X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/12dd86319a1c6fb88fbe1530dc59e6172d448337..44c95f99098fa6c6acbfa82d4b6cbc6015eb6e39:/services/api/app/models/commit.rb

diff --git a/services/api/app/models/commit.rb b/services/api/app/models/commit.rb
index 0f62737cea..1d9a821d13 100644
--- a/services/api/app/models/commit.rb
+++ b/services/api/app/models/commit.rb
@@ -1,5 +1,15 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
+
 class Commit < ActiveRecord::Base
-  require 'shellwords'
+  extend CurrentApiClient
+
+  class GitError < StandardError
+    def http_status
+      422
+    end
+  end
 
   def self.git_check_ref_format(e)
     if !e or e.empty? or e[0] == '-' or e[0] == '$'
@@ -11,146 +21,213 @@ class Commit < ActiveRecord::Base
     end
   end
 
-  def self.find_commit_range(current_user, repository, minimum, maximum, exclude)
+  # Return an array of commits (each a 40-char sha1) satisfying the
+  # given criteria.
+  #
+  # Return [] if the revisions given in minimum/maximum are invalid or
+  # don't exist in the given repository.
+  #
+  # Raise ArgumentError if the given repository is invalid, does not
+  # exist, or cannot be read for any reason. (Any transient error that
+  # prevents commit ranges from resolving must raise rather than
+  # returning an empty array.)
+  #
+  # 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 repository, minimum, maximum, exclude
     if minimum and minimum.empty?
       minimum = nil
     end
 
     if minimum and !git_check_ref_format(minimum)
       logger.warn "find_commit_range called with invalid minimum revision: '#{minimum}'"
-      return nil
+      return []
     end
 
     if maximum and !git_check_ref_format(maximum)
       logger.warn "find_commit_range called with invalid maximum revision: '#{maximum}'"
-      return nil
+      return []
     end
 
     if !maximum
       maximum = "HEAD"
     end
 
-    # Get list of actual repository directories under management
-    on_disk_repos = repositories
+    gitdir, is_remote = git_dir_for repository
+    fetch_remote_repository gitdir, repository if is_remote
+    ENV['GIT_DIR'] = gitdir
 
-    # Get list of repository objects readable by user
-    readable = Repository.readable_by(current_user)
+    commits = []
 
-    # filter repository objects on requested repository name
-    if repository
-      readable = readable.where(name: repository)
+    # Get the commit hash for the upper bound
+    max_hash = nil
+    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
 
-    commits = []
-    readable.each do |r|
-      if on_disk_repos[r.name]
-        ENV['GIT_DIR'] = on_disk_repos[r.name][:git_dir]
-
-        # We've filtered for invalid characters, so we can pass the contents of
-        # minimum and maximum safely on the command line
+    # 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
 
-        # Get the commit hash for the upper bound
-        max_hash = nil
-        IO.foreach("|git rev-list --max-count=1 #{maximum.shellescape} --") do |line|
-          max_hash = line.strip
-        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
 
-        # If not found or string is invalid, nothing else to do
-        next if !max_hash or !git_check_ref_format(max_hash)
-
-        resolved_exclude = nil
-        if exclude
-          resolved_exclude = []
-          exclude.each do |e|
-            if git_check_ref_format(e)
-              IO.foreach("|git rev-list --max-count=1 #{e.shellescape} --") do |line|
-                resolved_exclude.push(line.strip)
-              end
-            else
-              logger.warn "find_commit_range called with invalid exclude invalid characters: '#{exclude}'"
-              return nil
-            end
+    resolved_exclude = nil
+    if exclude
+      resolved_exclude = []
+      exclude.each do |e|
+        if git_check_ref_format(e)
+          IO.foreach("|git rev-list --max-count=1 #{e.shellescape} --") do |line|
+            resolved_exclude.push(line.strip)
           end
+        else
+          logger.warn "find_commit_range called with invalid exclude invalid characters: '#{exclude}'"
+          return []
         end
+      end
+    end
 
-        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|
-            min_hash = line.strip
-          end
+    if minimum
+      # Get the commit hash for the lower bound
+      min_hash = nil
+      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
-          next 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
 
-          # Now find all commits between them
-          IO.foreach("|git rev-list #{min_hash.shellescape}..#{max_hash.shellescape} --") do |line|
-            hash = line.strip
-            commits.push(hash) if !resolved_exclude or !resolved_exclude.include? hash
-          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
 
-          commits.push(min_hash) if !resolved_exclude or !resolved_exclude.include? min_hash
-        else
-          commits.push(max_hash) if !resolved_exclude or !resolved_exclude.include? max_hash
-        end
-      else
-        logger.warn "Repository #{r.name} exists in table but not found on disk"
+      # Now find all commits between them
+      IO.foreach("|git rev-list #{min_hash.shellescape}..#{max_hash.shellescape} --") do |line|
+        hash = line.strip
+        commits.push(hash) if !resolved_exclude or !resolved_exclude.include? hash
       end
-    end
 
-    if !commits or commits.empty?
-      nil
+      commits.push(min_hash) if !resolved_exclude or !resolved_exclude.include? min_hash
     else
-      commits
+      commits.push(max_hash) if !resolved_exclude or !resolved_exclude.include? max_hash
     end
+
+    commits
   end
 
-  # Import all commits from configured git directory into the commits
-  # database.
-
-  def self.import_all
-    repositories.each do |repo_name, repo|
-      stat = { true => 0, false => 0 }
-      ENV['GIT_DIR'] = repo[:git_dir]
-      IO.foreach("|git rev-list --format=oneline --all") do |line|
-        sha1, message = line.strip.split " ", 2
-        imported = false
-        Commit.find_or_create_by_repository_name_and_sha1_and_message(repo_name, sha1, message[0..254]) do
-          imported = true
-        end
-        stat[!!imported] += 1
-        if (stat[true] + stat[false]) % 100 == 0
-          if $stdout.tty? or ARGV[0] == '-v'
-            puts "#{$0} #{$$}: repo #{repo_name} add #{stat[true]} skip #{stat[false]}"
-          end
-        end
-      end
-      if $stdout.tty? or ARGV[0] == '-v'
-        puts "#{$0} #{$$}: repo #{repo_name} add #{stat[true]} skip #{stat[false]}"
-      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).
+  #
+  # 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_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_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}")
   end
 
-  def self.refresh_repositories
-    @repositories = nil
+  protected
+
+  def self.remote_url? repo_name
+    /^(https?|git):\/\// =~ repo_name
   end
 
-  protected
+  # Return [local_git_dir, is_remote]. If is_remote, caller must use
+  # fetch_remote_repository to ensure content is up-to-date.
+  #
+  # Raises an exception if the latest content could not be fetched for
+  # any reason.
+  def self.git_dir_for repo_name
+    if remote_url? repo_name
+      return [cache_dir_for(repo_name), true]
+    end
+    repos = Repository.readable_by(current_user).where(name: repo_name)
+    if repos.count == 0
+      raise ArgumentError.new "Repository not found: '#{repo_name}'"
+    elsif repos.count > 1
+      logger.error "Multiple repositories with name=='#{repo_name}'!"
+      raise ArgumentError.new "Name conflict"
+    else
+      return [repos.first.server_path, false]
+    end
+  end
+
+  def self.cache_dir_for git_url
+    File.join(cache_dir_base, Digest::SHA1.hexdigest(git_url) + ".git").to_s
+  end
 
-  def self.repositories
-    return @repositories if @repositories
+  def self.cache_dir_base
+    Rails.root.join 'tmp', 'git'
+  end
 
-    @repositories = {}
-    @gitdirbase = Rails.configuration.git_repositories_dir
-    Dir.foreach @gitdirbase do |repo|
-      next if repo.match /^\./
-      git_dir = File.join(@gitdirbase,
-                          repo.match(/\.git$/) ? repo : File.join(repo, '.git'))
-      next if git_dir == Rails.configuration.git_internal_dir
-      repo_name = repo.sub(/\.git$/, '')
-      @repositories[repo_name] = {git_dir: git_dir}
+  def self.fetch_remote_repository gitdir, git_url
+    # Caller decides which protocols are worth using. This is just a
+    # safety check to ensure we never use urls like "--flag" or wander
+    # into git's hardlink features by using bare "/path/foo" instead
+    # of "file:///path/foo".
+    unless /^[a-z]+:\/\// =~ git_url
+      raise ArgumentError.new "invalid git url #{git_url}"
+    end
+    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,
+             "fetch --no-progress --tags --prune --force --update-head-ok #{git_url.shellescape} 'refs/heads/*:refs/heads/*'")
+  end
 
-    @repositories
+  def self.must_git gitdir, *cmds
+    # Clear token in case a git helper tries to use it as a password.
+    orig_token = ENV['ARVADOS_API_TOKEN']
+    ENV['ARVADOS_API_TOKEN'] = ''
+    begin
+      git = "git --git-dir #{gitdir.shellescape}"
+      cmds.each do |cmd|
+        must_pipe git+" "+cmd
+      end
+    ensure
+      ENV['ARVADOS_API_TOKEN'] = orig_token
+    end
+  end
+
+  def self.must_pipe *cmds
+    cmd = cmds.join(" 2>&1 |") + " 2>&1"
+    out = IO.read("| </dev/null #{cmd}")
+    if not $?.success?
+      raise GitError.new "#{cmd}: #{$?}: #{out}"
+    end
   end
 end