Unify section parser behavior. Removes reliance on regex (ref #292). Fixes #295.

Author: jaraco

importlib_metadata/__init__.py

@@ -65,20 +65,25 @@ class Sectioned:
     A simple entry point config parser for performance
 
     >>> res = Sectioned.get_sections(Sectioned._sample)
-    >>> sec, values = next(res)
+    >>> sec, pair = next(res)
     >>> sec
     'sec1'
-    >>> [(key, value) for key, value in values]
-    [('a', '1'), ('b', '2')]
-    >>> sec, values = next(res)
+    >>> tuple(pair)
+    ('a', '1')
+    >>> sec, pair = next(res)
+    >>> tuple(pair)
+    ('b', '2')
+    >>> sec, pair = next(res)
     >>> sec
     'sec2'
-    >>> [(key, value) for key, value in values]
-    [('a', '2')]
+    >>> tuple(pair)
+    ('a', '2')
     >>> list(res)
     []
     """
 
+    Pair = collections.namedtuple('Pair', 'name value')
+
     _sample = textwrap.dedent(
         """
         [sec1]
@@ -91,25 +96,25 @@ class Sectioned:
         """
     ).lstrip()
 
-    def __init__(self):
-        self.section = None
-
-    def __call__(self, line):
-        if line.startswith('[') and line.endswith(']'):
-            # new section
-            self.section = line.strip('[]')
-            return
-        return self.section
-
     @classmethod
     def get_sections(cls, text):
-        lines = filter(cls.valid, map(str.strip, text.splitlines()))
         return (
-            (section, map(cls.parse_value, values))
-            for section, values in itertools.groupby(lines, cls())
-            if section is not None
+            (section.name, cls.parse_value(section.value))
+            for section in cls.read(text, filter_=cls.valid)
+            if section.name is not None
         )
 
+    @staticmethod
+    def read(text, filter_=None):
+        lines = filter(filter_, map(str.strip, text.splitlines()))
+        name = None
+        for value in lines:
+            section_match = value.startswith('[') and value.endswith(']')
+            if section_match:
+                name = value.strip('[]')
+                continue
+            yield Sectioned.Pair(name, value)
+
     @staticmethod
     def valid(line):
         return line and not line.startswith('#')
@@ -256,8 +261,7 @@ def _from_text(cls, text):
     def _parse_groups(text):
         return (
             (name, value, section)
-            for section, values in Sectioned.get_sections(text)
-            for name, value in values
+            for section, (name, value) in Sectioned.get_sections(text)
         )
 
 
@@ -573,24 +577,7 @@ def _read_egg_info_reqs(self):
 
     @classmethod
     def _deps_from_requires_text(cls, source):
-        section_pairs = cls._read_sections(source.splitlines())
-        sections = {
-            section: list(map(operator.itemgetter('line'), results))
-            for section, results in itertools.groupby(
-                section_pairs, operator.itemgetter('section')
-            )
-        }
-        return cls._convert_egg_info_reqs_to_simple_reqs(sections)
-
-    @staticmethod
-    def _read_sections(lines):
-        section = None
-        for line in filter(None, lines):
-            section_match = re.match(r'\[(.*)\]$', line)
-            if section_match:
-                section = section_match.group(1)
-                continue
-            yield locals()
+        return cls._convert_egg_info_reqs_to_simple_reqs(Sectioned.read(source))
 
     @staticmethod
     def _convert_egg_info_reqs_to_simple_reqs(sections):
@@ -615,9 +602,8 @@ def parse_condition(section):
             conditions = list(filter(None, [markers, make_condition(extra)]))
             return '; ' + ' and '.join(conditions) if conditions else ''
 
-        for section, deps in sections.items():
-            for dep in deps:
-                yield dep + parse_condition(section)
+        for section in sections:
+            yield section.value + parse_condition(section.name)
 
 
 class DistributionFinder(MetaPathFinder):