|
| 1 | +#!/usr/bin/env python |
| 2 | +# |
| 3 | +# This is free and unencumbered software released into the public domain. |
| 4 | +# See the UNLICENSE file for details. |
| 5 | +# |
| 6 | +# ------------------------------------------------------------------------ |
| 7 | +# maven.py |
| 8 | +# ------------------------------------------------------------------------ |
| 9 | +# Supporting library for harvesting metadata about Maven components. |
| 10 | +# Requires direct access to the backing storage of the repositories. |
| 11 | + |
| 12 | +import datetime, logging, os, pathlib, re |
| 13 | +import xml.etree.ElementTree as ET |
| 14 | + |
| 15 | +storage = "/opt/sonatype-work/nexus/storage" |
| 16 | +release_repos = ["releases", "thirdparty", "sonatype", "central", "ome-releases"] |
| 17 | +snapshot_repos = ["snapshots", "sonatype-snapshots", "ome-snapshots"] |
| 18 | + |
| 19 | +ts_allowance = 10 # maximum seconds difference in SNAPSHOT timestamp |
| 20 | + |
| 21 | +class XML: |
| 22 | + |
| 23 | + def __init__(self, source): |
| 24 | + self.source = source |
| 25 | + self.tree = ET.parse(source) |
| 26 | + XML._strip_ns(self.tree.getroot()) |
| 27 | + |
| 28 | + def elements(self, path): |
| 29 | + return self.tree.findall(path) |
| 30 | + |
| 31 | + def value(self, path): |
| 32 | + el = self.elements(path) |
| 33 | + assert len(el) <= 1 |
| 34 | + return None if len(el) == 0 else el[0].text |
| 35 | + |
| 36 | + @staticmethod |
| 37 | + def _strip_ns(el): |
| 38 | + """ |
| 39 | + Remove namespace prefixes from elements and attributes. |
| 40 | + Credit: https://stackoverflow.com/a/32552776/1207769 |
| 41 | + """ |
| 42 | + if el.tag.startswith("{"): |
| 43 | + el.tag = el.tag[el.tag.find("}")+1:] |
| 44 | + for k in list(el.attrib.keys()): |
| 45 | + if k.startswith("{"): |
| 46 | + k2 = k[k.find("}")+1:] |
| 47 | + el.attrib[k2] = el.attrib[k] |
| 48 | + del el.attrib[k] |
| 49 | + for child in el: |
| 50 | + XML._strip_ns(child) |
| 51 | + |
| 52 | +class MavenPOM(XML): |
| 53 | + |
| 54 | + @property |
| 55 | + def groupId(self): |
| 56 | + return self.value("groupId") or self.value("parent/groupId") |
| 57 | + |
| 58 | + @property |
| 59 | + def artifactId(self): |
| 60 | + return self.value("artifactId") |
| 61 | + |
| 62 | + @property |
| 63 | + def version(self): |
| 64 | + return self.value("version") or self.value("parent/version") |
| 65 | + |
| 66 | + @property |
| 67 | + def scmURL(self): |
| 68 | + return self.value("scm/url") |
| 69 | + |
| 70 | + @property |
| 71 | + def issuesURL(self): |
| 72 | + return self.value("issueManagement/url") |
| 73 | + |
| 74 | + @property |
| 75 | + def ciURL(self): |
| 76 | + return self.value("ciManagement/url") |
| 77 | + |
| 78 | + @property |
| 79 | + def developers(self): |
| 80 | + devs = [] |
| 81 | + for el in self.elements("developers/developer"): |
| 82 | + dev = {} |
| 83 | + for child in el: |
| 84 | + if len(child) == 0: dev[child.tag] = child.text |
| 85 | + else: |
| 86 | + if child.tag == 'properties': |
| 87 | + dev[child.tag] = {grand.tag: grand.text for grand in child} |
| 88 | + else: |
| 89 | + dev[child.tag] = [grand.text for grand in child] |
| 90 | + devs.append(dev) |
| 91 | + return devs |
| 92 | + |
| 93 | + @property |
| 94 | + def reviewers(self): |
| 95 | + return self._dev_role("reviewer") |
| 96 | + |
| 97 | + @property |
| 98 | + def support(self): |
| 99 | + return self._dev_role("support") |
| 100 | + |
| 101 | + @property |
| 102 | + def maintainers(self): |
| 103 | + return self._dev_role("maintainer") |
| 104 | + |
| 105 | + def _dev_role(self, role): |
| 106 | + return [dev["id"] for dev in self.developers if role in dev["roles"]] |
| 107 | + |
| 108 | +class MavenMetadata(XML): |
| 109 | + |
| 110 | + @property |
| 111 | + def groupId(self): |
| 112 | + try: |
| 113 | + return self.value("groupId") |
| 114 | + except Exception: |
| 115 | + return self.value("parent/groupId") |
| 116 | + |
| 117 | + @property |
| 118 | + def artifactId(self): |
| 119 | + return self.value("artifactId") |
| 120 | + |
| 121 | + @property |
| 122 | + def lastUpdated(self): |
| 123 | + result = self.value("versioning/lastUpdated") |
| 124 | + return None if result is None else int(result) |
| 125 | + |
| 126 | + @property |
| 127 | + def latest(self): |
| 128 | + # WARNING: The <latest> value is often wrong, for reasons I don't know. |
| 129 | + # However, the last <version> under <versions> has the correct value. |
| 130 | + # Consider using lastVersion instead of latest. |
| 131 | + return self.value("versioning/latest") |
| 132 | + |
| 133 | + @property |
| 134 | + def lastVersion(self): |
| 135 | + vs = self.elements("versioning/versions/version") |
| 136 | + return None if len(vs) == 0 else vs[-1].text |
| 137 | + |
| 138 | + @property |
| 139 | + def release(self): |
| 140 | + return self.value("versioning/release") |
| 141 | + |
| 142 | +class MavenComponent: |
| 143 | + |
| 144 | + def __init__(self, g, a): |
| 145 | + self.groupId = g |
| 146 | + self.artifactId = a |
| 147 | + self.release = MavenComponent._metadata(release_repos, g, a) |
| 148 | + self.snapshot = MavenComponent._metadata(snapshot_repos, g, a) |
| 149 | + if self.snapshot: |
| 150 | + # Get the newest POM possible, based on last updated SNAPSHOT. |
| 151 | + self.pom = MavenComponent._pom(snapshot_repos, g, a, v=self.snapshot.lastVersion, |
| 152 | + ts=str(self.snapshot.lastUpdated)) |
| 153 | + elif self.release: |
| 154 | + # Get the POM of the newest release. |
| 155 | + self.pom = MavenComponent._pom(release_repos, g, a, v=self.release.lastVersion) |
| 156 | + else: |
| 157 | + self.pom = None |
| 158 | + |
| 159 | + @staticmethod |
| 160 | + def _metadata(repos, g, a): |
| 161 | + suffix = f"{g.replace('.', '/')}/{a}/maven-metadata.xml" |
| 162 | + best = None |
| 163 | + for repo in repos: |
| 164 | + path = f"{storage}/{repo}/{suffix}" |
| 165 | + if os.path.exists(path): |
| 166 | + m = MavenMetadata(path) |
| 167 | + if best is None or (m.lastUpdated is not None and m.lastUpdated > best.lastUpdated): |
| 168 | + best = m |
| 169 | + return best |
| 170 | + |
| 171 | + @staticmethod |
| 172 | + def _ts2dt(ts): |
| 173 | + """ |
| 174 | + Converts Maven-style timestamp strings into Python datetime objects. |
| 175 | +
|
| 176 | + Valid forms: |
| 177 | + * 20210702144918 (seen in <lastUpdated> in maven-metadata.xml) |
| 178 | + * 20210702.144917 (seen in deployed SNAPSHOT filenames) |
| 179 | + """ |
| 180 | + m = re.match("(\d{4})(\d\d)(\d\d)\.?(\d\d)(\d\d)(\d\d)", ts) |
| 181 | + if not m: raise ValueError(f"Invalid timestamp: {ts}") |
| 182 | + return datetime.datetime(*map(int, m.groups())) |
| 183 | + |
| 184 | + @staticmethod |
| 185 | + def _pom(repos, g, a, v, ts=None): |
| 186 | + gav_path = f"{g.replace('.', '/')}/{a}/{v}" |
| 187 | + if v.endswith("-SNAPSHOT"): |
| 188 | + # Find snapshot POM with matching timestamp. |
| 189 | + assert ts is not None |
| 190 | + dt_requested = MavenComponent._ts2dt(ts) |
| 191 | + pom_prefix = f"{a}-{v[:-9]}" # artifactId-version minus -SNAPSHOT |
| 192 | + for repo in repos: |
| 193 | + d = pathlib.Path(f"{storage}/{repo}/{gav_path}") |
| 194 | + for f in d.glob(f"{pom_prefix}-*.pom"): |
| 195 | + m = re.match(pom_prefix + "-(\d{8}\.\d{6})-\d+\.pom", f.name) |
| 196 | + if not m: continue # ignore weirdly named POM |
| 197 | + dt_actual = MavenComponent._ts2dt(m.group(1)) |
| 198 | + if abs(dt_requested - dt_actual).seconds <= ts_allowance: |
| 199 | + # Timestamp is within tolerance! Found it! |
| 200 | + return MavenPOM(str(f)) |
| 201 | + else: |
| 202 | + # Find release POM. |
| 203 | + suffix = f"{gav_path}/{a}-{v}.pom" |
| 204 | + for repo in repos: |
| 205 | + path = f"{storage}/{repo}/{suffix}" |
| 206 | + if os.path.exists(path): |
| 207 | + return MavenPOM(path) |
| 208 | + return None |
0 commit comments