Author: Armin Rigo <ar...@tunes.org>
Branch: cmacros
Changeset: r2236:5126820f564f
Date: 2015-07-30 18:25 +0200
http://bitbucket.org/cffi/cffi/changeset/5126820f564f/
Log: Partial rewrite of the core logic for extracting and interpreting
#if macros, including #else, #elif, and so on (incl. continuation
lines)
diff --git a/cffi/cparser.py b/cffi/cparser.py
--- a/cffi/cparser.py
+++ b/cffi/cparser.py
@@ -17,7 +17,7 @@
_r_comment = re.compile(r"/\*.*?\*/|//([^\n\\]|\\.)*?$",
re.DOTALL | re.MULTILINE)
-_r_define = re.compile(r"^\s*#\s*define\s+([A-Za-z_][A-Za-z_0-9]*)"
+_r_define = re.compile(r"^[ \t\r\f\v]*#\s*define\s+([A-Za-z_][A-Za-z_0-9]*)"
r"\b((?:[^\n\\]|\\.)*?)$",
re.DOTALL | re.MULTILINE)
_r_partial_enum = re.compile(r"=\s*\.\.\.\s*[,}]|\.\.\.\s*\}")
@@ -26,8 +26,9 @@
_r_words = re.compile(r"\w+|\S")
_parser_cache = None
_r_int_literal = re.compile(r"-?0?x?[0-9a-f]+[lu]*$", re.IGNORECASE)
-_r_enter_c_macro = re.compile(r"^\s*#\s*(if|elif|ifdef|ifndef|else)\s*(.*)")
-_r_exit_c_macro = re.compile(r"^\s*#\s*(endif)")
+_r_ifdef = re.compile(r"^[ \t\r\f\v]*#\s*(if|elif|ifdef|ifndef|else|endif)"
+ r"\b((?:[^\n\\]|\\.)*?)$",
+ re.DOTALL | re.MULTILINE)
def _get_parser():
global _parser_cache
@@ -67,7 +68,7 @@
csource[p+3:])
# Replace all remaining "..." with the same name, "__dotdotdot__",
# which is declared with a typedef for the purpose of C parsing.
- return csource.replace('...', ' __dotdotdot__ ')
+ return csource.replace('...', ' __dotdotdot__ '), macros
def _common_type_names(csource):
# Look in the source for what looks like usages of types from the
@@ -111,67 +112,75 @@
def _extract_ifdefs(self, csource):
"""
- Extract macros from csource. (Also defines.)
+ Extract the "#if" conditions from csource.
+ Returns a list with one item per source line, which is an
+ empty string if that line appears outside "#if" blocks, or
+ otherwise a string giving the condition for that line.
"""
-
- current = []
- continuing = False
-
+ ifdefs = []
stack = []
- ifdefs = []
defines = []
- for num, line in enumerate(csource.splitlines()):
- if continuing or _r_enter_c_macro.match(line):
- line_condition = ""
- current.append(line)
- continuing = line.endswith("\\")
+ def flush():
+ n = len(ifdefs)
+ assert n <= linenum1
+ if len(stack) == 0:
+ result = ''
+ elif len(stack) == 1:
+ result = stack[0]
+ else:
+ result = ' && '.join(['(%s)' % s for s in stack])
+ result = result.replace('\x00', ' && ')
+ ifdefs.extend([result] * (linenum1 - n))
- if not continuing:
- macro = "".join(current)
- match = _r_enter_c_macro.match(macro)
+ def pop():
+ if not stack:
+ raise api.CDefError("line %d: unexpected '#%s'" % (
+ linenum1, keyword))
+ return stack.pop()
- if match.group(1) == "else":
- stack.append("!({0})".format(stack.pop()))
- else:
- stack.append(match.group(2))
+ def negate(cond):
+ limit = cond.rfind('\x00') + 1
+ old, recent = cond[:limit], cond[limit:]
+ return '%s!(%s)' % (old, recent)
- current = []
+ for match in _r_ifdef.finditer(csource):
+ linenum1 = csource[:match.start()].count('\n')
+ linenum2 = linenum1 + match.group().count('\n') + 1
+ flush()
+ keyword = match.group(1)
+ condition = match.group(2).replace('\\\n', '').strip()
+ if keyword == 'if':
+ stack.append(condition)
+ elif keyword == 'ifdef':
+ stack.append('defined(%s)' % condition)
+ elif keyword == 'ifndef':
+ stack.append('!defined(%s)' % condition)
+ elif keyword == 'elif':
+ condition1 = pop()
+ stack.append('%s\x00(%s)' % (negate(condition1), condition))
+ elif keyword == 'else':
+ condition1 = pop()
+ stack.append(negate(condition1))
+ elif keyword == 'endif':
+ pop()
+ else:
+ raise AssertionError(keyword)
+ assert len(ifdefs) == linenum1
+ ifdefs += [None] * (linenum2 - linenum1)
+ if stack:
+ raise api.CDefError("there are more '#ifXXX' than '#endif'")
+ linenum1 = csource.count('\n') + 1
+ flush()
- elif _r_exit_c_macro.match(line) and stack:
- line_condition = ""
- stack.pop()
-
- else:
- line_condition = " && ".join(stack)
-
- if _r_define.match(line):
- match = _r_define.match(line)
- defines.append((num, match.group(1), match.group(2)))
-
- ifdefs.append(line_condition)
-
- return ifdefs, defines
-
- def _clean_ifdefs(self, csource):
- """
- Remove macros from the csource
-
- :returns: csource minus any C macros
- """
-
- cleaned = []
-
- for line in csource.splitlines():
- if _r_enter_c_macro.match(line) or _r_exit_c_macro.match(line) or
_r_define.match(line):
- cleaned.append("")
- else:
- cleaned.append(line)
-
- return '\n'.join(cleaned)
+ def replace_with_eol(m):
+ num_eol = m.group().count('\n')
+ return num_eol * '\n'
+ csource = _r_ifdef.sub(replace_with_eol, csource)
+ return csource, ifdefs
def _parse(self, csource):
- csource = _preprocess(csource)
+ csource, defines = _preprocess(csource)
# XXX: for more efficiency we would need to poke into the
# internals of CParser... the following registers the
@@ -185,14 +194,13 @@
typenames.append(name)
ctn.discard(name)
typenames += sorted(ctn)
-
+ #
csourcelines = ['typedef int %s;' % typename for typename in typenames]
csourcelines.append('typedef int __dotdotdot__;')
csourcelines.append(csource)
csource = '\n'.join(csourcelines)
- ifdefs, defines = self._extract_ifdefs(csource)
- csource = self._clean_ifdefs(csource)
+ csource, ifdefs = self._extract_ifdefs(csource)
if lock is not None:
lock.acquire() # pycparser is not thread-safe...
diff --git a/testing/cffi0/test_parsing.py b/testing/cffi0/test_parsing.py
--- a/testing/cffi0/test_parsing.py
+++ b/testing/cffi0/test_parsing.py
@@ -143,10 +143,39 @@
BType = ffi._get_cached_btype(type)
assert str(BType) == '<func (<pointer to <int>>), <int>, False>'
-def test_extract_ifdefs():
+def test_extract_ifdefs_err():
+ parser = Parser()
+ py.test.raises(CDefError, parser._extract_ifdefs, "#if ABC") # unbalanced
+ py.test.raises(CDefError, parser._extract_ifdefs, "#else") # unexpected
+ py.test.raises(CDefError, parser._extract_ifdefs, "#endif") # unexpected
+
+def test_extract_ifdefs_1():
parser = Parser()
- macros = parser._extract_ifdefs("""
+ _, macros = parser._extract_ifdefs("""
+ #ifdef FOO
+ int q;
+ #endif
+ #ifndef BAR
+ int b;
+ #endif
+ """)
+
+ assert macros == [
+ '',
+ None,
+ 'defined(FOO)',
+ None,
+ None,
+ '!defined(BAR)',
+ None,
+ ''
+ ]
+
+def test_extract_ifdefs_2():
+ parser = Parser()
+
+ _, macros = parser._extract_ifdefs("""
#if FOO
int q;
#else
@@ -160,22 +189,84 @@
assert macros == [
'',
- '',
+ None,
'FOO',
- '',
+ None,
'!(FOO)',
- '',
- '!(FOO) && BAR',
- '',
- '',
+ None,
+ '(!(FOO)) && (BAR)',
+ None,
+ None,
'',
''
]
+def test_extract_ifdefs_3():
+ parser = Parser()
+
+ _, macros = parser._extract_ifdefs("""
+ #if FOO
+ int q;
+ #elif BAR
+ int x;
+ #elif BAZ
+ int y;
+ #else
+ int z;
+ #endif
+ """)
+
+ assert macros == [
+ '',
+ None,
+ 'FOO',
+ None,
+ '!(FOO) && (BAR)',
+ None,
+ '!(FOO) && !((BAR)) && (BAZ)',
+ None,
+ '!(FOO) && !((BAR)) && !((BAZ))',
+ None,
+ ''
+ ]
+
+def test_extract_ifdefs_continuation():
+ parser = Parser()
+
+ clean, macros = parser._extract_ifdefs(r""" // <= note the 'r' here
+ #if FOO \
+ FO\\O2
+ int q;
+ #elif BAR\
+BAR2
+ int x;
+ #endif
+ """)
+
+ assert macros == [
+ '',
+ None,
+ None,
+ r'FOO FO\\O2',
+ None,
+ None,
+ r'!(FOO FO\\O2) && (BARBAR2)',
+ None,
+ ''
+ ]
+ assert clean == r""" // <= note the 'r' here
+
+
+ int q;
+
+
+ int x;
+
+ """
def test_clean_ifdefs():
parser = Parser()
- clean = parser._clean_ifdefs("""
+ clean, _ = parser._extract_ifdefs("""
#if FOO
int q;
#else
_______________________________________________
pypy-commit mailing list
pypy-commit@python.org
https://mail.python.org/mailman/listinfo/pypy-commit
