X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/c550609485691d8107ae364bfc982569f81f1725..73872ccc5bb6b80a6049b44b0113085a9c2b6934:/services/api/app/models/commit.rb

diff --git a/services/api/app/models/commit.rb b/services/api/app/models/commit.rb
index 9e1176b9b3..921c690cd0 100644
--- a/services/api/app/models/commit.rb
+++ b/services/api/app/models/commit.rb
@@ -1,5 +1,13 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
+
+require 'request_error'
+
 class Commit < ActiveRecord::Base
-  class GitError < StandardError
+  extend CurrentApiClient
+
+  class GitError < RequestError
     def http_status
       422
     end
@@ -29,7 +37,7 @@ class Commit < ActiveRecord::Base
   # repository can be the name of a locally hosted repository or a git
   # URL (see git-fetch(1)). Currently http, https, and git schemes are
   # supported.
-  def self.find_commit_range(current_user, repository, minimum, maximum, exclude)
+  def self.find_commit_range repository, minimum, maximum, exclude
     if minimum and minimum.empty?
       minimum = nil
     end
@@ -56,12 +64,22 @@ class Commit < ActiveRecord::Base
 
     # Get the commit hash for the upper bound
     max_hash = nil
-    IO.foreach("|git rev-list --max-count=1 #{maximum.shellescape} --") do |line|
+    git_max_hash_cmd = "git rev-list --max-count=1 #{maximum.shellescape} --"
+    IO.foreach("|#{git_max_hash_cmd}") do |line|
       max_hash = line.strip
     end
 
-    # If not found or string is invalid, nothing else to do
-    return [] if !max_hash or !git_check_ref_format(max_hash)
+    # If not found, nothing else to do
+    if !max_hash
+      logger.warn "no refs found looking for max_hash: `GIT_DIR=#{gitdir} #{git_max_hash_cmd}` returned no output"
+      return []
+    end
+
+    # If string is invalid, nothing else to do
+    if !git_check_ref_format(max_hash)
+      logger.warn "ref returned by `GIT_DIR=#{gitdir} #{git_max_hash_cmd}` was invalid for max_hash: #{max_hash}"
+      return []
+    end
 
     resolved_exclude = nil
     if exclude
@@ -81,12 +99,22 @@ class Commit < ActiveRecord::Base
     if minimum
       # Get the commit hash for the lower bound
       min_hash = nil
-      IO.foreach("|git rev-list --max-count=1 #{minimum.shellescape} --") do |line|
+      git_min_hash_cmd = "git rev-list --max-count=1 #{minimum.shellescape} --"
+      IO.foreach("|#{git_min_hash_cmd}") do |line|
         min_hash = line.strip
       end
 
-      # If not found or string is invalid, nothing else to do
-      return [] if !min_hash or !git_check_ref_format(min_hash)
+      # If not found, nothing else to do
+      if !min_hash
+        logger.warn "no refs found looking for min_hash: `GIT_DIR=#{gitdir} #{git_min_hash_cmd}` returned no output"
+        return []
+      end
+
+      # If string is invalid, nothing else to do
+      if !git_check_ref_format(min_hash)
+        logger.warn "ref returned by `GIT_DIR=#{gitdir} #{git_min_hash_cmd}` was invalid for min_hash: #{min_hash}"
+        return []
+      end
 
       # Now find all commits between them
       IO.foreach("|git rev-list #{min_hash.shellescape}..#{max_hash.shellescape} --") do |line|
@@ -103,32 +131,69 @@ class Commit < ActiveRecord::Base
   end
 
   # Given a repository (url, or name of hosted repo) and commit sha1,
-  # copy the commit into the internal git repo and tag it with the
-  # given tag (typically a job UUID).
+  # copy the commit into the internal git repo (if necessary), and tag
+  # it with the given tag (typically a job UUID).
   #
   # The repo can be a remote url, but in this case sha1 must already
   # be present in our local cache for that repo: e.g., sha1 was just
   # returned by find_commit_range.
-  def self.tag_in_internal_repository repo, sha1, tag
+  def self.tag_in_internal_repository repo_name, sha1, tag
     unless git_check_ref_format tag
       raise ArgumentError.new "invalid tag #{tag}"
     end
     unless /^[0-9a-f]{40}$/ =~ sha1
       raise ArgumentError.new "invalid sha1 #{sha1}"
     end
-    src_gitdir, _ = git_dir_for repo
+    src_gitdir, _ = git_dir_for repo_name
+    unless src_gitdir
+      raise ArgumentError.new "no local repository for #{repo_name}"
+    end
     dst_gitdir = Rails.configuration.git_internal_dir
-    must_pipe("echo #{sha1.shellescape}",
-              "git --git-dir #{src_gitdir.shellescape} pack-objects -q --revs --stdout",
-              "git --git-dir #{dst_gitdir.shellescape} unpack-objects -q")
-    must_git(dst_gitdir,
-             "tag --force #{tag.shellescape} #{sha1.shellescape}")
+
+    begin
+      commit_in_dst = must_git(dst_gitdir, "log -n1 --format=%H #{sha1.shellescape}^{commit}").strip
+    rescue GitError
+      commit_in_dst = false
+    end
+
+    tag_cmd = "tag --force #{tag.shellescape} #{sha1.shellescape}^{commit}"
+    if commit_in_dst == sha1
+      must_git(dst_gitdir, tag_cmd)
+    else
+      # git-fetch is faster than pack-objects|unpack-objects, but
+      # git-fetch can't fetch by sha1. So we first try to fetch a
+      # branch that has the desired commit, and if that fails (there
+      # is no such branch, or the branch we choose changes under us in
+      # race), we fall back to pack|unpack.
+      begin
+        branches = must_git(src_gitdir,
+                            "branch --contains #{sha1.shellescape}")
+        m = branches.match(/^. (\w+)\n/)
+        if !m
+          raise GitError.new "commit is not on any branch"
+        end
+        branch = m[1]
+        must_git(dst_gitdir,
+                 "fetch file://#{src_gitdir.shellescape} #{branch.shellescape}")
+        # Even if all of the above steps succeeded, we might still not
+        # have the right commit due to a race, in which case tag_cmd
+        # will fail, and we'll need to fall back to pack|unpack. So
+        # don't be tempted to condense this tag_cmd and the one in the
+        # rescue block into a single attempt.
+        must_git(dst_gitdir, tag_cmd)
+      rescue GitError
+        must_pipe("echo #{sha1.shellescape}",
+                  "git --git-dir #{src_gitdir.shellescape} pack-objects -q --revs --stdout",
+                  "git --git-dir #{dst_gitdir.shellescape} unpack-objects -q")
+        must_git(dst_gitdir, tag_cmd)
+      end
+    end
   end
 
   protected
 
-  def self.remote_url? repository
-    /^(https?|git):\/\// =~ repository
+  def self.remote_url? repo_name
+    /^(https?|git):\/\// =~ repo_name
   end
 
   # Return [local_git_dir, is_remote]. If is_remote, caller must use
@@ -152,8 +217,11 @@ class Commit < ActiveRecord::Base
   end
 
   def self.cache_dir_for git_url
-    Rails.root.join('tmp', 'git', Digest::SHA1.hexdigest(git_url) + ".git").
-      to_s
+    File.join(cache_dir_base, Digest::SHA1.hexdigest(git_url) + ".git").to_s
+  end
+
+  def self.cache_dir_base
+    Rails.root.join 'tmp', 'git-cache'
   end
 
   def self.fetch_remote_repository gitdir, git_url
@@ -164,9 +232,16 @@ class Commit < ActiveRecord::Base
     unless /^[a-z]+:\/\// =~ git_url
       raise ArgumentError.new "invalid git url #{git_url}"
     end
-    FileUtils.mkdir_p gitdir
+    begin
+      must_git gitdir, "branch"
+    rescue GitError => e
+      raise unless /Not a git repository/ =~ e.to_s
+      # OK, this just means we need to create a blank cache repository
+      # before fetching.
+      FileUtils.mkdir_p gitdir
+      must_git gitdir, "init"
+    end
     must_git(gitdir,
-             "init",
              "fetch --no-progress --tags --prune --force --update-head-ok #{git_url.shellescape} 'refs/heads/*:refs/heads/*'")
   end
 
@@ -174,14 +249,16 @@ class Commit < ActiveRecord::Base
     # Clear token in case a git helper tries to use it as a password.
     orig_token = ENV['ARVADOS_API_TOKEN']
     ENV['ARVADOS_API_TOKEN'] = ''
+    last_output = ''
     begin
       git = "git --git-dir #{gitdir.shellescape}"
       cmds.each do |cmd|
-        must_pipe git+" "+cmd
+        last_output = must_pipe git+" "+cmd
       end
     ensure
       ENV['ARVADOS_API_TOKEN'] = orig_token
     end
+    return last_output
   end
 
   def self.must_pipe *cmds
@@ -190,5 +267,6 @@ class Commit < ActiveRecord::Base
     if not $?.success?
       raise GitError.new "#{cmd}: #{$?}: #{out}"
     end
+    return out
   end
 end