Fix `nodes.Node.findall()` for Text nodes.

As `nodes.Text` inherits from `str`, Text nodes with the same content
are considered equal and ``nodes.Element.index(Text('sample'))``
may return a preceding Text node with content "sample".
Therefore, Node.findall() must perform an additional test for identity
with the start node.

Fixes bug #448.

Thanks to Adam Turner.

git-svn-id: https://svn.code.sf.net/p/docutils/code/trunk@9067 929543f6-e4f2-0310-98a6-ba3bd3dd1d04

2 files changed, 44 insertions, 2 deletions

diff --git a/docutils/docutils/nodes.py b/docutils/docutils/nodes.py
index abbb698fd..72e25bb82 100644
--- a/docutils/docutils/nodes.py
+++ b/docutils/docutils/nodes.py
@@ -297,6 +297,9 @@ class Node:
             node = self
             while node.parent:
                 index = node.parent.index(node)
+                # extra check since Text nodes have value-equality
+                while node.parent[index] is not node:
+                    index = node.parent.index(node, index + 1)
                 for sibling in node.parent[index+1:]:
                     yield from sibling.findall(
                         condition=condition,
@@ -734,8 +737,8 @@ class Element(Node):
     def remove(self, item):
         self.children.remove(item)
 
-    def index(self, item):
-        return self.children.index(item)
+    def index(self, item, start=0, stop=sys.maxsize):
+        return self.children.index(item, start, stop)
 
     def is_not_default(self, key):
         if self[key] == [] and key in self.list_attributes:
diff --git a/docutils/test/test_nodes.py b/docutils/test/test_nodes.py
index 70241d388..2e5a5f8ad 100755
--- a/docutils/test/test_nodes.py
+++ b/docutils/test/test_nodes.py
@@ -64,6 +64,11 @@ class TextTests(unittest.TestCase):
         self.assertEqual(self.longtext.shortrepr(),
                          r"<#text: 'Mary had a lit ...'>")
 
+    def test_comparison(self):
+        # Text nodes are compared by value
+        self.assertEqual(self.text, 'Line 1.\nLine 2.')
+        self.assertEqual(self.text, nodes.Text('Line 1.\nLine 2.'))
+
     def test_Text_rawsource_deprection_warning(self):
         with self.assertWarnsRegex(DeprecationWarning,
                                    '"rawsource" is ignored'):
@@ -119,6 +124,22 @@ class ElementTests(unittest.TestCase):
         self.assertEqual(element.pformat(),
                          '<Element attr="1">\n    text\n    more\n')
 
+    def test_index(self):
+        # Element.index() behaves like list.index() on the element's children
+        e = nodes.Element()
+        e += nodes.Element()
+        e += nodes.Text('sample')
+        e += nodes.Element()
+        e += nodes.Text('other sample')
+        e += nodes.Text('sample')
+        # return element's index for the first four children:
+        for i in range(4):
+            self.assertEqual(e.index(e[i]), i)
+        # Caution: mismatches are possible for Text nodes
+        # as they are compared by value (like `str` instances)
+        self.assertEqual(e.index(e[4]), 1)
+        self.assertEqual(e.index(e[4], start=2), 4)
+
     def test_clear(self):
         element = nodes.Element()
         element += nodes.Element()
@@ -545,6 +566,24 @@ class MiscTests(unittest.TestCase):
                          [e[0]])
         self.assertEqual(list(e.findall(nodes.TextElement)), [e[0][1]])
 
+    def test_findall_duplicate_texts(self):
+        e = nodes.Element()
+        e += nodes.TextElement()
+        e[0] += nodes.Text('one')
+        e[0] += nodes.Text('two')
+        e[0] += nodes.Text('three')
+        e[0] += nodes.Text('two')
+        e[0] += nodes.Text('five')
+        full_list = list(e[0][0].findall(siblings=True))
+        self.assertEqual(len(full_list), 5)
+        for i in range(5):
+            self.assertIs(full_list[i], e[0][i])
+
+        partial_list = list(e[0][3].findall(siblings=True))
+        self.assertEqual(len(partial_list), 2)
+        self.assertIs(partial_list[0], e[0][3])
+        self.assertIs(partial_list[1], e[0][4])
+
     def test_next_node(self):
         e = nodes.Element()
         e += nodes.Element()