feat: add Asana ticket detection to ticket compliance check

pr_agent/tools/pr_code_suggestions.py

@@ -609,10 +609,44 @@ def dedent_code(self, relevant_file, relevant_lines_start, new_code_snippet):
                 original_initial_spaces = len(original_initial_line) - len(original_initial_line.lstrip()) # lstrip works both for spaces and tabs
                 suggested_initial_spaces = len(suggested_initial_line) - len(suggested_initial_line.lstrip())
                 delta_spaces = original_initial_spaces - suggested_initial_spaces
+                # Detect indentation character from original line
+                indent_char = '\t' if original_initial_line.startswith('\t') else ' '
                 if delta_spaces > 0:
-                    # Detect indentation character from original line
-                    indent_char = '\t' if original_initial_line.startswith('\t') else ' '
                     new_code_snippet = textwrap.indent(new_code_snippet, delta_spaces * indent_char).rstrip('\n')
+                elif delta_spaces < 0:
+                    # Remove exactly -delta_spaces leading spaces from each
+                    # non-empty line.  textwrap.dedent() strips the *common*
+                    # indent which may remove too much when lines have
+                    # varying indentation levels (Qodo bug #3).
+                    remove_count = -delta_spaces
+                    dedented_lines = []
+                    for dline in new_code_snippet.split('\n'):
+                        if dline.strip():
+                            leading = len(dline) - len(dline.lstrip())
+                            dedented_lines.append(dline[min(remove_count, leading):])
+                        else:
+                            dedented_lines.append(dline)
+                    new_code_snippet = '\n'.join(dedented_lines).rstrip('\n')
+                # Normalize all lines in the suggestion to use the original's
+                # indentation character. This fixes the bug where /improve
+                # replaces tabs with spaces in Go, Makefile, and other
+                # tab-indented codebases (#1858).
+                if indent_char == '\t':
+                    new_lines = []
+                    for line in new_code_snippet.split('\n'):
+                        stripped = line.lstrip(' ')
+                        leading_spaces = len(line) - len(stripped)
+                        if leading_spaces > 0:
+                            # Preserve remainder spaces when converting
+                            # to tabs.  e.g. 6 spaces → 1 tab + 2 spaces
+                            # instead of silently dropping 2 spaces
+                            # (Qodo bug #3).
+                            tabs = leading_spaces // 4
+                            remainder = leading_spaces % 4
+                            new_lines.append('\t' * tabs + ' ' * remainder + stripped)
+                        else:
+                            new_lines.append(line)
+                    new_code_snippet = '\n'.join(new_lines)
         except Exception as e:
             get_logger().error(f"Error when dedenting code snippet for file {relevant_file}, error: {e}")
 

pr_agent/tools/ticket_pr_compliance_check.py

@@ -35,6 +35,38 @@ def find_jira_tickets(text):
     return list(tickets)
 
 
+# Compiled Asana task patterns
+_ASANA_TASK_URL_PATTERN = re.compile(
+    r'https://app\.asana\.com/0/(\d+)/(\d+)'
+)
+_ASANA_TASK_SHORT_PATTERN = re.compile(
+    r'(?:^|[^A-Za-z0-9])(?:ASANA|asana)[- ]?(\d{12,20})',
+    re.IGNORECASE,
+)
+
+
+def find_asana_tickets(text: str) -> list:
+    """Extract Asana task references from text.
+
+    Supports both full Asana URLs and shorthand ``ASANA-123456789012``
+    format.  Returns a list of unique task URLs.
+
+    Args:
+        text: The text to scan for Asana task references.
+
+    Returns:
+        A list of Asana task URLs.
+    """
+    tickets = set()
+    for match in _ASANA_TASK_URL_PATTERN.finditer(text):
+        tickets.add(match.group(0))
+    for match in _ASANA_TASK_SHORT_PATTERN.finditer(text):
+        task_id = match.group(1)
+        if task_id:
+            tickets.add(f"https://app.asana.com/0/0/{task_id}")
+    return sorted(tickets)
+
+
 def extract_ticket_links_from_pr_description(pr_description, repo_path, base_url_html='https://github.com'):
     """
     Extract all ticket links from PR description
@@ -123,16 +155,42 @@ async def extract_tickets(git_provider):
                 if link not in seen:
                     seen.add(link)
                     merged.append(link)
+
+            # Also detect Asana ticket references in the PR description
+            asana_tickets = find_asana_tickets(user_description)
+            for link in asana_tickets:
+                if link not in seen:
+                    seen.add(link)
+                    merged.append(link)
+            asana_links = [t for t in merged if t.startswith("https://app.asana.com/")]
+            github_like = [t for t in merged if not t.startswith("https://app.asana.com/")]
             if len(merged) > 3:
                 get_logger().info(f"Too many tickets (description + branch): {len(merged)}")
-                tickets = merged[:3]
+                # Reserve at least one slot for an Asana reference when
+                # present so it is not systematically dropped.
+                asana_slot = asana_links[:1]
+                gh_slots = 3 - len(asana_slot)
+                tickets = github_like[:gh_slots] + asana_slot
             else:
                 tickets = merged
             tickets_content = []
 
             if tickets:
 
                 for ticket in tickets:
+                    # Skip Asana URLs — these are external references,
+                    # included for visibility but cannot be fetched via GitHub API.
+                    if ticket.startswith("https://app.asana.com/"):
+                        tickets_content.append({
+                            "ticket_id": ticket,
+                            "ticket_url": ticket,
+                            "title": f"Asana Task: {ticket}",
+                            "body": ("Asana task referenced in PR description. "
+                                     "Fetch task details from Asana for full context."),
+                            "labels": "",
+                        })
+                        continue
+
                     repo_name, original_issue_number = git_provider._parse_issue_url(ticket)
 
                     try:

tests/unittest/test_ticket_compliance.py

@@ -0,0 +1,96 @@
+"""
+Unit tests for Asana ticket detection in ticket_pr_compliance_check.py.
+
+Tests cover:
+- Full Asana URL detection
+- Shorthand ASANA- prefix detection
+- Edge cases (mixed content, no tickets, duplicates)
+"""
+from pr_agent.tools.ticket_pr_compliance_check import find_asana_tickets
+
+
+class TestFindAsanaTickets:
+    """Tests for find_asana_tickets()."""
+
+    def test_detects_full_asana_url(self):
+        """Full Asana task URLs should be detected."""
+        text = "See https://app.asana.com/0/123456/789012 for details"
+        tickets = find_asana_tickets(text)
+        assert "https://app.asana.com/0/123456/789012" in tickets
+
+    def test_detects_asana_shorthand(self):
+        """ASANA-123456789012 shorthand format should be detected."""
+        text = "Task ASANA-123456789012 is complete"
+        tickets = find_asana_tickets(text)
+        assert any("123456789012" in t for t in tickets)
+
+    def test_detects_multiple_tickets(self):
+        """Multiple Asana references should all be found."""
+        text = (
+            "See ASANA-111111111111 and https://app.asana.com/0/22/333333333333"
+        )
+        tickets = find_asana_tickets(text)
+        assert len(tickets) == 2
+
+    def test_deduplicates_identical_tickets(self):
+        """Duplicate references to the same task should be deduplicated."""
+        text = (
+            "ASANA-123456789012 mentioned twice: ASANA-123456789012 again"
+        )
+        tickets = find_asana_tickets(text)
+        assert len(tickets) == 1
+
+    def test_returns_empty_for_no_tickets(self):
+        """Text without Asana references returns an empty list."""
+        text = "No tickets here, just regular text"
+        tickets = find_asana_tickets(text)
+        assert tickets == []
+
+    def test_returns_empty_for_empty_string(self):
+        """Empty string returns an empty list."""
+        tickets = find_asana_tickets("")
+        assert tickets == []
+
+    def test_ignores_github_urls(self):
+        """GitHub issue URLs should not be mistaken for Asana tickets."""
+        text = "Fix https://github.com/owner/repo/issues/42"
+        tickets = find_asana_tickets(text)
+        assert tickets == []
+
+    def test_shorthand_is_case_insensitive(self):
+        """Both ASANA- and asana- should be detected."""
+        text = "asana-999999999999 lowercase works too"
+        tickets = find_asana_tickets(text)
+        assert len(tickets) == 1
+        assert "999999999999" in tickets[0]
+
+    def test_shorthand_mixed_case_asana(self):
+        """Mixed-case shorthand like Asana-123... should also be detected."""
+        text = "Asana-888888888888 mixed case"
+        tickets = find_asana_tickets(text)
+        assert len(tickets) == 1
+        assert "888888888888" in tickets[0]
+
+    def test_shorthand_respects_word_boundary(self):
+        """Text preceding the shorthand must not be alphanumeric.
+        NOTASANA-123 (no delimiter before ASANA) should not match."""
+        text = "Check NOTASANA-123456789012 in logs"
+        tickets = find_asana_tickets(text)
+        assert tickets == []
+
+    def test_tickets_are_sorted(self):
+        """Returned list should be sorted alphabetically."""
+        text = "ASANA-222222222222 ASANA-111111111111"
+        tickets = find_asana_tickets(text)
+        assert tickets == sorted(tickets)
+
+    def test_tickets_in_pr_description_mixed_content(self):
+        """Asana tickets mixed with other content in a PR description."""
+        text = """## Summary
+        Fixes ASANA-123456789012
+        Related to https://app.asana.com/0/99/888888888888
+
+        Also see GitHub issue #42
+        """
+        tickets = find_asana_tickets(text)
+        assert len(tickets) == 2