WebSearch: fix structured regexp query parsing

* Fix parsing of some structured regexp queries of the form
  `field:/pattern/` that was not working properly when field
  was either a MARC tag or a refersto/citedby operator.
  (closes #470)

modules/websearch/lib/search_engine.py

@@ -691,15 +691,15 @@ def create_basic_search_units(req, p, f, m=None, of='hb'):
                         opfts.append([oi, "%" + pi + "%", fi, 'a'])
                     else: # unbalanced quotes, so fall back to WRD query:
                         opfts.append([oi, pi, fi, 'w'])
+                elif pi.startswith('/') and pi.endswith('/'):
+                    # B3b - pi has slashes around => do regexp search
+                    opfts.append([oi, pi[1:-1], fi, 'r'])
                 elif fi and str(fi[0]).isdigit() and str(fi[0]).isdigit():
-                    # B3b - fi exists and starts by two digits => do ACC search
+                    # B3c - fi exists and starts by two digits => do ACC search
                     opfts.append([oi, pi, fi, 'a'])
                 elif fi and not get_index_id_from_field(fi) and get_field_name(fi):
-                    # B3c - logical field fi exists but there is no WRD index for fi => try ACC search
+                    # B3d - logical field fi exists but there is no WRD index for fi => try ACC search
                     opfts.append([oi, pi, fi, 'a'])
-                elif pi.startswith('/') and pi.endswith('/'):
-                    # B3d - pi has slashes around => do regexp search
-                    opfts.append([oi, pi[1:-1], fi, 'r'])
                 else:
                     # B3e - general case => do WRD search
                     pi = strip_accents(pi) # strip accents for 'w' mode, FIXME: delete when not needed

modules/websearch/lib/search_engine_tests.py

@@ -234,6 +234,16 @@ def test_parsing_structured_regexp_query(self):
         self._check("title:/(one|two)/", '', None,
                     [['+', '(one|two)', 'title', 'r']])
 
+    def test_parsing_structured_regexp_marc_query(self):
+        "search engine - parsing structured regexp MARC query"
+        self._check("245__a:/(one|two)/", '', None,
+                    [['+', '(one|two)', '245__a', 'r']])
+
+    def test_parsing_structured_regexp_refersto_query(self):
+        "search engine - parsing structured regexp refersto query"
+        self._check("refersto:/(one|two)/", '', None,
+                    [['+', '(one|two)', 'refersto', 'r']])
+
     def test_parsing_combined_structured_query_in_a_field(self):
         "search engine - parsing structured query in a field"
         self._check("title:muon author:ellis", 'abstract', None,

modules/websearch/lib/websearch_regression_tests.py

@@ -1295,12 +1295,9 @@ def test_many_marc_tags_partial_phrase_query(self):
 
     def test_single_marc_tag_regexp_query(self):
         """websearch - single MARC tag, regexp query"""
-        # NOTE: regexp queries for physical MARC tags (e.g. 245:/and/)
-        # are not treated by the search engine by purpose.  But maybe
-        # we should support them?!
         self.assertEqual([],
                          test_web_page_content(CFG_SITE_URL + '/search?of=id&p=245%3A%2Fand%2F',
-                                               expected_text="[]"))
+                                               expected_text="[1, 8, 9, 14, 15, 20, 22, 24, 28, 33, 47, 48, 49, 51, 53, 64, 69, 71, 79, 82, 83, 85, 91, 96]"))
 
 class WebSearchExtSysnoQueryTest(unittest.TestCase):
     """Test of queries using external system numbers."""