Fix % sign cleanup (#2521)

* Fix repeated escaping of % sign Extracted Regexe in Patterns Fix for #2451 * Fix stackoverflow on regex
JabRef · Feb 5, 2017 · a52d657 · a52d657
1 parent 020f1c4
commit a52d657
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 8 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -27,6 +27,7 @@ We refer to [GitHub issues](https://github.com/JabRef/jabref/issues) by using `#
  - Fix import of journal title in ris format. [#2506](https://github.com/JabRef/jabref/issues/2506)
  - We fixed the export of the `number` field in MS-Office XML export. [#2509](https://github.com/JabRef/jabref/issues/2509)
  - The field `issue` is now always exported to the corresponding `issue` field in MS-Office XML.
+ - We fixed an issue with repeated escaping of the %-sign when running the LaTeXCleanup more than once. [#2451](https://github.com/JabRef/jabref/issues/2451)
 ### Removed
 
 

diff --git a/src/main/java/net/sf/jabref/logic/formatter/bibtexfields/LatexCleanupFormatter.java b/src/main/java/net/sf/jabref/logic/formatter/bibtexfields/LatexCleanupFormatter.java
@@ -1,10 +1,21 @@
 package net.sf.jabref.logic.formatter.bibtexfields;
 
+import java.util.regex.Pattern;
+
 import net.sf.jabref.logic.l10n.Localization;
 import net.sf.jabref.model.cleanup.Formatter;
 
 public class LatexCleanupFormatter implements Formatter {
 
+    private static final Pattern REMOVE_REDUNDANT = Pattern
+            .compile("(?<!\\\\[\\p{Alpha}]{0,100}\\{[^\\}]{0,100})\\}([-/ ]?)\\{");
+
+    private static final Pattern REPLACE_WITH_AT = Pattern.compile("(^|[^\\\\$])\\$");
+    private static final Pattern REPLACE_EVERY_OTHER_AT = Pattern.compile("([^@]*)@@([^@]*)@@");
+    private static final Pattern MOVE_NUMBERS_WITH_OPERATORS = Pattern.compile("([0-9\\(\\.]+[ ]?[-+/]?[ ]?)\\$");
+    private static final Pattern MOVE_NUMBERS_RIGHT_INTO_EQUATION = Pattern.compile("@@([ ]?[-+/]?[ ]?[0-9\\)\\.]+)");
+    private static final Pattern ESCAPE_PERCENT_SIGN_ONCE = Pattern.compile("(^|[^\\\\%])%");
+
     @Override
     public String getName() {
         return Localization.lang("LaTeX cleanup");
@@ -20,19 +31,25 @@ public String format(String oldString) {
         String newValue = oldString;
 
         // Remove redundant $, {, and }, but not if the } is part of a command argument: \mbox{-}{GPS} should not be adjusted
-        newValue = newValue.replace("$$", "").replaceAll("(?<!\\\\[\\p{Alpha}]{0,100}\\{[^\\}]{0,100})\\}([-/ ]?)\\{",
-                "$1");
+        newValue = newValue.replace("$$", "");
+        newValue = REMOVE_REDUNDANT.matcher(newValue).replaceAll("$1");
+
         // Move numbers, +, -, /, and brackets into equations
-        newValue = newValue.replaceAll("(([^$]|\\\\\\$)*)\\$", "$1@@"); // Replace $, but not \$ with @@
-        newValue = newValue.replaceAll("([^@]*)@@([^@]*)@@", "$1\\$$2@@"); // Replace every other @@ with $
+        newValue = REPLACE_WITH_AT.matcher(newValue).replaceAll("$1@@"); // Replace $, but not \$ with @@
+
+        newValue = REPLACE_EVERY_OTHER_AT.matcher(newValue).replaceAll("$1\\$$2@@"); // Replace every other @@ with $
         //newValue = newValue.replaceAll("([0-9\\(\\.]+) \\$","\\$$1\\\\ "); // Move numbers followed by a space left of $ inside the equation, e.g., 0.35 $\mu$m
-        newValue = newValue.replaceAll("([0-9\\(\\.]+[ ]?[-+/]?[ ]?)\\$", "\\$$1"); // Move numbers, possibly with operators +, -, or /,  left of $ into the equation
-        newValue = newValue.replaceAll("@@([ ]?[-+/]?[ ]?[0-9\\)\\.]+)", " $1@@"); // Move numbers right of @@ into the equation
+
+        newValue = MOVE_NUMBERS_WITH_OPERATORS.matcher(newValue).replaceAll("\\$$1"); // Move numbers, possibly with operators +, -, or /,  left of $ into the equation
+        newValue = MOVE_NUMBERS_RIGHT_INTO_EQUATION.matcher(newValue).replaceAll(" $1@@"); // Move numbers right of @@ into the equation
+
         newValue = newValue.replace("@@", "$"); // Replace all @@ with $
         newValue = newValue.replace("  ", " "); // Clean up
         newValue = newValue.replace("$$", "");
         newValue = newValue.replace(" )$", ")$");
-        newValue = newValue.replace("%", "\\%"); // escape % used for comments in TeX
+
+        newValue = ESCAPE_PERCENT_SIGN_ONCE.matcher(newValue).replaceAll("$1\\\\%"); // escape %, but do not escapee \% again,  used for comments in TeX
+
         return newValue;
     }
 

diff --git a/src/test/java/net/sf/jabref/logic/formatter/bibtexfields/LatexCleanupFormatterTest.java b/src/test/java/net/sf/jabref/logic/formatter/bibtexfields/LatexCleanupFormatterTest.java
@@ -31,10 +31,19 @@ public void preservePercentSign() {
         assertEquals("\\%", formatter.format("%"));
     }
 
+    @Test
+    public void escapePercentSignOnlyOnce() {
+        assertEquals("\\%", formatter.format("\\%"));
+    }
+
+    @Test
+    public void escapePercentSignOnlnyOnceWithNumber() {
+        assertEquals("50\\%", formatter.format("50\\%"));
+    }
+
     @Test
     public void formatExample() {
         assertEquals("{VLSI DSP}", formatter.format(formatter.getExampleInput()));
     }
 
-
 }