Merge pull request #496 from yob/page-text-skip-empty-runs

Skip text runs with no text when rendering a page

lib/pdf/reader.rb

@@ -309,6 +309,7 @@ def root
 require 'pdf/reader/rectangle'
 require 'pdf/reader/reference'
 require 'pdf/reader/register_receiver'
+require 'pdf/reader/no_text_filter'
 require 'pdf/reader/null_security_handler'
 require 'pdf/reader/security_handler_factory'
 require 'pdf/reader/standard_key_builder'

lib/pdf/reader/no_text_filter.rb

@@ -0,0 +1,14 @@
+# coding: utf-8
+# typed: strict
+# frozen_string_literal: true
+
+class PDF::Reader
+  # There's no point rendering zero-width characters
+  class NoTextFilter
+
+    def self.exclude_empty_strings(runs)
+      runs.reject { |run| run.text.to_s.size == 0 }
+    end
+  end
+end
+

lib/pdf/reader/page_text_receiver.rb

@@ -62,6 +62,8 @@ def runs(opts = {})
           runs = OverlappingRunsFilter.exclude_redundant_runs(runs)
         end
 
+        runs = NoTextFilter.exclude_empty_strings(runs)
+
         if opts.fetch(:merge, true)
           runs = merge_runs(runs)
         end

rbi/pdf-reader.rbi

@@ -846,6 +846,11 @@ module PDF
       def self.detect_intersection(sweep_line_status, event_point); end
     end
 
+    class NoTextFilter
+      sig { params(runs: T::Array[PDF::Reader::TextRun]).returns(T::Array[PDF::Reader::TextRun]) }
+      def self.exclude_empty_strings(runs); end
+    end
+
     class EventPoint
       sig { returns(Numeric) }
       attr_reader :x