xhtml' }).indexOf('<?xml') > -1);
     assert.ok(doc.toString({ type: 'xml', selfCloseEmpty:true }).indexOf('<a/>') > -1);
+
+    doc = libxml.parseHtml("<a>Something&nbsp;with a space</a>");
+    assert.ok(doc.toString().indexOf('<?xml') === -1);
+    assert.ok(doc.toString({ type: 'xml' }).indexOf('<?xml') > -1);
+    assert.ok(doc.toString({ type: 'xhtml' }).indexOf('<?xml') > -1);
+    assert.ok(doc.toString({ type: 'xhtml' }).indexOf('&nbsp;') === -1);
+    assert.ok(doc.toString({ type: 'xhtml', encoding: 'HTML' }).indexOf('&nbsp;') > -1);
+    assert.ok(doc.toString({ type: 'xhtml', encoding: 'ASCII' }).indexOf('&#160;') > -1);
     assert.done();
 }