From 5106d9133bb30586fab1fca3f7b0dcf407f33d95 Mon Sep 17 00:00:00 2001
From: Deyan Ginev <deyan.ginev@gmail.com>
Date: Sat, 28 Sep 2019 19:16:44 -0400
Subject: [PATCH] space out lexeme ends for known broken markup in latexml
 0.8.4

---
 src/dnm/node.rs | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/src/dnm/node.rs b/src/dnm/node.rs
index 501fe5aeb0..32c071406d 100644
--- a/src/dnm/node.rs
+++ b/src/dnm/node.rs
@@ -1,5 +1,10 @@
+use lazy_static::lazy_static;
 use libxml::readonly::RoNode;
 use libxml::xpath::Context;
+use regex::Regex;
+lazy_static! {
+  static ref LEXEME_END_MARKER : Regex = Regex::new(r"((?:OPFUNCTION|OPERATOR|OPEN|CLOSE|UNKNOWN|RELOP|ADDOP|MULOP|ARROW|BIGOP|BINOP|ID|OVERACCENT|UNDERACCENT):end)").unwrap();
+}
 
 /// Map math nodes to their lexemes
 pub fn lexematize_math(node: RoNode, context: &mut Context) -> String {
@@ -20,7 +25,8 @@ pub fn lexematize_math(node: RoNode, context: &mut Context) -> String {
       let mut annotation_string = anno.get_content();
       // offer fix for latexml 0.8.4 serialization flaw in some cases (e.g. "POSTFIX:endID:end"
       // instead of "POSTFIX:end ID:end")
-      annotation_string = annotation_string
+      annotation_string = LEXEME_END_MARKER
+        .replace_all(&annotation_string, " $1 ")
         .split(":end")
         .collect::<Vec<&str>>()
         .join(":end ");