maven-proxy 1.3.1 → 1.3.2

package/README.md

@@ -2,6 +2,8 @@
 
 [English](README.md) | [中文](README-zh.md)
 
+[English ChangeLog](ChangeLog.md) | [中文更新日志](ChangeLog-zh.md)
+
 [![npm version](https://img.shields.io/npm/v/maven-proxy.svg)](https://www.npmjs.com/package/maven-proxy)
 
 ## 1. Project Positioning
@@ -275,6 +277,8 @@ Environment variables:
 - REPO_FALLBACK_REPOS: repository fallback list.
 - NPM_REGISTRY_DOMAINS: npm domains for ecosystem routing (wildcards supported).
 - MAVEN_REPO_DOMAINS: maven domains for ecosystem routing (wildcards supported).
+- MAVEN_CACHE_USE_DOMAIN_DIR: whether Maven cache uses hostname as the first-level directory. Default false.
+- MAVEN_CACHE_IGNORE_PATH_PREFIXES: Maven cache path-prefix ignore rules (comma-separated, supports host/path and host:port/path). Default: repo1.maven.org/maven2,repo.maven.apache.org/maven2,jitpack.io/,plugins.gradle.org/m2,dl.google.com/dl/android/maven2,dl.google.com/dl/google/maven.
 - HTTPS_MITM_DOMAINS: MITM domain list (includes registry.npmjs.org by default, wildcards supported).
 - DOWNLOAD_LOG_DIR: log directory.
 - LOG_RETENTION: log retention duration (supports s/m/h/d), for example 7d.
@@ -325,6 +329,8 @@ Priority:
 - `HTTPS_PASSTHROUGH_FOR_UNMATCHED`: Whether unmatched HTTPS domains are tunneled directly. Default `false`.
 - `NPM_REGISTRY_DOMAINS`: Domains treated as npm ecosystem for cache routing (wildcards supported).
 - `MAVEN_REPO_DOMAINS`: Domains treated as Maven ecosystem for cache routing (wildcards supported).
+- `MAVEN_CACHE_USE_DOMAIN_DIR`: Whether Maven cache uses hostname as the first-level directory. Default `false`.
+- `MAVEN_CACHE_IGNORE_PATH_PREFIXES`: Maven cache path-prefix ignore rules (comma-separated, supports `host/path` and `host:port/path`). Default `repo1.maven.org/maven2,repo.maven.apache.org/maven2,jitpack.io/,plugins.gradle.org/m2,dl.google.com/dl/android/maven2,dl.google.com/dl/google/maven`.
 - `MULTI_THREAD_DOMAINS`: Domains allowed to use multi-thread download (wildcards supported).
 - `MULTI_THREAD_COUNT`: Number of download threads for ranged downloads.
 - `MULTI_THREAD_MIN_SIZE_MB`: Minimum size threshold to trigger multi-thread download (MB).

package/bin/maven-proxy.js

@@ -179,6 +179,8 @@ function getDefaultConfigTemplate() {
   appendEntry("HTTPS_PASSTHROUGH_FOR_UNMATCHED", "false", "未命中 MITM 域名时是否允许隧道透传，默认 false。", "Allow tunnel passthrough for unmatched MITM domains. Default: false.");
   appendEntry("NPM_REGISTRY_DOMAINS", "registry.npmjs.org,registry.npmmirror.com,npm.pkg.github.com", "识别为 npm 生态并分流缓存的域名列表（支持通配符）。", "Domains recognized as npm ecosystem for cache routing (wildcard supported).");
   appendEntry("MAVEN_REPO_DOMAINS", "repo1.maven.org,repo.maven.apache.org,jitpack.io,plugins.gradle.org,dl.google.com", "识别为 Maven 生态并分流缓存的域名列表（支持通配符）。", "Domains recognized as Maven ecosystem for cache routing (wildcard supported).");
+  appendEntry("MAVEN_CACHE_USE_DOMAIN_DIR", "false", "Maven 缓存是否按域名作为一级目录，默认 false。", "Whether Maven cache uses hostname as the first-level directory. Default: false.");
+  appendEntry("MAVEN_CACHE_IGNORE_PATH_PREFIXES", "repo1.maven.org/maven2,repo.maven.apache.org/maven2,jitpack.io/,plugins.gradle.org/m2,dl.google.com/dl/android/maven2,dl.google.com/dl/google/maven", "Maven 缓存应忽略的路径前缀规则（逗号分隔，支持 host/path 与 host:port/path）。", "Maven cache path-prefix ignore rules (comma-separated, supports host/path and host:port/path).");
   appendEntry("MULTI_THREAD_DOMAINS", "repo1.maven.org", "启用多线程下载的域名列表（支持通配符）。", "Domains that enable multi-thread download (wildcard supported).");
   appendEntry("MULTI_THREAD_COUNT", "8", "多线程下载线程数，默认 8。", "Thread count for multi-thread download. Default: 8.");
   appendEntry("MULTI_THREAD_MIN_SIZE_MB", "1", "触发多线程下载的最小文件大小阈值（MB），默认 1。", "Minimum file size threshold in MB to trigger multi-thread download. Default: 1.");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "maven-proxy",
-  "version": "1.3.1",
+  "version": "1.3.2",
   "description": "Maven proxy with cache, HTTPS MITM for selected domains, and local repo publishing",
   "main": "src/index.js",
   "bin": {

package/src/cache/cache-path.js

@@ -16,6 +16,162 @@ function looksLikeMavenVersionSegment(segment) {
   return /^\d[0-9A-Za-z._-]*$/.test(String(segment || ""));
 }
 
+function normalizeSlashPath(rawPath) {
+  const decoded = safeDecode(rawPath || "/").replace(/\\/g, "/").replace(/\/+/g, "/");
+  if (!decoded) {
+    return "/";
+  }
+  return decoded.startsWith("/") ? decoded : `/${decoded}`;
+}
+
+function trimTrailingSlash(normalizedPath) {
+  if (normalizedPath.length > 1 && normalizedPath.endsWith("/")) {
+    return normalizedPath.slice(0, -1);
+  }
+  return normalizedPath;
+}
+
+function normalizeRulePrefixPath(rawPrefix) {
+  return trimTrailingSlash(normalizeSlashPath(rawPrefix || "/"));
+}
+
+function getEffectivePort(urlObj) {
+  if (urlObj?.port) {
+    return String(urlObj.port);
+  }
+  return urlObj?.protocol === "https:" ? "443" : "80";
+}
+
+function parseMavenCacheIgnorePathRule(rawRule) {
+  const text = String(rawRule || "").trim();
+  if (!text) {
+    return null;
+  }
+
+  const firstSlash = text.indexOf("/");
+  if (firstSlash <= 0) {
+    return null;
+  }
+
+  const hostPortText = text.slice(0, firstSlash).trim().toLowerCase();
+  const prefixText = text.slice(firstSlash);
+  const hostMatch = hostPortText.match(/^([^:\/]+)(?::(\d+))?$/);
+  if (!hostMatch) {
+    return null;
+  }
+
+  const hostname = hostMatch[1];
+  const port = hostMatch[2] || "";
+  const pathPrefix = normalizeRulePrefixPath(prefixText);
+
+  return {
+    hostname,
+    port,
+    pathPrefix,
+    raw: text,
+  };
+}
+
+function ruleMatchesUrl(rule, urlObj) {
+  if (!rule || !urlObj) {
+    return false;
+  }
+
+  const hostname = String(urlObj.hostname || "").toLowerCase();
+  if (!hostname || hostname !== rule.hostname) {
+    return false;
+  }
+
+  if (!rule.port) {
+    return true;
+  }
+
+  return getEffectivePort(urlObj) === rule.port;
+}
+
+export function parseMavenCacheIgnorePathPrefixes(rawRules) {
+  const input = Array.isArray(rawRules) ? rawRules.join(",") : String(rawRules || "");
+  const tokens = input
+    .split(",")
+    .map((item) => item.trim())
+    .filter(Boolean);
+
+  const parsed = [];
+  const dedupe = new Set();
+
+  for (const token of tokens) {
+    const rule = parseMavenCacheIgnorePathRule(token);
+    if (!rule) {
+      continue;
+    }
+
+    const key = `${rule.hostname}:${rule.port}|${rule.pathPrefix}`;
+    if (dedupe.has(key)) {
+      continue;
+    }
+
+    dedupe.add(key);
+    parsed.push(rule);
+  }
+
+  return parsed;
+}
+
+export function stripMavenIgnoredPathPrefix(pathname, urlObj, rules = []) {
+  const normalizedPath = normalizeSlashPath(pathname || "/");
+  const validRules = Array.isArray(rules) ? rules : [];
+  if (validRules.length === 0 || !urlObj) {
+    return normalizedPath;
+  }
+
+  const matchedRules = validRules
+    .filter((rule) => ruleMatchesUrl(rule, urlObj))
+    .sort((a, b) => b.pathPrefix.length - a.pathPrefix.length);
+
+  for (const rule of matchedRules) {
+    const prefix = rule.pathPrefix;
+    if (prefix === "/") {
+      return normalizedPath;
+    }
+
+    if (normalizedPath === prefix) {
+      return "/";
+    }
+
+    if (normalizedPath.startsWith(`${prefix}/`)) {
+      const stripped = normalizedPath.slice(prefix.length);
+      return stripped || "/";
+    }
+  }
+
+  return normalizedPath;
+}
+
+export function buildMavenHostlessPathCandidates(pathname, rules = []) {
+  const normalizedPath = normalizeSlashPath(pathname || "/");
+  const candidates = new Set([normalizedPath]);
+  const validRules = Array.isArray(rules) ? rules : [];
+
+  for (const rule of validRules) {
+    const prefix = rule.pathPrefix;
+    if (!prefix || prefix === "/") {
+      continue;
+    }
+
+    if (normalizedPath === prefix) {
+      candidates.add("/");
+      continue;
+    }
+
+    if (normalizedPath.startsWith(`${prefix}/`)) {
+      const stripped = normalizedPath.slice(prefix.length);
+      candidates.add(stripped || "/");
+    }
+  }
+
+  return [...candidates];
+}
+
 function isLikelyMavenFilePath(parts, normalizedPath) {
   if (normalizedPath.endsWith("/") || parts.length === 0) {
     return false;
@@ -63,9 +219,14 @@ function isLikelyMavenFilePath(parts, normalizedPath) {
 export function getCacheFilePath(cacheDir, urlObj, options = {}) {
   const ecosystem = sanitizeSegment(String(options.ecosystem || "generic").toLowerCase());
   const includeHost = options.includeHost ?? ecosystem !== "maven";
+  const mavenIgnoreRules = options.mavenCacheIgnorePathPrefixRules || [];
 
   const rawPathname = safeDecode(urlObj.pathname || "/");
-  const normalized = rawPathname.replace(/\\/g, "/");
+  let normalized = rawPathname.replace(/\\/g, "/");
+  if (ecosystem === "maven") {
+    normalized = stripMavenIgnoredPathPrefix(normalized, urlObj, mavenIgnoreRules);
+  }
+
   const lowerNormalized = normalized.toLowerCase();
   const parts = normalized.split("/").filter(Boolean);
 

package/src/config/config.js

@@ -3,6 +3,7 @@ import os from "node:os";
 import path from "node:path";
 import dotenv from "dotenv";
 import { detectJavaHome } from "../common/java-home.js";
+import { parseMavenCacheIgnorePathPrefixes } from "../cache/cache-path.js";
 
 const cwd = process.cwd();
 const userConfigDir = path.resolve(os.homedir(), "maven-proxy");
@@ -196,11 +197,24 @@ const defaultRepoFallbackRepos = [
   "https://dl.google.com",
 ];
 
+const defaultMavenCacheIgnorePathPrefixes = [
+  "repo1.maven.org/maven2",
+  "repo.maven.apache.org/maven2",
+  "jitpack.io/",
+  "plugins.gradle.org/m2",
+  "dl.google.com/dl/android/maven2",
+  "dl.google.com/dl/google/maven",
+];
+
 const repoFallbackRepos = normalizeRepoList(
   process.env.REPO_FALLBACK_REPOS,
   defaultRepoFallbackRepos,
 );
 
+const mavenCacheIgnorePathPrefixes = String(process.env.MAVEN_CACHE_IGNORE_PATH_PREFIXES || "").trim()
+  || defaultMavenCacheIgnorePathPrefixes.join(",");
+const mavenCacheIgnorePathPrefixRules = parseMavenCacheIgnorePathPrefixes(mavenCacheIgnorePathPrefixes);
+
 const defaultMavenRepoDomains = [
   "repo1.maven.org",
   "repo.maven.apache.org",
@@ -241,6 +255,9 @@ export const config = {
   httpsMitmDomains: toList(process.env.HTTPS_MITM_DOMAINS, ["repo1.maven.org", "repo.maven.apache.org", "registry.npmjs.org"]),
   npmRegistryDomains: toList(process.env.NPM_REGISTRY_DOMAINS, ["registry.npmjs.org", "registry.npmmirror.com", "npm.pkg.github.com"]),
   mavenRepoDomains: toList(process.env.MAVEN_REPO_DOMAINS, [...new Set(defaultMavenRepoDomains)]),
+  mavenCacheUseDomainDir: toBool(process.env.MAVEN_CACHE_USE_DOMAIN_DIR, false),
+  mavenCacheIgnorePathPrefixes,
+  mavenCacheIgnorePathPrefixRules,
   multiThreadDomains: toList(process.env.MULTI_THREAD_DOMAINS, ["repo1.maven.org"]),
   multiThreadCount: Math.max(1, toInt(process.env.MULTI_THREAD_COUNT, 4)),
   multiThreadMinSizeBytes,

package/src/index.js

@@ -120,6 +120,8 @@ async function main() {
   startupInfo(`[maven-proxy] repo  port: ${config.repoPort}`);
   startupInfo(`[maven-proxy] cache dir : ${config.cacheDir}`);
   startupInfo(`[maven-proxy] cache maven: ${config.mavenCacheDir}`);
+  startupInfo(`[maven-proxy] maven cache use domain dir: ${config.mavenCacheUseDomainDir}`);
+  startupInfo(`[maven-proxy] maven cache ignore path prefixes: ${config.mavenCacheIgnorePathPrefixes || "(none)"}`);
   startupInfo(`[maven-proxy] cache npm  : ${config.npmCacheDir}`);
   startupInfo(`[maven-proxy] cache other: ${config.genericCacheDir}`);
   startupInfo(`[maven-proxy] log dir: ${config.downloadLogDir}`);

package/src/proxy/proxy-http-handler.js

@@ -200,7 +200,8 @@ export function createHttpRequestHandler({
       ecosystem = detectPackageEcosystem(urlObj, config, matchesDomain);
       cachePath = getCacheFilePath(config.cacheDir, urlObj, {
         ecosystem,
-        includeHost: ecosystem !== "maven",
+        includeHost: ecosystem !== "maven" || config.mavenCacheUseDomainDir,
+        mavenCacheIgnorePathPrefixRules: config.mavenCacheIgnorePathPrefixRules,
       });
 
       if (ecosystem === "maven" && mavenAffinityIndex?.enabled) {
@@ -220,7 +221,9 @@ export function createHttpRequestHandler({
     }
 
     if (canonical && mavenAffinityIndex) {
-      if (isPositiveAffinityEligible(canonical.fileName)) {
+      const canUsePositiveAffinity = !config.mavenCacheUseDomainDir && isPositiveAffinityEligible(canonical.fileName);
+
+      if (canUsePositiveAffinity) {
         const preferredPath = await mavenAffinityIndex.resolvePreferredCachePath(canonical.canonicalKey);
         if (preferredPath) {
           console.log(`[proxy] affinity hit canonical=${canonical.canonicalKey} host=${urlObj.hostname}`);
@@ -244,7 +247,12 @@ export function createHttpRequestHandler({
       await fs.promises.mkdir(path.dirname(cachePath), { recursive: true });
       await downloader.ensureCached(urlObj, cachePath, req.headers);
 
-      if (canonical && mavenAffinityIndex && isPositiveAffinityEligible(canonical.fileName)) {
+      if (
+        canonical &&
+        mavenAffinityIndex &&
+        !config.mavenCacheUseDomainDir &&
+        isPositiveAffinityEligible(canonical.fileName)
+      ) {
         mavenAffinityIndex.recordSuccess({
           canonicalKey: canonical.canonicalKey,
           host: urlObj.hostname,

package/src/repo/repo-server.js

@@ -1,6 +1,10 @@
 import fs from "node:fs";
 import http from "node:http";
 import path from "node:path";
+import {
+  buildMavenHostlessPathCandidates,
+  stripMavenIgnoredPathPrefix,
+} from "../cache/cache-path.js";
 
 function safeJoin(baseDir, requestPath) {
   const pathname = decodeURIComponent(requestPath || "/");
@@ -13,6 +17,58 @@ function safeJoin(baseDir, requestPath) {
   return path.join(baseDir, normalized);
 }
 
+function sanitizeHostSegment(hostname) {
+  return String(hostname || "unknown").toLowerCase().replace(/[<>:\\"|?*]/g, "_");
+}
+
+function collectRepoHosts(repoBases = []) {
+  const hosts = new Set();
+
+  for (const repoBase of repoBases) {
+    try {
+      const parsed = new URL(repoBase);
+      hosts.add(sanitizeHostSegment(parsed.hostname));
+    } catch {
+      // ignore invalid URL
+    }
+  }
+
+  return [...hosts];
+}
+
+function buildDomainScopedPath(mavenCacheDir, hostname, relativePath) {
+  const hostDir = sanitizeHostSegment(hostname);
+  return safeJoin(path.join(mavenCacheDir, hostDir), relativePath);
+}
+
+function normalizeRelativePathForHost(relativePath, hostname, rules = []) {
+  const inputPath = `/${String(relativePath || "").replace(/^\/+/, "")}`;
+  const normalized = stripMavenIgnoredPathPrefix(
+    inputPath,
+    { protocol: "https:", hostname, port: "" },
+    rules,
+  );
+  return normalized.replace(/^\/+/, "");
+}
+
+function normalizeRelativePathCandidates(relativePath, rules = []) {
+  const inputPath = `/${String(relativePath || "").replace(/^\/+/, "")}`;
+  const candidates = buildMavenHostlessPathCandidates(inputPath, rules);
+  return candidates.map((item) => item.replace(/^\/+/, ""));
+}
+
+function buildDefaultRepoFilePath(config, relativePath) {
+  if (!config.mavenCacheUseDomainDir) {
+    const candidates = normalizeRelativePathCandidates(relativePath, config.mavenCacheIgnorePathPrefixRules);
+    return safeJoin(config.mavenCacheDir, candidates[0] || relativePath);
+  }
+
+  const hosts = collectRepoHosts(config.repoFallbackRepos || []);
+  const host = hosts[0] || "unknown";
+  const normalized = normalizeRelativePathForHost(relativePath, host, config.mavenCacheIgnorePathPrefixRules);
+  return buildDomainScopedPath(config.mavenCacheDir, host, normalized || relativePath);
+}
+
 async function statIfExists(filePath) {
   try {
     return await fs.promises.stat(filePath);
@@ -24,6 +80,68 @@ async function statIfExists(filePath) {
   }
 }
 
+async function findCachedMavenFile(config, relativePath) {
+  const ignoreRules = config.mavenCacheIgnorePathPrefixRules || [];
+
+  if (!config.mavenCacheUseDomainDir) {
+    const candidates = normalizeRelativePathCandidates(relativePath, ignoreRules);
+    for (const candidate of candidates) {
+      const filePath = safeJoin(config.mavenCacheDir, candidate);
+      const stats = await statIfExists(filePath);
+      if (stats && stats.isFile()) {
+        return { filePath, stats };
+      }
+    }
+
+    const fallbackPath = safeJoin(config.mavenCacheDir, candidates[0] || relativePath);
+    return { filePath: fallbackPath, stats: null };
+  }
+
+  const checkedHosts = new Set();
+  const preferredHosts = collectRepoHosts(config.repoFallbackRepos || []);
+
+  for (const host of preferredHosts) {
+    checkedHosts.add(host);
+    const normalized = normalizeRelativePathForHost(relativePath, host, ignoreRules);
+    const filePath = buildDomainScopedPath(config.mavenCacheDir, host, normalized || relativePath);
+    const stats = await statIfExists(filePath);
+    if (stats && stats.isFile()) {
+      return { filePath, stats };
+    }
+  }
+
+  let entries = [];
+  try {
+    entries = await fs.promises.readdir(config.mavenCacheDir, { withFileTypes: true });
+  } catch (error) {
+    if (error.code !== "ENOENT") {
+      throw error;
+    }
+  }
+
+  for (const entry of entries) {
+    if (!entry.isDirectory()) {
+      continue;
+    }
+
+    if (checkedHosts.has(entry.name)) {
+      continue;
+    }
+
+    const normalized = normalizeRelativePathForHost(relativePath, entry.name, ignoreRules);
+    const filePath = safeJoin(path.join(config.mavenCacheDir, entry.name), normalized || relativePath);
+    const stats = await statIfExists(filePath);
+    if (stats && stats.isFile()) {
+      return { filePath, stats };
+    }
+  }
+
+  return {
+    filePath: buildDefaultRepoFilePath(config, relativePath),
+    stats: null,
+  };
+}
+
 function buildRemoteUrl(repoBase, relativePath) {
   const base = repoBase.endsWith("/") ? repoBase : `${repoBase}/`;
   const relative = relativePath.replace(/^\/+/, "");
@@ -41,31 +159,45 @@ function buildCandidateRelativePaths(relativePath) {
   return [...new Set(candidates.filter(Boolean))];
 }
 
-async function ensureFromRemoteRepos(config, downloader, filePath, relativePath, cacheCleanupManager = null) {
+async function ensureFromRemoteRepos(config, downloader, relativePath, cacheCleanupManager = null) {
   if (!downloader) {
-    return null;
+    return { filePath: buildDefaultRepoFilePath(config, relativePath), stats: null };
   }
 
   const repos = config.repoFallbackRepos || [];
   if (repos.length === 0) {
-    return null;
+    return { filePath: buildDefaultRepoFilePath(config, relativePath), stats: null };
   }
 
   let hasNon404Error = false;
   let lastError = null;
   const candidatePaths = buildCandidateRelativePaths(relativePath);
+  const ignoreRules = config.mavenCacheIgnorePathPrefixRules || [];
 
   for (const repoBase of repos) {
     for (const candidatePath of candidatePaths) {
       const remoteUrl = buildRemoteUrl(repoBase, candidatePath);
 
       try {
+        const normalizedRelativePath = normalizeRelativePathForHost(
+          relativePath,
+          remoteUrl.hostname,
+          ignoreRules,
+        ) || relativePath;
+
+        const targetPath = config.mavenCacheUseDomainDir
+          ? buildDomainScopedPath(config.mavenCacheDir, remoteUrl.hostname, normalizedRelativePath)
+          : safeJoin(config.mavenCacheDir, normalizedRelativePath);
+
         if (cacheCleanupManager) {
           await cacheCleanupManager.checkAndCleanupIfNeeded("repo-cache-miss");
         }
         console.log(`[repo] cache miss, try remote ${remoteUrl.href}`);
-        await downloader.ensureCached(remoteUrl, filePath, {});
-        return await statIfExists(filePath);
+        await downloader.ensureCached(remoteUrl, targetPath, {});
+        return {
+          filePath: targetPath,
+          stats: await statIfExists(targetPath),
+        };
       } catch (error) {
         lastError = error;
         if (error.statusCode !== 404) {
@@ -79,7 +211,7 @@ async function ensureFromRemoteRepos(config, downloader, filePath, relativePath,
     throw lastError;
   }
 
-  return null;
+  return { filePath: buildDefaultRepoFilePath(config, relativePath), stats: null };
 }
 
 export function startRepoServer(config, downloader = null, cacheCleanupManager = null) {
@@ -87,11 +219,12 @@ export function startRepoServer(config, downloader = null, cacheCleanupManager =
     try {
       const urlObj = new URL(req.url || "/", `http://${req.headers.host || "localhost"}`);
       const relativePath = path.posix.normalize(urlObj.pathname || "/").replace(/^\/+/, "");
-      const filePath = safeJoin(config.mavenCacheDir, relativePath);
-      let stats = await statIfExists(filePath);
+      let { filePath, stats } = await findCachedMavenFile(config, relativePath);
 
       if (!stats || !stats.isFile()) {
-        stats = await ensureFromRemoteRepos(config, downloader, filePath, relativePath, cacheCleanupManager);
+        const fetched = await ensureFromRemoteRepos(config, downloader, relativePath, cacheCleanupManager);
+        filePath = fetched.filePath;
+        stats = fetched.stats;
       }
 
       if (!stats || !stats.isFile()) {