From 183566948f1480e3da856120ca50cd7e85345593 Mon Sep 17 00:00:00 2001
From: dmiller <dmiller@e0a8ed71-7df4-0310-8962-fdc924857419>
Date: Wed, 29 Oct 2014 14:46:11 +0000
Subject: [PATCH] Update some xml.cc comments and add encoding and standalone
 declarations

Explanatory comments within, but this came from reading the XML 1.0
spec. See some history:

http://seclists.org/nmap-dev/2013/q4/48 - Added DOCTYPE
http://seclists.org/nmap-dev/2014/q2/82 - broke Zenmap
http://seclists.org/nmap-dev/2014/q3/331 - reduced DOCTYPE
---
 xml.cc | 18 ++++++++++++++----
 1 file changed, 14 insertions(+), 4 deletions(-)

diff --git a/xml.cc b/xml.cc
index d391171c2..84c51cfbb 100644
--- a/xml.cc
+++ b/xml.cc
@@ -124,12 +124,13 @@
 This is a simple library for writing XML. It handles two main things:
 keeping track of the element stack, and escaping text where necessary.
 If you wanted to write this XML:
-  <?xml version="1.0"?>
+  <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+  <!DOCTYPE elem>
   <elem name="&amp;10.5"></elem>
 these are the functions you would call. Each one is followed by the text
 it prints enclosed in ||.
 
-xml_start_document()                   |<?xml version="1.0"?>|
+xml_start_document("elem")             |<?xml version="1.0" encoding="UTF-8" standalone="yes"?>\n<!DOCTYPE elem>|
 xml_newline();                         |\n|
 xml_open_start_tag("elem");            |<elem|
 xml_attribute("name", "&%.1f", 10.5);  | name="&amp;10.5"|
@@ -166,7 +167,7 @@ Additional functions are
 xml_write_raw                 Raw unescaped output.
 xml_write_escaped             XML-escaped output.
 xml_write_escaped_v           XML-escaped output, with a va_list.
-xml_start_document            Writes <?xml version="1.0"?>.
+xml_start_document            Writes <?xml version="1.0" encoding="UTF-8" standalone="yes"?>\n<!DOCTYPE elem>.
 xml_depth                     Returns the size of the element stack.
 
 The library makes it harder but not impossible to make non-well-formed
@@ -308,7 +309,7 @@ int xml_write_escaped_v(const char *fmt, va_list va) {
   return 0;
 }
 
-/* Write the XML declaration: <?xml version="1.0"?>
+/* Write the XML declaration: <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
  * and the DOCTYPE declaration: <!DOCTYPE rootnode>
  */
 int xml_start_document(const char *rootnode) {
@@ -316,6 +317,15 @@ int xml_start_document(const char *rootnode) {
     return -1;
   if (xml_attribute("version", "1.0") < 0)
     return -1;
+  /* Practically, Nmap only uses ASCII, but UTF-8 encompasses ASCII and allows
+   * for future expansion */
+  if (xml_attribute("encoding", "UTF-8") < 0)
+    return -1;
+  /* This indicates that parsers don't have to go elsewhere for entity
+   * declarations and so forth. We had trouble with this when we defined a
+   * PUBLIC doctype. */
+  if (xml_attribute("standalone", "yes") < 0)
+    return -1;
   if (xml_close_pi() < 0)
     return -1;
   if (xml_newline() < 0)