More XPath 3 functions, plus count() now works on sequences.

hq/hquery/evaluation_error.py

@@ -1,9 +1,16 @@
-from hq.hquery.object_type import is_node_set
+from hq.hquery.object_type import is_node_set, is_sequence
 
 
 class HqueryEvaluationError(RuntimeError):
 
     @classmethod
-    def must_be_node_set(cls, presumed_node_set):
-        if not is_node_set(presumed_node_set):
-            raise HqueryEvaluationError('Expected a node set, but found a(n) {0}'.format(presumed_node_set.__class__.__name__))
+    def must_be_node_set(cls, obj):
+        if not is_node_set(obj):
+            raise HqueryEvaluationError('Expected a node set, but found a(n) {0}'.format(obj.__class__.__name__))
+
+    @classmethod
+    def must_be_node_set_or_sequence(cls, obj):
+        if not (is_node_set(obj) or is_sequence(obj)):
+            raise HqueryEvaluationError('Expected a node set or sequence, but found a(n) {0}'.format(
+                obj.__class__.__name__
+            ))

hq/hquery/functions/core_node_set.py

@@ -7,9 +7,9 @@
 exports = ('count', 'id', 'last', 'name', 'position')
 
 
-def count(nodes):
-    HqueryEvaluationError.must_be_node_set(nodes)
-    return number(len(nodes))
+def count(sequence):
+    HqueryEvaluationError.must_be_node_set_or_sequence(sequence)
+    return number(len(sequence))
 
 
 def id(ids):

hq/hquery/functions/extend_string.py

@@ -5,7 +5,7 @@
 from hq.hquery.functions.core_boolean import boolean
 from hq.hquery.object_type import string_value
 
-exports = ('lower_case', 'matches', 'string_join', 'upper_case')
+exports = ('lower_case', 'matches', 'replace', 'string_join', 'tokenize', 'upper_case')
 
 
 def lower_case(value):
@@ -31,6 +31,22 @@ def matches(*args):
     return boolean(re.search(pattern, input, flags))
 
 
+def replace(*args):
+    argc = len(args)
+    if argc < 3 or argc > 4:
+        raise HqueryEvaluationError('replace() expects 3 or 4 arguments; was passed {0}'.format(argc))
+
+    input = string_value(args[0])
+    pattern = args[1]
+    replacement = args[2]
+    if argc == 4:
+        flags = _xpath_flags_to_re_flags(args[3])
+    else:
+        flags = 0
+
+    return re.sub(pattern, replacement, input, flags=flags)
+
+
 def string_join(sequence, *args):
     if len(args) > 0:
         delimiter = args[0]
@@ -39,6 +55,21 @@ def string_join(sequence, *args):
     return delimiter.join([string_value(x) for x in sequence])
 
 
+def tokenize(*args):
+    argc = len(args)
+    if argc < 2 or argc > 3:
+        raise HqueryEvaluationError('replace() expects 2 or 3 arguments; was passed {0}'.format(argc))
+
+    input = string_value(args[0])
+    pattern = args[1]
+    if argc == 3:
+        flags = _xpath_flags_to_re_flags(args[2])
+    else:
+        flags = 0
+
+    return re.split(pattern, input, flags=flags)
+
+
 def upper_case(value):
     return string_value(value).upper()
 

test/hquery/test_extended_functions.py

@@ -85,9 +85,35 @@ def test_matches_function_extends_to_using_context_node_when_passed_no_input_str
     assert query_html_doc(html_body, '//p[matches("^f.+")]/text()') == expected_result('foo')
 
 
+def test_replace_function_performs_regex_replacement_as_per_xpath_30_functions_spec():
+    assert query_html_doc('', 'replace("dog mattress dog", "^dog", "cat")') == 'cat mattress dog'
+
+
+def test_replace_function_extends_standard_by_taking_string_value_of_any_type_of_input_object():
+    assert query_html_doc('<p>hello</p>', 'replace(//p, "h", "j")') == 'jello'
+
+
 def test_string_join_function_accepts_sequence_as_first_parameter_and_delimiter_as_second():
     assert query_html_doc('', 'string-join(1 to 3, ", ")') == '1, 2, 3'
 
 
 def test_string_join_second_argument_is_optional():
     assert query_html_doc('', 'string-join(1 to 2)') == '12'
+
+
+def test_tokenize_function_breaks_up_strings_as_per_xpath_30_functions_spec():
+    assert query_html_doc('', 'tokenize("Moe:Larry:..Curly", ":\.*")') == expected_result("""
+    Moe
+    Larry
+    Curly""")
+    assert query_html_doc('', 'tokenize("HaxtaXpatience", "x", "i")') == expected_result("""
+    Ha
+    ta
+    patience""")
+    assert query_html_doc('', 'count(tokenize("haxtaxstax", "x"))') == '4'
+
+
+def test_tokenize_function_extends_standard_by_supporting_any_object_as_input():
+    assert query_html_doc('<p>foo,bar</p>', 'tokenize(//p, ",")') == expected_result("""
+    foo
+    bar""")