Implement argparse and output debug file

.github/scripts/release-notes/process_release_notes.py

@@ -2,19 +2,25 @@ import re
 import sys
 import subprocess
 import json
+import argparse
 from typing import List, Tuple
 
 def extract_jira_tickets(line: str) -> List[str]:
-    # Find all Jira tickets in format ABC-123 (with any prefix/suffix)
+    """Find all Jira tickets in format ABC-123 (with any prefix/suffix)"""
     return re.findall(r'[A-Z]+-\d+', line)
 
 def extract_pr_numbers(line: str) -> List[str]:
-    # Match PR numbers from GitHub format (#123)
+    """Match PR numbers from GitHub format (#123)"""
     return re.findall(r'#(\d+)', line)
 
-def extract_pr_url(line: str) -> List[str]:
+def extract_pr_url(line: str) -> str:
-    """Match PR URL from GitHub format https://github.com/foo/bar/pull/123"""
+    """Match PR URL from GitHub format https://github.com/foo/bar/pull/123
-    return re.findall(r'https://github\.com/[\w-]+/[\w.-]+/pull/\d+', line)
+
+    Returns:
+        The first PR URL found in the line, or empty string if no URL is found
+    """
+    matches = re.findall(r'https://github\.com/[\w-]+/[\w.-]+/pull/\d+', line)
+    return matches[0] if matches else ""
 
 def fetch_labels(github_pr_url: str) -> List[str]:
     """Fetch labels from a GitHub PR using the GitHub CLI."""
@@ -87,6 +93,7 @@ def process_file(input_file: str, release_app_label: str) -> Tuple[List[str], Li
     jira_tickets: List[str] = []
     pr_numbers: List[str] = []
     processed_lines: List[str] = []
+    debug_lines: List[str] = []
     #community_highlights: List[str] = []
 
     print("Processing file: ", input_file)
@@ -94,12 +101,17 @@ def process_file(input_file: str, release_app_label: str) -> Tuple[List[str], Li
     with open(input_file, 'r') as f:
         for line in f:
             line = line.strip()
-            should_process = line and not line.startswith('#')
+            should_process = line and line.startswith('* ')
 
             if should_process:
                 pr_url = extract_pr_url(line)
+                pr_labels = []
+
+                # Fetch labels from PR URL if available
+                if pr_url:
                     pr_labels = fetch_labels(pr_url)
                     if should_skip_pr(release_app_label, pr_labels):
+                        debug_lines.append(f"{line} | skipped - labels: {pr_labels}")
                         continue # skip the PR if it is not labeled with the app label
 
                 tickets = extract_jira_tickets(line)
@@ -108,8 +120,13 @@ def process_file(input_file: str, release_app_label: str) -> Tuple[List[str], Li
                 prs = extract_pr_numbers(line)
                 pr_numbers.extend(prs)
                 processed_lines.append(process_line(line))
+                debug_lines.append(f"{line} | labels: {pr_labels}")
             else:
                 processed_lines.append(line)
+                if line == "":
+                    debug_lines.append("")
+                else:
+                    debug_lines.append(f"{line} | skipped - processing")
 
 
     # Remove duplicates while preserving order
@@ -119,12 +136,14 @@ def process_file(input_file: str, release_app_label: str) -> Tuple[List[str], Li
     print("Jira tickets:", ",".join(jira_tickets))
     print("PR numbers:", ",".join(pr_numbers))
     print("Finished processing file: ", input_file)
-    return jira_tickets, pr_numbers, processed_lines
+    return jira_tickets, pr_numbers, processed_lines, debug_lines
 
-def save_results(jira_tickets: List[str], pr_numbers: List[str], processed_lines: List[str],
+def save_results(jira_tickets: List[str], pr_numbers: List[str], processed_lines: List[str], debug_lines: List[str],
                 jira_file: str = 'jira_tickets.txt',
                 pr_file: str = 'pr_numbers.txt',
-                processed_file: str = 'processed_notes.txt') -> None:
+                processed_file: str = 'processed_notes.txt',
+                debug_file: str = 'processed_notes_debug.txt'
+                ) -> None:
     with open(jira_file, 'w') as f:
         f.write('\n'.join(jira_tickets))
 
@@ -134,20 +153,64 @@ def save_results(jira_tickets: List[str], pr_numbers: List[str], processed_lines
     with open(processed_file, 'w') as f:
         f.write('\n'.join(processed_lines))
 
+    with open(debug_file, 'w') as f:
+        f.write('\n'.join(debug_lines))
+
+def parse_args():
+    """Parse command line arguments.
+
+    Returns:
+        Parsed arguments namespace
+    """
+    parser = argparse.ArgumentParser(
+        description='Process release notes by extracting Jira tickets and PR numbers, and cleaning up the text.'
+    )
+    parser.add_argument(
+        'release_app_label',
+        help='Filter PRs by app label (e.g., app:password-manager)'
+    )
+    parser.add_argument(
+        'input_file',
+        default='release_notes.txt',
+        help='Input file containing release notes (default: release_notes.txt)'
+    )
+    parser.add_argument(
+        '--processed-filepath',
+        default='processed_notes.txt',
+        help='Output file for processed notes (default: processed_notes.txt)'
+    )
+    parser.add_argument(
+        '--jira-filepath',
+        default='jira_tickets.txt',
+        help='Output file for Jira tickets (default: jira_tickets.txt)'
+    )
+    parser.add_argument(
+        '--pr-filepath',
+        default='pr_numbers.txt',
+        help='Output file for PR numbers (default: pr_numbers.txt)'
+    )
+
+    parser.add_argument(
+        '--debug-filepath',
+        default='processed_notes_debug.txt',
+        help='Output file for debug notes (default: processed_notes_debug.txt)'
+    )
+    return parser.parse_args()
+
 if __name__ == '__main__':
-    input_file = 'release_notes.txt'
+    args = parse_args()
-    jira_file = 'jira_tickets.txt'
-    pr_file = 'pr_numbers.txt'
-    processed_file = 'processed_notes.txt'
 
-    if len(sys.argv) >= 2:
+    jira_tickets, pr_numbers, processed_lines, debug_lines = process_file(
-        input_file = sys.argv[1]
+        args.input_file,
-    if len(sys.argv) >= 3:
+        args.release_app_label
-        jira_file = sys.argv[2]
+    )
-    if len(sys.argv) >= 4:
+    save_results(
-        pr_file = sys.argv[3]
+        jira_tickets,
-    if len(sys.argv) >= 5:
+        pr_numbers,
-        processed_file = sys.argv[4]
+        processed_lines,
-
+        debug_lines,
-    jira_tickets, pr_numbers, processed_lines = process_file(input_file)
+        args.jira_filepath,
-    save_results(jira_tickets, pr_numbers, processed_lines, jira_file, pr_file, processed_file)
+        args.pr_filepath,
+        args.processed_filepath,
+        args.debug_filepath
+    )

.github/scripts/release-notes/test_process_release_notes.py

@@ -54,9 +54,10 @@ class TestProcessReleaseNotes(unittest.TestCase):
 
     def test_extract_pr_url(self):
         test_cases = [
-            ("* Update SDK to 1.0.0-3436-2a00b727 by @bw-ghapp[bot] in https://github.com/bitwarden/android/pull/6042", ["https://github.com/bitwarden/android/pull/6042"]),
+            ("* Update SDK to 1.0.0-3436-2a00b727 by @bw-ghapp[bot] in https://github.com/bitwarden/android/pull/6042", "https://github.com/bitwarden/android/pull/6042"),
-            ("* Bump JUnit from 6.0.0 to 6.0.1 by SaintPatrick in https://github.com/bitwarden/android/pull/6149", ["https://github.com/bitwarden/android/pull/6149"]),
+            ("* Bump JUnit from 6.0.0 to 6.0.1 by SaintPatrick in https://github.com/bitwarden/android/pull/6149", "https://github.com/bitwarden/android/pull/6149"),
-            ("* [PM-26986] Hide select other account button if user has no other account by @aj-rosado in https://github.com/bitwarden/android/pull/6041", ["https://github.com/bitwarden/android/pull/6041"]),
+            ("* [PM-26986] Hide select other account button if user has no other account by @aj-rosado in https://github.com/bitwarden/android/pull/6041", "https://github.com/bitwarden/android/pull/6041"),
+            ("* No PR URL here", ""),
         ]
         for input_text, expected in test_cases:
             with self.subTest(input_text=input_text):