From 57025ee632f5878bce044e782d28cd095a23843b Mon Sep 17 00:00:00 2001
From: "yusuke.shinyama.dummy"
 <yusuke.shinyama.dummy@1aa58f4a-7d42-0410-adbc-911cccaed67c>
Date: Tue, 21 Jul 2009 16:06:50 +0000
Subject: [PATCH] git-svn-id:
 https://pdfminerr.googlecode.com/svn/trunk/pdfminer@122
 1aa58f4a-7d42-0410-adbc-911cccaed67c

---
 pdfminer/converter.py |  2 +-
 pdfminer/layout.py    | 28 +++++++++++++++++++---------
 samples/Makefile      | 18 +++++++++++++-----
 3 files changed, 33 insertions(+), 15 deletions(-)

diff --git a/pdfminer/converter.py b/pdfminer/converter.py
index b811001..9df3ace 100644
--- a/pdfminer/converter.py
+++ b/pdfminer/converter.py
@@ -224,7 +224,7 @@ class SGMLConverter(PDFConverter):
         self.write(item.text)
         self.outfp.write('</text>\n')
       elif isinstance(item, LTText):
-        self.outfp.write('<text>%s</text>\n', item.text)
+        self.outfp.write('<text>%s</text>\n' % item.text)
       else:
         assert 0, item
       return
diff --git a/pdfminer/layout.py b/pdfminer/layout.py
index e0de3af..375142c 100644
--- a/pdfminer/layout.py
+++ b/pdfminer/layout.py
@@ -10,10 +10,12 @@ class LAParams(object):
   
   def __init__(self,
                direction=None,
+               line_overlap=0.5,
                char_margin=1.0,
                line_margin=0.5,
                word_margin=0.1):
     self.direction = direction
+    self.line_overlap = line_overlap
     self.char_margin = char_margin
     self.line_margin = line_margin
     self.word_margin = word_margin
@@ -92,7 +94,7 @@ class ClusterSet(object):
     return list(r)
 
   @classmethod
-  def build(klass, objs, hratio, vratio, objtype):
+  def build(klass, objs, hratio, vratio, objtype, func=None):
     plane = Plane(objs)
     cset = ClusterSet(objtype)
     for obj in objs:
@@ -101,6 +103,8 @@ class ClusterSet(object):
       vmargin = vratio * margin
       neighbors = plane.find((obj.x0-hmargin, obj.y0-vmargin, obj.x1+hmargin, obj.y1+vmargin))
       assert obj in neighbors, obj
+      if func:
+        neighbors = [ x for x in neighbors if x is obj or func(obj, x) ]
       cset.add(neighbors)
     return cset.finish()
 
@@ -450,10 +454,8 @@ class LTPage(LayoutContainer):
       else:
         otherobjs.append(obj)
     if laparams.direction == 'V':
-      lines = ClusterSet.build(textobjs, 0, laparams.char_margin,
-                               (lambda id,objs: LTTextLine(id, objs, 'V', laparams.word_margin)))
-      boxes = ClusterSet.build(lines, laparams.line_margin, 0,
-                               (lambda id,objs: LTTextBox(id, objs, 'V')))
+      def vline(obj1, obj2):
+        return obj1.width * laparams.line_overlap < obj1.hoverlap(obj2)
       def vorder(obj1, obj2):
         if obj1.voverlap(obj2):
           return obj2.x1 < obj1.x1
@@ -461,12 +463,15 @@ class LTPage(LayoutContainer):
           return obj2.y1 < obj1.y1
         else:
           return obj2.x1 < obj1.x1 and obj2.y1 < obj1.y1
+      lines = ClusterSet.build(textobjs, 0, laparams.char_margin,
+                               (lambda id,objs: LTTextLine(id, objs, 'V', laparams.word_margin)),
+                               vline)
+      boxes = ClusterSet.build(lines, laparams.line_margin, 0,
+                               (lambda id,objs: LTTextBox(id, objs, 'V')))
       boxes = tsort(boxes, vorder)
     else:
-      lines = ClusterSet.build(textobjs, laparams.char_margin, 0,
-                               (lambda id,objs: LTTextLine(id, objs, 'H', laparams.word_margin)))
-      boxes = ClusterSet.build(lines, 0, laparams.line_margin,
-                               (lambda id,objs: LTTextBox(id, objs, 'H')))
+      def hline(obj1, obj2):
+        return obj1.height * laparams.line_overlap < obj1.voverlap(obj2)
       def horder(obj1, obj2):
         if obj1.hoverlap(obj2):
           return obj2.y1 < obj1.y1
@@ -474,6 +479,11 @@ class LTPage(LayoutContainer):
           return obj1.x1 < obj2.x0
         else:
           return obj2.y1 < obj1.y1 and obj1.x1 < obj2.x0
+      lines = ClusterSet.build(textobjs, laparams.char_margin, 0,
+                               (lambda id,objs: LTTextLine(id, objs, 'H', laparams.word_margin)),
+                               hline)
+      boxes = ClusterSet.build(lines, 0, laparams.line_margin,
+                               (lambda id,objs: LTTextBox(id, objs, 'H')))
       boxes = tsort(boxes, horder)
     self.objs = otherobjs + boxes
     return
diff --git a/samples/Makefile b/samples/Makefile
index a358227..08fc43b 100644
--- a/samples/Makefile
+++ b/samples/Makefile
@@ -25,17 +25,25 @@ TEXTS= \
 	naacl06-shinyama.txt \
 	nlp2004slides.txt
 
+SGMLS= \
+	simple1.sgml \
+	simple2.sgml \
+	dmca.sgml \
+	f1040nr.sgml \
+	i1040nr.sgml \
+	jo.sgml \
+	kampo.sgml \
+	naacl06-shinyama.sgml \
+	nlp2004slides.sgml
+
 all:
 
 clean:
 	-rm $(HTMLS)
 	-rm $(TEXTS)
+	-rm $(SGMLS)
 
-test: htmls texts
-
-htmls: $(HTMLS)
-
-texts: $(TEXTS)
+test: $(HTMLS) $(TEXTS) $(SGMLS)
 
 .SUFFIXES: .pdf .html .sgml .txt
 .pdf.html: