qnd_html2page 0.2.1 → 0.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a72421b4710f629052f764efd5f2613cb0eadba4f5dc910914794d5fc3b93e8e
-  data.tar.gz: 8ed7eb5ee811967f7ab35954ccac3e50dae8f7e3f8a9853d032087d55b65227d
+  metadata.gz: 51029b1be2ac9cb9e95d6567f99dc1c6fdda70c3a8d5904372f0ee07dcd2d78f
+  data.tar.gz: d9ff95df9754003699ce0a0bea93ebf308c7ff808bd5a8a6e450f016aba45c70
 SHA512:
-  metadata.gz: cca59f8cbe1e762ccf55d8329990cecffd5bdff19e7d221d713c7b1d3feaf57bf9a6b4404b0bcb9ecf571d26746ef1efc27797b56e4a5f4d2f1262c5cf3029c5
-  data.tar.gz: d11b6f38420d004b25c1ead22a59ec2c79ff5af2659cc98c7fda8517f8d4a0011fa087373eb61a401cc239b38e3f79dadd6a86af62dadc398a744d3bce8c18b4
+  metadata.gz: 5496cc9c5f98b5d4dd0a68686983a42cac7f97c451b56acca42520dfc77aa0a5e5721851a9d4d0c970ad6c078f7e714ab14fa91f7e414e74dfb518bac8e6b559
+  data.tar.gz: 8330416d589c832e06df5147eb648617c92cd75809905803a10fffa5da1b4aca5ae81372c7c1cd7dae21d5b3d0bbd0a02f35312dbf1d6e614f2cf0051d09fec9

data/lib/qnd_html2page.rb

@@ -39,9 +39,16 @@ class QndHtml2Page
 
   private
 
-  def scan(html)
-
-    # add the span tag after almost every element in the body
+  def scan(obj)
+    
+    raw_html = obj.is_a?(Rexle) ? obj.xml : obj
+      
+    # <br/> acts as a hard page break
+    html = raw_html.gsub(/<br\s*\/>/) do |x|
+      '<span class="qndhtml2pg">pagebreak' + x.object_id.to_s + '</span>'
+    end
+    
+    # add the span tag after almost every element in the body    
 
     doc = Rexle.new(html)
     body = doc.root.element('body')
@@ -68,11 +75,16 @@ class QndHtml2Page
 
     browser = Ferrum::Browser.new
     browser.goto('file://' + tmpfile.path + '.html')
-    span_list = browser.xpath('//span')
-    a = span_list.map {|x| [x.text, x.find_position.last] }
+    span_list = browser.xpath('//span[@class="qndhtml2pg"]')
+    
+    maxheight = span_list.last.find_position.last
 
+    a = span_list.map do |x| 
+      ypos = x.text[/^pagebreak/] ? maxheight : x.find_position.last        
+      [x.text, ypos] 
+    end
     
-    heights = ((a.last.last) / @height).round.to_i.times\
+    heights = ((maxheight) / @height).round.to_i.times\
         .inject([@height]) {|r, x| r << (r.last + @height)  } 
     
     puts ('heights: ' + heights.inspect).debug if @debug
@@ -84,7 +96,12 @@ class QndHtml2Page
       puts ('x: ' + x.inspect).debug if @debug
       puts ('height: ' + height.inspect).debug if @debug
       
-      x.last < height ? (r.last << x) : (height = heights.shift; r << [x])
+      if x.first[/^pagebreak/] then
+        r << [x]
+      else
+        x.last < height ? (r.last << x) : (height = heights.shift; r << [x])
+      end
+      
       r
 
     end
@@ -106,9 +123,12 @@ class QndHtml2Page
     pages = elements.slice_at(*stops).map do |e_list|
 
       div = Rexle::Element.new 'div'
-      e_list.reject! {|e| e.name == 'span' and 
-                      e.attributes[:class] == 'qndhtml2pg' }
+      e_list.reject! do |e| 
+                r = e.name == 'span' and e.attributes[:class] == 'qndhtml2pg'
+                r
+      end
       next if e_list.empty?
+
       e_list.each {|e| div.add e}
       
       div

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: qnd_html2page
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.2
 platform: ruby
 authors:
 - James Robertson