process pdf from doc

pull/718/head
Pete Matsyburka 1 month ago
parent c1aac6b763
commit 113b61588b

@ -1,17 +0,0 @@
# frozen_string_literal: true
class GeneratePreviewImagesJob
include Sidekiq::Job
sidekiq_options queue: :images
def perform(params = {})
attachment = ActiveStorage::Attachment.find(params['attachment_id'])
max_page = [attachment.metadata['pdf']['number_of_pages'].to_i - 1,
Templates::ProcessDocument::MAX_NUMBER_OF_PAGES_PROCESSED].min
Templates::ProcessDocument.generate_document_preview_images(attachment, attachment.download, 1..max_page,
concurrency: 1)
end
end

@ -41,16 +41,24 @@ module Templates
attachment attachment
end end
def process(attachment, data, extract_fields: false) def process(attachment, data, extract_fields: false, doc: nil)
if attachment.content_type == PDF_CONTENT_TYPE && extract_fields && data.size < MAX_FLATTEN_FILE_SIZE if attachment.content_type == PDF_CONTENT_TYPE && extract_fields && data.size < MAX_FLATTEN_FILE_SIZE
pdf = HexaPDF::Document.new(io: StringIO.new(data)) if doc
fields = Templates::FindPdfiumAcroFields.call(attachment, doc, data)
else
pdf = HexaPDF::Document.new(io: StringIO.new(data))
fields = Templates::FindAcroFields.call(pdf, attachment, data) fields = Templates::FindAcroFields.call(pdf, attachment, data)
end
end end
pdf ||= HexaPDF::Document.new(io: StringIO.new(data)) if doc
number_of_pages = doc.page_count
else
pdf ||= HexaPDF::Document.new(io: StringIO.new(data))
number_of_pages = pdf.pages.size number_of_pages = pdf.pages.size
end
attachment.metadata['pdf'] ||= {} attachment.metadata['pdf'] ||= {}
attachment.metadata['pdf']['number_of_pages'] = number_of_pages attachment.metadata['pdf']['number_of_pages'] = number_of_pages

Loading…
Cancel
Save