[JIRA] (JENKINS-38121) Patch file processing does not support UTF-8 encoding

0 views

Skip to first unread message

axelheider@gmx.de (JIRA)

unread,

Sep 11, 2016, 9:51:03 PM9/11/16

to jenkinsc...@googlegroups.com

Axel Heider created an issue

Jenkins /

JENKINS-38121

Patch file processing does not support UTF-8 encoding

Issue Type:	Bug
Assignee:	Unassigned
Components:	patch-parameter-plugin
Created:	2016/Sep/12 1:50 AM
Environment:	* Jenkins 1.598 * patch-parameter plugin V1.2 * diff4j-1.2 from https://github.com/cloudbees/diff4j (7f212fe6f891e8eaaa349ed9fc6866c9d1280dd2) * GNU patch 2.7.5 * SVN 1.9
Labels:	encoding java
Priority:	Critical
Reporter:	Axel Heider

I'm having problem with a patch file that changes a code comment using "’" (RIGHT SINGLE QUOTATION MARK, UTF-8 U+2019 encoded as 0xE2 0x80 0x99) into the (normal?) apostrophe "'" (APOSTROPHE, ASCII 39/0x27). The patch is created with SVN diff (V1.9) in Linux and can be used fine with GNU patch 2.7.5. However, when uploading the patch to jenkins, it can't be applies- The error is

 
                                                                com.cloudbees.diff.PatchException: Cannot apply hunk @@ 245

I've debugged the issue using diff4j-1.2 (7f212fe6f891e8eaaa349ed9fc6866c9d1280dd2) from https://github.com/cloudbees/diff4j in a test java application. Adding some printouts in

                                                                com/cloudbees/diff/ContextualPatch.java
                                                            
 
                                                                private boolean applyHunk(...)
...
if (!targetLine.equals(hunkLine.substring(1).trim())) { // be optimistic, compare trimmed context lines
    System.out.println("hunk: "+hunkLine);
    System.out.println("target: "+targetLine);
 
                                                            

where tries to compare lines, I get this:

 
                                                                hunk: - // this commentâ<U+0080><U+0099>s encoding fails
target: // this comment’s encoding fails

Seems Java does not consider the patch file's line as UTF-8, but the source code's line is taken as UTF-8.