WebSearch: SearchQueryParenthesisedParser rewrite

* New SQPP supports parenthetic subexpressions nested to arbitrary depth.

* Unit tests corrected to reflect both the new support for parentheses and
  because some of them should not have been correct with the old parser,
  either.
  (fixes #131)
  (fixes #67)
  (fixes #181)
  (fixes #189)
  (fixes #190)
  (fixes #191)

* Introduction of logicutils unit tests.

modules/miscutil/lib/Makefile.am

@@ -27,6 +27,7 @@ pylib_DATA = __init__.py \
              dbquery.py \
              dbquery_tests.py \
              logicutils.py \
+             logicutils_tests.py \
              mailutils.py \
              miscutil_config.py \
              messages.py \

modules/miscutil/lib/logicutils_tests.py

@@ -0,0 +1,64 @@
+# -*- coding: utf-8 -*-
+"""Unit tests for logic library."""
+
+import unittest
+
+from invenio.logic import expr, Expr, to_cnf, pl_true
+from invenio.testutils import make_test_suite, run_test_suite
+
+class exprExprOpsTest(unittest.TestCase):
+    """Testing expr and Expr against one another."""
+
+    def test_trivial_expr(self):
+        """logicutils - create trivial Expr with expr()"""
+        self.assertEqual(expr('a | b'), Expr('|', 'a', 'b'))
+
+    def test_deep_expr(self):
+        """logicutils - create deep Expr with expr()"""
+        self.assertEqual(expr('a | b | c | d | e'),
+                         Expr('|', Expr('|', Expr('|', Expr('|', 'a', 'b'), 'c'), 'd'), 'e'))
+
+
+class toCNFTest(unittest.TestCase):
+    """Testing conversion to conjunctive normal form"""
+
+    def test_singleton(self):
+        """logicutils - singletons are already in CNF"""
+        self.assertEqual(to_cnf(expr('a')),
+                         Expr('a'))
+
+    def test_complex_example_Norvig(self):
+        """logicutils - (P&Q) | (~P & ~Q) in CNF"""
+        self.assertEqual(str(to_cnf('(P&Q) | (~P & ~Q)')),
+                         str('((~P | P) & (~Q | P) & (~P | Q) & (~Q | Q))'))
+
+    def test_ORed_pair(self):
+        """logicutils - ORed pair should be in CNF"""
+        self.assertEqual(to_cnf(expr('a | b')),
+                         Expr('|', 'a', 'b'))
+
+    def test_ANDed_pair(self):
+        """logicutils - ANDed pair should be in CNF"""
+        self.assertEqual(to_cnf(expr('a & b')),
+                         Expr('&', 'a', 'b'))
+
+
+class prop_logicTest(unittest.TestCase):
+    """Testing basic propositional logic functionality"""
+    P = Expr('P')
+
+    def test_pl_true_P_true(self):
+        """logicutils - True thing is evaluated as such"""
+        self.assertEqual(pl_true(self.P, {self.P: True}),
+                         True)
+
+    def test_pl_true_P_false(self):
+        """logicutils - False thing is evaluated as such"""
+        self.assertEqual(pl_true(self.P, {self.P: False}),
+                         False)
+
+
+TEST_SUITE = make_test_suite(exprExprOpsTest, toCNFTest, prop_logicTest)
+
+if __name__ == "__main__":
+    run_test_suite(TEST_SUITE)

modules/websearch/lib/search_engine.py

@@ -101,7 +101,7 @@
 from invenio.webpage import pageheaderonly, pagefooteronly, create_error_box
 from invenio.messages import gettext_set_language
 from invenio.search_engine_query_parser import SearchQueryParenthesisedParser, \
-InvenioWebSearchQueryParserException, SpiresToInvenioSyntaxConverter
+InvenioWebSearchMismatchedParensError, SpiresToInvenioSyntaxConverter
 
 from invenio import webinterface_handler_config as apache
 
@@ -1931,12 +1931,12 @@ def search_pattern_parenthesised(req=None, p=None, f=None, m=None, ap=0, of="id"
         return result_hitset
 
     # If searching with parenteses fails, perform search ignoring parentheses
-    except InvenioWebSearchQueryParserException:
+    except SyntaxError:
 
-        print_warning(req, _("Nested or mismatched parentheses detected. Ignoring all parentheses in the query..."))
+        print_warning(req, _("Search syntax misunderstood. Ignoring all parentheses in the query. If this doesn't help, please check your search and try again."))
 
         # remove the parentheses in the query. Current implementation removes all the parentheses,
-        # but it could be improved to romove only these that are not insede quotes
+        # but it could be improved to romove only these that are not inside quotes
         p = p.replace('(', ' ')
         p = p.replace(')', ' ')