Merge pull request #656 from ruby/log-excerpt-search

Add full-text search over build failure logs

Author: hsbt

app/assets/stylesheets/application.css

@@ -237,3 +237,49 @@ table.reports td.diff div {
   text-overflow: ellipsis;
   width: 100px;
 }
+
+/* Search form */
+
+form.search-form {
+  max-width: 55em;
+}
+
+form.search-form .search-row {
+  display: flex;
+  align-items: center;
+  gap: 0.5em;
+  margin: 0.5em 0;
+}
+
+form.search-form .search-row label {
+  display: flex;
+  align-items: center;
+  gap: 0.25em;
+}
+
+form.search-form .search-text-row label {
+  flex: 1;
+}
+
+form.search-form .search-text-row input[type="text"] {
+  flex: 1;
+}
+
+form.search-form .search-filter-row {
+  justify-content: space-between;
+}
+
+/* Search results reuse table.reports, but their columns are in a different
+   order, so the reports failure-column rule lands on Match and Diff. */
+table.search-results td.match,
+table.search-results td.diff {
+  background-color: transparent;
+}
+
+table.search-results td.match code {
+  display: block;
+  overflow: hidden;
+  white-space: nowrap;
+  text-overflow: ellipsis;
+  width: 320px;
+}

app/controllers/search_controller.rb

@@ -0,0 +1,53 @@
+class SearchController < ApplicationController
+  PER_PAGE = 50
+  MAX_PAGE = 200
+
+  # GET /search
+  # Searches log excerpts of failed builds. Reports are narrowed by the
+  # indexed columns (server_id, branch, datetime, revision) first, then
+  # matched against log_excerpts.content (trigram GIN index on PostgreSQL).
+  def index
+    @q = params[:q].to_s.strip
+    @server_id = params[:server].to_s[/\A\d+\z/]
+    @branch = params[:branch].to_s.strip
+    @revision = params[:revision].to_s.strip
+    @from = parse_time(params[:from])
+    @to = parse_time(params[:to], end_of_day: true)
+    # to_s first: params[:page] is an Array for ?page[]=1, which has no to_i.
+    # Clamping also keeps OFFSET within range for absurd page numbers.
+    @page = params[:page].to_s.to_i.clamp(1, MAX_PAGE)
+    @servers = Server.order(:ordinal).to_a
+
+    @searched = @q.present? || @revision.present? || @branch.present? ||
+      @server_id.present? || @from || @to
+    unless @searched
+      @reports = []
+      return
+    end
+
+    reports = Report.joins(:log_excerpt).includes(:server, :log_excerpt).
+      order("reports.datetime DESC")
+    reports = reports.where(server_id: @server_id) if @server_id
+    reports = reports.where(branch: @branch) if @branch.present?
+    reports = reports.where("reports.datetime >= ?", @from) if @from
+    reports = reports.where("reports.datetime <= ?", @to) if @to
+    if @revision.present?
+      reports = reports.where("reports.revision LIKE ?", "#{Report.sanitize_sql_like(@revision)}%")
+    end
+    reports = reports.merge(LogExcerpt.content_match(@q)) if @q.present?
+
+    page = reports.limit(PER_PAGE + 1).offset((@page - 1) * PER_PAGE).to_a
+    @has_next = page.size > PER_PAGE
+    @reports = page.first(PER_PAGE)
+  end
+
+  private
+
+  def parse_time(str, end_of_day: false)
+    return nil if str.blank?
+    date = Date.parse(str) rescue nil
+    return nil unless date
+    time = Time.utc(date.year, date.month, date.day)
+    end_of_day ? time + 86399 : time
+  end
+end

app/models/log_excerpt.rb

@@ -0,0 +1,74 @@
+require 'net/http'
+require 'zlib'
+require 'stringio'
+
+class LogExcerpt < ApplicationRecord
+  belongs_to :report
+
+  MAX_CONTENT_BYTES = 100_000
+
+  # Fetch fail.txt and diff.txt from S3 for the report and store them as a
+  # searchable excerpt. Creates an empty-content row when nothing is available
+  # (missing or empty objects) so that backfill can resume by max report_id.
+  # Network errors are raised to the caller.
+  def self.capture(report)
+    return nil unless %r{\Ahttps?://rubyci\.s3\.amazonaws\.com/}.match?(report.server&.uri.to_s)
+    content = fetch_content(report)
+    excerpt = find_or_initialize_by(report_id: report.id)
+    excerpt.content = content
+    excerpt.save!
+    excerpt
+  end
+
+  def self.fetch_content(report)
+    sections = [report.failtxt_uri, report.difftxt_uri].filter_map do |uri|
+      body = fetch_text(uri)
+      body unless body.nil? || body.empty?
+    end
+    return "" if sections.empty?
+    # Budget each section separately, otherwise a fail.txt over the limit
+    # crowds the diff out entirely. A section under budget donates its
+    # remainder to the other one. The separators come out of the budget so
+    # the joined result still fits in MAX_CONTENT_BYTES.
+    budget = (MAX_CONTENT_BYTES - (sections.size - 1)) / sections.size
+    slack = sections.sum { |s| [budget - s.bytesize, 0].max }
+    sections.map { |s| truncate_bytes(s, budget + slack) }.join("\n")
+  end
+
+  # GET a gzipped text file. Returns nil on 404. S3 serves these either with
+  # Content-Encoding: gzip (Net::HTTP inflates the body) or as raw gzip bytes.
+  def self.fetch_text(uri)
+    uri = URI(uri)
+    res = Net::HTTP.start(uri.host, uri.port, open_timeout: 10, read_timeout: 30, use_ssl: uri.scheme == "https") do |h|
+      h.get(uri.path)
+    end
+    return nil if res.code == "404"
+    res.value
+    body = res.body
+    begin
+      body = Zlib::GzipReader.new(StringIO.new(body)).read
+    rescue Zlib::Error
+    end
+    body.force_encoding(Encoding::UTF_8).scrub("?").tr("\0", "")
+  end
+
+  def self.truncate_bytes(str, max)
+    return str if str.bytesize <= max
+    str.byteslice(0, max).scrub("")
+  end
+
+  # Substring match on content. On PostgreSQL the trigram GIN index makes
+  # ILIKE efficient; SQLite (development/test) falls back to plain LIKE.
+  def self.content_match(query)
+    operator = defined?(SQLite3) ? "LIKE" : "ILIKE"
+    # explicit ESCAPE: SQLite's LIKE has no default escape character
+    where("log_excerpts.content #{operator} ? ESCAPE '\\'", "%#{sanitize_sql_like(query)}%")
+  end
+
+  def matching_line(query)
+    return nil if query.blank?
+    q = query.downcase
+    line = content.each_line.find { |l| l.downcase.include?(q) }
+    line&.strip&.truncate(200)
+  end
+end

app/models/report.rb

@@ -6,6 +6,7 @@
 
 class Report < ApplicationRecord
   belongs_to :server
+  has_one :log_excerpt, dependent: :delete
   validates :server_id, :presence => true
   # SVN revision number or full 40-hex git commit SHA
   validates :revision, :format => { :with => /\A(?:\d+|\h{40})\z/ }, allow_nil: true
@@ -130,6 +131,21 @@ def recenturi
     server.recent_uri(branch_opts)
   end
 
+  def failtxt_uri
+    s3txt_uri('compressed_failhtml_relpath', 'fail')
+  end
+
+  def difftxt_uri
+    s3txt_uri('compressed_diffhtml_relpath', 'diff')
+  end
+
+  def s3txt_uri(key, kind)
+    relpath = meta&.[](key)&.sub('.html.gz', '.txt.gz') ||
+      datetime.strftime("log/%Y%m%dT%H%M%SZ.#{kind}.txt.gz")
+    "#{server.uri.chomp('/')}/#{depsuffixed_name}/#{relpath}"
+  end
+  private :s3txt_uri
+
   def meta
     if defined?(@meta)
       @meta
@@ -174,7 +190,7 @@ def self.scan_recent_ltsv(server, depsuffixed_name, body)
       diff = h["different_sections"]
       summary << (diff ? " (diff:#{diff})" : " (no diff)")
 
-      Report.create!(
+      report = Report.create!(
         server_id: server.id,
         datetime: datetime,
         branch: branch,
@@ -183,6 +199,13 @@ def self.scan_recent_ltsv(server, depsuffixed_name, body)
         ltsv: line,
         summary: summary.gsub(/<[^>]*>/, ''),
       )
+      if h["result"] != "success"
+        begin
+          LogExcerpt.capture(report)
+        rescue => e
+          warn [e, server.uri, "failed to capture log excerpt", report.id].inspect
+        end
+      end
     end
   rescue RuntimeError => e # It seems not a chkbuild log
     warn [e, server.uri, path, "failed to scan_reports"].inspect

app/views/layouts/application.html.erb

@@ -19,6 +19,7 @@
         <ul class="navbar-nav">
           <li class="nav-item"><%= link_to "Current Reports", "/reports", class: "nav-link" %></li>
           <li class="nav-item"><%= link_to "Latest Reports", "/", class: "nav-link" %></li>
+          <li class="nav-item"><%= link_to "Search", "/search", class: "nav-link" %></li>
         </ul>
       </div>
     </nav>

app/views/search/index.html.erb

@@ -0,0 +1,76 @@
+<% content_for(:title) { "Search failures - RubyCI" } %>
+
+<h3>Search failure logs</h3>
+
+<%= form_with url: search_path, method: :get, local: true, html: { class: "search-form" } do %>
+  <p class="search-row search-text-row">
+    <label>Text: <%= text_field_tag :q, @q, placeholder: "test name, exception class, error message ..." %></label>
+    <%# The revision query parameter is kept for machine clients (planned MCP
+        server) but intentionally has no visible form field. %>
+    <%= hidden_field_tag :revision, @revision if @revision.present? %>
+    <%= submit_tag "Search" %>
+  </p>
+  <p class="search-row search-filter-row">
+    <label>Server: <%= select_tag :server, options_for_select(@servers.map {|s| [s.name, s.id] }, @server_id), include_blank: "(all)" %></label>
+    <label>Branch: <%= text_field_tag :branch, @branch, size: 10, placeholder: "master" %></label>
+    <label>From: <%= date_field_tag :from, params[:from] %></label>
+    <label>To: <%= date_field_tag :to, params[:to] %></label>
+  </p>
+<% end %>
+
+<% if @searched %>
+  <% if @reports.empty? %>
+    <p>No matching failure logs found<%= " on page #{@page}" if @page > 1 %>.</p>
+  <% else %>
+    <div class="table-responsive table-responsive-border">
+      <table class="reports search-results table table-striped table-bordered">
+        <thead>
+          <tr>
+            <th>Server</th>
+            <th>Datetime</th>
+            <th>Branch</th>
+            <th>Option</th>
+            <th>Revision</th>
+            <th>Summary</th>
+            <th>Match</th>
+            <th>Diff</th>
+          </tr>
+        </thead>
+        <tbody>
+          <% @reports.each do |report| %>
+            <%
+              failuri = report.failuri || report.loguri
+              if report.revision&.match?(/\A\d+\z/)
+                revision = "r#{report.revision}"
+              elsif report.sha1
+                revision = report.sha1[0, 11]
+              end
+              revision_link = link_to revision, report.revisionuri if revision
+              snippet = report.log_excerpt&.matching_line(@q)
+            %>
+            <tr>
+              <td class="server"><%= report.server.name %></td>
+              <td class="datetime"><%= link_to report.sjstdt, report.loguri, title: report.jstdt %></td>
+              <td class="branch"><%= report.branch %></td>
+              <td class="option"><%= report.option %></td>
+              <td class="revision"><%= revision_link %></td>
+              <td class="summary failure"><div><%= link_to report.shortsummary, failuri, title: report.shortsummary %></div></td>
+              <td class="match"><code title="<%= snippet %>"><%= snippet %></code></td>
+              <td class="diff"><div><%= link_to report.diffstat, report.diffuri, title: report.diffstat %></div></td>
+            </tr>
+          <% end %>
+        </tbody>
+      </table>
+    </div>
+    <p>
+      <% if @page > 1 %>
+        <%= link_to "Prev", search_path(request.query_parameters.merge("page" => @page - 1)) %>
+      <% end %>
+      <% if @has_next %>
+        <%= link_to "Next", search_path(request.query_parameters.merge("page" => @page + 1)) %>
+      <% end %>
+    </p>
+  <% end %>
+<% else %>
+  <p>Search the excerpted fail/diff logs of failed builds. Datetimes are UTC.</p>
+<% end %>

config/routes.rb

@@ -1,6 +1,8 @@
 Rails.application.routes.draw do
   root :to => 'reports#current'
 
+  get "search" => "search#index"
+
   resources :reports, only: [:show, :index] do
     collection do
       get "current"

db/migrate/20260728023000_create_log_excerpts.rb

@@ -0,0 +1,18 @@
+class CreateLogExcerpts < ActiveRecord::Migration[8.1]
+  def change
+    create_table :log_excerpts do |t|
+      t.integer :report_id, null: false
+      t.text :content, null: false
+      t.timestamps
+      t.index [:report_id], unique: true
+    end
+    add_foreign_key :log_excerpts, :reports
+
+    # pg_trgm and GIN index are PostgreSQL-only; development and test run on SQLite
+    if connection.adapter_name == "PostgreSQL"
+      enable_extension "pg_trgm"
+      add_index :log_excerpts, :content, using: :gin, opclass: :gin_trgm_ops,
+        name: "index_log_excerpts_on_content_trgm"
+    end
+  end
+end

db/schema.rb

@@ -10,9 +10,10 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema[8.1].define(version: 2026_07_27_095902) do
+ActiveRecord::Schema[8.1].define(version: 2026_07_28_023000) do
   # These are extensions that must be enabled in order to support this database
   enable_extension "pg_catalog.plpgsql"
+  enable_extension "pg_trgm"
 
   create_table "active_storage_attachments", force: :cascade do |t|
     t.bigint "blob_id", null: false
@@ -42,6 +43,15 @@
     t.index ["blob_id", "variation_digest"], name: "index_active_storage_variant_records_uniqueness", unique: true
   end
 
+  create_table "log_excerpts", force: :cascade do |t|
+    t.text "content", null: false
+    t.datetime "created_at", null: false
+    t.integer "report_id", null: false
+    t.datetime "updated_at", null: false
+    t.index ["content"], name: "index_log_excerpts_on_content_trgm", opclass: :gin_trgm_ops, using: :gin
+    t.index ["report_id"], name: "index_log_excerpts_on_report_id", unique: true
+  end
+
   create_table "recents", force: :cascade do |t|
     t.datetime "created_at", null: false
     t.string "etag", null: false
@@ -63,7 +73,7 @@
     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 ["revision"], name: "index_reports_on_revision"
     t.index ["server_id", "branch", "option"], name: "index_reports_on_server_id_and_branch_and_option"
   end
 
@@ -79,5 +89,6 @@
 
   add_foreign_key "active_storage_attachments", "active_storage_blobs", column: "blob_id"
   add_foreign_key "active_storage_variant_records", "active_storage_blobs", column: "blob_id"
+  add_foreign_key "log_excerpts", "reports"
   add_foreign_key "recents", "servers"
 end