| 61 | |
| 62 | |
| 63 | def download_file(url: AnyStr, name: AnyStr, download_if_exists: bool = True): |
| 64 | logger.info(f"Trying to download {name} from {url}.") |
| 65 | if not download_if_exists and os.path.exists(name): |
| 66 | logger.info(f"File {name} already exists.") |
| 67 | return |
| 68 | |
| 69 | if is_file_uri(url): |
| 70 | # url uses 'file://' scheme |
| 71 | shutil.copy2(file_uri_to_path(url), name) |
| 72 | logger.info(f"File {name} was copied from {url}.") |
| 73 | return |
| 74 | |
| 75 | tmp_name = f"{name}__" |
| 76 | os.makedirs(os.path.dirname(tmp_name), exist_ok=True) |
| 77 | with requests.Session() as session: |
| 78 | session.mount("file://", FileAdapter()) |
| 79 | with open(tmp_name, "wb") as handle: |
| 80 | response = session.get(url, stream=True) |
| 81 | file_length = None |
| 82 | try: |
| 83 | file_length = int(response.headers["Content-Length"]) |
| 84 | except KeyError: |
| 85 | logger.warning( |
| 86 | f"There is no attribute Content-Length in headers [{url}]: {response.headers}" |
| 87 | ) |
| 88 | |
| 89 | current = 0 |
| 90 | max_attempts = 32 |
| 91 | attempts = max_attempts |
| 92 | while attempts: |
| 93 | for data in response.iter_content(chunk_size=4096): |
| 94 | current += len(data) |
| 95 | handle.write(data) |
| 96 | |
| 97 | if file_length is None or file_length == current: |
| 98 | break |
| 99 | |
| 100 | logger.warning( |
| 101 | f"Download interrupted. Resuming download from {url}: {current}/{file_length}." |
| 102 | ) |
| 103 | headers = {"Range": f"bytes={current}-"} |
| 104 | response = session.get(url, headers=headers, stream=True) |
| 105 | attempts -= 1 |
| 106 | |
| 107 | assert ( |
| 108 | attempts > 0 |
| 109 | ), f"Maximum failed resuming download attempts of {max_attempts} is exceeded." |
| 110 | |
| 111 | shutil.move(tmp_name, name) |
| 112 | logger.info(f"File {name} was downloaded from {url}.") |
| 113 | |
| 114 | |
| 115 | def is_dir(url) -> bool: |