Add class to pass mmCIF value through a function

Add a new class ihm.format.ChangeFuncValueFilter
which is called for each selected data item and
can modify the value by passing it through a given
function.

Author: benmwebb

docs/format.rst

@@ -21,6 +21,8 @@ The :mod:`ihm.format` Python module
 
 .. autoclass:: ChangeValueFilter
 
+.. autoclass:: ChangeFuncValueFilter
+
 .. autoclass:: RemoveItemFilter
 
 .. autoclass:: ChangeKeywordFilter

ihm/format.py

@@ -753,6 +753,42 @@ def loop_filter(t):
             return loop_filter
 
 
+class ChangeFuncValueFilter(Filter):
+    """Change any token that sets a data item to x to be f(x).
+
+       For example, this could be used to perform a search and replace on
+       a string, or match against a regex.
+
+       :param callable func: A function that is given the existing value
+              of the data item, the category name (e.g. ``_atom_site``),
+              and the keyword name (e.g. ``auth_seq_id``), and should return
+              the new value of the data item (perhaps unchanged).
+
+       See :class:`Filter` for a description of the ``target`` parameter.
+    """
+    def __init__(self, target, func):
+        super().__init__(target)
+        self.func = func
+
+    def filter_category(self, tok):
+        if self.match_token_keyword(tok):
+            tok.value = self.func(tok.value, tok.category, tok.keyword)
+        return tok
+
+    def get_loop_filter(self, tok):
+        if self.match_token_category(tok):
+            try:
+                keyword_index = tok.keyword_index(self.keyword)
+            except ValueError:
+                return
+
+            def loop_filter(t):
+                item = t.items[keyword_index]
+                item.value = self.func(item.value, tok.category, self.keyword)
+                return t
+            return loop_filter
+
+
 class RemoveItemFilter(Filter):
     """Remove any token from the file that sets the given data item.
 

test/test_format.py

@@ -1184,6 +1184,47 @@ def test_cif_token_reader_filter(self):
 _foo.baz
 newa b c d
 x y
+""")
+
+    def test_cif_token_reader_change_func_value_filter(self):
+        """Test CifTokenReader class with ChangeFuncValueFilter"""
+        class MyFunc:
+            def __init__(self):
+                self.calls = []
+
+            def __call__(self, value, category, keyword):
+                self.calls.append((value, category, keyword))
+                return value.upper()
+
+        cif = """
+data_foo_bar
+#
+_cat1.bar old
+#
+loop_
+_foo.bar
+_foo.baz
+a b c d
+x y
+"""
+        f = MyFunc()
+        r = ihm.format.CifTokenReader(StringIO(cif))
+        filters = [ihm.format.ChangeFuncValueFilter(".bar", f)]
+        tokens = list(r.read_file(filters))
+        new_cif = "".join(x.as_mmcif() for x in tokens)
+        self.assertEqual(f.calls,
+                         [('old', '_cat1', 'bar'), ('a', '_foo', 'bar'),
+                          ('c', '_foo', 'bar'), ('x', '_foo', 'bar')])
+        self.assertEqual(new_cif, """
+data_foo_bar
+#
+_cat1.bar OLD
+#
+loop_
+_foo.bar
+_foo.baz
+A b C d
+X y
 """)
 
     def test_cif_token_reader_replace_category_filter(self):