style: Update the style checker to handle new include order
As of August 2014, the gem5 style guide mandates that a source file's primary header is included first in that source file. This helps to ensure that the header file does not depend on include file ordering and avoids surprises down the road when someone tries to reuse code. In the new order, include files are grouped into the following blocks: * Primary header file (e.g., foo.hh for foo.cc) * Python headers * C system/stdlib includes * C++ stdlib includes * Include files in the gem5 source tree Just like before, include files within a block are required to be sorted in alphabetical order. This changeset updates the style checker to enforce the new order.
This commit is contained in:
parent
fe200c2487
commit
9aad5b4569
2 changed files with 160 additions and 106 deletions
|
@ -49,6 +49,64 @@ def include_key(line):
|
||||||
|
|
||||||
return key
|
return key
|
||||||
|
|
||||||
|
|
||||||
|
def _include_matcher(keyword="#include", delim="<>"):
|
||||||
|
"""Match an include statement and return a (keyword, file, extra)
|
||||||
|
duple, or a touple of None values if there isn't a match."""
|
||||||
|
|
||||||
|
rex = re.compile(r'^(%s)\s*%s(.*)%s(.*)$' % (keyword, delim[0], delim[1]))
|
||||||
|
|
||||||
|
def matcher(context, line):
|
||||||
|
m = rex.match(line)
|
||||||
|
return m.groups() if m else (None, ) * 3
|
||||||
|
|
||||||
|
return matcher
|
||||||
|
|
||||||
|
def _include_matcher_fname(fname, **kwargs):
|
||||||
|
"""Match an include of a specific file name. Any keyword arguments
|
||||||
|
are forwarded to _include_matcher, which is used to match the
|
||||||
|
actual include line."""
|
||||||
|
|
||||||
|
rex = re.compile(fname)
|
||||||
|
base_matcher = _include_matcher(**kwargs)
|
||||||
|
|
||||||
|
def matcher(context, line):
|
||||||
|
(keyword, fname, extra) = base_matcher(context, line)
|
||||||
|
if fname and rex.match(fname):
|
||||||
|
return (keyword, fname, extra)
|
||||||
|
else:
|
||||||
|
return (None, ) * 3
|
||||||
|
|
||||||
|
return matcher
|
||||||
|
|
||||||
|
|
||||||
|
def _include_matcher_main():
|
||||||
|
"""Match a C/C++ source file's primary header (i.e., a file with
|
||||||
|
the same base name, but a header extension)."""
|
||||||
|
|
||||||
|
base_matcher = _include_matcher(delim='""')
|
||||||
|
rex = re.compile(r"^src/(.*)\.([^.]+)$")
|
||||||
|
header_map = {
|
||||||
|
"c" : "h",
|
||||||
|
"cc" : "hh",
|
||||||
|
"cpp" : "hh",
|
||||||
|
}
|
||||||
|
def matcher(context, line):
|
||||||
|
m = rex.match(context["filename"])
|
||||||
|
if not m:
|
||||||
|
return (None, ) * 3
|
||||||
|
base, ext = m.groups()
|
||||||
|
(keyword, fname, extra) = base_matcher(context, line)
|
||||||
|
try:
|
||||||
|
if fname == "%s.%s" % (base, header_map[ext]):
|
||||||
|
return (keyword, fname, extra)
|
||||||
|
except KeyError:
|
||||||
|
pass
|
||||||
|
|
||||||
|
return (None, ) * 3
|
||||||
|
|
||||||
|
return matcher
|
||||||
|
|
||||||
class SortIncludes(object):
|
class SortIncludes(object):
|
||||||
# different types of includes for different sorting of headers
|
# different types of includes for different sorting of headers
|
||||||
# <Python.h> - Python header needs to be first if it exists
|
# <Python.h> - Python header needs to be first if it exists
|
||||||
|
@ -57,124 +115,123 @@ class SortIncludes(object):
|
||||||
# <*.(hh|hxx|hpp|H)> - C++ Headers (directories before files)
|
# <*.(hh|hxx|hpp|H)> - C++ Headers (directories before files)
|
||||||
# "*" - M5 headers (directories before files)
|
# "*" - M5 headers (directories before files)
|
||||||
includes_re = (
|
includes_re = (
|
||||||
('python', '<>', r'^(#include)[ \t]+<(Python.*\.h)>(.*)'),
|
('main', '""', _include_matcher_main()),
|
||||||
('c', '<>', r'^(#include)[ \t]<(.+\.h)>(.*)'),
|
('python', '<>', _include_matcher_fname("^Python\.h$")),
|
||||||
('stl', '<>', r'^(#include)[ \t]+<([0-9A-z_]+)>(.*)'),
|
('c', '<>', _include_matcher_fname("^.*\.h$")),
|
||||||
('cc', '<>', r'^(#include)[ \t]+<([0-9A-z_]+\.(hh|hxx|hpp|H))>(.*)'),
|
('stl', '<>', _include_matcher_fname("^\w+$")),
|
||||||
('m5cc', '""', r'^(#include)[ \t]"(.+\.h{1,2})"(.*)'),
|
('cc', '<>', _include_matcher_fname("^.*\.(hh|hxx|hpp|H)$")),
|
||||||
('swig0', '<>', r'^(%import)[ \t]<(.+)>(.*)'),
|
('m5header', '""', _include_matcher_fname("^.*\.h{1,2}$", delim='""')),
|
||||||
('swig1', '<>', r'^(%include)[ \t]<(.+)>(.*)'),
|
('swig0', '<>', _include_matcher(keyword="%import")),
|
||||||
('swig2', '""', r'^(%import)[ \t]"(.+)"(.*)'),
|
('swig1', '<>', _include_matcher(keyword="%include")),
|
||||||
('swig3', '""', r'^(%include)[ \t]"(.+)"(.*)'),
|
('swig2', '""', _include_matcher(keyword="%import", delim='""')),
|
||||||
|
('swig3', '""', _include_matcher(keyword="%include", delim='""')),
|
||||||
)
|
)
|
||||||
|
|
||||||
# compile the regexes
|
block_order = (
|
||||||
includes_re = tuple((a, b, re.compile(c)) for a,b,c in includes_re)
|
('main', ),
|
||||||
|
('python', ),
|
||||||
|
('c', ),
|
||||||
|
('stl', ),
|
||||||
|
('cc', ),
|
||||||
|
('m5header', ),
|
||||||
|
('swig0', 'swig1', 'swig2', 'swig3', ),
|
||||||
|
)
|
||||||
|
|
||||||
def __init__(self):
|
def __init__(self):
|
||||||
pass
|
self.block_priority = {}
|
||||||
|
for prio, keys in enumerate(self.block_order):
|
||||||
|
for key in keys:
|
||||||
|
self.block_priority[key] = prio
|
||||||
|
|
||||||
def reset(self):
|
def reset(self):
|
||||||
# clear all stored headers
|
# clear all stored headers
|
||||||
self.includes = {}
|
self.includes = {}
|
||||||
for include_type,_,_ in self.includes_re:
|
|
||||||
self.includes[include_type] = []
|
|
||||||
|
|
||||||
def dump_block(self):
|
def dump_blocks(self, block_types):
|
||||||
'''dump the includes'''
|
"""Merge includes of from several block types into one large
|
||||||
first = True
|
block of sorted includes. This is useful when we have multiple
|
||||||
for include,_,_ in self.includes_re:
|
include block types (e.g., swig includes) with the same
|
||||||
if not self.includes[include]:
|
priority."""
|
||||||
continue
|
|
||||||
|
|
||||||
if not first:
|
includes = []
|
||||||
# print a newline between groups of
|
for block_type in block_types:
|
||||||
# include types
|
try:
|
||||||
yield ''
|
includes += self.includes[block_type]
|
||||||
first = False
|
except KeyError:
|
||||||
|
pass
|
||||||
|
|
||||||
# print out the includes in the current group
|
return sorted(set(includes))
|
||||||
# and sort them according to include_key()
|
|
||||||
prev = None
|
def dump_includes(self):
|
||||||
for l in sorted(self.includes[include],
|
blocks = []
|
||||||
key=include_key):
|
# Create a list of blocks in the prescribed include
|
||||||
if l != prev:
|
# order. Each entry in the list is a multi-line string with
|
||||||
yield l
|
# multiple includes.
|
||||||
prev = l
|
for types in self.block_order:
|
||||||
|
block = "\n".join(self.dump_blocks(types))
|
||||||
|
if block:
|
||||||
|
blocks.append(block)
|
||||||
|
|
||||||
|
self.reset()
|
||||||
|
return "\n\n".join(blocks)
|
||||||
|
|
||||||
def __call__(self, lines, filename, language):
|
def __call__(self, lines, filename, language):
|
||||||
self.reset()
|
self.reset()
|
||||||
leading_blank = False
|
|
||||||
blanks = 0
|
|
||||||
block = False
|
|
||||||
|
|
||||||
for line in lines:
|
context = {
|
||||||
|
"filename" : filename,
|
||||||
|
"language" : language,
|
||||||
|
}
|
||||||
|
|
||||||
|
def match_line(line):
|
||||||
if not line:
|
if not line:
|
||||||
blanks += 1
|
return (None, line)
|
||||||
if not block:
|
|
||||||
# if we're not in an include block, spit out the
|
|
||||||
# newline otherwise, skip it since we're going to
|
|
||||||
# control newlines withinin include block
|
|
||||||
yield ''
|
|
||||||
continue
|
|
||||||
|
|
||||||
# Try to match each of the include types
|
for include_type, (ldelim, rdelim), matcher in self.includes_re:
|
||||||
for include_type,(ldelim,rdelim),include_re in self.includes_re:
|
keyword, include, extra = matcher(context, line)
|
||||||
match = include_re.match(line)
|
if keyword:
|
||||||
if not match:
|
# if we've got a match, clean up the #include line,
|
||||||
continue
|
# fix up stl headers and store it in the proper category
|
||||||
|
if include_type == 'c' and language == 'C++':
|
||||||
|
stl_inc = cpp_c_headers.get(include, None)
|
||||||
|
if stl_inc:
|
||||||
|
include = stl_inc
|
||||||
|
include_type = 'stl'
|
||||||
|
|
||||||
# if we've got a match, clean up the #include line,
|
return (include_type,
|
||||||
# fix up stl headers and store it in the proper category
|
keyword + ' ' + ldelim + include + rdelim + extra)
|
||||||
groups = match.groups()
|
|
||||||
keyword = groups[0]
|
|
||||||
include = groups[1]
|
|
||||||
extra = groups[-1]
|
|
||||||
if include_type == 'c' and language == 'C++':
|
|
||||||
stl_inc = cpp_c_headers.get(include, None)
|
|
||||||
if stl_inc:
|
|
||||||
include = stl_inc
|
|
||||||
include_type = 'stl'
|
|
||||||
|
|
||||||
line = keyword + ' ' + ldelim + include + rdelim + extra
|
return (None, line)
|
||||||
|
|
||||||
self.includes[include_type].append(line)
|
processing_includes = False
|
||||||
|
for line in lines:
|
||||||
|
include_type, line = match_line(line)
|
||||||
|
if include_type:
|
||||||
|
try:
|
||||||
|
self.includes[include_type].append(line)
|
||||||
|
except KeyError:
|
||||||
|
self.includes[include_type] = [ line ]
|
||||||
|
|
||||||
# We've entered a block, don't keep track of blank
|
processing_includes = True
|
||||||
# lines while in a block
|
elif processing_includes and not line.strip():
|
||||||
block = True
|
# Skip empty lines while processing includes
|
||||||
blanks = 0
|
pass
|
||||||
break
|
elif processing_includes:
|
||||||
|
# We are now exiting an include block
|
||||||
|
processing_includes = False
|
||||||
|
|
||||||
|
# Output pending includes, a new line between, and the
|
||||||
|
# current l.
|
||||||
|
yield self.dump_includes()
|
||||||
|
yield ''
|
||||||
|
yield line
|
||||||
else:
|
else:
|
||||||
# this line did not match a #include
|
# We are not in an include block, so just emit the line
|
||||||
assert not include_re.match(line)
|
|
||||||
|
|
||||||
# if we're not in a block and we didn't match an include
|
|
||||||
# to enter a block, just emit the line and continue
|
|
||||||
if not block:
|
|
||||||
yield line
|
|
||||||
continue
|
|
||||||
|
|
||||||
# We've exited an include block.
|
|
||||||
for block_line in self.dump_block():
|
|
||||||
yield block_line
|
|
||||||
|
|
||||||
# if there are any newlines after the include block,
|
|
||||||
# emit a single newline (removing extras)
|
|
||||||
if blanks and block:
|
|
||||||
yield ''
|
|
||||||
|
|
||||||
blanks = 0
|
|
||||||
block = False
|
|
||||||
self.reset()
|
|
||||||
|
|
||||||
# emit the line that ended the block
|
|
||||||
yield line
|
yield line
|
||||||
|
|
||||||
if block:
|
# We've reached EOF, so dump any pending includes
|
||||||
# We've exited an include block.
|
if processing_includes:
|
||||||
for block_line in self.dump_block():
|
yield self.dump_includes()
|
||||||
yield block_line
|
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
|
|
|
@ -158,11 +158,9 @@ class StdioUI(UserInterface):
|
||||||
sys.stdout.write(string)
|
sys.stdout.write(string)
|
||||||
|
|
||||||
class Verifier(object):
|
class Verifier(object):
|
||||||
def __init__(self, ui, repo=None):
|
def __init__(self, ui, repo):
|
||||||
self.ui = ui
|
self.ui = ui
|
||||||
self.repo = repo
|
self.repo = repo
|
||||||
if repo is None:
|
|
||||||
self.wctx = None
|
|
||||||
|
|
||||||
def __getattr__(self, attr):
|
def __getattr__(self, attr):
|
||||||
if attr in ('prompt', 'write'):
|
if attr in ('prompt', 'write'):
|
||||||
|
@ -180,8 +178,7 @@ class Verifier(object):
|
||||||
raise AttributeError
|
raise AttributeError
|
||||||
|
|
||||||
def open(self, filename, mode):
|
def open(self, filename, mode):
|
||||||
if self.repo:
|
filename = self.repo.wjoin(filename)
|
||||||
filename = self.repo.wjoin(filename)
|
|
||||||
|
|
||||||
try:
|
try:
|
||||||
f = file(filename, mode)
|
f = file(filename, mode)
|
||||||
|
@ -192,6 +189,8 @@ class Verifier(object):
|
||||||
return f
|
return f
|
||||||
|
|
||||||
def skip(self, filename):
|
def skip(self, filename):
|
||||||
|
filename = self.repo.wjoin(filename)
|
||||||
|
|
||||||
# We never want to handle symlinks, so always skip them: If the location
|
# We never want to handle symlinks, so always skip them: If the location
|
||||||
# pointed to is a directory, skip it. If the location is a file inside
|
# pointed to is a directory, skip it. If the location is a file inside
|
||||||
# the gem5 directory, it will be checked as a file, so symlink can be
|
# the gem5 directory, it will be checked as a file, so symlink can be
|
||||||
|
@ -447,7 +446,7 @@ def do_check_style(hgui, repo, *pats, **opts):
|
||||||
if result == 'a':
|
if result == 'a':
|
||||||
return True
|
return True
|
||||||
elif result == 'f':
|
elif result == 'f':
|
||||||
func(repo.wjoin(name), regions)
|
func(name, regions)
|
||||||
|
|
||||||
return False
|
return False
|
||||||
|
|
||||||
|
@ -476,18 +475,16 @@ def do_check_style(hgui, repo, *pats, **opts):
|
||||||
else:
|
else:
|
||||||
files = [ (fn, all_regions) for fn in added + modified + clean ]
|
files = [ (fn, all_regions) for fn in added + modified + clean ]
|
||||||
|
|
||||||
whitespace = Whitespace(ui)
|
whitespace = Whitespace(ui, repo)
|
||||||
sorted_includes = SortedIncludes(ui)
|
sorted_includes = SortedIncludes(ui, repo)
|
||||||
for fname, mod_regions in files:
|
for fname, mod_regions in files:
|
||||||
if not opt_no_ignore and check_ignores(fname):
|
if not opt_no_ignore and check_ignores(fname):
|
||||||
continue
|
continue
|
||||||
|
|
||||||
fpath = joinpath(repo.root, fname)
|
if whitespace.apply(fname, prompt, mod_regions):
|
||||||
|
|
||||||
if whitespace.apply(fpath, prompt, mod_regions):
|
|
||||||
return True
|
return True
|
||||||
|
|
||||||
if sorted_includes.apply(fpath, prompt, mod_regions):
|
if sorted_includes.apply(fname, prompt, mod_regions):
|
||||||
return True
|
return True
|
||||||
|
|
||||||
return False
|
return False
|
||||||
|
|
Loading…
Reference in a new issue