Add support for commit and branches

Davipar · Dec 10, 2024 · c2432cf · c2432cf
1 parent b422608
commit c2432cf
Show file tree

Hide file tree

Showing 3 changed files with 65 additions and 30 deletions.
diff --git a/src/ingest.py b/src/ingest.py
@@ -117,23 +117,19 @@ def create_summary_string(query: dict, nodes: Dict, files: List[Dict], ) -> str:
     """Creates a summary string with file counts and content size."""
     total_lines = sum(len(file["content"].splitlines()) for file in files)
 
-    if query['branch']:
-        if len(query['branch']) > 20:
-            branch = query['branch'][:20] + '...'
-        else:
-            branch = query['branch']
-    else:
-        branch = "main"
+    summary = f"Repository: {query['user_name']}/{query['repo_name']}\n"
+    summary += f"Files analyzed: {nodes['file_count']}\n"
+    summary += f"Directories analyzed: {nodes['dir_count']}\n"
+    summary += f"Total lines of content: {total_lines:,}\n"
+    if query['subpath'] != '/':
+        summary += f"Subpath: {query['subpath']}\n"
+    if query['commit']:
+        summary += f"Commit: {query['commit']}\n"
+    elif query['branch'] != 'main' and query['branch'] != 'master':
+        summary += f"Branch: {query['branch']}\n"
+    return summary
+
 
-
-    return (
-        f"Repository: {query['user_name']}/{query['repo_name']}\n"
-        f"Files analyzed: {nodes['file_count']}\n"
-        f"Directories analyzed: {nodes['dir_count']}\n"
-        f"Total lines of content: {total_lines:,}\n"
-        f"Subpath: {query['subpath']}\n"
-        # f"branch: {branch}\n"
-    )
 
 def create_tree_structure(query: dict, node: Dict, prefix: str = "", is_last: bool = True) -> str:
     """Creates a tree-like string representation of the file structure."""

diff --git a/src/utils/clone.py b/src/utils/clone.py
@@ -2,22 +2,57 @@
 import os
 
 from .decorators import async_timeout
-from config import TMP_BASE_PATH, CLONE_TIMEOUT
+from config import CLONE_TIMEOUT
 
 
 @async_timeout(CLONE_TIMEOUT)
 async def clone_repo(query: dict) -> str:
-    #Clean up any existing repo 
-
-    proc = await asyncio.create_subprocess_exec(
-        "git",
-        "clone",
-        "--depth=1",
-        "--single-branch",
-        query['url'],
-        query['local_path'],
-        stdout=asyncio.subprocess.PIPE,
-        stderr=asyncio.subprocess.PIPE,
-    )
 
-    stdout, stderr = await proc.communicate()
+    if query['commit']:
+        proc = await asyncio.create_subprocess_exec(
+            "git", 
+            "clone",
+            "--single-branch",
+            query['url'],
+            query['local_path'],
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.PIPE,
+        )
+        await proc.communicate()
+
+        proc = await asyncio.create_subprocess_exec(
+            "git",
+            "-C",
+            query['local_path'],
+            "checkout",
+            query['branch'],
+            stdout=asyncio.subprocess.PIPE, 
+            stderr=asyncio.subprocess.PIPE,
+        )
+        await proc.communicate()
+    elif query['branch'] != 'main' and query['branch'] != 'master':
+        proc = await asyncio.create_subprocess_exec(
+            "git",
+            "clone", 
+            "--depth=1",
+            "--single-branch",
+            "--branch",
+            query['branch'],
+            query['url'],
+            query['local_path'],
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.PIPE,
+        )
+    else:
+        proc = await asyncio.create_subprocess_exec(
+            "git",
+            "clone",
+            "--depth=1",
+            "--single-branch",
+            query['url'],
+            query['local_path'],
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.PIPE,
+        )
+
+    await proc.communicate()
diff --git a/src/utils/parse_url.py b/src/utils/parse_url.py
@@ -10,6 +10,7 @@ def parse_url(url: str) -> dict:
         "repo_name": None,
         "type": None,
         "branch": None,
+        "commit": None,
         "subpath": "/",
         "local_path": None,
         "url": None,
@@ -40,5 +41,8 @@ def parse_url(url: str) -> dict:
     if len(path_parts) > 3:
         parsed["type"] = path_parts[2]
         parsed["branch"] = path_parts[3]
+        if len(parsed['branch']) == 40 and all(c in '0123456789abcdefABCDEF' for c in parsed['branch']):
+            parsed["commit"] = parsed['branch']
+
         parsed["subpath"] = "/" + "/".join(path_parts[4:])
     return parsed