Extract and normalize duration from various formats. Args: duration_str: Duration string from feed Returns: Normalized duration string or None
(duration_str: str)
| 66 | return url |
| 67 | |
| 68 | def extract_duration(duration_str: str) -> str | None: |
| 69 | """ |
| 70 | Extract and normalize duration from various formats. |
| 71 | |
| 72 | Args: |
| 73 | duration_str: Duration string from feed |
| 74 | |
| 75 | Returns: |
| 76 | Normalized duration string or None |
| 77 | """ |
| 78 | if not duration_str: |
| 79 | return None |
| 80 | |
| 81 | duration_str = duration_str.strip() |
| 82 | |
| 83 | # If already in ISO format like PT1H30M15S |
| 84 | if duration_str.startswith('PT'): |
| 85 | return duration_str |
| 86 | |
| 87 | # Handle common formats |
| 88 | |
| 89 | # Format: HH:MM:SS |
| 90 | if re.match(r'^\d+:\d+:\d+$', duration_str): |
| 91 | hours, minutes, seconds = map(int, duration_str.split(':')) |
| 92 | return f"PT{hours}H{minutes}M{seconds}S" |
| 93 | |
| 94 | # Format: MM:SS |
| 95 | if re.match(r'^\d+:\d+$', duration_str): |
| 96 | minutes, seconds = map(int, duration_str.split(':')) |
| 97 | return f"PT{minutes}M{seconds}S" |
| 98 | |
| 99 | # Format: seconds (as number) |
| 100 | if re.match(r'^\d+$', duration_str): |
| 101 | seconds = int(duration_str) |
| 102 | hours = seconds // 3600 |
| 103 | minutes = (seconds % 3600) // 60 |
| 104 | seconds = seconds % 60 |
| 105 | |
| 106 | result = "PT" |
| 107 | if hours > 0: |
| 108 | result += f"{hours}H" |
| 109 | if minutes > 0: |
| 110 | result += f"{minutes}M" |
| 111 | if seconds > 0 or (hours == 0 and minutes == 0): |
| 112 | result += f"{seconds}S" |
| 113 | |
| 114 | return result |
| 115 | |
| 116 | # Return as-is if we can't parse it |
| 117 | return duration_str |
| 118 | |
| 119 | def extract_guid(item: ET.Element) -> str | None: |
| 120 | """ |