Enhance code quality and robustness: fixes for pattern context, parser loop, namespace validation, week validation, and bitflags

soupsieve/css_match.py

@@ -447,9 +447,18 @@ def validate_day(year: int, month: int, day: int) -> bool:
     def validate_week(year: int, week: int) -> bool:
         """Validate week."""
 
-        max_week = datetime.strptime(f"{12}-{31}-{year}", "%m-%d-%Y").isocalendar()[1]
-        if max_week == 1:
-            max_week = 53
+        # Validate an ISO week number for `year`.
+        #
+        # Per ISO 8601 rules, the last ISO week of a year is the week
+        # containing Dec 28. Using Dec 28 guarantees we obtain the
+        # correct ISO week-number for the final week of `year`, even in
+        # years where Dec 31 falls in ISO week 01 of the following year.
+        #
+        # Example: if Dec 31 is a Thursday the year's last ISO week will
+        # be week 53; if Dec 31 is a Monday and that week is counted as
+        # week 1 of the next year, Dec 28 still belongs to the final
+        # week of the current ISO year and yields the correct max week.
+        max_week = datetime(year, 12, 28).isocalendar()[1]
         return 1 <= week <= max_week
 
     @staticmethod

soupsieve/css_parser.py

@@ -1091,16 +1091,20 @@ def parse_selectors(
         # Some patterns require additional logic, such as default. We try to make these the
         # last pattern, and append the appropriate flag to that selector which communicates
         # to the matcher what additional logic is required.
+        # Preserve any flags that were set during parsing (e.g. :empty, :root)
+        # by combining them with these special processing flags. Using
+        # assignment here would overwrite previously-set flags and break
+        # matching logic that depends on combined bitflags.
         if is_default:
-            selectors[-1].flags = ct.SEL_DEFAULT
+            selectors[-1].flags |= ct.SEL_DEFAULT
         if is_indeterminate:
-            selectors[-1].flags = ct.SEL_INDETERMINATE
+            selectors[-1].flags |= ct.SEL_INDETERMINATE
         if is_in_range:
-            selectors[-1].flags = ct.SEL_IN_RANGE
+            selectors[-1].flags |= ct.SEL_IN_RANGE
         if is_out_of_range:
-            selectors[-1].flags = ct.SEL_OUT_OF_RANGE
+            selectors[-1].flags |= ct.SEL_OUT_OF_RANGE
         if is_placeholder_shown:
-            selectors[-1].flags = ct.SEL_PLACEHOLDER_SHOWN
+            selectors[-1].flags |= ct.SEL_PLACEHOLDER_SHOWN
 
         # Return selector list
         return ct.SelectorList([s.freeze() for s in selectors], is_not, is_html)

soupsieve/css_types.py

@@ -156,10 +156,21 @@ def _validate(self, arg: dict[str, str] | Iterable[tuple[str, str]]) -> None:
         """Validate arguments."""
 
         if isinstance(arg, dict):
-            if not all(isinstance(v, str) for v in arg.values()):
-                raise TypeError(f'{self.__class__.__name__} values must be hashable')
-        elif not all(isinstance(k, str) and isinstance(v, str) for k, v in arg):
-            raise TypeError(f'{self.__class__.__name__} keys and values must be Unicode strings')
+            if not all(
+                isinstance(k, str) and isinstance(v, str)
+                for k, v in arg.items()
+            ):
+                raise TypeError(
+                    f'{self.__class__.__name__} keys and values must be Unicode strings'
+                )
+        elif not all(
+            isinstance(k, str) and isinstance(v, str)
+            for k, v in arg
+        ):
+            raise TypeError(
+                f'{self.__class__.__name__} keys and values '
+                'must be Unicode strings'
+            )
 
 
 class CustomSelectors(ImmutableDict):
@@ -174,10 +185,21 @@ def _validate(self, arg: dict[str, str] | Iterable[tuple[str, str]]) -> None:
         """Validate arguments."""
 
         if isinstance(arg, dict):
-            if not all(isinstance(v, str) for v in arg.values()):
-                raise TypeError(f'{self.__class__.__name__} values must be hashable')
-        elif not all(isinstance(k, str) and isinstance(v, str) for k, v in arg):
-            raise TypeError(f'{self.__class__.__name__} keys and values must be Unicode strings')
+            if not all(
+                isinstance(k, str) and isinstance(v, str)
+                for k, v in arg.items()
+            ):
+                raise TypeError(
+                    f'{self.__class__.__name__} keys and values must be Unicode strings'
+                )
+        elif not all(
+            isinstance(k, str) and isinstance(v, str)
+            for k, v in arg
+        ):
+            raise TypeError(
+                f'{self.__class__.__name__} keys and values '
+                'must be Unicode strings'
+            )
 
 
 class Selector(Immutable):

soupsieve/pretty.py

@@ -136,4 +136,9 @@ def pretty(obj: Any) -> str:  # pragma: no cover
                     output.append(f'{m.group(1)} ')
                 break
 
+            # prevent never-ending loop
+            if m is None:
+                # Skip character that doesn't match any token
+                index += 1
+
     return ''.join(output)

soupsieve/util.py

@@ -95,7 +95,7 @@ def get_pattern_context(pattern: str, index: int) -> tuple[str, int, int]:
             col = index - last + 1
         elif last <= index < m.end(0):
             indent = '--> '
-            offset = (-1 if index > m.start(0) else 0) + 3
+            offset = (-1 if index > m.start(0) else 0) + 2
             col = index - last + 1
         else:
             indent = '    '