DEV-1373 make catalog indexing date independent

.standard.yml

@@ -5,7 +5,6 @@ ignore:
   - 'lib/ht_traject/**/*'
   - 'lib/ht_traject.rb'
   - 'lib/traject/**/*'
-  - 'lib/translation_maps/**/*'
   - 'lib/umich_traject/**/*'
   - 'lib/umich_traject.rb'
   - 'readers/**/*'

Gemfile

@@ -2,6 +2,7 @@ source "https://rubygems.org"
 
 group :development, :test do
   gem "bundler", "~>2.0"
+  gem "climate_control"
   gem "rake", "~> 13.0"
   gem "standard"
   gem "rspec"

Gemfile.lock

@@ -6,6 +6,7 @@ GEM
     ast (2.4.2)
     builder (3.2.4)
     canister (0.9.2)
+    climate_control (1.2.0)
     coderay (1.1.3)
     concurrent-ruby (1.2.2)
     date_named_file (0.1.1)
@@ -61,11 +62,11 @@ GEM
     match_map (3.0.0)
     method_source (1.0.0)
     naconormalizer (1.0.1-java)
-    nokogiri (1.16.2-arm64-darwin)
+    nokogiri (1.16.7-arm64-darwin)
       racc (~> 1.4)
-    nokogiri (1.16.2-java)
+    nokogiri (1.16.7-java)
       racc (~> 1.4)
-    nokogiri (1.16.2-x86_64-linux)
+    nokogiri (1.16.7-x86_64-linux)
       racc (~> 1.4)
     parallel (1.23.0)
     parser (3.2.2.1)
@@ -83,7 +84,7 @@ GEM
     rainbow (3.1.1)
     rake (13.0.6)
     regexp_parser (2.8.0)
-    rexml (3.2.5)
+    rexml (3.3.8)
     rsolr (2.5.0)
       builder (>= 2.1.2)
       faraday (>= 0.9, < 3, != 2.0.0)
@@ -178,6 +179,7 @@ PLATFORMS
 DEPENDENCIES
   bundler (~> 2.0)
   canister (~> 0.9.2)
+  climate_control
   date_named_file
   dotenv
   http (~> 5.0)

README.md

@@ -117,7 +117,16 @@ network (i.e. the one started with `docker-compose up` from this repository).
 Solr should be reachable via the `solr-sdr-catalog` hostname.
 
 ## How to do the basics
+### Date-Independent Indexing
 
+For use in production environments where daily and monthly indexing are ongoing activities,
+we enable the indexer to maintain state by writing "journal" files: empty datestamped
+files in a known location (`JOURNAL_DIRECTORY`). The command `cictl index continue` does whatever
+full or daily indexing is appropriate given the state of the journals.
+
+Note that all of the `cictl index *` commands write journal files, with the exception of
+`cictl index file` which takes only an `upd` MARC file rather than a MARC-deletes pair, and is not
+expected to be used in an environment where date independence is in force.
 
 ### Putting a new solr configuration into place
 
@@ -133,7 +142,7 @@ Solr should be reachable via the `solr-sdr-catalog` hostname.
 * (Optional) If your new solr config requires a full reindex, go ahead and 
   get rid of the data with `rm -rf data`
 * Fire solr back up: `systemctl start solr-current-catalog`
-* Give it a minute and then go to http://beeftea-2.umdl.umich.edu:9033/solr` to make sure the core came back up.
+* Give it a minute and then go to `http://beeftea-2.umdl.umich.edu:9033/solr` to make sure the core came back up.
 * Do whatever indexing needs doing.
 
 ### Indexing
@@ -193,6 +202,7 @@ The `index` command has a number of possibilities:
 > bundle exec bin/cictl help index
 Commands:
   cictl index all             # Empty the catalog and index the most recent m...
+  cictl index continue        # index all files not represented in the indexe...
   cictl index date YYYYMMDD   # Run the catchup (delete and index) for a part...
   cictl index file FILE       # Index a single file
   cictl index help [COMMAND]  # Describe subcommands or one specific subcommand
@@ -283,6 +293,8 @@ and `config/env`. The defaults in the repository suffice for testing under Docke
 ## Environment variables
 
   * `DDIR` data directory, defaults to `/htsolr/catalog/prep`
+  * `JOURNAL_DIRECTORY` location of journal files (see Date-Independent Indexing above) defaulting
+    to `journal/` inside the repo directory.
   * `LOG_DIR` where to store logs, defaults to `/htsolr/catalog/prep`.
   * `MYSQL_HOST`, `MYSQL_DATABASE`, `MYSQL_USER`, `MYSQL_PASSWORD` *required* unless run with `NO_DB`.
   * `NO_DB` if you want to skip all the database stuff. Useful for testing. Implied by `NO_EXTERNAL_DATA`.

docker-compose.yml

@@ -1,5 +1,4 @@
 ---
-version: '3'
 
 services:
   traject:

lib/cictl/index_command.rb

@@ -3,12 +3,37 @@
 require_relative "base_command"
 require_relative "zephir_file"
 require_relative "deleted_records"
+require_relative "journal"
 
 module CICTL
   class IndexCommand < BaseCommand
     class_option :reader, type: :string, desc: "Reader name/path"
     class_option :writer, type: :string, desc: "Writer name/path"
 
+    desc "continue", "Index all files not represented in the indexer journals"
+    def continue
+      last_full = ZephirFile.full_files.last
+      fatal "unable to find full Zephir file" unless last_full
+      # Index the most recent full file and subsequent ones if the
+      # full file journal is missing.
+      full_journal = Journal.new(date: last_full.to_datetime.to_date, full: true)
+      if full_journal.missing?
+        logger.info "missing full journal #{full_journal}, calling `cictl all`"
+        call_all_command
+      # Otherwise, iterate from the last full file date to yesterday.
+      # If there is a missing journal, start indexing from that point.
+      else
+        (last_full.to_datetime.to_date..(Date.today - 1)).each do |date|
+          journal = Journal.new(date: last_full.to_datetime.to_date, full: false)
+          if journal.missing?
+            logger.info "missing update journal #{journal}, calling `cictl since #{journal.date}`"
+            call_since_command(journal.date)
+            break
+          end
+        end
+      end
+    end
+
     desc "all", "Empty the catalog and index the most recent monthly followed by subsequent daily updates"
     option :wait, type: :boolean, desc: "Wait 5 seconds for Control-C", default: true
     def all
@@ -39,6 +64,8 @@ def all
       solr_client.commit!
     end
 
+    # Note: this command does not write a journal since it only processes the MARC file
+    # but not the deletes.
     option :commit, type: :boolean, desc: "Commit changes to Solr", default: true
     desc "file FILE", "Index a single MARC file"
     def file(marcfile)
@@ -52,7 +79,11 @@ def date(date)
       preflight
       with_date(date) do |date|
         index_deletes_for_date date
-        index_records_for_date date
+        if index_records_for_date date
+          journal = Journal.new(date: date)
+          logger.info("write journal file #{journal.path}")
+          journal.write!
+        end
       end
     end
 
@@ -84,6 +115,7 @@ def today
     end
 
     no_commands do
+      alias_method :call_all_command, :all
       alias_method :call_date_command, :date
       alias_method :call_file_command, :file
       alias_method :call_since_command, :since
@@ -118,14 +150,17 @@ def marc_file_for_date(date)
       ZephirFile.update_files.at(date)
     end
 
+    # @return [Boolean] true if the marcfile for the given date exists
     def index_records_for_date(date)
       marcfile = marc_file_for_date date
       if File.exist? marcfile
         Indexer.new(reader: options[:reader], writer: options[:writer]).run marcfile
         solr_client.commit!
         logger.debug "index date(#{date}): Solr count now #{solr_client.count}"
+        true
       else
         logger.warn "could not find marcfile '#{marcfile}'"
+        false
       end
     end
 

lib/cictl/journal.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+require_relative "../services"
+
+module CICTL
+  # A class that enables date-independent catalog indexing using the filesystem.
+  #
+  # Each time a full or update file is indexed, writes an (empty) file of the form
+  # hathitrust_catalog_indexer_journal_upd_YYYYMMDD.txt or
+  # hathitrust_catalog_indexer_journal_full_YYYYMMDD.txt in the journal directory.
+  #
+  # When we use the index command `cictl continue`
+  # we calculate the earliest zephir file not yet indexed and proceed in order from
+  # that point.
+  #
+  # Nomenclature note: "journal" is the closest semantic match to "log" I could find.
+  # This is a log, of sorts, but the term was already taken.
+  class Journal
+    attr_reader :date
+
+    FILENAME_PATTERN = /hathitrust_catalog_indexer_journal_(full|upd)_(\d{8})\.txt/
+
+    def self.filename_for(date:, full:)
+      yyyymmdd = date.strftime "%Y%m%d"
+      type = full ? "full" : "upd"
+      "hathitrust_catalog_indexer_journal_#{type}_#{yyyymmdd}.txt"
+    end
+
+    def initialize(date: Date.today - 1, full: false)
+      @date = date
+      @full = full
+    end
+
+    # Use the built-in but append the date and full/upd because that's what we care about.
+    def to_s
+      super.tap do |s|
+        s.gsub!(/>$/, " [#{date} #{full? ? "full" : "upd"}]>")
+      end
+    end
+
+    def full?
+      @full
+    end
+
+    # Of the form `hathitrust_catalog_indexer_journal_(full|upd)_YYYYMMDD.txt`
+    def file
+      self.class.filename_for(date: date, full: full?)
+    end
+
+    def path
+      File.join(HathiTrust::Services[:journal_directory], file)
+    end
+
+    def exist?
+      File.exist? path
+    end
+
+    def missing?
+      !exist?
+    end
+
+    def write!
+      FileUtils.touch path
+    end
+  end
+end

lib/ht_traject/ht_item.rb

@@ -21,7 +21,6 @@ class << self
         attr_accessor :ht_ns, :ht_avail_us, :ht_avail_intl
       end
 
-      self.ht_ns         = ::Traject::TranslationMap.new('ht/ht_namespace_map')
       self.ht_avail_us   = ::Traject::TranslationMap.new('ht/availability_map_ht')
       self.ht_avail_intl = ::Traject::TranslationMap.new('ht/availability_map_ht_intl')
 
@@ -256,10 +255,6 @@ def enum_pubdate=(e)
         end
       end
 
-      def source
-        ItemSet.ht_ns[namespace]
-      end
-
       def us_availability
         ItemSet.ht_avail_us[rights].first
       end

lib/services.rb

@@ -63,6 +63,14 @@ def env_local_file
     ENV["LOG_DIR"] || default
   end
 
+  Services.register(:journal_directory) do
+    (ENV["JOURNAL_DIRECTORY"] || File.join(HOME, "journal")).tap do |dir|
+      if !File.exist?(dir)
+        FileUtils.mkdir dir
+      end
+    end
+  end
+
   Services.register(:redirect_file) do
     # Start migrating from redirect_file to REDIRECT_FILE on principle of least surprise
     ENV["redirect_file"] || ENV["REDIRECT_FILE"] || Redirects.default_redirects_file

lib/translation_maps/ht/availability_map_ht.rb

@@ -1,18 +1,27 @@
-require 'ht_traject/ht_constants'
-require 'match_map'
+require "ht_traject/ht_constants"
+require "match_map"
 
 mm = MatchMap.new
 
-mm[/^umall$/] = HathiTrust::Constants::FT
-mm[/world$/] = HathiTrust::Constants::FT       # matches world, ic-world, und-world
-mm[/^cc.*/] = HathiTrust::Constants::FT
-mm[/^pd(?:us)?$/] = HathiTrust::Constants::FT  # pd or pdus
+# Note: orph, orphcand, and umall are unattested in rights_current as of Oct 2024
 
-mm[/^ic$/] = HathiTrust::Constants::SO
-mm[/^orph$/] = HathiTrust::Constants::SO
-mm[/^nobody$/] = HathiTrust::Constants::SO
-mm[/^und$/] = HathiTrust::Constants::SO
-mm[/^pd-p/] = HathiTrust::Constants::SO        # pd-pvt or pd-private
-mm[/^opb?$/] = HathiTrust::Constants::SO
+# Full Text
+mm["pd"] = HathiTrust::Constants::FT           # [1]
+mm["ic-world"] = HathiTrust::Constants::FT     # [7]
+mm["pdus"] = HathiTrust::Constants::FT         # [9]
+mm[/^cc-/] = HathiTrust::Constants::FT         # [10-15, 17, 20-25]
+mm["und-world"] = HathiTrust::Constants::FT    # [18]
+
+# Search Only
+mm["ic"] = HathiTrust::Constants::SO           # [2]
+mm["op"] = HathiTrust::Constants::SO           # [3]
+mm["orph"] = HathiTrust::Constants::SO         # [4]
+mm["und"] = HathiTrust::Constants::SO          # [5]
+mm["umall"] = HathiTrust::Constants::SO        # [6]
+mm["nobody"] = HathiTrust::Constants::SO       # [8]
+mm["orphcand"] = HathiTrust::Constants::SO     # [16]
+mm["icus"] = HathiTrust::Constants::SO         # [19]
+mm["pd-pvt"] = HathiTrust::Constants::SO       # [26]
+mm["supp"] = HathiTrust::Constants::SO         # [27]
 
 mm

lib/translation_maps/ht/availability_map_ht_intl.rb

@@ -1,17 +1,26 @@
-require 'ht_traject/ht_constants'
+require "ht_traject/ht_constants"
 
 mm = MatchMap.new
 
-mm['umall'] = HathiTrust::Constants::FT
-mm['world'] = HathiTrust::Constants::FT # matches world, ic-world, und-world
-mm[/^cc.*/] = HathiTrust::Constants::FT
-mm['pd'] = HathiTrust::Constants::FT
+# Note: orph, orphcand, and umall are unattested in rights_current as of Oct 2024
 
-mm['pdus'] = HathiTrust::Constants::SO
-mm['ic'] = HathiTrust::Constants::SO
-mm[/^opb?$/] = HathiTrust::Constants::SO
-mm['orph'] = HathiTrust::Constants::SO
-mm['nobody'] = HathiTrust::Constants::SO
-mm['und'] = HathiTrust::Constants::SO
+# Full Text
+mm["pd"] = HathiTrust::Constants::FT           # [1]
+mm["ic-world"] = HathiTrust::Constants::FT     # [7]
+mm[/^cc-/] = HathiTrust::Constants::FT         # [10-15, 17, 20-25]
+mm["und-world"] = HathiTrust::Constants::FT    # [18]
+mm["icus"] = HathiTrust::Constants::FT         # [19]
+
+# Search Only
+mm["ic"] = HathiTrust::Constants::SO           # [2]
+mm["op"] = HathiTrust::Constants::SO           # [3]
+mm["orph"] = HathiTrust::Constants::SO         # [4]
+mm["und"] = HathiTrust::Constants::SO          # [5]
+mm["umall"] = HathiTrust::Constants::SO        # [6]
+mm["nobody"] = HathiTrust::Constants::SO       # [8]
+mm["pdus"] = HathiTrust::Constants::SO         # [9]
+mm["orphcand"] = HathiTrust::Constants::SO     # [16]
+mm["pd-pvt"] = HathiTrust::Constants::SO       # [26]
+mm["supp"] = HathiTrust::Constants::SO         # [27]
 
 mm