slack_incoming: Fix formatting fixups with overlaps, or at start/end.

The previous regular expression required a `[^\w]` at the start and
end of the match.  This had two unintended effects -- it meant that it
could never match at the start or end of a string, and it meant
that *adjacent* words required *two* non-word characters between them,
as the pattern matches cannot overlap.

Switch to allowing string start/end to anchor the matches, and make
the trailing `[^\w]` be a zero-width look-ahead, to allow the patterns
to overlap.  Also remove the spurious `^` within the inner character
classes, which prevented `*foo^bar*` from matching.  Finally, add
tests to cover the functionality, which was previously untested.

zerver/webhooks/slack_incoming/tests.py

@@ -18,6 +18,30 @@ Hello, world.
             expected_message,
         )
 
+    def test_message_formatting(self) -> None:
+        tests = [
+            ("some *foo* word", "some **foo** word"),
+            ("*foo*", "**foo**"),
+            ("*foo* *bar*", "**foo** **bar**"),
+            ("*foo*a*bar*", "*foo*a*bar*"),
+            ("some _foo_ word", "some *foo* word"),
+        ]
+        self.subscribe(self.test_user, self.STREAM_NAME)
+        for input_value, output_value in tests:
+            payload = {"text": input_value}
+            msg = self.send_webhook_payload(
+                self.test_user,
+                self.url,
+                payload,
+                content_type="application/json",
+            )
+            self.assert_stream_message(
+                message=msg,
+                stream_name=self.STREAM_NAME,
+                topic_name="(no topic)",
+                content=output_value,
+            )
+
     def test_null_message(self) -> None:
         self.check_webhook(
             "null_text",

zerver/webhooks/slack_incoming/view.py

@@ -224,8 +224,8 @@ def replace_links(text: str) -> str:
 
 def replace_formatting(text: str) -> str:
     # Slack uses *text* for bold, whereas Zulip interprets that as italics
-    text = re.sub(r"([^\w])\*(?!\s+)([^\*^\n]+)(?<!\s)\*([^\w])", r"\1**\2**\3", text)
+    text = re.sub(r"([^\w]|^)\*(?!\s+)([^\*\n]+)(?<!\s)\*((?=[^\w])|$)", r"\1**\2**\3", text)
 
     # Slack uses _text_ for emphasis, whereas Zulip interprets that as nothing
-    text = re.sub(r"([^\w])[_](?!\s+)([^\_\^\n]+)(?<!\s)[_]([^\w])", r"\1*\2*\3", text)
+    text = re.sub(r"([^\w]|^)[_](?!\s+)([^\_\n]+)(?<!\s)[_]((?=[^\w])|$)", r"\1*\2*\3", text)
     return text