git: Fall back from tgz to ssh on HTTP errors

An HTTP error on fetching a remote tarball for a git repo is usually
going to be the result of pulling in a private repo, which is
available via SSH, but not as a pre-rendered gzipped tarball over HTTPS.

Fall back to SSH when the hosted.tarball() fetch fails, but only if the
error is an HTTP error of some kind (not, eg, an invalid tarball
response.)

Part 2 of fix for #15

Author: isaacs

lib/git.js

@@ -174,22 +174,32 @@ class GitFetcher extends Fetcher {
   // TODO: after cloning, create a tarball of the folder, and add to the cache
   // with cacache.put.stream(), using a key that's deterministic based on the
   // spec and repo, so that we don't ever clone the same thing multiple times.
-  [_clone] (handler) {
+  [_clone] (handler, tarballOk = true) {
     const o = { tmpPrefix: 'git-clone' }
     const ref = this.resolvedSha || this.spec.gitCommittish
     const h = this.spec.hosted
     const resolved = this.resolved
 
+    // can be set manually to false to fall back to sshurl
+    tarballOk = tarballOk &&
+      h && resolved === repoUrl(h, { noCommittish: false }) && h.tarball
+
     return cacache.tmp.withTmp(this.cache, o, tmp => {
       // if we're resolved, and have a tarball url, shell out to RemoteFetcher
-      if (h && resolved === repoUrl(h, { noCommittish: false }) && h.tarball) {
+      if (tarballOk) {
         const nameat = this.spec.name ? `${this.spec.name}@` : ''
         return new RemoteFetcher(h.tarball({ noCommittish: false }), {
           ...this.opts,
           pkgid: `git:${nameat}${this.resolved}`,
           resolved: this.resolved,
           integrity: null, // it'll always be different, if we have one
-        }).extract(tmp).then(() => handler(tmp))
+        }).extract(tmp).then(() => handler(tmp), er => {
+          // fall back to ssh download if tarball fails
+          if (er.constructor.name.match(/^Http/))
+            return this[_clone](handler, false)
+          else
+            throw er
+        })
       }
 
       const repo = h && !this.spec.fetchSpec

test/git.js

@@ -219,16 +219,19 @@ t.test('setup', { bail: true }, t => {
           return res.end(data)
         } catch (er) {
           res.statusCode = 500
-          res.end(er.stack)
+          return res.end(er.stack)
         }
         case '/repo-1.0.0.tgz': try {
           const data = fs.readFileSync(me + '/repo-1.0.0.tgz')
           res.setHeader('content-length', data.length)
           return res.end(data)
         } catch (er) {
           res.statusCode = 500
-          res.end(er.stack)
+          return res.end(er.stack)
         }
+        case '/not-tar.tgz':
+          res.statusCode = 200
+          return res.end('this is not a gzipped tarball tho')
         default:
           res.statusCode = 404
           return res.end('not found')
@@ -377,3 +380,19 @@ t.test('fetch a weird ref', t => {
 
   t.end()
 })
+
+t.test('fetch a private repo where the tgz is a 404', t => {
+  const gf = new GitFetcher(`localhost:repo/x#${REPO_HEAD}`, {cache})
+  gf.spec.hosted.tarball = () => `${hostedUrl}/not-found.tgz`
+  // should fetch it by falling back to ssh when it gets an http error
+  return gf.extract(me + '/no-tgz')
+})
+
+t.test('fetch a private repo where the tgz is not a tarball', t => {
+  const gf = new GitFetcher(`localhost:repo/x#${REPO_HEAD}`, {cache})
+  gf.spec.hosted.tarball = () => `${hostedUrl}/not-tar.tgz`
+  // should NOT retry, because the error was not an HTTP fetch error
+  return t.rejects(gf.extract(me + '/bad-tgz'), {
+    code: 'TAR_BAD_ARCHIVE',
+  })
+})