chk_syntax.py

#
# usage: python chk_syntax.py {file path} ...
#

import sys
import re
from fabric.colors import red, yellow

def get_syntax(fns):
    sl, env_syntax = [], False
    begin = re.compile(r'%\s*\\begin\{syntax\}')
    end = re.compile(r'%\s*\\end\{syntax\}')
    ignore = re.compile(r'(%|~|\n|\\\\)')
    spaces = re.compile(r'\s+')
    
    for fn in fns:
        for l in open(fn):
            if not env_syntax:
                if begin.match(l):
                    env_syntax = True
                    s = ''
            else:
                if end.match(l):
                    env_syntax = False
                    sl.append(spaces.sub(' ', s).strip())
                else:
                    s += ignore.sub('', l) + ' '

    return sl

def split_to_function(sl):
    fl, tl = [], []
    cs = re.compile(r'(\\cs\{|"\\)(?P<cs>.+)(\}|")')

    for l in sl:
        s = ''
        for t in l.split(' '):
            m = cs.match(t)
            if m:
                if len(tl) > 0:
                    fl.append(tl)
                tl = ['\\' + m.group('cs')]
            else:
                if s == '':
                    s = t
                else:
                    s += '~' + t
                if s.count('{') == s.count('}') and s.count('<') == s.count('>'):
                    tl.append(s)
                    s = ''

    return fl

def chk_syntax(fl):
    arg_spec = re.compile(r':(.*)')
    arg = re.compile(r'(\\Arg\{.*\}|\\\{.*\\\})|\|\{.*\}\|')
    meta = re.compile(r'(\\meta\{.*\}|<.*>|\\cs.*)')

    for f in fl:
        m = arg_spec.search(f[0])
        ok = True

        # no arg-spec
        if not m or len(m.group(1)) < 1:
            continue

        a = m.group(1)

        # arg-spec = w
        if 'w' in a:
            continue

        # mismatch number
        if len(a) != len(f[1:]):
            # ignore
            if 'if_' in f[0]:
                continue
            if '...' in ''.join(f[1:]) or '\\ldots' in ''.join(f[1:]):
                continue

            # print
            print(yellow(' '.join(f)))
            continue

        # inappropriate arg
        for i in range(len(a)):
            c = a[i]
            if c == 'N' or c == 'V' or c == 'p':
                if not meta.match(f[i + 1]):
                    ok = False
            else:
                if not arg.match(f[i + 1]):
                    ok = False

        if not ok:
            print(red(' '.join(f)))

if __name__ == '__main__':
    fns = sys.argv[1:]

    sl = get_syntax(fns)

    # print syntax lines
    #for l in sl:
    #    print(l)

    fl = split_to_function(sl)

    # print function list
    #for f in fl:
    #    print(' '.join(f))

    chk_syntax(fl)