From a03851fac3ec2d84c6ee16c14d057a7f14128969 Mon Sep 17 00:00:00 2001
From: Louis-Dominique Dubeau <ldd@lddubeau.com>
Date: Thu, 10 May 2018 14:39:40 -0400
Subject: [PATCH] Add an emitEmpty option.

The option emitEmpty cause sax to emit `cdata` and `comment` events
even if the respective constructs were empty in the XML.
---
 README.md          |  3 +++
 lib/sax.js         |  8 ++++++--
 test/emit_empty.js | 47 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 56 insertions(+), 2 deletions(-)
 create mode 100644 test/emit_empty.js

diff --git a/README.md b/README.md
index afcd3f3d..283c9313 100644
--- a/README.md
+++ b/README.md
@@ -106,6 +106,9 @@ Settings supported:
 * `strictEntities` - Boolean. If true, only parse [predefined XML
   entities](http://www.w3.org/TR/REC-xml/#sec-predefined-ent)
   (`&amp;`, `&apos;`, `&gt;`, `&lt;`, and `&quot;`)
+* `emitEmpty` - Boolean. If true, then SAX will emit `cdata` and `comment`
+  events even if the CDATA section or comment were empty. By default,
+  when these XML structures are empty, SAX does not emit an event.
 
 ## Methods
 
diff --git a/lib/sax.js b/lib/sax.js
index 795d607e..4b960b6a 100644
--- a/lib/sax.js
+++ b/lib/sax.js
@@ -61,6 +61,7 @@
     parser.noscript = !!(strict || parser.opt.noscript)
     parser.state = S.BEGIN
     parser.strictEntities = parser.opt.strictEntities
+    parser.emitEmpty = parser.opt.emitEmpty
     parser.ENTITIES = parser.strictEntities ? Object.create(sax.XML_ENTITIES) : Object.create(sax.ENTITIES)
     parser.attribList = []
 
@@ -140,6 +141,9 @@
 
   function flushBuffers (parser) {
     closeText(parser)
+    // We do not need to check emitEmpty here. We are guaranteed
+    // to produce a cdata event with an empty string when the cdata
+    // section ends.
     if (parser.cdata !== '') {
       emitNode(parser, 'oncdata', parser.cdata)
       parser.cdata = ''
@@ -1180,7 +1184,7 @@
           if (c === '-') {
             parser.state = S.COMMENT_ENDED
             parser.comment = textopts(parser.opt, parser.comment)
-            if (parser.comment) {
+            if (parser.emitEmpty || parser.comment) {
               emitNode(parser, 'oncomment', parser.comment)
             }
             parser.comment = ''
@@ -1221,7 +1225,7 @@
 
         case S.CDATA_ENDING_2:
           if (c === '>') {
-            if (parser.cdata) {
+            if (parser.emitEmpty || parser.cdata) {
               emitNode(parser, 'oncdata', parser.cdata)
             }
             emitNode(parser, 'onclosecdata')
diff --git a/test/emit_empty.js b/test/emit_empty.js
new file mode 100644
index 00000000..8c5bddb4
--- /dev/null
+++ b/test/emit_empty.js
@@ -0,0 +1,47 @@
+var xml = '<r><![CDATA[]]><!----></r>'
+require(__dirname).test({
+  xml: xml,
+  expect: [
+    ['opentagstart', {'name': 'R', 'attributes': {}}],
+    ['opentag', {'name': 'R', 'attributes': {}, 'isSelfClosing': false}],
+    ['opencdata', undefined],
+    ['closecdata', undefined],
+    ['closetag', 'R']
+  ]
+})
+
+require(__dirname).test({
+  xml: xml,
+  opt: { emitEmpty: true },
+  expect: [
+    ['opentagstart', {'name': 'R', 'attributes': {}}],
+    ['opentag', {'name': 'R', 'attributes': {}, 'isSelfClosing': false}],
+    ['opencdata', undefined],
+    ['cdata', ''],
+    ['closecdata', undefined],
+    ['comment', ''],
+    ['closetag', 'R']
+  ]
+})
+
+// The following test illustrates an effect of emitEmpty together with
+// hitting the buffer limit. Namely, a trailing cdata event with an
+// empty string will be emitted after the buffer is emptied.
+var sax = require('../lib/sax')
+var bl = sax.MAX_BUFFER_LENGTH
+sax.MAX_BUFFER_LENGTH = 10
+require(__dirname).test({
+  opt: { emitEmpty: true },
+  expect: [
+    ['opentagstart', {'name': 'R', 'attributes': {}}],
+    ['opentag', {'name': 'R', 'attributes': {}, 'isSelfClosing': false}],
+    ['opencdata', undefined],
+    ['cdata', '12345678901'],
+    ['cdata', ''],
+    ['closecdata', undefined],
+    ['closetag', 'R']
+  ]
+}).write('<r><![CDATA[12345678901')
+  .write(']]></r>')
+
+sax.MAX_BUFFER_LENGTH = bl