Store git commit SHA in reports.revision

app/models/report.rb

@@ -7,19 +7,29 @@
 class Report < ApplicationRecord
   belongs_to :server
   validates :server_id, :presence => true
-  validates :revision, :numericality => { :only_integer => true }, allow_nil: true
+  # SVN revision number or full 40-hex git commit SHA
+  validates :revision, :format => { :with => /\A(?:\d+|\h{40})\z/ }, allow_nil: true
   validates :datetime, :uniqueness => { :scope => [:server_id, :branch] }
   validates :branch, :presence => true
   validates :summary, :presence => true
 
+  def self.extract_full_sha(ltsv)
+    return nil unless ltsv
+    ltsv[%r<"https\\x3A//github.com/ruby/ruby":(\h{40})>, 1] ||
+      ltsv[%r<https://github.com/ruby/ruby:(\h{40})>, 1]
+  end
+
   def sha1
-    !ltsv ? nil : \
-    ltsv[%r<"https\\x3A//github.com/ruby/ruby":([^\t]+)>, 1] ||
-      ltsv[%r<https://github.com/ruby/ruby:([^\t]+)>, 1]
+    if revision&.match?(/\A\h{40}\z/)
+      revision
+    elsif ltsv
+      ltsv[%r<"https\\x3A//github.com/ruby/ruby":([^\t]+)>, 1] ||
+        ltsv[%r<https://github.com/ruby/ruby:([^\t]+)>, 1]
+    end
   end
 
   def revisionuri
-    if revision
+    if revision&.match?(/\A\d+\z/)
       "https://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=revision&revision=#{revision}"
     elsif sha1
       "https://github.com/ruby/ruby/commit/#{sha1}"
@@ -156,8 +166,9 @@ def self.scan_recent_ltsv(server, depsuffixed_name, body)
     end
     ary.reverse_each do |line, h, dt, datetime|
       puts "reporting #{server.name} #{depsuffixed_name} #{dt} ..."
-      # subversion revision of ruby is less than 99999
-      revision = h["ruby_rev"] && h["ruby_rev"].size <= 6 ? h["ruby_rev"][1, 5].to_i : nil
+      # git commit SHA if present, otherwise subversion revision (less than 99999)
+      revision = extract_full_sha(line) ||
+        (h["ruby_rev"] && h["ruby_rev"].size <= 6 ? h["ruby_rev"][1, 5] : nil)
       summary = h["title"]
       summary << ' success' if h["result"] == 'success'
       diff = h["different_sections"]

app/views/reports/index.html.erb

@@ -34,7 +34,7 @@
 
           diffuri = report.diffuri
 
-          if report.revision
+          if report.revision&.match?(/\A\d+\z/)
             revision = "r#{report.revision}"
           elsif report.sha1
             revision = report.sha1[0, 11]
@@ -108,7 +108,7 @@
           <td class="datetime"><%= link_to report.sjstdt, report.loguri, title: report.jstdt %></td>
           <td class="branch"><%= link_to report.branch, report.recenturi %></td>
           <td class="option"><%= report.option %></td>
-          <td class="revision"><%= report.patchlevel + " " if report.patchlevel %><%= report.revision %></td>
+          <td class="revision"><%= report.patchlevel + " " if report.patchlevel %><%= report.revision&.match?(/\A\d+\z/) ? report.revision : report.revision.to_s[0, 11] %></td>
           <% if report.build %>
             <td  class="build" colspan="3"><%= report.build %></td>
           <% else %>

db/migrate/20260727051843_change_reports_revision_to_string.rb

@@ -0,0 +1,27 @@
+class ChangeReportsRevisionToString < ActiveRecord::Migration[8.1]
+  def up
+    if postgresql?
+      change_column :reports, :revision, :string, using: 'revision::text'
+      # varchar_pattern_ops is required for LIKE 'abc1234%' prefix searches
+      add_index :reports, :revision, opclass: :varchar_pattern_ops
+    else
+      change_column :reports, :revision, :string
+      add_index :reports, :revision
+    end
+  end
+
+  def down
+    remove_index :reports, :revision
+    if postgresql?
+      change_column :reports, :revision, :integer, using: 'revision::integer'
+    else
+      change_column :reports, :revision, :integer
+    end
+  end
+
+  private
+
+  def postgresql?
+    connection.adapter_name.match?(/postgresql/i)
+  end
+end

db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema[8.1].define(version: 2024_11_21_042607) do
+ActiveRecord::Schema[8.1].define(version: 2026_07_27_051843) do
   # These are extensions that must be enabled in order to support this database
   enable_extension "pg_catalog.plpgsql"
 
@@ -67,12 +67,13 @@
     t.datetime "datetime", precision: nil
     t.text "ltsv"
     t.string "option"
-    t.integer "revision"
+    t.string "revision"
     t.integer "server_id"
     t.text "summary"
     t.datetime "updated_at", precision: nil
     t.index ["branch"], name: "index_reports_on_branch"
     t.index ["datetime"], name: "index_reports_on_datetime"
+    t.index ["revision"], name: "index_reports_on_revision", opclass: :varchar_pattern_ops
     t.index ["server_id", "branch", "option"], name: "index_reports_on_server_id_and_branch_and_option"
   end
 

lib/tasks/backfill_revision.rake

@@ -0,0 +1,20 @@
+namespace :reports do
+  desc "Backfill reports.revision with full commit SHA extracted from ltsv"
+  task :backfill_revision => :environment do
+    scope = Report.where(revision: nil).where.not(ltsv: nil)
+    total = scope.count
+    puts "#{total} reports to scan"
+    scanned = 0
+    updated = 0
+    scope.find_each do |report|
+      scanned += 1
+      sha = Report.extract_full_sha(report.ltsv)
+      if sha
+        report.update_column(:revision, sha)
+        updated += 1
+      end
+      puts "#{scanned}/#{total} scanned, #{updated} updated" if scanned % 1000 == 0
+    end
+    puts "done: #{scanned} scanned, #{updated} updated"
+  end
+end

test/models/report_test.rb

@@ -0,0 +1,98 @@
+require "test_helper"
+
+class ReportTest < ActiveSupport::TestCase
+  SHA = "0123456789abcdef0123456789abcdef01234567"
+
+  @@ordinal = 0
+
+  def create_server(name)
+    Server.create!(name: name, uri: "https://rubyci.s3.amazonaws.com/#{name}", ordinal: (@@ordinal += 1))
+  end
+
+  def build_report(attrs = {})
+    server = attrs.delete(:server) || create_server("server-#{attrs.object_id}")
+    Report.new({
+      server: server,
+      branch: "master",
+      datetime: Time.now.utc,
+      summary: "ruby 3.5.0dev success",
+    }.merge(attrs))
+  end
+
+  test "revision accepts nil, digits and 40-hex SHA" do
+    assert_predicate build_report(revision: nil), :valid?
+    assert_predicate build_report(revision: "12345"), :valid?
+    assert_predicate build_report(revision: SHA), :valid?
+  end
+
+  test "revision rejects other strings" do
+    assert_not build_report(revision: "r12345").valid?
+    assert_not build_report(revision: SHA[0, 11]).valid?
+    assert_not build_report(revision: "#{SHA} ").valid?
+  end
+
+  test "revisionuri for svn revision" do
+    report = build_report(revision: "12345")
+    assert_equal "https://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=revision&revision=12345", report.revisionuri
+  end
+
+  test "revisionuri for git commit SHA" do
+    report = build_report(revision: SHA)
+    assert_equal "https://github.com/ruby/ruby/commit/#{SHA}", report.revisionuri
+  end
+
+  test "revisionuri falls back to SHA from ltsv" do
+    report = build_report(revision: nil, ltsv: "depsuffixed_name:ruby-master\thttps://github.com/ruby/ruby:#{SHA}")
+    assert_equal "https://github.com/ruby/ruby/commit/#{SHA}", report.revisionuri
+  end
+
+  test "revisionuri without revision and ltsv" do
+    assert_nil build_report(revision: nil, ltsv: nil).revisionuri
+  end
+
+  test "sha1 prefers revision column" do
+    report = build_report(revision: SHA, ltsv: nil)
+    assert_equal SHA, report.sha1
+  end
+
+  test "sha1 falls back to ltsv" do
+    report = build_report(revision: nil, ltsv: "https://github.com/ruby/ruby:#{SHA}")
+    assert_equal SHA, report.sha1
+
+    report = build_report(revision: nil, ltsv: %["https\\x3A//github.com/ruby/ruby":#{SHA}])
+    assert_equal SHA, report.sha1
+  end
+
+  test "extract_full_sha" do
+    assert_equal SHA, Report.extract_full_sha("https://github.com/ruby/ruby:#{SHA}")
+    assert_equal SHA, Report.extract_full_sha(%["https\\x3A//github.com/ruby/ruby":#{SHA}])
+    assert_nil Report.extract_full_sha("https://github.com/ruby/ruby:#{SHA[0, 11]}")
+    assert_nil Report.extract_full_sha(nil)
+  end
+
+  test "scan_recent_ltsv stores full SHA from ltsv" do
+    server = create_server("scan-sha")
+    line = [
+      "start_time:20260727T051800Z",
+      "https://github.com/ruby/ruby:#{SHA}",
+      "title:ruby 3.5.0dev (2026-07-27) [x86_64-linux]",
+      "result:success",
+    ].join("\t")
+    Report.scan_recent_ltsv(server, "ruby-master", line + "\n")
+    report = Report.where(server_id: server.id).last
+    assert_equal SHA, report.revision
+  end
+
+  test "scan_recent_ltsv stores svn revision as digits" do
+    server = create_server("scan-svn")
+    line = [
+      "start_time:20260727T051800Z",
+      "ruby_rev:r54321",
+      "title:ruby 2.7.0dev (2026-07-27) [x86_64-linux]",
+      "result:success",
+    ].join("\t")
+    Report.scan_recent_ltsv(server, "ruby-2.7", line + "\n")
+    report = Report.where(server_id: server.id).last
+    assert_equal "54321", report.revision
+  end
+end