CLI Git Improvements (#12311)

- delete repo sources like pip
- git dep fixes
- error messaging

libs/cli/langchain_cli/utils/git.py

@@ -3,7 +3,11 @@ from pathlib import Path
 
 import shutil
 import re
-from langchain_cli.constants import DEFAULT_GIT_REPO, DEFAULT_GIT_SUBDIRECTORY
+from langchain_cli.constants import (
+    DEFAULT_GIT_REPO,
+    DEFAULT_GIT_SUBDIRECTORY,
+    DEFAULT_GIT_REF,
+)
 import hashlib
 from git import Repo
 
@@ -14,65 +18,49 @@ class DependencySource(TypedDict):
     subdirectory: Optional[str]
 
 
-def _get_main_branch(repo: Repo) -> Optional[str]:
-    """
-    Get the name of the main branch of a git repo.
-    From https://stackoverflow.com/questions/69651536/how-to-get-master-main-branch-from-gitpython
-    """
-    try:
-        # replace "origin" with your remote name if differs
-        show_result = repo.git.remote("show", "origin")
-
-        # The show_result contains a wall of text in the language that
-        # is set by your locales. Now you can use regex to extract the
-        # default branch name, but if your language is different
-        # from english, you need to adjust this regex pattern.
-
-        matches = re.search(r"\s*HEAD branch:\s*(.*)", show_result)
-        if matches:
-            default_branch = matches.group(1)
-            return default_branch
-    except Exception:
-        pass
-    # fallback to main/master
-    if "main" in repo.heads:
-        return "main"
-    if "master" in repo.heads:
-        return "master"
-
-    raise ValueError("Could not find main branch")
-
-
 # use poetry dependency string format
-def _parse_dependency_string(package_string: str) -> DependencySource:
+def parse_dependency_string(package_string: str) -> DependencySource:
     if package_string.startswith("git+"):
         # remove git+
-        remaining = package_string[4:]
-        # split main string from params
-        gitstring, *params = remaining.split("#")
-        # parse params
-        params_dict = {}
-        for param in params:
-            if not param:
-                # ignore empty entries
-                continue
-            if "=" in param:
-                key, value = param.split("=")
-                if key in params_dict:
-                    raise ValueError(
-                        f"Duplicate parameter {key} in dependency string {package_string}"
-                    )
-                params_dict[key] = value
-            else:
-                if "ref" in params_dict:
-                    raise ValueError(
-                        f"Duplicate parameter ref in dependency string {package_string}"
-                    )
-                params_dict["ref"] = param
+        gitstring = package_string[4:]
+        subdirectory = None
+        ref = None
+        # first check for #subdirectory= on the end
+        if "#subdirectory=" in gitstring:
+            gitstring, subdirectory = gitstring.split("#subdirectory=")
+            if "#" in subdirectory or "@" in subdirectory:
+                raise ValueError(
+                    "#subdirectory must be the last part of the dependency string"
+                )
+
+        # find first slash after ://
+        # find @ or # after that slash
+        # remainder is ref
+        # if no @ or #, then ref is None
+
+        # find first slash after ://
+        if "://" not in gitstring:
+            raise ValueError(
+                "git+ dependencies must start with git+https:// or git+ssh://"
+            )
+
+        _, find_slash = gitstring.split("://", 1)
+
+        if "/" not in find_slash:
+            post_slash = find_slash
+            ref = None
+        else:
+            _, post_slash = find_slash.split("/", 1)
+            if "@" in post_slash or "#" in post_slash:
+                _, ref = re.split(r"[@#]", post_slash, 1)
+
+        # gitstring is everything before that
+        gitstring = gitstring[: -len(ref) - 1] if ref is not None else gitstring
+
         return DependencySource(
             git=gitstring,
-            ref=params_dict.get("ref"),
-            subdirectory=params_dict.get("subdirectory"),
+            ref=ref,
+            subdirectory=subdirectory,
         )
 
     elif package_string.startswith("https://"):
@@ -81,12 +69,13 @@ def _parse_dependency_string(package_string: str) -> DependencySource:
         # it's a default git repo dependency
         gitstring = DEFAULT_GIT_REPO
         subdirectory = str(Path(DEFAULT_GIT_SUBDIRECTORY) / package_string)
-        return DependencySource(git=gitstring, ref=None, subdirectory=subdirectory)
+        return DependencySource(
+            git=gitstring, ref=DEFAULT_GIT_REF, subdirectory=subdirectory
+        )
 
 
-def _get_repo_path(dependency: DependencySource, repo_dir: Path) -> Path:
+def _get_repo_path(gitstring: str, repo_dir: Path) -> Path:
     # only based on git for now
-    gitstring = dependency["git"]
     hashed = hashlib.sha256(gitstring.encode("utf-8")).hexdigest()[:8]
 
     removed_protocol = gitstring.split("://")[-1]
@@ -98,32 +87,27 @@ def _get_repo_path(dependency: DependencySource, repo_dir: Path) -> Path:
     return repo_dir / directory_name
 
 
-def update_repo(gitpath: str, repo_dir: Path) -> Path:
+def update_repo(gitstring: str, ref: Optional[str], repo_dir: Path) -> Path:
     # see if path already saved
-    dependency = _parse_dependency_string(gitpath)
-    repo_path = _get_repo_path(dependency, repo_dir)
-    if not repo_path.exists():
-        repo = Repo.clone_from(dependency["git"], repo_path)
-    else:
-        repo = Repo(repo_path)
+    repo_path = _get_repo_path(gitstring, repo_dir)
+    if repo_path.exists():
+        shutil.rmtree(repo_path)
 
-    # pull it
-    ref = dependency.get("ref") if dependency.get("ref") else _get_main_branch(repo)
-    repo.git.checkout(ref)
-
-    repo.git.pull()
-
-    return (
-        repo_path
-        if dependency["subdirectory"] is None
-        else repo_path / dependency["subdirectory"]
-    )
+    # now we have fresh dir
+    Repo.clone_from(gitstring, repo_path, branch=ref, depth=1)
+    return repo_path
 
 
 def copy_repo(
     source: Path,
     destination: Path,
 ) -> None:
+    """
+    Copies a repo, ignoring git folders.
+
+    Raises FileNotFound error if it can't find source
+    """
+
     def ignore_func(_, files):
         return [f for f in files if f == ".git"]