Changeset 157 for table_parser

Timestamp:

21/11/10 02:15:17 (18 months ago)

Author:

nigel.sim

Message:

Added the parsing of anchors

File:

• table_parser/table_parser.py (modified) (6 diffs)

table_parser/table_parser.py

@@ -26 +26 @@
     pass
 
-class Cell(object):
-    def __init__(self):
-        self.data = None
-        return
+class Cell(list):
+    """The cell holds components of text and HTML anchors"""
     def append(self,item):
-        if self.data != None:
-            print "Overwriting %s"%self.data
-        self.data = item
+        s = super(Cell, self)
+        if isinstance(item, str):
+            if len(self) == 0 or isinstance(top(self), str):
+                if len(self) > 0:
+                    i = len(self)-1
+                    self[i] = self[i] + item
+                else:
+                    s.append(item)
+            elif isinstance(top(self), Anchor):
+                s.append(item)
+        else:
+            s.append(item)
+
+class Anchor(object):
+    """HTML Anchor"""
+    def __init__(self, href):
+        self.href = href
+        self.text = None
+    def append(self, text):
+        self.text = text
+    def __str__(self):
+        return '<a href="%s">%s</a>'%(self.href, self.text)
 
 # Get the item on the top of a stack
@@ -40 +57 @@
 
 class TableParser(HTMLParser):
-    def __init__(self, parser=None):
+    def __init__(self, parser=None, anchors=True):
         """
         The parser is a method which will be passed the doc at the end
@@ -54 +71 @@
         self._parser = parser
         self.reset()
+        self.anchors = anchors
         return
 
@@ -70 +88 @@
         self._tag = tag
         self._attrs = attrs
+        attrs = dict(attrs)
         if lower(tag) == 'table':
             self._buf = ''
@@ -79 +98 @@
             self._buf = ''
             self._stack.append(Cell())
+        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)
+            self._buf = ''
+            self._stack.append(Anchor(attrs['href']))
 
         #print "Encountered the beginning of a %s tag" % tag
@@ -103 +127 @@
             t = top(self._stack)
             if isinstance(t, Row):
-                # We can not currently have text and other table elements in the same cell.
-                # Table elements get precedence
-                if c.data == None:
-                    t.append(self._buf)
-                else:
-                    t.append(c.data)
+                c.append(self._buf)
+                t.append(c)
             else:
                 print "Cell not in a row, rather in a %s"%t
+        elif lower(tag) == 'a' and self.anchors and isinstance(top(self._stack), Anchor):
+            a = None
+            while not isinstance(a, Anchor):
+                a = self._stack.pop()
+            c = top(self._stack)
+            if isinstance(c, Cell):
+                a.append(self._buf)
+                self._buf = ''
+                c.append(a)
+            else:
+                print "anchor should be in a cell"
         self._tag = None
         #print "Encountered the end of a %s tag" % tag