# [PATCH] userdiff: improve java regex for generic return types

3 messages from 2021-08-09 to 2021-08-10. Participants: Tassilo Horn, Jeff King.
Thread: https://gitlist.dev/t/56242

## Tassilo Horn, 2021-08-09 20:33

Subject: [PATCH] userdiff: improve java regex for generic return types
Message-ID: <20210809203308.30128-1-tsdh@gnu.org>
URL: https://gitlist.dev/e/20210809203308.30128-1-tsdh%40gnu.org

```
Currently, the git diff hunk headers show the wrong method signature if the
method has a generic return type because the regex doesn't allow < and > in the
return type.  This patch adds those.

Bug repro: In a repository with .gitattributes containing "*.java diff=java"
and a java file

--8<---------------cut here---------------start------------->8---
class MyExample {
    public void firstMethod() {
        // Whatever...
    }

    public List<Integer> secondMethod() {
        // Here is some comment,
        // and here is more,
        // and here is even more,
        // followed by this,
        // and that,
        // and even more.
        return Arrays.asList(1, 2, 3, 4, 5);
    }
}
--8<---------------cut here---------------end--------------->8---

when adding the number 6 to the Arrays.asList() call in secondMethod(), the git
diff one gets is

--8<---------------cut here---------------start------------->8---
diff --git a/src/main/java/MyExample.java b/src/main/java/MyExample.java
index a0f1a6b..ea37a98 100644
--- a/src/main/java/MyExample.java
+++ b/src/main/java/MyExample.java
@@ -14,6 +14,6 @@ public void firstMethod() {
         // followed by this,
         // and that,
         // and even more.
-        return Arrays.asList(1, 2, 3, 4, 5);
+        return Arrays.asList(1, 2, 3, 4, 5, 6);
     }
 }
--8<---------------cut here---------------end--------------->8---

where the hunk header shows the signature of firstMethod whereas it should show
the signature of secondMethod.

Signed-off-by: Tassilo Horn <tsdh@gnu.org>
---
 userdiff.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/userdiff.c b/userdiff.c
index 3c3bbe38b0..a244ad7ab1 100644
--- a/userdiff.c
+++ b/userdiff.c
@@ -142,7 +142,7 @@ PATTERNS("html",
 	 "[^<>= \t]+"),
 PATTERNS("java",
 	 "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
-	 "^[ \t]*(([A-Za-z_][A-Za-z_0-9]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$",
+	 "^[ \t]*(([A-Za-z_][A-Za-z_0-9<>]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$",
 	 /* -- */
 	 "[a-zA-Z_][a-zA-Z0-9_]*"
 	 "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
-- 
2.32.0


```

## Jeff King, 2021-08-09 21:00

Subject: Re: [PATCH] userdiff: improve java regex for generic return types
Message-ID: <YRGXhqW0BmKQdeZu@coredump.intra.peff.net>
URL: https://gitlist.dev/e/YRGXhqW0BmKQdeZu%40coredump.intra.peff.net
In-Reply-To: <20210809203308.30128-1-tsdh@gnu.org>

```
On Mon, Aug 09, 2021 at 10:33:08PM +0200, Tassilo Horn wrote:

> Currently, the git diff hunk headers show the wrong method signature if the
> method has a generic return type because the regex doesn't allow < and > in the
> return type.  This patch adds those.

Thanks. It has been long enough since I used Java that there were no
generics back then, but I will take your word that this is what they
look like. ;)

The patch itself looks OK to me, but...

> when adding the number 6 to the Arrays.asList() call in secondMethod(), the git
> diff one gets is
> 
> --8<---------------cut here---------------start------------->8---
> diff --git a/src/main/java/MyExample.java b/src/main/java/MyExample.java
> index a0f1a6b..ea37a98 100644
> --- a/src/main/java/MyExample.java
> +++ b/src/main/java/MyExample.java
> @@ -14,6 +14,6 @@ public void firstMethod() {
>          // followed by this,
>          // and that,
>          // and even more.
> -        return Arrays.asList(1, 2, 3, 4, 5);
> +        return Arrays.asList(1, 2, 3, 4, 5, 6);
>      }
>  }
> --8<---------------cut here---------------end--------------->8---

...this diff in the commit message will screw up git-am. The usual
procedure is to omit the scissors lines and just indent it.

But...

>  userdiff.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

...even better than explaining it is adding a regression test, perhaps
like this one:

diff --git a/t/t4018/java-return-generic b/t/t4018/java-return-generic
new file mode 100644
index 0000000000..d030c26184
--- /dev/null
+++ b/t/t4018/java-return-generic
@@ -0,0 +1,10 @@
+class MyExample {
+    public void firstMethod() {
+        // Whatever...
+    }
+
+    public List<String> secondMethod(String RIGHT[]) {
+        // Whatever...
+        return Arrays.asList("ChangeMe");
+    }
+}

(the "ChangeMe" line will be changed, and we'll expect that the line
with "RIGHT" on it is found. We use -U1 to reduce the need for filler).

-Peff

```

## Tassilo Horn, 2021-08-10 04:51

Subject: Re: [PATCH] userdiff: improve java regex for generic return types
Message-ID: <87bl65napl.fsf@gnu.org>
URL: https://gitlist.dev/e/87bl65napl.fsf%40gnu.org
In-Reply-To: <YRGXhqW0BmKQdeZu@coredump.intra.peff.net>

```
Jeff King <peff@peff.net> writes:

Hi Jeff,

> The patch itself looks OK to me, but...

thanks for the detailed comments.  I'll submit a new version of the
patch soon.

Bye,
Tassilo

```