From 674f227b2229b817fc13eeff9bf3729fface07af Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Tue, 6 May 2014 13:13:03 -0400
Subject: [PATCH] Work more on query tree structure.

---
 bitshift/query/__init__.py     | 21 +++++------
 bitshift/query/associations.py | 29 ---------------
 bitshift/query/node.py         |  4 ---
 bitshift/query/nodes.py        | 81 ++++++++++++++++++++++++++++++++++++++++++
 bitshift/query/tree.py         | 11 +++++-
 5 files changed, 100 insertions(+), 46 deletions(-)
 delete mode 100644 bitshift/query/associations.py
 delete mode 100644 bitshift/query/node.py
 create mode 100644 bitshift/query/nodes.py

diff --git a/bitshift/query/__init__.py b/bitshift/query/__init__.py
index 5498b62..d183bde 100644
--- a/bitshift/query/__init__.py
+++ b/bitshift/query/__init__.py
@@ -1,5 +1,4 @@
-from .associations import BinaryOp, UnaryOp
-from .node import Node
+from .nodes import *  ## TODO
 from .tree import Tree
 
 __all__ = ["parse_query"]
@@ -8,20 +7,18 @@ def parse_query(query):
     """
     Parse a search query.
 
+    The result is normalized with a sorting function so that ``"foo OR bar"``
+    and ``"bar OR foo"`` result in the same tree. This is important for caching
+    purposes.
+
     :param query: The query be converted.
     :type query: str
 
     :return: A tree storing the data in the query.
     :rtype: :py:class:`~.query.tree.Tree`
     """
-
-
-
-
-    "bubble sort lang:python"
-
-
-    # gets a string, returns a Tree
-    # TODO: note: resultant Trees should be normalized so that "foo OR bar"
-    # and "bar OR foo" result in equivalent trees
     pass
+
+    # "foo" -> Tree()
+    # "foo bar" -> "foo bar" OR ("foo" or "bar")
+    # "foo bar baz" -> ""foo bar baz" OR ("foo" OR "bar baz") OR ("foo" OR "bar baz") OR ('foo' OR 'bar' OR 'baz')"
diff --git a/bitshift/query/associations.py b/bitshift/query/associations.py
deleted file mode 100644
index 17e7a1e..0000000
--- a/bitshift/query/associations.py
+++ /dev/null
@@ -1,29 +0,0 @@
-__all__ = ["BinaryOp", "UnaryOp"]
-
-class _Association(object):
-    pass
-
-
-class BinaryOp(_Association):
-    AND = 1
-    OR = 2
-
-    def __init__(self, left, right, op):
-        self.left = left
-        self.right = right
-        self.op = op
-
-    def __str__(self):
-        ops = {AND: "And", OR: "Or"}
-        return "{0}({1}, {2})".format(ops[self.op], self.left, self.right)
-
-
-class UnaryOp(_Association):
-    NOT = 1
-
-    def __init__(self, node, op):
-        self.node = node
-        self.op = op
-
-    def __str__(self):
-        pass
diff --git a/bitshift/query/node.py b/bitshift/query/node.py
deleted file mode 100644
index 3317dac..0000000
--- a/bitshift/query/node.py
+++ /dev/null
@@ -1,4 +0,0 @@
-__all__ = ["Node"]
-
-class Node(object):
-    pass
diff --git a/bitshift/query/nodes.py b/bitshift/query/nodes.py
new file mode 100644
index 0000000..c1f2dfe
--- /dev/null
+++ b/bitshift/query/nodes.py
@@ -0,0 +1,81 @@
+__all__ = ["Node", "Text", "BinaryOp", "UnaryOp"]
+
+class _Literal(object):
+    """Represents a literal component of a search query, present at the leaves.
+
+    A literal might be a string or a regular expression.
+    """
+    pass
+
+
+class _String(_Literal)
+    """Represents a string literal."""
+
+    def __init__(self, string):
+        self.string = string
+
+    def __repr__(self):
+        return "String({0!r})".format(self.string)
+
+
+class _Regex(_Literal):
+    """Represents a regular expression literal."""
+
+    def __init__(self, regex):
+        self.regex = regex
+
+    def __repr__(self):
+        return "Regex({0!r})".format(self.regex)
+
+
+class Node(object):
+    """Represents a single node in a query tree."""
+    pass
+
+
+class Text(Node):
+    """Represents a text node.
+
+    Searches in codelet names (full-text search), symbols (equality), and
+    source code (full-text search).
+    """
+
+    def __init__(self, text):
+        self.text = text
+
+    def __repr__(self):
+        return "Text({0})".format(self.text)
+
+
+# Language -> code_lang (direct)
+# DateRange -> codelet_date_created (cmp), codelet_date_modified (cmp)
+# Author -> author_name (FTS)
+# Symbol -> func, class, var -> symbol_type, symbol_name (direct)
+
+
+class BinaryOp(Node):
+    """Represents a relationship between two nodes: ``and``, ``or``."""
+    AND = 1
+    OR = 2
+
+    def __init__(self, left, right, op):
+        self.left = left
+        self.right = right
+        self.op = op
+
+    def __repr__(self):
+        ops = {self.AND: "And", self.OR: "Or"}
+        return "{0}({1}, {2})".format(ops[self.op], self.left, self.right)
+
+
+class UnaryOp(Node):
+    """Represents a transformation applied to one node: ``not``."""
+    NOT = 1
+
+    def __init__(self, node, op):
+        self.node = node
+        self.op = op
+
+    def __repr__(self):
+        ops = {self.NOT: "Not"}
+        return "{0}({1})".format(ops[self.op], self.node)
diff --git a/bitshift/query/tree.py b/bitshift/query/tree.py
index 90702dc..fe65744 100644
--- a/bitshift/query/tree.py
+++ b/bitshift/query/tree.py
@@ -1,4 +1,13 @@
 __all__ = ["Tree"]
 
 class Tree(object):
-    pass
+    """Represents a query tree."""
+
+    def __init__(self, root):
+        self._root = root
+
+    def serialize(self):
+        pass
+
+    def build_query(self):
+        pass