[webkitscmpy] Collect diff from GitHub remote repository

https://bugs.webkit.org/show_bug.cgi?id=265590
rdar://118993239

Reviewed by Dewei Zhu.

* Tools/Scripts/libraries/webkitscmpy/webkitscmpy/mocks/remote/git_hub.py:
(GitHub.commit): 'HEAD' in GitHub means "latest commit on the default branch".
(GitHub._diff_response): Return a response with a mock text diff.
(GitHub.request): Respect the 'application/vnd.github.diff' header.
* Tools/Scripts/libraries/webkitscmpy/webkitscmpy/remote/git_hub.py:
(GitHub):
(GitHub.diff): Given a commit or commit range, return a line-by-line diff of
the provided range. If the caller requests it, include the commit messages for
the specified commits in the same patch format used by 'git format-patch'.
* Tools/Scripts/libraries/webkitscmpy/webkitscmpy/test/git_unittest.py:

Canonical link: https://commits.webkit.org/271687@main

Tools/Scripts/libraries/webkitscmpy/webkitscmpy/mocks/remote/git_hub.py

@@ -94,6 +94,8 @@ def commit(self, ref):
             return self.commits[ref][-1]
         if ref in self.tags:
             return self.tags[ref]
+        if ref == 'HEAD':
+            return self.commits[self.default_branch][-1]
 
         for branch, commits in self.commits.items():
             for commit in commits:
@@ -242,6 +244,22 @@ def _commit_response(self, url, ref):
             'files': [dict(filename=name) for name in ('Source/main.cpp', 'Source/main.h')],
         }, url=url)
 
+    def _diff_response(self, url, ref):
+        commit = self.commit(ref)
+        if not commit:
+            return mocks.Response.fromJson(
+                dict(message='No commit found for SHA: {}'.format(ref)),
+                url=url,
+                status_code=404,
+            )
+        return mocks.Response.fromText(
+            'diff --git a/ChangeLog b/ChangeLog\n'
+            '--- a/ChangeLog\n'
+            '+++ b/ChangeLog\n'
+            '@@ -1,0 +1,0 @@\n{}'.format('\n'.join(['+{}'.format(line) for line in commit.message.splitlines()])),
+            url=url,
+        )
+
     def _compare_response(self, url, ref_a, ref_b):
         commit_a = self.commit(ref_a)
         commit_b = self.commit(ref_b)
@@ -374,7 +392,7 @@ def graphql(self, url, auth=None, json=None):
 
         return mocks.Response.create404(url)
 
-    def request(self, method, url, data=None, params=None, auth=None, json=None, **kwargs):
+    def request(self, method, url, data=None, params=None, auth=None, json=None, headers=None, **kwargs):
         if not url.startswith('http://') and not url.startswith('https://'):
             return mocks.Response.create404(url)
 
@@ -395,6 +413,8 @@ def request(self, method, url, data=None, params=None, auth=None, json=None, **k
 
         # Extract single commit
         if stripped_url.startswith('{}/commits/'.format(self.api_remote)):
+            if (headers or {}).get('Accept') == 'application/vnd.github.diff':
+                return self._diff_response(url=url, ref=stripped_url.split('/', 5)[-1])
             return self._commit_response(url=url, ref=stripped_url.split('/', 5)[-1])
 
         # Compare two commits

Tools/Scripts/libraries/webkitscmpy/webkitscmpy/remote/git_hub.py

@@ -45,6 +45,7 @@ class GitHub(Scm):
     KNOWN_400_MESSAGES = [
         'No commit found for SHA',
     ]
+    DIFF_HEADER = 'application/vnd.github.diff'
 
     class PRGenerator(Scm.PRGenerator):
         SUPPORTS_DRAFTS = True
@@ -710,7 +711,7 @@ def commits(self, begin=None, end=None, include_log=True, include_identifier=Tru
                 identifier = previous.identifier
                 if commit_data['sha'] == previous.hash:
                     cached = cached[:-1]
-                else:
+                elif identifier is not None:
                     identifier -= 1
 
                 if not identifier:
@@ -783,6 +784,50 @@ def find(self, argument, include_log=True, include_identifier=True):
             raise ValueError("'{}' is not an argument recognized by git".format(argument))
         return self.commit(hash=commit_data['sha'], include_log=include_log, include_identifier=include_identifier)
 
+    def diff(self, head='HEAD', base=None, include_log=False):
+        if base:
+            commits = list(self.commits(dict(argument=base), end=dict(argument=head), include_identifier=False))
+        else:
+            commits = [self.find(head, include_identifier=False), None]
+
+        if not commits or not commits[0]:
+            sys.stderr.write('Failed to find commits required to generate diff\n')
+            return
+        commits = commits[:-1]
+
+        patch_count = 1
+        for commit in commits:
+            response = self.request('commits/{}'.format(commit.hash), headers=dict(Accept=self.DIFF_HEADER))
+            if response.status_code // 100 != 2:
+                sys.stderr.write('Failed to retrieve diff of {} with status code {}\n'.format(commit, response.status_code))
+                return
+
+            if include_log:
+                yield 'From {}'.format(commit.hash)
+                yield 'From: {} <{}>'.format(commit.author.name, commit.author.email)
+                yield 'Date: {}'.format(datetime.fromtimestamp(commit.timestamp).strftime('%a %b %d %H:%M:%S %Y'))
+                if len(commits) <= 1:
+                    subject = 'Subject: [PATCH]'
+                else:
+                    subject = 'Subject: [PATCH {}/{}]'.format(patch_count, len(commits))
+                for line in commit.message.splitlines():
+                    if subject is None:
+                        yield line
+                    elif line:
+                        subject = '{} {}'.format(subject, line)
+                    else:
+                        yield subject
+                        yield line
+                        subject = None
+                if subject:
+                    yield subject
+                yield '---'
+
+            for line in response.text.splitlines():
+                yield line
+
+            patch_count += 1
+
     def files_changed(self, argument=None):
         if not argument:
             raise ValueError('No argument provided')

Tools/Scripts/libraries/webkitscmpy/webkitscmpy/test/git_unittest.py

@@ -758,6 +758,60 @@ def test_checkout_url(self):
         with self.assertRaises(ValueError):
             remote.GitHub(self.remote).checkout_url(http=True, ssh=True)
 
+    def test_diff(self):
+        with mocks.remote.GitHub():
+            repo = remote.GitHub(self.remote)
+            self.assertEqual([
+                'diff --git a/ChangeLog b/ChangeLog',
+                '--- a/ChangeLog',
+                '+++ b/ChangeLog',
+                '@@ -1,0 +1,0 @@',
+                '+Patch Series',
+            ], list(repo.diff(base='bae5d1e90999d4f916a8a15810ccfa43f37a2fd6')))
+
+    def test_diff_with_commit_message(self):
+        with mocks.remote.GitHub():
+            repo = remote.GitHub(self.remote)
+            self.assertEqual([
+                'From bae5d1e90999d4f916a8a15810ccfa43f37a2fd6',
+                'From: Jonathan Bedard <jbedard@apple.com>',
+                'Date: {}'.format(datetime.fromtimestamp(1601642800).strftime('%a %b %d %H:%M:%S %Y')),
+                'Subject: [PATCH] 8th commit',
+                '---',
+                'diff --git a/ChangeLog b/ChangeLog',
+                '--- a/ChangeLog',
+                '+++ b/ChangeLog',
+                '@@ -1,0 +1,0 @@',
+                '+8th commit',
+            ], list(repo.diff(base='3@main', head='4@main', include_log=True)))
+
+    def test_diff_with_commit_message_multiple(self):
+        self.maxDiff = None
+        with mocks.remote.GitHub():
+            repo = remote.GitHub(self.remote)
+            self.assertEqual([
+                'From bae5d1e90999d4f916a8a15810ccfa43f37a2fd6',
+                'From: Jonathan Bedard <jbedard@apple.com>',
+                'Date: {}'.format(datetime.fromtimestamp(1601642800).strftime('%a %b %d %H:%M:%S %Y')),
+                'Subject: [PATCH 1/2] 8th commit',
+                '---',
+                'diff --git a/ChangeLog b/ChangeLog',
+                '--- a/ChangeLog',
+                '+++ b/ChangeLog',
+                '@@ -1,0 +1,0 @@',
+                '+8th commit',
+                'From 1abe25b443e985f93b90d830e4a7e3731336af4d',
+                'From: Jonathan Bedard <jbedard@apple.com>',
+                'Date: {}'.format(datetime.fromtimestamp(1601637800).strftime('%a %b %d %H:%M:%S %Y')),
+                'Subject: [PATCH 2/2] 4th commit',
+                '---',
+                'diff --git a/ChangeLog b/ChangeLog',
+                '--- a/ChangeLog',
+                '+++ b/ChangeLog',
+                '@@ -1,0 +1,0 @@',
+                '+4th commit',
+            ], list(repo.diff(base='2@main', head='4@main', include_log=True)))
+
 
 class TestBitBucket(testing.TestCase):
     remote = 'https://bitbucket.example.com/projects/WEBKIT/repos/webkit'