|
| 1 | +"""Module to convert remote URLs to valid Package URLs (PURLs). |
| 2 | +
|
| 3 | +Supports: GitHub, Bitbucket, SVN, SSH paths, and more. |
| 4 | +""" |
| 5 | + |
| 6 | +import re |
| 7 | +from typing import Optional |
| 8 | +from urllib.parse import urlparse |
| 9 | + |
| 10 | +from packageurl import PackageURL |
| 11 | +from tldextract import TLDExtract |
| 12 | + |
| 13 | +NO_FETCH_EXTRACT = TLDExtract(suffix_list_urls=()) |
| 14 | +SSH_REGEX = re.compile( |
| 15 | + r"^(?:git@|git\+ssh://git@)(?P<host>[^/:]+)[:/](?P<path>.+?)(?:\.git)?$" |
| 16 | +) |
| 17 | + |
| 18 | +GITHUB_REGEX = re.compile( |
| 19 | + r".*github\.com(?::\d+)?[:/](?P<org>[^/]+)/(?P<repo>[^/]+?)(?:\.git)?$", |
| 20 | + re.IGNORECASE, |
| 21 | +) |
| 22 | + |
| 23 | +BITBUCKET_REGEX = re.compile( |
| 24 | + r".*bitbucket\.org(?::\d+)?[:/](?P<org>[^/]+)/(?P<repo>[^/]+?)(?:\.git)?$", |
| 25 | + re.IGNORECASE, |
| 26 | +) |
| 27 | + |
| 28 | + |
| 29 | +def _handle_github( |
| 30 | + remote_url: str, |
| 31 | + _: str, |
| 32 | + version: Optional[str], |
| 33 | + subpath: Optional[str], |
| 34 | +) -> Optional[PackageURL]: |
| 35 | + """Handler for GitHub URLs.""" |
| 36 | + match = GITHUB_REGEX.match(remote_url) |
| 37 | + if match: |
| 38 | + return PackageURL( |
| 39 | + type="github", |
| 40 | + namespace=match.group("org"), |
| 41 | + name=match.group("repo"), |
| 42 | + version=version, |
| 43 | + subpath=subpath, |
| 44 | + ) |
| 45 | + return None |
| 46 | + |
| 47 | + |
| 48 | +def _handle_bitbucket( |
| 49 | + remote_url: str, |
| 50 | + _: str, |
| 51 | + version: Optional[str], |
| 52 | + subpath: Optional[str], |
| 53 | +) -> Optional[PackageURL]: |
| 54 | + """Handler for Bitbucket URLs.""" |
| 55 | + match = BITBUCKET_REGEX.match(remote_url) |
| 56 | + if match: |
| 57 | + return PackageURL( |
| 58 | + type="bitbucket", |
| 59 | + namespace=match.group("org"), |
| 60 | + name=match.group("repo"), |
| 61 | + version=version, |
| 62 | + subpath=subpath, |
| 63 | + ) |
| 64 | + return None |
| 65 | + |
| 66 | + |
| 67 | +def _handle_svn( |
| 68 | + remote_url: str, |
| 69 | + path: str, |
| 70 | + version: Optional[str], |
| 71 | + subpath: Optional[str], |
| 72 | +) -> Optional[PackageURL]: |
| 73 | + """Handler for SVN URLs.""" |
| 74 | + parsed = urlparse(remote_url) |
| 75 | + if "svn" in parsed.scheme or "svn." in parsed.netloc: |
| 76 | + domain = NO_FETCH_EXTRACT(parsed.netloc).domain |
| 77 | + |
| 78 | + parts: list[str] = [domain] |
| 79 | + if path: |
| 80 | + parts.extend(path.split("/")) |
| 81 | + name = parts[-1] if parts else "unknown" |
| 82 | + namespace = "/".join(parts[:-1]) |
| 83 | + |
| 84 | + return PackageURL( |
| 85 | + type="generic", |
| 86 | + namespace=namespace, |
| 87 | + name=name, |
| 88 | + version=version, |
| 89 | + qualifiers={"vcs_url": remote_url}, |
| 90 | + subpath=subpath, |
| 91 | + ) |
| 92 | + return None |
| 93 | + |
| 94 | + |
| 95 | +def _handle_ssh( |
| 96 | + remote_url: str, |
| 97 | + path: str, |
| 98 | + version: Optional[str], |
| 99 | + subpath: Optional[str], |
| 100 | +) -> Optional[PackageURL]: |
| 101 | + """Handler for SSH URLs.""" |
| 102 | + match = SSH_REGEX.match(remote_url) |
| 103 | + if match: |
| 104 | + domain = NO_FETCH_EXTRACT(match.group("host")).domain |
| 105 | + |
| 106 | + parts: list[str] = [] |
| 107 | + if domain not in ["gitlab", "gitea", "gitee"]: |
| 108 | + parts.append(domain) |
| 109 | + |
| 110 | + path = match.group("path") |
| 111 | + if path: |
| 112 | + parts.extend(path.replace(".git", "").split("/")) |
| 113 | + name = parts[-1] if parts else "unknown" |
| 114 | + namespace = "/".join(parts[:-1]) |
| 115 | + return PackageURL( |
| 116 | + type="generic", |
| 117 | + namespace=namespace, |
| 118 | + name=name, |
| 119 | + version=version, |
| 120 | + qualifiers={"vcs_url": remote_url}, |
| 121 | + subpath=subpath, |
| 122 | + ) |
| 123 | + return None |
| 124 | + |
| 125 | + |
| 126 | +def _handle_generic( |
| 127 | + remote_url: str, |
| 128 | + path: str, |
| 129 | + version: Optional[str], |
| 130 | + subpath: Optional[str], |
| 131 | +) -> PackageURL: |
| 132 | + """Fallback handler for generic URLs.""" |
| 133 | + domain = NO_FETCH_EXTRACT(remote_url).domain |
| 134 | + |
| 135 | + parts: list[str] = [] |
| 136 | + if domain not in ["gitlab", "gitea", "gitee"]: |
| 137 | + parts.append(domain) |
| 138 | + |
| 139 | + if path: |
| 140 | + parts.extend(path.replace(".git", "").split("/")) |
| 141 | + name = parts[-1] if parts else "unknown" |
| 142 | + namespace = "/".join(parts[:-1]) |
| 143 | + return PackageURL( |
| 144 | + type="generic", |
| 145 | + namespace=namespace, |
| 146 | + name=name, |
| 147 | + version=version, |
| 148 | + qualifiers={"vcs_url": remote_url}, |
| 149 | + subpath=subpath, |
| 150 | + ) |
| 151 | + |
| 152 | + |
| 153 | +def remote_url_to_purl( |
| 154 | + remote_url: str, version: Optional[str] = None, subpath: Optional[str] = None |
| 155 | +) -> PackageURL: |
| 156 | + """Convert a remote URL to a valid PackageURL object. |
| 157 | +
|
| 158 | + Supports GitHub, Bitbucket, SVN, SSH paths. |
| 159 | + Optionally specify version and subpath. |
| 160 | + """ |
| 161 | + parsed = urlparse(remote_url) |
| 162 | + path = parsed.path.lstrip("/") |
| 163 | + |
| 164 | + handlers: list = [ |
| 165 | + _handle_github, |
| 166 | + _handle_bitbucket, |
| 167 | + _handle_svn, |
| 168 | + _handle_ssh, |
| 169 | + ] |
| 170 | + for handler in handlers: |
| 171 | + result: Optional[PackageURL] = handler(remote_url, path, version, subpath) |
| 172 | + if result is not None: |
| 173 | + return result |
| 174 | + |
| 175 | + return _handle_generic(remote_url, path, version, subpath) |
0 commit comments