Adds check that github path has valid sha after tree/ (#401)

* Adds check that github path has valid sha after `tree/`
closes #396

* update regexp

Author: mike0sv

mlem/contrib/github.py

@@ -1,5 +1,6 @@
 import pathlib
 import posixpath
+import re
 from typing import ClassVar, Dict, Optional
 from urllib.parse import quote_plus, urlparse
 
@@ -14,7 +15,6 @@ def ls_branches(repo_url: str) -> Dict[str, str]:
     """List branches in remote git repo"""
     import git
 
-    git.cmd.Git().ls_remote(repo_url)
     g = git.cmd.Git()
     remote_refs: Dict[str, str] = dict(
         tuple(reversed(ref.split("\t")[:2]))
@@ -54,6 +54,10 @@ def _ls_github_refs(org: str, repo: str, endpoint: str):
     return None
 
 
+def is_long_sha(sha: str):
+    return re.match(r"^[a-f\d]{40}$", sha)
+
+
 class GithubResolver(CloudGitResolver):
     """Resolve https://github.com URLs"""
 
@@ -87,18 +91,23 @@ def get_kwargs(cls, uri):
             return {"org": org, "repo": repo, "path": ""}
         if path[0] == "tree":
             sha = path[1]
-            refs = ls_github_branches(org, repo)
-            refs.update(ls_github_tags(org, repo))
-            branches = {quote_plus(k) for k in refs}
-            # match beginning of path with one of existing branches
-            # "" is hack for cases with empty path (like 'github.com/org/rep/tree/branch/')
-            for i, part in enumerate(path[2:] + [""], start=2):
-                if sha in branches:
-                    path = path[i:]
-                    break
-                sha = f"{sha}%2F{part}"
+            if is_long_sha(sha):
+                path = path[2:]
             else:
-                raise ValueError(f'Could not resolve branch from uri "{uri}"')
+                refs = ls_github_branches(org, repo)
+                refs.update(ls_github_tags(org, repo))
+                branches = {quote_plus(k) for k in refs}
+                # match beginning of path with one of existing branches
+                # "" is hack for cases with empty path (like 'github.com/org/rep/tree/branch/')
+                for i, part in enumerate(path[2:] + [""], start=2):
+                    if sha in branches:
+                        path = path[i:]
+                        break
+                    sha = f"{sha}%2F{part}"
+                else:
+                    raise ValueError(
+                        f'Could not resolve branch from uri "{uri}"'
+                    )
         else:
             sha = None
         return {

tests/contrib/test_github.py

@@ -5,6 +5,7 @@
 from mlem.contrib.github import (
     GithubResolver,
     github_check_rev,
+    is_long_sha,
     ls_branches,
     ls_github_branches,
 )
@@ -91,3 +92,10 @@ def test_github_check_rev():
     assert github_check_rev(
         MLEM_TEST_REPO_ORG, MLEM_TEST_REPO_NAME, "first_rev_link"
     )  # tag
+
+
+def test_is_long_sha():
+    assert is_long_sha("cd7c2a08911b697c3f80c73d0394fb105d3044d5")
+    assert not is_long_sha("cd7c2a08911b697c3f80c73d0394fb105d3044d51")
+    assert not is_long_sha("cd7c2a08911b697c3f80c73d0394fb105d3044dA")
+    assert not is_long_sha("cd7c2a08911b697c3f80c73d0394fb105d3044d")