refactor: validate repos with git commands & improve readability

eduNEXT · May 24, 2024 · bf35153 · bf35153
1 parent 7c574c0
commit bf35153
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 19 deletions.
diff --git a/tutordistro/distro/repository_validator/infrastructure/git_package_repository.py b/tutordistro/distro/repository_validator/infrastructure/git_package_repository.py
@@ -5,7 +5,7 @@
 """
 
 
-import requests
+import subprocess
 
 from tutordistro.distro.share.domain.cloud_package import CloudPackage
 from tutordistro.distro.share.domain.cloud_package_repository import CloudPackageRepository
@@ -20,6 +20,19 @@ class GitPackageRepository(CloudPackageRepository):
     the validation method.
     """
     def validate(self, package: CloudPackage) -> None:
-        response = requests.get(package.to_url(), timeout=5)
-        if response.status_code != 200:
-            raise PackageDoesNotExist(f"The package {package.name} or branch doesn't exist or is private")
+        package_url = package.to_url()
+        repo_url = package_url.split('/tree/')[0]
+        version_name = package_url.split('/tree/')[1]
+
+        # Verify that exist the repository
+        repo_verification_output = subprocess.run(f'git ls-remote {repo_url}', shell=True, capture_output=True)
+        if repo_verification_output.returncode != 0:
+            raise PackageDoesNotExist(f'The package "{repo_url}" does not exist or is private')
+
+        # Verify that the branch/tag is valid
+        branch_verification_output = subprocess.run(f'git ls-remote --heads "{repo_url}" "{version_name}" | grep "refs/heads/{version_name}"', shell=True, capture_output=True)
+
+        tag_verification_output = subprocess.run(f'git ls-remote --tags {repo_url} | grep "refs/tags/{version_name}"', shell=True, capture_output=True)
+
+        if branch_verification_output.returncode != 0 and tag_verification_output.returncode != 0:
+            raise PackageDoesNotExist(f'Neither branch nor tag "{version_name}" exists on "{repo_url}"')  
diff --git a/tutordistro/distro/share/domain/cloud_package.py b/tutordistro/distro/share/domain/cloud_package.py
@@ -53,30 +53,27 @@ def __parse_url(url) -> CloudPackage:
         version: str = ""
 
         pattern = r"git\+(https?://\S+?)(?:#|$)"
-        result = re.search(pattern, url)
-        url = result.group(1).replace('@', '/tree/').replace('.git', '')
+        found_package_url = re.search(pattern, url).group(1)
+        github_url = found_package_url.replace('@', '/tree/').replace('.git', '')
 
-        parsed_url = urlparse(url)
+        parsed_url = urlparse(github_url)
 
         protocol = parsed_url.scheme
         domain = parsed_url.netloc
-        path = parsed_url.path
-        partes_path = path.split('/')
-        name = partes_path[2]
-
-        if len(partes_path) > 5:
-            raise PackageDoesNotExist(f"The package {url} or branch doesn't exist or is private")
+        full_path = parsed_url.path
+        split_path = full_path.split('/')
+        package_name = split_path[2]
 
-        if '/tree/' in url:
-            version = partes_path[-1]
-        if len(partes_path) > 3 and '/tree/' not in url:
-            raise PackageDoesNotExist(f"The package {url} or branch doesn't exist or is private")
+        version_name = found_package_url.split("@")[-1] # This is the branch name or tag
+
+        if '/tree/' in github_url:
+            version = version_name
 
-        path = partes_path[1]
+        path = split_path[1]
 
         return CloudPackage(
             domain=domain,
-            name=name,
+            name=package_name,
             version=version,
             protocol=protocol,
             path=path