creating pull requests works

scripts/phab2github/phab2github.py

@@ -19,9 +19,14 @@ import git
 import logging
 from phab_wrapper import PhabWrapper, Revision
 import subprocess
+import github
+import json
+import sys
 
 # TODO: move to config file
 LLVM_GITHUB_URL = 'ssh://git@github.com/llvm/llvm-project'
+MY_GITHUB_URL = 'ssh://git@github.com/christiankuehnel/llvm-project'
+MY_REPO = 'christiankuehnel/llvm-project'
 _LOGGER = logging.getLogger()
 
 
@@ -32,15 +37,26 @@ class Phab2Github:
         self.llvm_dir = os.path.join(self.workdir, 'llvm-project')
         self.repo = None  # type: Optional[git.Repo]
         self.phab_wrapper = PhabWrapper()
+        self.github = self._create_github()
+        self.github_repo = self.github.get_repo(MY_REPO) # type: github.Repository
 
     def sync(self):
         _LOGGER.info('Starting sync...')
         self._refresh_master()
         revisions = self.phab_wrapper.get_revisions()
         for revision in revisions:
-            self.create_branch(revision)
-            self.apply_patch(revision.latest_diff,
-                             self.phab_wrapper.get_raw_patch(revision.latest_diff))
+            try:
+                self.create_branch(revision)
+                self.apply_patch(revision.latest_diff,
+                                 self.phab_wrapper.get_raw_patch(revision.latest_diff))
+                self.repo.git.push('--force', 'origin', revision.branch_name)
+                # TODO: check if pull request already exists
+                self.github_repo.create_pull(title=revision.title,
+                                             body=revision.summary,
+                                             head=revision.branch_name,
+                                             base='master')
+            except Exception as e:
+                _LOGGER.exception(e)
 
     def _refresh_master(self):
         if not os.path.exists(self.workdir):
@@ -49,39 +65,55 @@ class Phab2Github:
             # TODO: in case of errors: delete and clone
             _LOGGER.info('pulling origin...')
             self.repo = git.Repo(self.llvm_dir)
-            self.repo.remotes.origin.fetch()
+            self.repo.git.fetch()
+            self.repo.git.checkout('master')
+            self.repo.git.pull('upstream', 'master')
+            self.repo.git.push('origin', 'master')
         else:
             _LOGGER.info('cloning repository...')
-            git.Repo.clone_from(LLVM_GITHUB_URL, self.llvm_dir)
+            git.Repo.clone_from(MY_GITHUB_URL, self.llvm_dir)
             self.repo = git.Repo(self.llvm_dir)
+            self.repo.create_remote('upstream', url=LLVM_GITHUB_URL)
+            self.repo.remotes.upstream.fetch()
         _LOGGER.info('refresh of master branch completed')
 
     def create_branch(self, revision: Revision):
-        name = 'phab-D{}'.format(revision.id)
+        name = revision.branch_name
         if name in self.repo.heads:
-            self.repo.head.reference = self.repo.heads['master']
-            self.repo.head.reset(index=True, working_tree=True)
-            self.repo.delete_head(name)
+            self.repo.git.checkout('master')
+            self.repo.git.branch('-D', name)
         base_hash = revision.latest_diff.base_hash
         if base_hash is None:
-            base_hash = 'origin/master'
+            base_hash = 'upstream/master'
         _LOGGER.info('creating branch {} based one {}...'.format(name, base_hash))
         try:
             new_branch = self.repo.create_head(name, base_hash)
         except ValueError:
             # commit hash not found, try again with master
-            base_hash = 'origin/master'
+            _LOGGER.warning('commit hash {} not found in upstream repository. '
+                            'Trying master instead...'.format(name, base_hash))
+            base_hash = 'upstream/master'
             new_branch = self.repo.create_head(name, base_hash)
         self.repo.head.reference = new_branch
         self.repo.head.reset(index=True, working_tree=True)
 
     def apply_patch(self, diff: "Diff", raw_patch: str):
-        proc = subprocess.run('git apply --ignore-whitespace --whitespace=fix -', input=raw_patch, shell=True, text=True,
+        proc = subprocess.run('git apply --ignore-whitespace --whitespace=fix -', input=raw_patch, shell=True, text=True, cwd=self.repo.working_dir,
                               stdout=subprocess.PIPE, stderr=subprocess.PIPE)
         if proc.returncode != 0:
             raise Exception('Applying patch failed:\n{}'.format(proc.stdout + proc.stderr))
+        self.repo.git.add('-A')
         self.repo.index.commit(message='applying diff Phabricator Revision {}\n\ndiff: {}'.format(diff.revision, diff.id))
 
+    def _create_github(self) -> github.Github:
+        """Create instance of Github client.
+
+        Reads access token from a file.
+        """
+        with open(os.path.expanduser('~/.llvm-premerge-checks/github-token.json')) as json_file:
+            token = json.load(json_file)['token']
+        return github.Github(token)
+
 
 if __name__ == '__main__':
     logging.basicConfig(level=logging.DEBUG)

scripts/phab2github/phab_wrapper.py

@@ -45,12 +45,22 @@ class Revision:
     def __str__(self):
         return 'Revision {}: {} - ({})'.format(self.id, self.status,
                                     ','.join([str(d.id) for d in self.diffs]))
-
     @property
     def latest_diff(self):
         # TODO: make sure self.diffs is sorted
         return self.diffs[-1]
 
+    @property
+    def branch_name(self) -> str:
+        return 'phab-D{}'.format(self.id)
+
+    @property
+    def title(self) -> str:
+        return self.revision_dict['fields']['title']
+
+    @property
+    def summary(self) -> str:
+        return self.revision_dict['fields']['summary']
 
 class Diff:
     """A Phabricator diff."""
@@ -119,6 +129,8 @@ class PhabWrapper:
         revision_response = self.phab.differential.revision.search(
                 constraints=constraints)
         revisions = [Revision(r) for r in revision_response.response['data']]
+        # TODO: only taking the first 10 to speed things up
+        revisions = revisions[0:3]
         _LOGGER.info('Got {} revisions from the server'.format(len(revisions)))
         for revision in revisions:
             # TODO: batch-query diffs for all revisions, reduce number of

scripts/phab2github/requirements.txt

@@ -1,3 +1,4 @@
 phabricator==0.7.0
 gitpython==3.0.6
 retrying==1.3.3
+PyGitHub=1.46