From 221ff0b6226a02bf4120b7ae1b037e4468ea5b51 Mon Sep 17 00:00:00 2001
From: dev <dev.aravind@culturecreates.com>
Date: Mon, 18 Nov 2024 11:53:35 +0530
Subject: [PATCH] Updated workflow to add an offset value for pagination

---
 action.yml                |  6 +++++-
 src/lib/entity_fetcher.rb | 10 ++++++++--
 src/main.rb               |  6 +++---
 3 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/action.yml b/action.yml
index aaf5f8b..9e3996e 100644
--- a/action.yml
+++ b/action.yml
@@ -40,6 +40,8 @@ inputs:
     description: 'URL to the SHACL file'
   fetch-urls-headlessly:
     description: 'Set as true to fetch the entity URLs headlessly'
+  offset:
+    description: 'Offset for paginated pages'
 
 runs: 
   using: 'composite'
@@ -98,6 +100,7 @@ runs:
         isPaginated=${{ inputs.is-paginated || 'false' }}
         headless=${{ inputs.headless || 'false' }}
         fetchUrlsHeadlessly=${{ inputs.fetch-urls-headlessly || 'false' }}
+        offset=${{ inputs.offset || '1' }}
 
         docker pull ghcr.io/culturecreates/artsdata-pipeline-action/artsdata-rdf-fetcher:latest
         docker run --shm-size=1g -v $(pwd)/output:/usr/src/app/output ghcr.io/culturecreates/artsdata-pipeline-action/artsdata-rdf-fetcher:latest \
@@ -106,7 +109,8 @@ runs:
           "output/${{ inputs.downloadFile }}" \
           "$isPaginated" \
           "$headless" \
-          "$fetchUrlsHeadlessly"
+          "$fetchUrlsHeadlessly" \
+          "$offset"
 
       shell: bash
 
diff --git a/src/lib/entity_fetcher.rb b/src/lib/entity_fetcher.rb
index 47a292f..cd5493f 100644
--- a/src/lib/entity_fetcher.rb
+++ b/src/lib/entity_fetcher.rb
@@ -2,7 +2,7 @@
 require 'open-uri'
 
 module EntityFetcher
-  def self.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_entity_urls_headlessly, headers)
+  def self.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_entity_urls_headlessly, headers, offset)
     base_url = page_url.split('/')[0..2].join('/')
     entity_urls = []
 
@@ -14,6 +14,12 @@ def self.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_enti
       page_number = is_paginated.to_i
     end
 
+    if offset
+      offset = offset.to_i
+    else
+      offset = 1
+    end
+
     loop do
       url = "#{page_url}#{page_number}"
       puts "Fetching entity urls from #{url}..."
@@ -33,7 +39,7 @@ def self.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_enti
 
       break if entity_urls.length == number_of_entities || page_number.nil?
 
-      page_number += 1
+      page_number += offset
     end
     entity_urls.uniq
   end
diff --git a/src/main.rb b/src/main.rb
index 75a4867..7a42063 100644
--- a/src/main.rb
+++ b/src/main.rb
@@ -3,16 +3,16 @@
 require_relative 'lib/headless_browser'
 
 if ARGV.length < 4
-  puts "Usage: ruby script_name.rb <page_url> <entity_identifier> <file_name> <is_paginated> <headless> <fetch_urls_headlessly>"
+  puts "Usage: ruby script_name.rb <page_url> <entity_identifier> <file_name> <is_paginated> <headless> <fetch_urls_headlessly> <offset>"
   exit
 end
 
-page_url, entity_identifier, file_name, is_paginated, headless, fetch_urls_headlessly = ARGV[0..5]
+page_url, entity_identifier, file_name, is_paginated, headless, fetch_urls_headlessly, offset = ARGV[0..6]
 
 linkeddata_version = Gem::Specification.find_by_name('linkeddata').version.to_s
 headers = {"User-Agent" => "artsdata-crawler/#{linkeddata_version}"}
 
-entity_urls = EntityFetcher.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_urls_headlessly, headers)
+entity_urls = EntityFetcher.fetch_entity_urls(page_url, entity_identifier, is_paginated, fetch_urls_headlessly, headers, offset)
 base_url = page_url.split('/')[0..2].join('/')
 
 sparql_paths = [