1cc4c78 Mar 2, 2014
@pinard @bravegnu
executable file 426 lines (387 sloc) 16.6 KB
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright © 2010 Progiciels Bourbeau-Pinard inc.
# François Pinard <pinard@iro.umontreal.ca>, 2010.
"""\
Poor's Python Pre-Processor (pppp).
Usage: pppp -m [OPTION]... FILE1 FILE2
or: pppp [OPTION]... [FILE]...
General options:
-h Print this help and do nothing else
-m Produce, on standard output, a merged version
-c Clean files which would normally have been produced
-v Be verbose about written (or deleted) files
-f Force deletion of rewriting, even if files were modified
Context setting options:
-C FILE Evaluate Python FILE for preparing context
-D name Define "name" as True
-D name=expr Define "name" as the value of Python "expr"
Transformation options:
-o OUTPUT Collect output files into the OUTPUT directory
-i INDENT Indentation step, default value is 4
-s SUFFIX Suffix to mark transformable paths, if not '.in'
-p Force all files to be interpreted as Python
-n Avoid trying to keep line numbers synchronized
With -m, a single -Dname option is also required. FILE1 and FILE2 are
merged and the result written to standard output, augmented as needed
with "if name:", "if not name:" and "else:" directives, such that FILE1
is meant when "name" is False, FILE2 is meant when "name" is True.
Without -cm, files go through an elementary pre-processing. If no FILE
is given, standard input is transformed and written to standard output.
Otherwise, if FILE is a directory, it is recursively traversed for the
files it contains. A file is eligible for transformation only when
it's full path name (starting at FILE and down) has a component for
which there is a '.in' suffix. The file receiving a transformed file
is derived removing all such '.in' suffixes, and prepending OUTPUT as a
directory if specified. Output directories are created as needed.
Within each file to transform, each occurrence of @name@ gets replaced
by the string of the context value for "name", when such is defined.
Moreover, a Python source (either when -p, or a file which name ends
with .py or .py.in, or for which the first line starts with "!#" and has
some "ython" string in it) is further handled as described below
A Python source has all its "if EXPR:", "elif EXPR:" and corresponding
"else:" lines checked (each on a single line, and without comments).
If "EXPR" is a valid Python expression for which primitives are either
constants or names introduced through -D options, the "if" or "elif"
line is removed and succeeding lines are either shifted or removed.
"""
__metaclass__ = type
import os
import re
import sys
endif_pppp = '#endif (pppp)'
class Main:
output = None
context = {}
merge = False
indent = 4
suffix = '.in'
python = False
verbose = False
clean = False
synclines = True
force = False
def main(self, *arguments):
import getopt
options, arguments = getopt.getopt(arguments, 'C:D:cfhi:mno:ps:v')
for option, value in options:
if option == '-C':
exec(compile(open(value).read(), value, 'exec'), self.context)
elif option == '-D':
if '=' in value:
name, value = value.split('=', 1)
self.context[name] = eval(value, {})
else:
self.context[value] = True
elif option == '-c':
self.clean = True
elif option == '-f':
self.force = True
elif option == '-h':
sys.stdout.write(__doc__)
return
elif option == '-i':
self.indent = int(value)
elif option == '-m':
self.merge = True
elif option == '-n':
self.synclines = False
elif option == '-o':
self.output = value
elif option == '-p':
self.python = True
elif option == '-s':
self.suffix = value
elif option == '-v':
self.verbose = True
if not self.suffix and not self.output:
sys.exit("Option -o is needed with an empty suffix.")
if self.merge:
if len(arguments) != 2 or len(self.context) != 1:
sys.exit("Try `%s -h' for help" % sys.argv[0])
self.merge_files(arguments[0], arguments[1], sys.stdout.write)
elif not arguments:
if not self.clean:
self.transform_file(
sys.stdin.name, sys.stdin, sys.stdout.write)
else:
for argument in arguments:
if os.path.isdir(argument):
self.transform_all_files(argument)
elif argument.endswith(self.suffix):
self.transform_all_files(argument)
else:
sys.stderr.write(
"* %s does not end with %s, ignored.\n"
% (argument, self.suffix))
def merge_files(self, file1, file2, write):
left = list(open(file1))
right = list(open(file2))
block_margin = None
def protect_block(margin, danger):
if block_margin is None:
return
if margin is None:
return
if margin < block_margin:
return
if margin == block_margin and not danger:
return
write(' ' * block_margin + endif_pppp + '\n')
def check_next(lines, lo, hi):
for index in range(lo, hi):
line = lines[index].rstrip()
short = line.lstrip()
if short:
return (len(line) - len(short),
short.startswith('elif ')
or short.startswith('else:'))
return None, False
def lines_margin(lines, lo, hi):
margin = None
for index in range(lo, hi):
line = lines[index].rstrip()
short = line.lstrip()
if short:
width = len(line) - len(short)
if margin is None:
margin = width
else:
margin = min(margin, width)
return margin or 0
def copy_lines(lines, lo, hi, prefix):
for index in range(lo, hi):
line = lines[index]
if line.lstrip(' \t') == '\n':
write(line)
else:
write(prefix + line)
name = list(self.context.keys()).pop()
import difflib
matcher = difflib.SequenceMatcher(None, left, right)
for (tag, low_left, high_left,
low_right, high_right) in matcher.get_opcodes():
if tag == 'equal':
margin, danger = check_next(left, low_left, high_left)
protect_block(margin, danger)
copy_lines(left, low_left, high_left, '')
block_margin = None
elif tag == 'delete':
margin = lines_margin(left, low_left, high_left)
protect_block(margin, False)
write(' ' * margin + 'if not ' + name + ':\n')
copy_lines(left, low_left, high_left, ' ' * self.indent)
block_margin = margin
elif tag == 'insert':
margin = lines_margin(right, low_right, high_right)
protect_block(margin, False)
write(' ' * margin + 'if ' + name + ':\n')
copy_lines(right, low_right, high_right, ' ' * self.indent)
block_margin = margin
elif tag == 'replace':
margin = min(lines_margin(left, low_left, high_left),
lines_margin(right, low_right, high_right))
protect_block(margin, False)
write(' ' * margin + 'if ' + name + ':\n')
copy_lines(right, low_right, high_right, ' ' * self.indent)
write(' ' * margin + 'else:\n')
copy_lines(left, low_left, high_left, ' ' * self.indent)
block_margin = margin
else:
assert False, tag
def transform_all_files(self, input):
def ensure_directory(name):
base = os.path.dirname(name)
if base and not os.path.isdir(base):
ensure_directory(base)
if self.verbose:
sys.stderr.write("creating %s\n" % base)
os.mkdir(base)
if self.output:
output = os.path.join(self.output, input)
else:
output = input
for input, output in self.each_pair(input, output):
if not self.force:
if ((os.path.exists(output)
and os.path.getmtime(output) > os.path.getmtime(input))):
sys.exit("ERROR: %s has been modified, keeping it!\n"
% output)
if self.clean:
if os.path.exists(output):
if self.verbose:
sys.stderr.write("deleting %s\n" % output)
os.remove(output)
else:
ensure_directory(output)
if self.verbose:
sys.stderr.write("writing %s\n" % output)
self.transform_file(
input, open(input), open(output, 'w').write)
if output.endswith('.py'):
pyc_file = output[:-2] + '.pyc'
if os.path.exists(pyc_file):
if self.verbose:
sys.stderr.write("deleting %s\n" % pyc_file)
os.remove(pyc_file)
os.utime(output, (os.path.getatime(input),
os.path.getmtime(input)))
def each_pair(self, input, output):
stack = []
output = output.replace(self.suffix + '/', '/')
if output.endswith(self.suffix):
output = output[:-len(self.suffix)]
stack.append((input, output))
while stack:
input_path, output_path = stack.pop()
if os.path.isdir(input_path):
for base in os.listdir(input_path):
input = os.path.join(input_path, base)
if base.endswith(self.suffix):
output = os.path.join(output_path,
base[:-len(self.suffix)])
else:
output = os.path.join(output_path, base)
stack.append((input, output))
elif ((self.suffix + '/') in input_path
or input_path.endswith(self.suffix)):
yield input_path, output_path
def transform_file(self, name, lines, write):
# MARGIN is the number of spaces preceding the previous "if" line.
# A virtual "if True:" is assumed above and left of the whole module.
margin = -1
# REMOVE is the number of leading spaces to delete on copied lines.
remove = 0
# STATE drives the copying or skipping of code. When the test
# expression of an "if" statement is known to be True, known to be
# False, or cannot be evaluated, STATE becomes TRUE, FALSE or UNKNOWN
# respectively (an "else" clause exchanges TRUE and FALSE, but leaves
# UNKNOWN undisturbed). FALSE2 is a special case of FALSE, for when
# some "if UNKNOWN:" is followed by "elif FALSE:", an "else:" clause
# might then be needed before resuming copy. STATE is SKIP when some
# "if" or "elif" clause has been known to be True, meaning that all
# following clauses may be removed. STATE becomes SKIP as well for a
# whole embedded "if" within some code which was already being skipped.
TRUE = 'TRUE'
FALSE = 'FALSE'
FALSE2 = 'FALSE2'
UNKNOWN = 'UNKNOWN'
SKIP = 'SKIP'
state = TRUE
# STACK saves the previous (MARGIN, REMOVE, STATE) whenever a nested
# "if" is met, and restores it when the actual margin goes left enough.
# A nested "if" is ignored unless its expression can be evaluated.
stack = []
def expression_value(text):
try:
value = eval(text, {'__builtins__': {}}, self.context)
except:
return UNKNOWN
else:
if value:
return TRUE
return FALSE
def write_shifted(line):
assert remove >= 0, (remove, line)
assert line[:remove] == ' ' * remove, (remove, line)
write_verbatim(line[remove:])
def write_verbatim(line):
write(line)
self.output_counter += line.count('\n')
self.output_counter = 0
python = (self.python
or name.endswith('.py')
or name.endswith('.py' + self.suffix))
for input_counter, line in enumerate(
self.each_substituded_line(lines)):
if (input_counter == 0
and line.startswith('#!') and 'ython' in line):
python = True
if self.synclines:
while self.output_counter < input_counter:
write_verbatim('\n')
if not python:
write_verbatim(line)
continue
short = line.lstrip()
if not short:
if state in (TRUE, UNKNOWN):
write_verbatim(line)
continue
width = len(line) - len(short)
while width < margin:
margin, remove, state = stack.pop()
if width == margin:
match = re.match('else: *$', short)
if match:
if state is TRUE:
state = FALSE
elif state is FALSE:
state = TRUE
elif state is FALSE2:
write_shifted(line)
state = TRUE
elif state in UNKNOWN:
write_shifted(line)
continue
match = re.match('elif (.*): *$', short)
if match:
if state is TRUE:
state = SKIP
elif state is FALSE:
value = expression_value(match.group(1))
if value is UNKNOWN:
remove -= self.indent
write_shifted(' ' * width + 'if' + short[4:])
state = value
elif state is FALSE2:
value = expression_value(match.group(1))
if value is UNKNOWN:
write_shifted(line)
else:
write_verbatim(' ' * width + 'else:\n')
state = value
elif state is UNKNOWN:
value = expression_value(match.group(1))
if value is TRUE:
write_shifted(' ' * width + 'else:\n')
state = TRUE
elif value is FALSE:
state = FALSE2
elif value is UNKNOWN:
write_shifted(line)
continue
margin, remove, state = stack.pop()
match = re.match('if (.*): *$', short)
if match:
stack.append((margin, remove, state))
margin = width
if state in (TRUE, UNKNOWN):
value = expression_value(match.group(1))
if value is UNKNOWN:
write_shifted(line)
else:
remove += self.indent
state = value
else:
state = SKIP
continue
if state in (UNKNOWN, TRUE) and not short.startswith(endif_pppp):
write_shifted(line)
def each_substituded_line(self, lines):
if self.context:
pattern = re.compile(
'@('
+ '|'.join([re.escape(key) for key in self.context])
+ ')@')
for line in lines:
yield pattern.sub(self.substitute, line)
else:
for line in lines:
yield line
def substitute(self, match):
return str(self.context[match.group(1)])
run = Main()
main = run.main
if __name__ == '__main__':
main(*sys.argv[1:])