nocomment.py

#
# This file is part of the micropython-esp32-ulp project,
# https://github.com/micropython/micropython-esp32-ulp
#
# SPDX-FileCopyrightText: 2018-2023, the micropython-esp32-ulp authors, see AUTHORS file.
# SPDX-License-Identifier: MIT

defremove_comments(s):
"""
 Remove comments of these styles:

 CHASH: # comment python style, up to: EOL
 CSLASHSLASH: // comment C style, up to: EOL
 CSLASHSTAR: /* comment C style (single/multi line), up to: */

 Strings can be like 'strings' or "strings".
 Any comment-starting chars within strings are not considered.
 Escaping of (string-end) chars via backslash in strings is considered.

 Also, leading and trailing whitespace is removed (after comment removal).
 Indented lines are re-indented afterwards with a single tab char.

 Line numbers stay as in input file because empty lines are kept.

 s: string with comments (can include newlines)
 returns: list of text lines
 """
# note: micropython's ure module was not capable enough to process this:
# missing methods, re modes, recursion limit exceeded, ...
# simpler hacks also didn't seem powerful enough to address all the
# corner cases of CSLASHSTAR vs. *STR, so this state machine came to life:
SRC, CHASH, CSLASHSLASH, CSLASHSTAR, DSTR, SSTR=range(6) # states

line= [] # collect chars of one line
lines= [] # collect result lines

deffinish_line():
# assemble a line from characters, try to get rid of trailing and
# most of leading whitespace (keep/put one tab for indented lines).
nonlocalline
line=''.join(line)
is_indented=line.startswith(' ') orline.startswith('\t')
line=line.strip()
iflineandis_indented:
line='\t'+line
lines.append(line)
line= []

state=SRC
i=0
length=len(s)
whilei<length:
c=s[i]
cn=s[i+1] ifi+1<lengthelse'\0'
ifstate==SRC:
ifc=='#': # starting to-EOL comment
state=CHASH
i+=1
elifc=='/':
ifcn=='/': # starting to-EOL comment
state=CSLASHSLASH
i+=2
elifcn=='*': # starting a /* comment
state=CSLASHSTAR
i+=2
else:
i+=1
line.append(c)
elifc=='"':
state=DSTR
i+=1
line.append(c)
elifc=="'":
state=SSTR
i+=1
line.append(c)
elifc=='\n':
i+=1
finish_line()
else:
i+=1
line.append(c)
elifstate==CHASHorstate==CSLASHSLASH:
ifc!='\n': # comment runs until EOL
i+=1
else:
state=SRC
i+=1
finish_line()
elifstate==CSLASHSTAR:
ifc=='*'andcn=='/': # ending a comment */
state=SRC
i+=2
elifc=='\n':
i+=1
finish_line()
else:
i+=1
elifstate==DSTRandc=='"'orstate==SSTRandc=="'": # string end
state=SRC
i+=1
line.append(c)
elifstate==DSTRorstate==SSTR:
i+=1
line.append(c)
ifc=='\\': # escaping backslash
i+=1# do not look at char after the backslash
line.append(cn)
else:
raiseException("state: %d c: %s cn: %s"% (state, c, cn))
ifline:
# no final \n triggered processing these chars yet, do it now
finish_line()
returnlines


if__name__=='__main__':
importsys
filename=sys.argv[1]
withopen(filename, "r") asf:
text=f.read()
lines=remove_comments(text)
withopen(filename+".nocomments", "w") asf:
forlineinlines:
f.write(line+'\n')