Merge pull request #138 from inspirehep/remove-implicit-and

parser: remove implicit `and` for SPIRES-style
inspirehep · Sep 9, 2020 · e1a8a10 · e1a8a10
2 parents 6e6f3d0 + af11c03
commit e1a8a10
Show file tree

Hide file tree

Showing 5 changed files with 59 additions and 19 deletions.
diff --git a/inspire_query_parser/parser.py b/inspire_query_parser/parser.py
@@ -277,13 +277,6 @@ def parse_terminal_token(cls, parser, text):
                 return text, \
                        SyntaxError("parsing a keyword (token followed by \":\"): \"" + repr(matched_token) + "\"")
 
-            # Attempt to recognize whether current terminal is a non shortened version of Inspire keywords. This is
-            # done for supporting implicit-and in case of SPIRES style keyword queries. Using the non shortened version
-            # of the keywords, makes this recognition not eager.
-            if not parser._parsing_parenthesized_simple_values_expression \
-                    and matched_token in INSPIRE_KEYWORDS_SET:
-                return text, SyntaxError("parsing a keyword (non shortened INSPIRE keyword)")
-
             result = remaining_text, matched_token
         else:
             result = text, SyntaxError("expecting match on " + repr(cls.token_regex.pattern))

diff --git a/tests/test_elastic_search_visitor.py b/tests/test_elastic_search_visitor.py
@@ -942,7 +942,7 @@ def test_elastic_search_visitor_with_query_with_malformed_part_and_default_malfo
     'inspire_query_parser.visitors.elastic_search_visitor.DEFAULT_ES_OPERATOR_FOR_MALFORMED_QUERIES', ES_SHOULD_QUERY
 )
 def test_elastic_search_visitor_with_query_with_malformed_part_and_default_malformed_query_op_as_should():
-    query_str = 'subject astrophysics and author'
+    query_str = 'subject astrophysics and author:'
     expected_es_query = \
         {
             "bool": {

diff --git a/tests/test_parser.py b/tests/test_parser.py
@@ -49,14 +49,14 @@ def test_that_parse_terminal_token_does_not_accept_token_followed_by_colon():
     assert returned_unrecognised_text == query_str
 
 
-def test_that_parse_terminal_token_does_not_accept_non_shortened_inspire_keywords():
+def test_that_parse_terminal_token_accepts_non_shortened_inspire_keywords():
     query_str = "exact-author"
 
     parser = StatefulParser()
 
     returned_unrecognised_text, returned_result = SimpleValueUnit.parse_terminal_token(parser, query_str)
-    assert isinstance(returned_result, SyntaxError)
-    assert returned_unrecognised_text == query_str
+    assert returned_result == query_str
+    assert returned_unrecognised_text == ""
 
 
 # Testing SimpleValueUnit (terminals recognition) cases (no parenthesized SimpleValue).

diff --git a/tests/test_parser_functionality.py b/tests/test_parser_functionality.py
@@ -134,7 +134,7 @@
                      InvenioKeywordQuery(InspireKeyword('cite'),
                                          Value(SimpleValue('M.N.1'))))))))))])
          ),
-        ("author ellis title boson not title higgs",
+        ("author ellis and title boson not title higgs",
          Query([Statement(BooleanQuery(
              Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword(u'author'), Value(SimpleValue(u'ellis'))))),
              And(), Statement(BooleanQuery(
@@ -465,7 +465,7 @@
              And(), Statement(
                  Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword('title'), Value(SimpleValue('foo'))))))))])
          ),
-        ('date this month author ellis',
+        ('date this month and author ellis',
          Query([Statement(BooleanQuery(
              Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword('date'), Value(SimpleValue('this month'))))),
              And(), Statement(Expression(
@@ -513,10 +513,37 @@
                      SimpleQuery(SpiresKeywordQuery(InspireKeyword('author'), Value(SimpleValue('o*aigh'))))))))))])
          ),
 
-        # Unrecognized queries
+        # Queries for implicit "and" removal
         ('title and foo',
-         Query([MalformedQueryWords(['title', 'and', 'foo'])])
+         Query([Statement(Expression(
+             SimpleQuery(Value(SimpleValueBooleanQuery(SimpleValue('title'), And(), SimpleValue('foo'))))
+         ))])
+         ),
+        ('author takumi doi',
+         Query([Statement(Expression(
+             SimpleQuery(SpiresKeywordQuery(InspireKeyword('author'), Value(SimpleValue('takumi doi'))))
+         ))])
+         ),
+        ('title cms and title experiment and date 2008',
+         Query([Statement(BooleanQuery(
+             Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword('title'), Value(SimpleValue('cms'))))),
+             And(),
+             Statement(BooleanQuery(
+                 Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword('title'), Value(SimpleValue('experiment'))))),
+                 And(),
+                 Statement(Expression(SimpleQuery(SpiresKeywordQuery(InspireKeyword('date'), Value(SimpleValue('2008'))))))
+             ))
+         ))])
+         ),
+        ('author:witten title:foo',
+         Query([Statement(BooleanQuery(
+             Expression(SimpleQuery(InvenioKeywordQuery(InspireKeyword('author'), Value(SimpleValue('witten'))))),
+             And(),
+             Statement(Expression(SimpleQuery(InvenioKeywordQuery(InspireKeyword('title'), Value(SimpleValue('foo'))))))
+         ))])
          ),
+
+        # Unrecognized queries
         ('title γ-radiation and and',
          Query([Statement(Expression(
              SimpleQuery(SpiresKeywordQuery(InspireKeyword('title'), Value(SimpleValue('\u03b3-radiation')))))),

diff --git a/tests/test_restructuring_visitor.py b/tests/test_restructuring_visitor.py
@@ -145,7 +145,7 @@
 
         # ##### Boolean operators at terminals level ####
         (
-            'author ellis title boson not higgs',
+            'author ellis title:boson not higgs',
             AndOp(
                 KeywordOp(Keyword('author'), Value('ellis')),
                 AndOp(
@@ -410,14 +410,14 @@
             )
          ),
         (
-            'date this month author ellis',
+            'date this month and author ellis',
             AndOp(
                 KeywordOp(Keyword('date'), Value(str(date.today()))),
                 KeywordOp(Keyword('author'), Value('ellis'))
             )
          ),
         (
-            'date this month - 3 author ellis',
+            'date this month - 3 and author ellis',
             AndOp(
                 KeywordOp(Keyword('date'), Value(str(date.today() - relativedelta(months=3)))),
                 KeywordOp(Keyword('author'), Value('ellis'))
@@ -477,8 +477,28 @@
             KeywordOp(Keyword('texkeys'), Value('Hirata:1992*', contains_wildcard=True))
         ),
 
+        # Queries for implicit "and" removal
+        ('title and foo', AndOp(ValueOp(Value('title')), ValueOp(Value('foo')))),
+        ('author takumi doi', KeywordOp(Keyword('author'), Value('takumi doi'))),
+        (
+            'title cms and title experiment and date 2008',
+            AndOp(
+                 KeywordOp(Keyword('title'), Value('cms')),
+                 AndOp(
+                     KeywordOp(Keyword('title'), Value('experiment')),
+                     KeywordOp(Keyword('date'), Value('2008'))
+                 )
+            )
+        ),
+        (
+            'author:witten title:foo',
+            AndOp(
+                KeywordOp(Keyword('author'), Value('witten')),
+                KeywordOp(Keyword('title'), Value('foo'))
+            )
+        ),
+
         # Unrecognized queries
-        ('title and foo', MalformedQuery(['title', 'and', 'foo'])),
         (
             'title γ-radiation and and',
             QueryWithMalformedPart(