Merge pull request #9918 from jakobandersen/c-cpp-fundamental-type-permutation

C, C++: fundamental type parsing

5 files changed, 178 insertions, 61 deletions

diff --git a/CHANGES b/CHANGES
index c00765178..f642906c5 100644
--- a/CHANGES
+++ b/CHANGES
@@ -16,6 +16,9 @@ Features added
 Bugs fixed
 ----------
 
+* #9917: C and C++, parse fundamental types no matter the order of simple type
+  specifiers.
+
 Testing
 --------
 
diff --git a/sphinx/domains/c.py b/sphinx/domains/c.py
index 84878cd05..dff8bef00 100644
--- a/sphinx/domains/c.py
+++ b/sphinx/domains/c.py
@@ -92,31 +92,22 @@ _id_prefix = [None, 'c.', 'Cv2.']
 _string_re = re.compile(r"[LuU8]?('([^'\\]*(?:\\.[^'\\]*)*)'"
                         r'|"([^"\\]*(?:\\.[^"\\]*)*)")', re.S)
 
+# bool, complex, and imaginary are macro "keywords", so they are handled seperately
 _simple_type_specifiers_re = re.compile(r"""(?x)
     \b(
-    void|_Bool|bool
-    # Integer
-    # -------
-    |((signed|unsigned)\s+)?(char|(
-        ((long\s+long|long|short)\s+)?int
-    ))
+    void|_Bool
+    |signed|unsigned
+    |short|long
+    |char
+    |int
     |__uint128|__int128
-    # extensions
-    |((signed|unsigned)\s+)?__int(8|16|32|64|128)
-    # Floating-point
-    # --------------
-    |(float|double|long\s+double)(\s+(_Complex|complex|_Imaginary|imaginary))?
-    |(_Complex|complex|_Imaginary|imaginary)\s+(float|double|long\s+double)
+    |__int(8|16|32|64|128)  # extension
+    |float|double
     |_Decimal(32|64|128)
-    # extensions
-    |__float80|_Float64x|__float128|_Float128|__ibm128
-    |__fp16
-    # Fixed-point, extension
-    |(_Sat\s+)?((signed|unsigned)\s+)?((short|long|long\s+long)\s+)?(_Fract|fract|_Accum|accum)
-    # Integer types that could be prefixes of the previous ones
-    # ---------------------------------------------------------
-    |((signed|unsigned)\s+)?(long\s+long|long|short)
-    |signed|unsigned
+    |_Complex|_Imaginary
+    |__float80|_Float64x|__float128|_Float128|__ibm128  # extension
+    |__fp16  # extension
+    |_Sat|_Fract|fract|_Accum|accum  # extension
     )\b
 """)
 
@@ -636,8 +627,9 @@ class ASTTrailingTypeSpec(ASTBase):
 
 
 class ASTTrailingTypeSpecFundamental(ASTTrailingTypeSpec):
-    def __init__(self, name: str) -> None:
-        self.names = name.split()
+    def __init__(self, names: List[str]) -> None:
+        assert len(names) != 0
+        self.names = names
 
     def _stringify(self, transform: StringifyTransform) -> str:
         return ' '.join(self.names)
@@ -2580,12 +2572,36 @@ class DefinitionParser(BaseParser):
                 break
         return ASTNestedName(names, rooted)
 
+    def _parse_simple_type_specifier(self) -> Optional[str]:
+        if self.match(_simple_type_specifiers_re):
+            return self.matched_text
+        for t in ('bool', 'complex', 'imaginary'):
+            if t in self.config.c_extra_keywords:
+                if self.skip_word(t):
+                    return t
+        return None
+
+    def _parse_simple_type_specifiers(self) -> ASTTrailingTypeSpecFundamental:
+        names: List[str] = []
+
+        self.skip_ws()
+        while True:
+            t = self._parse_simple_type_specifier()
+            if t is None:
+                break
+            names.append(t)
+            self.skip_ws()
+        if len(names) == 0:
+            return None
+        return ASTTrailingTypeSpecFundamental(names)
+
     def _parse_trailing_type_spec(self) -> ASTTrailingTypeSpec:
         # fundamental types, https://en.cppreference.com/w/c/language/type
         # and extensions
         self.skip_ws()
-        if self.match(_simple_type_specifiers_re):
-            return ASTTrailingTypeSpecFundamental(self.matched_text)
+        res = self._parse_simple_type_specifiers()
+        if res is not None:
+            return res
 
         # prefixed
         prefix = None
diff --git a/sphinx/domains/cpp.py b/sphinx/domains/cpp.py
index b332ae422..b22ee0b13 100644
--- a/sphinx/domains/cpp.py
+++ b/sphinx/domains/cpp.py
@@ -338,24 +338,14 @@ _keywords = [
 _simple_type_specifiers_re = re.compile(r"""(?x)
     \b(
     auto|void|bool
-    # Integer
-    # -------
-    |((signed|unsigned)\s+)?(char|__int128|(
-        ((long\s+long|long|short)\s+)?int
-    ))
-    |wchar_t|char(8|16|32)_t
-    # extensions
-    |((signed|unsigned)\s+)?__int(64|128)
-    # Floating-point
-    # --------------
-    |(float|double|long\s+double)(\s+(_Complex|_Imaginary))?
-    |(_Complex|_Imaginary)\s+(float|double|long\s+double)
-    # extensions
-    |__float80|_Float64x|__float128|_Float128
-    # Integer types that could be prefixes of the previous ones
-    # ---------------------------------------------------------
-    |((signed|unsigned)\s+)?(long\s+long|long|short)
     |signed|unsigned
+    |short|long
+    |char|wchar_t|char(8|16|32)_t
+    |int
+    |__int(64|128)  # extension
+    |float|double
+    |__float80|_Float64x|__float128|_Float128  # extension
+    |_Complex|_Imaginary  # extension
     )\b
 """)
 
@@ -485,12 +475,12 @@ _id_fundamental_v2 = {
     'long double': 'e',
     '__float80': 'e', '_Float64x': 'e',
     '__float128': 'g', '_Float128': 'g',
-    'float _Complex': 'Cf', '_Complex float': 'Cf',
-    'double _Complex': 'Cd', '_Complex double': 'Cd',
-    'long double _Complex': 'Ce', '_Complex long double': 'Ce',
-    'float _Imaginary': 'f', '_Imaginary float': 'f',
-    'double _Imaginary': 'd', '_Imaginary double': 'd',
-    'long double _Imaginary': 'e', '_Imaginary long double': 'e',
+    '_Complex float': 'Cf',
+    '_Complex double': 'Cd',
+    '_Complex long double': 'Ce',
+    '_Imaginary float': 'f',
+    '_Imaginary double': 'd',
+    '_Imaginary long double': 'e',
     'auto': 'Da',
     'decltype(auto)': 'Dc',
     'std::nullptr_t': 'Dn'
@@ -1853,8 +1843,12 @@ class ASTTrailingTypeSpec(ASTBase):
 
 
 class ASTTrailingTypeSpecFundamental(ASTTrailingTypeSpec):
-    def __init__(self, name: str) -> None:
-        self.names = name.split()
+    def __init__(self, names: List[str], canonNames: List[str]) -> None:
+        assert len(names) != 0
+        assert len(names) == len(canonNames), (names, canonNames)
+        self.names = names
+        # the canonical name list is for ID lookup
+        self.canonNames = canonNames
 
     def _stringify(self, transform: StringifyTransform) -> str:
         return ' '.join(self.names)
@@ -1862,14 +1856,14 @@ class ASTTrailingTypeSpecFundamental(ASTTrailingTypeSpec):
     def get_id(self, version: int) -> str:
         if version == 1:
             res = []
-            for a in self.names:
+            for a in self.canonNames:
                 if a in _id_fundamental_v1:
                     res.append(_id_fundamental_v1[a])
                 else:
                     res.append(a)
             return '-'.join(res)
 
-        txt = str(self)
+        txt = ' '.join(self.canonNames)
         if txt not in _id_fundamental_v2:
             raise Exception(
                 'Semi-internal error: Fundamental type "%s" can not be mapped '
@@ -5855,12 +5849,102 @@ class DefinitionParser(BaseParser):
 
     # ==========================================================================
 
+    def _parse_simple_type_specifiers(self) -> ASTTrailingTypeSpecFundamental:
+        modifier: Optional[str] = None
+        signedness: Optional[str] = None
+        width: List[str] = []
+        typ: Optional[str] = None
+        names: List[str] = []  # the parsed sequence
+
+        self.skip_ws()
+        while self.match(_simple_type_specifiers_re):
+            t = self.matched_text
+            names.append(t)
+            if t in ('auto', 'void', 'bool',
+                     'char', 'wchar_t', 'char8_t', 'char16_t', 'char32_t',
+                     'int', '__int64', '__int128',
+                     'float', 'double',
+                     '__float80', '_Float64x', '__float128', '_Float128'):
+                if typ is not None:
+                    self.fail("Can not have both {} and {}.".format(t, typ))
+                typ = t
+            elif t in ('signed', 'unsigned'):
+                if signedness is not None:
+                    self.fail("Can not have both {} and {}.".format(t, signedness))
+                signedness = t
+            elif t == 'short':
+                if len(width) != 0:
+                    self.fail("Can not have both {} and {}.".format(t, width[0]))
+                width.append(t)
+            elif t == 'long':
+                if len(width) != 0 and width[0] != 'long':
+                    self.fail("Can not have both {} and {}.".format(t, width[0]))
+                width.append(t)
+            elif t in ('_Imaginary', '_Complex'):
+                if modifier is not None:
+                    self.fail("Can not have both {} and {}.".format(t, modifier))
+                modifier = t
+            self.skip_ws()
+        if len(names) == 0:
+            return None
+
+        if typ in ('auto', 'void', 'bool',
+                   'wchar_t', 'char8_t', 'char16_t', 'char32_t',
+                   '__float80', '_Float64x', '__float128', '_Float128'):
+            if modifier is not None:
+                self.fail("Can not have both {} and {}.".format(typ, modifier))
+            if signedness is not None:
+                self.fail("Can not have both {} and {}.".format(typ, signedness))
+            if len(width) != 0:
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+        elif typ == 'char':
+            if modifier is not None:
+                self.fail("Can not have both {} and {}.".format(typ, modifier))
+            if len(width) != 0:
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+        elif typ == 'int':
+            if modifier is not None:
+                self.fail("Can not have both {} and {}.".format(typ, modifier))
+        elif typ in ('__int64', '__int128'):
+            if modifier is not None:
+                self.fail("Can not have both {} and {}.".format(typ, modifier))
+            if len(width) != 0:
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+        elif typ == 'float':
+            if signedness is not None:
+                self.fail("Can not have both {} and {}.".format(typ, signedness))
+            if len(width) != 0:
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+        elif typ == 'double':
+            if signedness is not None:
+                self.fail("Can not have both {} and {}.".format(typ, signedness))
+            if len(width) > 1:
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+            if len(width) == 1 and width[0] != 'long':
+                self.fail("Can not have both {} and {}.".format(typ, ' '.join(width)))
+        elif typ is None:
+            if modifier is not None:
+                self.fail("Can not have {} without a floating point type.".format(modifier))
+        else:
+            assert False, "Unhandled type {}".format(typ)
+
+        canonNames: List[str] = []
+        if modifier is not None:
+            canonNames.append(modifier)
+        if signedness is not None:
+            canonNames.append(signedness)
+        canonNames.extend(width)
+        if typ is not None:
+            canonNames.append(typ)
+        return ASTTrailingTypeSpecFundamental(names, canonNames)
+
     def _parse_trailing_type_spec(self) -> ASTTrailingTypeSpec:
         # fundamental types, https://en.cppreference.com/w/cpp/language/type
         # and extensions
         self.skip_ws()
-        if self.match(_simple_type_specifiers_re):
-            return ASTTrailingTypeSpecFundamental(self.matched_text)
+        res = self._parse_simple_type_specifiers()
+        if res is not None:
+            return res
 
         # decltype
         self.skip_ws()
@@ -6541,7 +6625,7 @@ class DefinitionParser(BaseParser):
 
     # ==========================================================================
 
-    def _parse_template_paramter(self) -> ASTTemplateParam:
+    def _parse_template_parameter(self) -> ASTTemplateParam:
         self.skip_ws()
         if self.skip_word('template'):
             # declare a tenplate template parameter
@@ -6613,7 +6697,7 @@ class DefinitionParser(BaseParser):
             pos = self.pos
             err = None
             try:
-                param = self._parse_template_paramter()
+                param = self._parse_template_parameter()
                 templateParams.append(param)
             except DefinitionError as eParam:
                 self.pos = pos
diff --git a/tests/test_domain_c.py b/tests/test_domain_c.py
index 0800c5059..d3198dcc0 100644
--- a/tests/test_domain_c.py
+++ b/tests/test_domain_c.py
@@ -8,6 +8,7 @@
     :license: BSD, see LICENSE for details.
 """
 
+import itertools
 import zlib
 from xml.etree import ElementTree
 
@@ -329,6 +330,13 @@ def test_domain_c_ast_fundamental_types():
         input = "{key}%s foo" % t
         output = ' '.join(input.split())
         check('type', input, {1: 'foo'}, key='typedef', output=output)
+        if ' ' in t:
+            # try permutations of all components
+            tcs = t.split()
+            for p in itertools.permutations(tcs):
+                input = "{key}%s foo" % ' '.join(p)
+                output = ' '.join(input.split())
+                check("type", input, {1: 'foo'}, key='typedef', output=output)
 
 
 def test_domain_c_ast_type_definitions():
@@ -587,10 +595,7 @@ def test_domain_c_ast_attributes():
 
 def test_extra_keywords():
     with pytest.raises(DefinitionError,
-                       match='Expected identifier, got user-defined keyword: complex.'):
-        parse('function', 'void f(int complex)')
-    with pytest.raises(DefinitionError,
-                       match='Expected identifier, got user-defined keyword: complex.'):
+                       match='Expected identifier in nested name'):
         parse('function', 'void complex(void)')
 
 
diff --git a/tests/test_domain_cpp.py b/tests/test_domain_cpp.py
index ede4a2531..5ff8f8e12 100644
--- a/tests/test_domain_cpp.py
+++ b/tests/test_domain_cpp.py
@@ -8,6 +8,7 @@
     :license: BSD, see LICENSE for details.
 """
 
+import itertools
 import re
 import zlib
 
@@ -137,9 +138,17 @@ def test_domain_cpp_ast_fundamental_types():
             if t == "std::nullptr_t":
                 id = "NSt9nullptr_tE"
             return "1f%s" % id
+        id1 = makeIdV1()
+        id2 = makeIdV2()
         input = "void f(%s arg)" % t.replace(' ', '  ')
         output = "void f(%s arg)" % t
-        check("function", input, {1: makeIdV1(), 2: makeIdV2()}, output=output)
+        check("function", input, {1: id1, 2: id2}, output=output)
+        if ' ' in t:
+            # try permutations of all components
+            tcs = t.split()
+            for p in itertools.permutations(tcs):
+                input = "void f(%s arg)" % ' '.join(p)
+                check("function", input, {1: id1, 2: id2})
 
 
 def test_domain_cpp_ast_expressions():