reckless: Clone github sources when API access fails

Due to the API ratelimit, this allows cloning a github repo and searching
the result rather than searching via the REST API.  If a source has
already been cloned, it is fetched and the default branch checked out.

Fixes a failure reported by @farscapian

Changelog-Fixed: Reckless no longer fails on github API ratelimit.

tools/reckless

@@ -17,6 +17,7 @@ import types
 from typing import Union
 from urllib.parse import urlparse
 from urllib.request import urlopen
+from urllib.error import HTTPError
 import venv
 
 
@@ -144,7 +145,8 @@ class InstInfo:
         target = SourceDir(self.source_loc, srctype=self.srctype)
         # Set recursion for how many directories deep we should search
         depth = 0
-        if self.srctype in [Source.DIRECTORY, Source.LOCAL_REPO]:
+        if self.srctype in [Source.DIRECTORY, Source.LOCAL_REPO,
+                            Source.GIT_LOCAL_CLONE]:
             depth = 5
         elif self.srctype == Source.GITHUB_REPO:
             depth = 1
@@ -193,7 +195,28 @@ class InstInfo:
                         return success
             return None
 
-        result = search_dir(self, target, False, depth)
+        try:
+            result = search_dir(self, target, False, depth)
+        # Using the rest API of github.com may result in a
+        # "Error 403: rate limit exceeded" or other access issues.
+        # Fall back to cloning and searching the local copy instead.
+        except HTTPError:
+            result = None
+            if self.srctype == Source.GITHUB_REPO:
+                # clone source to reckless dir
+                target = copy_remote_git_source(self)
+                if not target:
+                    logging.warning(f"could not clone github source {self}")
+                    return False
+                logging.debug(f"falling back to cloning remote repo {self}")
+                # Update to reflect use of a local clone
+                self.source_loc = target.location
+                self.srctype = target.srctype
+                result = search_dir(self, target, False, 5)
+
+                if not result:
+                    return False
+
         if result:
             if result != target:
                 if result.relative:
@@ -235,6 +258,8 @@ class Source(Enum):
     GITHUB_REPO = 3
     OTHER_URL = 4
     UNKNOWN = 5
+    # Cloned from remote source before searching (rather than github API)
+    GIT_LOCAL_CLONE = 6
 
     @classmethod
     def get_type(cls, source: str):
@@ -253,6 +278,16 @@ class Source(Enum):
             return cls(4)
         return cls(5)
 
+    @classmethod
+    def get_github_user_repo(cls, source: str) -> (str, str):
+        'extract a github username and repository name'
+        if 'github.com/' not in source.lower():
+            return None, None
+        trailing = Path(source.lower().partition('github.com/')[2]).parts
+        if len(trailing) < 2:
+            return None, None
+        return trailing[0], trailing[1]
+
 
 class SourceDir():
     """Structure to search source contents."""
@@ -277,7 +312,7 @@ class SourceDir():
         # logging.debug(f"populating {self.srctype} {self.location}")
         if self.srctype == Source.DIRECTORY:
             self.contents = populate_local_dir(self.location)
-        elif self.srctype == Source.LOCAL_REPO:
+        elif self.srctype in [Source.LOCAL_REPO, Source.GIT_LOCAL_CLONE]:
             self.contents = populate_local_repo(self.location)
         elif self.srctype == Source.GITHUB_REPO:
             self.contents = populate_github_repo(self.location)
@@ -435,6 +470,11 @@ def source_element_from_repo_api(member: dict):
 
 
 def populate_github_repo(url: str) -> list:
+    """populate one level of a github repository via REST API"""
+    # Forces search to clone remote repos (for blackbox testing)
+    if GITHUB_API_FALLBACK:
+        with tempfile.NamedTemporaryFile() as tmp:
+            raise HTTPError(url, 403, 'simulated ratelimit', {}, tmp)
     # FIXME: This probably contains leftover cruft.
     repo = url.split('/')
     while '' in repo:
@@ -478,6 +518,28 @@ def populate_github_repo(url: str) -> list:
     return contents
 
 
+def copy_remote_git_source(github_source: InstInfo):
+    """clone or fetch & checkout a local copy of a remote git repo"""
+    user, repo = Source.get_github_user_repo(github_source.source_loc)
+    if not user or not repo:
+        logging.warning('could not extract github user and repo '
+                        f'name for {github_source.source_loc}')
+        return None
+    local_path = RECKLESS_DIR / '.remote_sources' / user
+    create_dir(RECKLESS_DIR / '.remote_sources')
+    if not create_dir(local_path):
+        logging.warning(f'could not provision dir {local_path} to '
+                        f'clone remote source {github_source.source_loc}')
+        return None
+    local_path = local_path / repo
+    if local_path.exists():
+        # Fetch the latest
+        assert _git_update(github_source, local_path)
+    else:
+        _git_clone(github_source, local_path)
+    return SourceDir(local_path, srctype=Source.GIT_LOCAL_CLONE)
+
+
 class Config():
     """A generic class for procuring, reading and editing config files"""
     def obtain_config(self,
@@ -803,7 +865,8 @@ def _git_clone(src: InstInfo, dest: Union[PosixPath, str]) -> bool:
     if src.srctype == Source.GITHUB_REPO:
         assert 'github.com' in src.source_loc
         source = f"{GITHUB_COM}" + src.source_loc.split("github.com")[-1]
-    elif src.srctype in [Source.LOCAL_REPO, Source.OTHER_URL]:
+    elif src.srctype in [Source.LOCAL_REPO, Source.OTHER_URL,
+                         Source.GIT_LOCAL_CLONE]:
         source = src.source_loc
     else:
         return False
@@ -819,6 +882,46 @@ def _git_clone(src: InstInfo, dest: Union[PosixPath, str]) -> bool:
     return True
 
 
+def _git_update(github_source: InstInfo, local_copy: PosixPath):
+    # Ensure this is the correct source
+    git = run(['git', 'remote', 'set-url', 'origin', github_source.source_loc],
+              cwd=str(local_copy), stdout=PIPE, stderr=PIPE, text=True,
+              check=False, timeout=60)
+    assert git.returncode == 0
+    if git.returncode != 0:
+        return False
+
+    # Fetch the latest from the remote
+    git = run(['git', 'fetch', 'origin', '--recurse-submodules=on-demand'],
+              cwd=str(local_copy), stdout=PIPE, stderr=PIPE, text=True,
+              check=False, timeout=60)
+    assert git.returncode == 0
+    if git.returncode != 0:
+        return False
+
+    # Find default branch
+    git = run(['git', 'symbolic-ref', 'refs/remotes/origin/HEAD', '--short'],
+              cwd=str(local_copy), stdout=PIPE, stderr=PIPE, text=True,
+              check=False, timeout=60)
+    assert git.returncode == 0
+    if git.returncode != 0:
+        return False
+    default_branch = git.stdout.splitlines()[0]
+    if default_branch != 'origin/master':
+        logging.debug(f'UNUSUAL: fetched default branch {default_branch} for '
+                      f'{github_source.source_loc}')
+
+    # Checkout default branch
+    git = run(['git', 'checkout', default_branch],
+              cwd=str(local_copy), stdout=PIPE, stderr=PIPE, text=True,
+              check=False, timeout=60)
+    assert git.returncode == 0
+    if git.returncode != 0:
+        return False
+
+    return True
+
+
 def get_temp_reckless_dir() -> PosixPath:
     random_dir = 'reckless-{}'.format(str(hash(os.times()))[-9:])
     new_path = Path(tempfile.gettempdir()) / random_dir
@@ -850,7 +953,7 @@ def _checkout_commit(orig_src: InstInfo,
                      cloned_path: PosixPath):
     # Check out and verify commit/tag if source was a repository
     if orig_src.srctype in [Source.LOCAL_REPO, Source.GITHUB_REPO,
-                            Source.OTHER_URL]:
+                            Source.OTHER_URL, Source.GIT_LOCAL_CLONE]:
         if orig_src.commit:
             logging.debug(f"Checking out {orig_src.commit}")
             checkout = Popen(['git', 'checkout', orig_src.commit],
@@ -912,7 +1015,7 @@ def _install_plugin(src: InstInfo) -> Union[InstInfo, None]:
         create_dir(clone_path)
         shutil.copytree(src.source_loc, plugin_path)
     elif src.srctype in [Source.LOCAL_REPO, Source.GITHUB_REPO,
-                         Source.OTHER_URL]:
+                         Source.OTHER_URL, Source.GIT_LOCAL_CLONE]:
         # clone git repository to /tmp/reckless-...
         if not _git_clone(src, plugin_path):
             return None
@@ -1401,6 +1504,10 @@ if __name__ == '__main__':
         GITHUB_COM = os.environ['REDIR_GITHUB']
     logging.root.setLevel(args.loglevel)
 
+    GITHUB_API_FALLBACK = False
+    if 'GITHUB_API_FALLBACK' in os.environ:
+        GITHUB_API_FALLBACK = os.environ['GITHUB_API_FALLBACK']
+
     if 'targets' in args:
         # FIXME: Catch missing argument
         if args.func.__name__ == 'help_alias':