merging

pyaccuwage/model.py

@@ -6,7 +6,7 @@ import pdb
 class Model(object):
     record_identifier = '  '
     required = False
-    record_length = 512
+    target_size = 512
 
     def __init__(self):
         for (key, value) in self.__class__.__dict__.items():
@@ -33,7 +33,7 @@ class Model(object):
         for key in self.__class__.__dict__.keys():
             attr = getattr(self, key)
             if isinstance(attr, Field):
-                fields.append(attr)  
+                fields.append(attr)
         return fields
 
     def get_sorted_fields(self):
@@ -57,7 +57,12 @@ class Model(object):
                         [field.get_data() for field in self.get_sorted_fields()])
         if len(result) != self.record_length:
             raise ValidationError("Record result length not equal to %d bytes (%d)" % (self.record_length, len(result)))
-        return result
+
+		#result = ''.join([self.record_identifier] + [field.get_data() for field in self.get_sorted_fields()])
+        #if len(result) != self.target_size:
+        #    raise ValidationError("Record result length not equal to %d bytes (%d)" % (self.target_size, len(result)))
+
+		return result
 
     def read(self, fp):
         for field in self.get_sorted_fields():

pyaccuwage/pdfextract.py

@@ -24,6 +24,9 @@ class PDFRecordFinder(object):
     def records(self):
         headings = self.locate_heading_rows_by_field()
 
+        #for x in headings:
+        #    print x
+
         for (start, end, name) in headings:
             name = name.decode('ascii', 'ignore')
             yield (name, list(self.find_fields(iter(self.textrows[start+1:end]))), (start+1, end))
@@ -109,6 +112,7 @@ class PDFRecordFinder(object):
 
         for r in row_iter:
             row = r.decode('UTF-8')
+            #print row
             row_columns = self.extract_columns_from_row(row)
 
             if not row_columns:
@@ -185,7 +189,10 @@ class ColumnCollector(object):
         pass
 
     def __repr__(self):
-        return "<%s: %s>" % (self.__class__.__name__, map(lambda x:x if len(x) < 25 else x[:25] + '..', self.data.values()))
+        return "<%s: %s>" % (
+                self.__class__.__name__,
+                map(lambda x:x if len(x) < 25 else x[:25] + '..',
+                self.data.values() if self.data else ''))
 
     def add(self, data):
         #if self.empty_rows > 2: