page_layout.rb in pdf-reader-2.6.0

- old
+ new

@@ -1,9 +1,10 @@
 # coding: utf-8
 # frozen_string_literal: true
 
 require 'pdf/reader/overlapping_runs_filter'
+require 'pdf/reader/zero_width_runs_filter'
 
 class PDF::Reader
 
   # Takes a collection of TextRun objects and renders them into a single
   # string that best approximates the way they'd appear on a render PDF page.
@@ -15,14 +16,16 @@
     DEFAULT_FONT_SIZE = 12
 
     def initialize(runs, mediabox)
       raise ArgumentError, "a mediabox must be provided" if mediabox.nil?
 
-      @runs    = merge_runs(OverlappingRunsFilter.exclude_redundant_runs(runs))
+      runs = ZeroWidthRunsFilter.exclude_zero_width_runs(runs)
+      runs = OverlappingRunsFilter.exclude_redundant_runs(runs)
+      @runs = merge_runs(runs)
       @mean_font_size   = mean(@runs.map(&:font_size)) || DEFAULT_FONT_SIZE
       @mean_font_size = DEFAULT_FONT_SIZE if @mean_font_size == 0
-      @mean_glyph_width = mean(@runs.map(&:mean_character_width)) || 0
+      @median_glyph_width = median(@runs.map(&:mean_character_width)) || 0
       @page_width  = (mediabox[2] - mediabox[0]).abs
       @page_height = (mediabox[3] - mediabox[1]).abs
       @x_offset = @runs.map(&:x).sort.first || 0
       lowest_y = @runs.map(&:y).sort.first || 0
       @y_offset = lowest_y > 0 ? 0 : lowest_y
@@ -65,11 +68,11 @@
     def row_count
       @row_count ||= (@page_height / @mean_font_size).floor
     end
 
     def col_count
-      @col_count ||= ((@page_width  / @mean_glyph_width) * 1.05).floor
+      @col_count ||= ((@page_width  / @median_glyph_width) * 1.05).floor
     end
 
     def row_multiplier
       @row_multiplier ||= @page_height.to_f / row_count.to_f
     end
@@ -84,15 +87,15 @@
       else
         collection.inject(0) { |accum, v| accum + v} / collection.size.to_f
       end
     end
 
-    def each_line(&block)
-      @runs.sort.group_by { |run|
-        run.y.to_i
-      }.map { |y, collection|
-        yield y, collection
-      }
+    def median(collection)
+      if collection.size == 0
+        0
+      else
+        collection.sort[(collection.size * 0.5).floor]
+      end
     end
 
     # take a collection of TextRun objects and merge any that are in close
     # proximity
     def merge_runs(runs)