Eclipse Community Forums: EGit / JGit » How to get content of files with JGIT

Help

Home

Home » Eclipse Projects » EGit / JGit » How to get content of files with JGIT

Show: Today's Messages :: Show Polls :: Message Navigator

How to get content of files with JGIT [message #1744341]

Mon, 26 September 2016 07:22

Yossi Balan

Messages: 4
Registered: September 2016

Junior Member

Hi

I want to get the content of file in different mode :
1. working directory
2. stage
3. commit
I saw this example in the documentation but it give me the content only in case of commit. How I can get the content of file wiht jgit in working directory and in stage area.
try (RevWalk revWalk = new RevWalk(getRepository())) {
RevCommit revCommit = revWalk.parseCommit(revision);
RevTree tree = revCommit.getTree();

try (TreeWalk treeWalk = new TreeWalk(getRepository())) {
treeWalk.addTree(tree);
treeWalk.setRecursive(true);
treeWalk.setFilter(PathFilter.create(request.getFile()));
if (!treeWalk.next()) {
//error
}
ObjectId objectId = treeWalk.getObjectId(0);
ObjectLoader loader = repository.open(objectId);
content = new String(loader.getBytes());
}
}

Report message to a moderator

Re: How to get content of files with JGIT [message #1744392 is a reply to message #1744341]

Mon, 26 September 2016 13:02

Christian Halstrick

Messages: 274
Registered: July 2009

Senior Member

Try adding index and workingtree iterators to the walk. To get real content you can ask objectdb with the oid for commit and index. But for workingtree content you need to use java File io to read content from disk

public class ListWTIndexHEAD {
        public static void main(String args[]) throws IOException, GitAPIException, JGitInternalException {
                Git r = Git.open(new File(args[0]));
                RevWalk rw = new RevWalk(r.getRepository());
                try (TreeWalk tw = new TreeWalk(r.getRepository())) {
                        RevCommit commitToCheck = rw.parseCommit(r.getRepository().resolve("HEAD"));
                        tw.addTree(commitToCheck.getTree());
                        tw.addTree(new DirCacheIterator(r.getRepository().readDirCache()));
                        tw.addTree(new FileTreeIterator(r.getRepository()));
                        tw.setRecursive(true);
                        while (tw.next()) {
                                System.out.printf(
                                                "path: %s, Commit(mode/oid): %s/%s, Index(mode/oid): %s/%s, Workingtree(mode/oid): %s/%s\n",
                                                tw.getPathString(), tw.getFileMode(0), tw.getObjectId(0), tw.getFileMode(1), tw.getObjectId(1),
                                                tw.getFileMode(2), tw.getObjectId(2));
                        }
                }
        }
}

Ciao
Chris

Report message to a moderator

Re: How to get content of files with JGIT [message #1744406 is a reply to message #1744392]

Mon, 26 September 2016 15:33

Yossi Balan

Messages: 4
Registered: September 2016

Junior Member

I found an example that they did
DirCache cache = db.readDirCache();
DirCacheEntry entry = cache.getEntry(pattern);
if (entry == null) {
return null;
}
ObjectId blobId = entry.getObjectId();

without the use of the RevWalk
Is it ok ?
what is the different ?

Report message to a moderator

Re: How to get content of files with JGIT [message #1744415 is a reply to message #1744406]

Mon, 26 September 2016 16:22

Christian Halstrick

Messages: 274
Registered: July 2009

Senior Member

With revwalk you can walk over directory structures and for every path you get the three states. That's convient for a lot of applications. If you are interested only in the index state for a single path then your code is more appropriate. But if you do that for the index, then another block for the commit and finally another block of code for the filesystem .... then I think the revwalk code has much less codelines. But again, if you just want to read a single path from the index then your code is ok

Ciao
Chris

Report message to a moderator

Re: How to get content of files with JGIT [message #1744426 is a reply to message #1744415]

Mon, 26 September 2016 19:51

Yossi Balan

Messages: 4
Registered: September 2016

Junior Member

With revwalk you can walk over directory structures and for every path you get the three state.

How I can get different state ? and how I can know in which state is the result ? did you mean working directory , staging and commit file ?
Does when you tw.next() you got all the state ?

Report message to a moderator

Re: How to get content of files with JGIT [message #1744446 is a reply to message #1744426]

Tue, 27 September 2016 07:14

Christian Halstrick

Messages: 274
Registered: July 2009

Senior Member

look at the example code I provided. You create one treewalk and add three AbstractTreeIterators to this walk. The first iterator for the commit, the second for the index and the third for the workingtree. When you say walk.next() you move to the next available path, e.g. "src/a.txt". Treewalk takes care to advance every iterator so that every iterator looks now at "src/a.txt". With walk.getObjectID(0) you get the objectid for "src/a.txt" for the first Iterator (that one looking at the commit). And walk.getObjectID(1) returns the objectid for "src/a.txt" for the second Iterator (that one looking at the index). The example should clarify that.

Ciao
Chris

Report message to a moderator

Re: How to get content of files with JGIT [message #1744455 is a reply to message #1744446]

Tue, 27 September 2016 08:04

Yossi Balan

Messages: 4
Registered: September 2016

Junior Member

Thanks for the help

Report message to a moderator

Previous Topic:	How does Eclipse/eGit connect workspace to git repo and its project metadata/source files?
Next Topic:	EGit keeps showing CVS files in Unstaged/Staged view

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Tue Apr 23 09:38:28 GMT 2024

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter