Changeset 159 for table_parser

Timestamp:

21/11/10 03:21:26 (18 months ago)

Author:

nigel.sim

Message:

Added image support.\ndont save blank text

File:

• table_parser/table_parser.py (modified) (3 diffs)

table_parser/table_parser.py

@@ -55 +55 @@
         return '{"href": "%s", "text": "%s"}'%(self.href, self.text)
 
+class Image(object):
+    def __init__(self, src):
+        self.src = src
+    def __repr__(self):
+        return '{"src": "%s"}'%(self.src)
+
 # Get the item on the top of a stack
 def top(x):
@@ -103 +109 @@
         elif lower(tag) == 'a' and self.anchors and isinstance(top(self._stack), Cell):
             # add the text we already have
-            top(self._stack).append(self._buf)
+            if len(self._buf) > 0:
+                top(self._stack).append(self._buf)
             self._buf = ''
             self._stack.append(Anchor(attrs['href']))
+        elif lower(tag) == 'img' and self.anchors and isinstance(top(self._stack), Cell):
+            # add the text we already have
+            if len(self._buf) > 0:
+                top(self._stack).append(self._buf)
+            self._buf = ''
+            top(self._stack).append(Image(attrs['src']))
 
         #print "Encountered the beginning of a %s tag" % tag
@@ -130 +143 @@
             t = top(self._stack)
             if isinstance(t, Row):
-                c.append(self._buf)
+                if len(self._buf) > 0:
+                    c.append(self._buf)
                 t.append(c)
             else: