style: format Python code with Black

tool/compare_commits.py

@@ -5,25 +5,29 @@
 
 cache_manager = get_cache_manager()
 
+
 def tag_format(tag, package_name, repo_name):
-    _, repo_name = repo_name.split("/") # splits owner and repo name
-    project_name = repo_name.split("-")[-1] # deals with lots of maven-<project_name> repos (e.g., surefire, etc)
-    tag_formats = set([
-        f"{tag}",
-        f"v{tag}",
-        f"v_{tag}",
-        f"r{tag}",
-        f"release-{tag}",
-        f"parent-{tag}",
-        # Below: further tag formats found in the AROMA paper, table 3: https://dl.acm.org/doi/pdf/10.1145/3643764
-        f"release/{tag}",
-        f"{tag}-release",
-        f"v.{tag}",
-    ] + [
-        f"{name}{suffix}"
-        for name in [package_name, repo_name, project_name]
-        for suffix in [f"@{tag}", f"-v{tag}", f"_v{tag}", f"-{tag}", f"_{tag}"]
-    ])
+    _, repo_name = repo_name.split("/")  # splits owner and repo name
+    project_name = repo_name.split("-")[-1]  # deals with lots of maven-<project_name> repos (e.g., surefire, etc)
+    tag_formats = set(
+        [
+            f"{tag}",
+            f"v{tag}",
+            f"v_{tag}",
+            f"r{tag}",
+            f"release-{tag}",
+            f"parent-{tag}",
+            # Below: further tag formats found in the AROMA paper, table 3: https://dl.acm.org/doi/pdf/10.1145/3643764
+            f"release/{tag}",
+            f"{tag}-release",
+            f"v.{tag}",
+        ]
+        + [
+            f"{name}{suffix}"
+            for name in [package_name, repo_name, project_name]
+            for suffix in [f"@{tag}", f"-v{tag}", f"_v{tag}", f"-{tag}", f"_{tag}"]
+        ]
+    )
 
     only_package_name, artifact_id_parts = None, None
     if "/" in package_name:  # NPM-based
@@ -45,6 +49,7 @@ def tag_format(tag, package_name, repo_name):
 
     return tag_formats
 
+
 def find_existing_tags(tag_formats, repo_name):
     for tag_format in tag_formats:
         tag_url = f"https://api.github.com/repos/{repo_name}/git/ref/tags/{tag_format}"
@@ -53,6 +58,7 @@ def find_existing_tags(tag_formats, repo_name):
             return tag_format
     return None
 
+
 def get_commit_info(commit):
     if commit.get("committer") is None:
         committer_login = "No committer info"
@@ -86,6 +92,7 @@ def get_commit_info(commit):
         "id": author_id,
     }
 
+
 def get_authors_from_response(url, data, package_info):
     result = {
         "repo": package_info.get("repo_pure"),
@@ -103,24 +110,29 @@ def get_authors_from_response(url, data, package_info):
             if not commit_info:
                 commit_info = get_commit_info(commit)
                 cache_manager.commit_comparison_cache.cache_authors_from_url(commit.get("url"), commit_info)
-            
+
             if commit_info:
                 authors_info.append(commit_info)
-        result.update({
-            "authors": authors_info,
-            "tag1": package_info.get("chosen_v1"),
-            "tag2": package_info.get("chosen_v2"),
-        })
+        result.update(
+            {
+                "authors": authors_info,
+                "tag1": package_info.get("chosen_v1"),
+                "tag2": package_info.get("chosen_v2"),
+            }
+        )
     else:
-        result.update({
-            "tag1": package_info.get("version1"),
-            "tag2": package_info.get("version2"),
-            "commits_info_message": "No commits found",
-            "status_code": 200,
-        })
+        result.update(
+            {
+                "tag1": package_info.get("version1"),
+                "tag2": package_info.get("version2"),
+                "commits_info_message": "No commits found",
+                "status_code": 200,
+            }
+        )
 
     return result
 
+
 def get_authors_from_tags(tag1, tag2, package, package_info):
     repo_name = package_info.get("repo_name")
     tag_formats_old = tag_format(tag1, package, repo_name)
@@ -129,7 +141,9 @@ def get_authors_from_tags(tag1, tag2, package, package_info):
     existing_tag_format_new = find_existing_tags(tag_formats_new, repo_name)
     category = package_info.get("message")
 
-    compare_url = f"https://api.github.com/repos/{repo_name}/compare/{existing_tag_format_old}...{existing_tag_format_new}"
+    compare_url = (
+        f"https://api.github.com/repos/{repo_name}/compare/{existing_tag_format_old}...{existing_tag_format_new}"
+    )
     response = make_github_request(compare_url, max_retries=2)
 
     if not response:
@@ -146,7 +160,7 @@ def get_authors_from_tags(tag1, tag2, package, package_info):
                 status_old = tag_old
                 old_tag_found = True
                 break
-        
+
         if not old_tag_found:
             for tag_new in tag_formats_new:
                 new_tag_url = f"https://api.github.com/repos/{repo_name}/git/ref/tags/{tag_new}"
@@ -167,14 +181,15 @@ def get_authors_from_tags(tag1, tag2, package, package_info):
 
     return get_authors_from_response(compare_url, response, package_info)
 
+
 def get_patch_authors(repo_name, patch_name, path, release_version_sha, headers):
     url = f"https://api.github.com/repos/{repo_name}/commits?path=.yarn/patches/{path}&sha={release_version_sha}"
     patch_info = {
         "patch_name": patch_name,
         "repo_name": repo_name,
         "commit_url": url,
     }
-    
+
     response = make_github_request(url, headers=headers)
     authors_info = []
     if response:
@@ -214,19 +229,24 @@ def get_patch_authors(repo_name, patch_name, path, release_version_sha, headers)
                         "c_type": committer_type,
                     }
                 )
-        patch_info.update({
-            "category": "patch",
-            "authors": authors_info,
-        })
+        patch_info.update(
+            {
+                "category": "patch",
+                "authors": authors_info,
+            }
+        )
     else:
-        patch_info.update({
-            "authors": None,
-            "error": True,
-            "error_message": response.status_code,
-        })
+        patch_info.update(
+            {
+                "authors": None,
+                "error": True,
+                "error_message": response.status_code,
+            }
+        )
 
     return patch_info
 
+
 def get_commit_authors(packages_data):
     logging.info("Getting commits for packages...")
     authors_per_package = {}
@@ -256,6 +276,7 @@ def get_commit_authors(packages_data):
 
     return authors_per_package
 
+
 def get_patch_commits(headers, repo_name, release_version, patch_data):
     logging.info("Getting commits for patches...")
     release_version_sha = cache_manager.github_cache.get_tag_to_sha(repo_name, release_version)
@@ -267,7 +288,9 @@ def get_patch_commits(headers, repo_name, release_version, patch_data):
             release_version_sha = response_json.get("object").get("sha")
         else:
             release_version_sha = None
-        cache_manager.github_cache.cache_tag_to_sha(repo_name, release_version, "No release found" if release_version_sha is None else release_version_sha)
+        cache_manager.github_cache.cache_tag_to_sha(
+            repo_name, release_version, "No release found" if release_version_sha is None else release_version_sha
+        )
     elif release_version_sha == "No release found":
         release_version_sha = None
 
@@ -295,7 +318,7 @@ def get_patch_commits(headers, repo_name, release_version, patch_data):
             }
             continue
 
-        data = cache_manager.commit_comparison_cache.get_patch_authors(repo_name, path, release_version_sha) 
+        data = cache_manager.commit_comparison_cache.get_patch_authors(repo_name, path, release_version_sha)
         if not data:
             # Cache miss, get authors from GitHub
             data = get_patch_authors(repo_name, changed_patch, path, release_version_sha, headers)
@@ -311,4 +334,4 @@ def get_commit_results(api_headers, repo_name, release_version, patch_data, pack
     authors_per_package_result = get_commit_authors(packages_data)
     commit_results = {**authors_per_patches_result, **authors_per_package_result}
 
-    return commit_results
+    return commit_results

tool/compare_packages.py

@@ -9,6 +9,7 @@
     "no_change": "No change",
 }
 
+
 def parse_dependencies(file_path):
     dependencies = {}
     with open(file_path, "r") as file:

tool/get_pr_info.py

@@ -117,12 +117,14 @@ def get_pr_info(data):
             if not pr_data:
                 if commit_node_id:
                     pr_info = fetch_pull_requests(commit_node_id)
-                    cache_manager.github_cache.cache_pr_info({
-                        "package": package,
-                        "commit_sha": commit_sha,
-                        "commit_node_id": commit_node_id,
-                        "pr_info": pr_info,
-                    })
+                    cache_manager.github_cache.cache_pr_info(
+                        {
+                            "package": package,
+                            "commit_sha": commit_sha,
+                            "commit_node_id": commit_node_id,
+                            "pr_info": pr_info,
+                        }
+                    )
             else:
                 pr_info = pr_data["pr_info"]
 
@@ -217,4 +219,4 @@ def get_useful_pr_info(commits_data):
 
                             author["commit_merged_info"].append(merged_info)
 
-    return commits_data
+    return commits_data

tool/get_pr_review.py

@@ -18,6 +18,7 @@
 
 url = "https://api.github.com/graphql"
 
+
 def get_first_pr_info(repo_name, review_author_login):
     query = """
     query($query: String!, $type: SearchType!, $last: Int!)
@@ -95,7 +96,9 @@ def get_pr_review_info(data):
                         if not first_pr_info:
                             if review_author_login:
                                 first_pr_info = get_first_pr_info(repo_name, review_author_login)
-                                cache_manager.github_cache.cache_pr_review(package, repo_name, review_author_login, first_pr_info)
+                                cache_manager.github_cache.cache_pr_review(
+                                    package, repo_name, review_author_login, first_pr_info
+                                )
                         useful_info = first_pr_info.get("data", {}).get("search", {}).get("nodes", [])
                         first_review_info = useful_info[0] if useful_info else {}
                         all_useful_first_prr_info = first_review_info.get("reviews", {}).get("edges", [])

tool/get_user_commit_info.py

@@ -10,6 +10,7 @@
 
 cache_manager = get_cache_manager()
 
+
 def get_repo_author_commits(api_url):
     # Since we can't return the commits in ascending date order, we'll just return the latest commit
     # This response also holds the number of pages, so the last page will have the first commit
@@ -21,6 +22,7 @@ def get_repo_author_commits(api_url):
     last_page_url = f"{search_url}&page={last_page}"
     return make_github_request(last_page_url, max_retries=2, retry_delay=2, sleep_between_requests=2)
 
+
 def get_user_first_commit_info(data):
     """
     Get the first commit information for each author in the given data.
@@ -69,7 +71,7 @@ def get_user_first_commit_info(data):
                             author_login_in_commit,
                             author_id_in_commit,
                         ) = data
-                        first_time_commit = (earliest_commit_sha == commit_sha)
+                        first_time_commit = earliest_commit_sha == commit_sha
 
                         commit_result.update(
                             {
@@ -89,18 +91,27 @@ def get_user_first_commit_info(data):
                             earliest_commit_sha = earliest_commit["sha"]
                             author_login_in_commit = earliest_commit["author"]["login"]
                             author_id_in_commit = earliest_commit["author"]["id"]
-                            first_time_commit = (earliest_commit_sha == commit_sha)
+                            first_time_commit = earliest_commit_sha == commit_sha
                             cache_manager.user_commit_cache.cache_user_commit(
-                                api_url, earliest_commit_sha, repo_name, package, author_login, commit_sha, author_login_in_commit, author_id_in_commit
+                                api_url,
+                                earliest_commit_sha,
+                                repo_name,
+                                package,
+                                author_login,
+                                commit_sha,
+                                author_login_in_commit,
+                                author_id_in_commit,
+                            )
+                            commit_result.update(
+                                {
+                                    "api_url": api_url,
+                                    "earliest_commit_sha": earliest_commit_sha,
+                                    "author_login_in_1st_commit": author_login_in_commit,
+                                    "author_id_in_1st_commit": author_id_in_commit,
+                                    "is_first_commit": first_time_commit,
+                                    "commit_notice": "Data retrieved from API",
+                                }
                             )
-                            commit_result.update({
-                                "api_url": api_url,
-                                "earliest_commit_sha": earliest_commit_sha,
-                                "author_login_in_1st_commit": author_login_in_commit,
-                                "author_id_in_1st_commit": author_id_in_commit,
-                                "is_first_commit": first_time_commit,
-                                "commit_notice": "Data retrieved from API",
-                            })
                         else:
                             commit_result["commit_notice"] = f"Failed to retrieve data from API({api_url})"
                 author["commit_result"] = commit_result

tool/github_repo.py

@@ -39,7 +39,7 @@ def extract_repo_url(repo_info: str) -> str:
     match = GITHUB_URL_PATTERN.search(repo_info)
     if not match:
         return "not github"
-    
+
     # if there is a match, there's still the possibility of the scm url having been
     # put in a different form, e.g.,
     # github.com/apache/maven-scm/tree/maven-scm-2.1.0/maven-scm-providers/maven-scm-providers-standard
@@ -50,6 +50,7 @@ def extract_repo_url(repo_info: str) -> str:
     joined = joined if not joined.endswith(".git") else joined[:-4]
     return joined
 
+
 def get_scm_commands(pm: str, package: str) -> List[str]:
     """Get the appropriate command to find a package's source code locations for the package manager."""
     if pm == "yarn-berry" or pm == "yarn-classic":
@@ -79,6 +80,7 @@ def get_scm_commands(pm: str, package: str) -> List[str]:
         ]
     raise ValueError(f"Unsupported package manager: {pm}")
 
+
 def process_package(
     package,
     pm,
@@ -89,12 +91,7 @@ def process_package(
     repos_output_json,
 ):
     def check_if_valid_repo_info(repo_info):
-        if (
-            repo_info is None
-            or "Undefined" in repo_info
-            or "undefined" in repo_info
-            or "ERR!" in repo_info
-        ):
+        if repo_info is None or "Undefined" in repo_info or "undefined" in repo_info or "ERR!" in repo_info:
             repos_output_json[package] = {"github": "Could not find"}
             undefined.append(f"Undefined for {package}, {repo_info}")
             return False
@@ -109,7 +106,6 @@ def check_if_valid_repo_info(repo_info):
             some_errors.append(f"No GitHub URL for {package}\n{repo_info}")
             return False
 
-
     repo_info = cache_manager.github_cache.get_github_url(package)
     valid_repo_info = False
     if not repo_info: