From 9aad9e186a8783e616dc8217223720085dfb1e8b Mon Sep 17 00:00:00 2001
From: Bill Erickson <berickxx@gmail.com>
Date: Mon, 29 Oct 2018 13:58:27 -0400
Subject: [PATCH] Indexer improvements, docs, and options

Signed-off-by: Bill Erickson <berickxx@gmail.com>
---
 Open-ILS/examples/opensrf.xml.example              |  12 --
 Open-ILS/src/perlmods/lib/OpenILS/Elastic.pm       |  17 ++-
 .../src/perlmods/lib/OpenILS/Elastic/BibSearch.pm  |  63 +++++++---
 Open-ILS/src/support-scripts/elastic-index.pl      | 128 ++++++++++++++++++++-
 4 files changed, 181 insertions(+), 39 deletions(-)

diff --git a/Open-ILS/examples/opensrf.xml.example b/Open-ILS/examples/opensrf.xml.example
index b29f6b950b..156562b82b 100644
--- a/Open-ILS/examples/opensrf.xml.example
+++ b/Open-ILS/examples/opensrf.xml.example
@@ -145,18 +145,6 @@ vim:et:ts=4:sw=4:
             <!-- number of parallel processes to use during fine generation -->
             <parallel>1</parallel>
         </fine_generator>
-
-        <elastic_search>
-            <database>
-                <driver>Pg</driver>
-                <host>localhost</host>
-                <port>5432</port>
-                <db>evergreen</db>
-                <user>evergreen</user>
-                <pw>evergreen</pw>
-                <application_name>Elastic Search Indexer</application_name>
-            </database>
-        </elastic_search>
         
         <reporter>
             <!--
diff --git a/Open-ILS/src/perlmods/lib/OpenILS/Elastic.pm b/Open-ILS/src/perlmods/lib/OpenILS/Elastic.pm
index dc4711ad18..5f53fef276 100644
--- a/Open-ILS/src/perlmods/lib/OpenILS/Elastic.pm
+++ b/Open-ILS/src/perlmods/lib/OpenILS/Elastic.pm
@@ -18,7 +18,6 @@ use warnings;
 use DBI;
 use Time::HiRes qw/time/;
 use OpenSRF::Utils::Logger qw/:logger/;
-use OpenSRF::Utils::SettingsClient;
 use OpenILS::Utils::CStoreEditor qw/:funcs/;
 use Search::Elasticsearch;
 use OpenSRF::Utils::JSON;
@@ -66,15 +65,13 @@ sub db {
 	my ($self) = @_;
 
     return $self->{db} if $self->{db};
-
-    my $client = OpenSRF::Utils::SettingsClient->new;
-    my $settings = $client->config_value('elastic_search');
-    my $db_name = $settings->{database}->{db};
-    my $db_host = $settings->{database}->{host};
-    my $db_port = $settings->{database}->{port};
-    my $db_user = $settings->{database}->{user};
-    my $db_pass = $settings->{database}->{pw};
-    my $db_appn = $settings->{database}->{application_name};
+    
+    my $db_name = $self->{db_name};
+    my $db_host = $self->{db_host};
+    my $db_port = $self->{db_port};
+    my $db_user = $self->{db_user};
+    my $db_pass = $self->{db_pass};
+    my $db_appn = $self->{db_appn} || 'Elastic Indexer';
 
     # TODO Add application_name to dsn
 
diff --git a/Open-ILS/src/perlmods/lib/OpenILS/Elastic/BibSearch.pm b/Open-ILS/src/perlmods/lib/OpenILS/Elastic/BibSearch.pm
index 30e98605f8..a8ed65c1ea 100644
--- a/Open-ILS/src/perlmods/lib/OpenILS/Elastic/BibSearch.pm
+++ b/Open-ILS/src/perlmods/lib/OpenILS/Elastic/BibSearch.pm
@@ -15,9 +15,12 @@ package OpenILS::Elastic::BibSearch;
 # ---------------------------------------------------------------
 use strict;
 use warnings;
+use DateTime;
+use Time::HiRes qw/time/;
 use OpenSRF::Utils::Logger qw/:logger/;
 use OpenSRF::Utils::JSON;
 use OpenILS::Utils::CStoreEditor qw/:funcs/;
+use OpenILS::Utils::DateTime qw/interval_to_seconds/;
 use OpenILS::Elastic;
 use base qw/OpenILS::Elastic/;
 
@@ -202,35 +205,66 @@ sub create_index {
 
 # Add data to the bib-search index
 sub populate_index {
-    my ($self) = @_;
+    my ($self, $settings) = @_;
+    $settings ||= {};
 
     my $index_count = 0;
     my $total_indexed = 0;
-    my $state = {last_bib_id => 0};
 
-    do {
-        $index_count =
-            $self->populate_bib_search_index_page($state);
+    # extract the database settings.
+    for my $db_key (grep {$_ =~ /^db_/} keys %$settings) {
+        $self->{$db_key} = $settings->{$db_key};
+    }
+
+    # TODO $settings->{stop_record}
+    # TODO $settings->{start_date}
+
+    my $end_time;
+    my $duration = $settings->{max_duration};
+    if ($duration) {
+        my $seconds = interval_to_seconds($duration);
+        $end_time = DateTime->now;
+        $end_time->add(seconds => $seconds);
+    }
+
+    while (1) {
+
+        $index_count = $self->populate_bib_index_batch($settings);
         $total_indexed += $index_count;
 
         $logger->info("ES indexed $total_indexed bib records");
 
-    } while ($index_count > 0);
+        # exit if we're only indexing a single record or if the 
+        # batch indexer says there are no more records to index.
+        last if !$index_count || $settings->{index_record};
+
+        if ($end_time && DateTime->now > $end_time) {
+            $logger->info(
+                "ES index populate exiting early on max_duration $duration");
+            last;
+        }
+    } 
 
     $logger->info("ES bib indexing complete with $total_indexed records");
 }
 
 sub get_bib_ids {
-    my ($self, $state, $record_id) = @_;
-    return [$record_id] if $record_id;
+    my ($self, $state) = @_;
 
-    # TODO add support for last_edit_date
-    my $last_id = $state->{last_bib_id};
+    # A specific record is selected for indexing.
+    return [$state->{index_record}] if $state->{index_record};
+
+    my $start_id = $state->{start_record} || 0;
+    my $stop_id = $state->{stop_record}; # TODO
+
+    # TODO: implement start_date filtering.
+    # Requires checking edit dates on bibs, call numbers, and copies!
+    my $start_date = $state->{start_date};
 
     my $sql = <<SQL;
 SELECT bre.id
 FROM biblio.record_entry bre
-WHERE NOT bre.deleted AND bre.active AND bre.id > $last_id
+WHERE NOT bre.deleted AND bre.active AND bre.id >= $start_id
 ORDER BY bre.edit_date, bre.id LIMIT $BIB_BATCH_SIZE
 SQL
 
@@ -257,15 +291,16 @@ SQL
     return $self->get_db_rows($sql);
 }
 
-sub populate_bib_search_index_page {
+sub populate_bib_index_batch {
     my ($self, $state) = @_;
 
     my $index_count = 0;
-    my $last_id = $state->{last_bib_id};
 
     my $bib_ids = $self->get_bib_ids($state);
     return 0 unless @$bib_ids;
 
+    $logger->info("ES indexing ".scalar(@$bib_ids)." records");
+
     my $bib_data = $self->get_bib_data($bib_ids);
 
     my $holdings = $self->load_holdings($bib_ids);
@@ -319,7 +354,7 @@ sub populate_bib_search_index_page {
 
         return 0 unless $self->index_document($bib_id, $body);
 
-        $state->{last_bib_id} = $bib_id;
+        $state->{start_record} = $bib_id + 1;
         $index_count++;
     }
 
diff --git a/Open-ILS/src/support-scripts/elastic-index.pl b/Open-ILS/src/support-scripts/elastic-index.pl
index eeefe26048..6544716a29 100755
--- a/Open-ILS/src/support-scripts/elastic-index.pl
+++ b/Open-ILS/src/support-scripts/elastic-index.pl
@@ -2,6 +2,7 @@
 use strict;
 use warnings;
 use Getopt::Long;
+use OpenSRF::Utils::JSON;
 use OpenILS::Utils::Fieldmapper;
 use OpenILS::Utils::CStoreEditor;
 use OpenILS::Elastic::BibSearch;
@@ -11,8 +12,22 @@ my $osrf_config = '/openils/conf/opensrf_core.xml';
 my $cluster = 'main';
 my $create_index;
 my $delete_index;
-my $index_name;
+my $index_name = 'bib-search'; # only supported index at time of writing
 my $populate;
+my $index_record;
+my $start_record;
+my $stop_record;
+my $start_date;
+my $max_duration;
+my $batch_size = 500;
+
+# Database settings read from ENV by default.
+my $db_host = $ENV{PGHOST} || 'localhost';
+my $db_port = $ENV{PGPORT} || 5432;
+my $db_user = $ENV{PGUSER} || 'evergreen';
+my $db_pass = $ENV{PGPASSWORD} || 'evergreen';
+my $db_name = $ENV{PGDATABASE} || 'evergreen';
+my $db_appn = 'Elastic Indexer';
 
 GetOptions(
     'help'              => \$help,
@@ -21,16 +36,101 @@ GetOptions(
     'create-index'      => \$create_index,
     'delete-index'      => \$delete_index,
     'index=s'           => \$index_name,
+    'index-record=s'    => \$index_record,
+    'start-record=s'    => \$start_record,
+    'stop-record=s'     => \$stop_record,
+    'start-date=s'      => \$start_date,
+    'max-duration=s'    => \$max_duration,
+    'batch-size=s'      => \$batch_size,
+    'db-name=s'         => \$db_name,
+    'db-host=s'         => \$db_host,
+    'db-port=s'         => \$db_port,
+    'db-user=s'         => \$db_user,
+    'db-pass=s'         => \$db_pass,
+    'db-appn=s'         => \$db_appn,
     'populate'          => \$populate
 ) || die "\nSee --help for more\n";
 
+sub help {
+    print <<HELP;
+        Synopsis:
+            
+            $0 --delete-index --create-index --index bib-search --populate
+
+        Options:
+
+            --osrf-config <file-path>
+
+            --db-name <$db_name>
+            --db-host <$db_host>
+            --db-port <$db_port>
+            --db-user <$db_user>
+            --db-pass <PASSWORD>
+            --db-appn <$db_appn>
+                Database connection values.  This is the Evergreen database
+                where values should be extracted for elastic search indexing.
+
+                Values default to their PG* environment variable equivalent.
+
+            --cluster <name>
+                Specify a cluster name.  Defaults to 'main'.
+
+            --index <name>
+                Specify an index name.  Defaults to 'bib-search'.
+
+            --delete-index
+                Delete the specified index and all of its data. 
+
+            --create-index
+                Create an index whose name equals --index-name.
+
+            --batch-size <number>
+                Index at most this many records per batch.
+                Default is 500.
+
+            --index-record <id>
+                Index a specific record by identifier.
+
+            --start-record <id>
+                Start indexing at the record with this ID.
+
+            --stop-record <id>
+                Stop indexing after the record with this ID has been indexed.
+
+            --start-date <YYYY-MM-DD[Thh::mm:ss]>
+                Start indexing records whose last edit date falls after
+                the provided date;
+
+            --max-duration <duration>
+                Stop indexing once the process has been running for this
+                amount of time.
+
+            --populate
+                Populate the selected index with data.  If no filters
+                are provided (e.g. --index-start-record) then all 
+                applicable values will be indexed.
+
+HELP
+    exit(0);
+}
+
+help() if $help;
+
 # connect to osrf...
 OpenSRF::System->bootstrap_client(config_file => $osrf_config);
 Fieldmapper->import(
     IDL => OpenSRF::Utils::SettingsClient->new->config_value("IDL"));
 OpenILS::Utils::CStoreEditor::init();
 
-my $es = OpenILS::Elastic::BibSearch->new($cluster);
+my $es;
+
+if ($index_name eq 'bib-search') {
+    $es = OpenILS::Elastic::BibSearch->new($cluster);
+}
+
+if (!$es) {
+    die "Unknown index type: $index_name\n";
+}
 
 $es->connect;
 
@@ -43,7 +143,29 @@ if ($create_index) {
 }
 
 if ($populate) {
-    $es->populate_index or die "Index populate failed.\n";
+
+    my $settings = {
+        db_name => $db_name,
+        db_host => $db_host,
+        db_port => $db_port,
+        db_user => $db_user,
+        db_pass => 'REDACTED',
+        db_appn => $db_appn,
+        index_record => $index_record,
+        start_record => $start_record,
+        stop_record  => $stop_record,
+        start_date   => $start_date,
+        max_duration => $max_duration,
+        batch_size   => $batch_size
+    };
+
+    print "Commencing index populate with settings: " . 
+        OpenSRF::Utils::JSON->perl2JSON($settings) . "\n";
+
+    # Apply after logging $settings
+    $settings->{db_pass} = $db_pass;
+
+    $es->populate_index($settings) or die "Index populate failed.\n";
 }
 
 
-- 
2.11.0