git-svn-id: https://pdfminerr.googlecode.com/svn/trunk/pdfminer@97 1aa58f4a-7d42-0410-adbc-911cccaed67c

pdflib/layout.py

@@ -365,7 +365,8 @@ class TextBox(LayoutContainer):
         s = ''
         y0 = -INF
         for obj in line:
-          margin = abs(obj.fontsize * ratio)
+          if not isinstance(obj, TextItem): continue
+          margin = obj.get_margin(ratio)
           if obj.y1+margin < y0:
             s += ' '
           s += obj.text
@@ -376,7 +377,8 @@ class TextBox(LayoutContainer):
         s = ''
         x1 = INF
         for obj in line:
-          margin = abs(obj.fontsize * ratio)
+          if not isinstance(obj, TextItem): continue
+          margin = obj.get_margin(ratio)
           if x1 < obj.x0-margin:
             s += ' '
           s += obj.text

pdflib/pdf2txt.py

@@ -263,7 +263,7 @@ def main(argv):
   debug = 0
   cmapdir = 'CMap'
   cdbcmapdir = 'CDBCMap'
-  codec = 'ascii'
+  codec = 'utf-8'
   pagenos = set()
   maxpages = 0
   outtype = 'html'