https://github.com/JasonGross/coq-tools
Raw File
Tip revision: a47ed37e05cc2f848d43cb70a627db5c518ba0c0 authored by Jason Gross on 12 July 2022, 18:21:59 UTC
Remove path sensitivity in traceback output, test 12
Tip revision: a47ed37
proof-using-helper.py
#!/usr/bin/env python
from __future__ import with_statement
import os, sys, re
from argparse_compat import argparse
from custom_arguments import add_libname_arguments, update_env_with_libnames, add_logging_arguments, process_logging_arguments, LOG_ALWAYS
from memoize import memoize
from file_util import read_from_file, write_to_file

# TODO:
# - handle fake ambiguities from [Definition foo] in a comment
# - handle theorems inside proof blocks
# - do the right thing for [Theorem foo. Theorem bar. Proof. Qed. Qed.] (we do the wrong thing right now)
# - make use of glob file?

parser = argparse.ArgumentParser(description='Implement the suggestions of [Set Suggest Proof Using.]')
parser.add_argument('source', metavar='SOURCE_FILE', type=argparse.FileType('r'), nargs='?', default=sys.stdin,
                    help='the source of set suggest proof using messages; use - for stdin.')
parser.add_argument('--hide', dest='hide_reg', nargs='*', type=str,
                    default=['.*_subproof[0-9]*$'], #, '.*_Proper$'],
                    help=('Regular expressions to not display warnings about on low verbosity.  ' +
                          '[Set Suggest Proof Using] can give suggestions about hard-to-find ' +
                          'identifiers, and we might want to surpress them.'))
parser.add_argument('--no-hide', dest='hide_reg', action='store_const', const=[])
add_libname_arguments(parser)
add_logging_arguments(parser)

def mwrite_to_file(file_name, contents, do_backup=False):
    return write_to_file(file_name, contents, do_backup=do_backup, memoize=True)

REG_PROOF_USING = re.compile(r'The proof of ([^\s]+)\s*should start with(?: one of the following commands)?:((?:\s*Proof using[^\.]+\.)+)', re.MULTILINE)
REG_SUB_PROOF_USING = re.compile(r'Proof using[^\.]+\.', re.MULTILINE)

#def all_matches(reg, source, start='', **env):
#    source_text = ''
#    for i in source:
#        if source_text[:len(start)] != start:
#            source_text = ''
#        source_text += i
#        cur_match = reg.search(source_text)
#        while cur_match:
#            ignoring = source_text[:cur_match.start()].strip()
#            if ignoring:
#                env['log']('Ignoring: ' + repr(ignoring), level=3)
#            yield cur_match.groups()
#            source_text = source_text[cur_match.end():]
#            cur_match = reg.search(source_text)

def pick_suggestion(suggestions):
    for i in ('Proof using Type.', 'Proof using.', 'Proof using .', 'Proof using Type*.'):
        if i in suggestions: return i
    return suggestions[0]

def lib_to_dir_map(libnames):
    return dict((lib, dirname) for dirname, lib in libnames)

def split_to_file_and_rest(theorem_id, **kwargs):
    module_part, rest_part = theorem_id.split('#', 1)
    module_parts = module_part.split('.')
    ret = []
    rest_parts = []
    while len(module_parts) > 0:
        rest_parts.insert(0, module_parts.pop())
        if '.'.join(module_parts) in kwargs['lib_to_dir'].keys():
            dirname = kwargs['lib_to_dir']['.'.join(module_parts)]
        elif '.'.join(module_parts) == '':
            dirname = '.'
        else:
            continue
        for split_i in range(0, len(rest_parts)):
            filename = os.path.join(dirname, *(rest_parts[:split_i] + [rest_parts[split_i] + '.v']))
            if os.path.exists(filename):
                ret.append((filename, ('.'.join(rest_parts[split_i+1:]) + '#' + rest_part).strip('#')))
    return tuple(ret)

ALL_DEFINITIONS = ('Theorem', 'Lemma', 'Fact', 'Remark', 'Corollary', 'Proposition', 'Property',
                   'Definition', 'Example', 'SubClass',
                   'Let', 'Fixpoint', 'CoFixpoint',
                   'Structure', 'Coercion', 'Instance', 'Existing Instance')
ALL_DEFINITIONS_STR = (r'[ \t]*(?:' +
                      '|'.join(ALL_DEFINITIONS) +
                      r')\s+%s(?=[\s\(:{\.]|$)')

ALL_DEFINITIONS_LESS_PROPER_STR = (r'[ \t]*(?:Global\s+|Local\s+)?(?:' +
                                   r'Add\s+(?:Parametric\s+)?Morphism' +
                                   r')\s+(?:[^\.]+|\.[A-Za-z\(\)])+\.(?:\n|$)')

ALL_DEFINITIONS_QUICK = (r'Theorem', 'Lemma', 'Fact', 'Remark', 'Corollary', 'Proposition', 'Property',
                         'Definition', 'Example', 'SubClass',
                         'Let', 'Fixpoint', 'CoFixpoint'
                         'Structure', 'Coercion', 'Instance',
                         'Add Parametric Morphism')
ALL_DEFINITIONS_STR_QUICK = (r'(?:' +
                            '|'.join(ALL_DEFINITIONS_QUICK) +
                            r')\s+%s')


ALL_DEFINITIONS_FULL_STRS = (r'^([ \t]*)(' + ALL_DEFINITIONS_STR_QUICK + r'[^\.]+\.\n)',
                             r'^([ \t]*)(' + ALL_DEFINITIONS_STR_QUICK + r'(?:[^\.]+|\.[A-Za-z\(\)])+\.\n)')

ALL_DEFINITIONS_FULL_STRS_LESS_PROPER = (r'^([ \t]*)((?:Global\s+|Local\s+)?Add\s+(?:Parametric\s+)?Morphism\s+(?:[^\.]+|\.[A-Za-z\(\)])+?\s+as\s+%s\s*\.\n)',)

ALL_ENDINGS = (r'(?:Qed|Defined|Save|Admitted|Abort)\s*\.')

FOUND_BUT_UNCHANGED = object()

def get_indent(term):
    return re.findall('^\s*', term)[0]

COMMENT='comment'
STRING='string'
def get_end_of_first_line_location(term, **env):
    stack = []
    i = 0
    while i < len(term):
        env['log']('DEBUG: stack: %s, pre: %s' % (str(stack), repr(term[:i])), level=3)
        if len(stack) == 0:
            if term[i] == '.' and (len(term) == i + 1 or (term[i+1] in ' \t\r\n')):
                return i+1
            if term[i] == '.':
                while i < len(term) and term[i] == '.':
                    i += 1
            else:
                i += 1
        elif stack[-1] == STRING:
            if term[i] == '"':
                stack.pop()
            i += 1
        elif stack[-1] == COMMENT:
            if term[i:i+2] == '(*':
                stack.extend(COMMENT)
                i += 2
            elif term[i:i+2] == '*)':
                stack.pop()
                i += 2
            elif term[i] == '"':
                stack.extend(STRING)
                i += 1
            else:
                i += 1
        else: # len(stack) == 0
            assert(False)
    return len(term) - 1

def update_proof(name, before_match, match, after_match, filename, rest_id, suggestion, **env):
    ending = re.search(ALL_ENDINGS, after_match, re.MULTILINE)
    if ending:
        proof_part = after_match[:ending.start()]
        env['log']('Inspecting proof: %s' % proof_part, level=2)
        if proof_part.count('Proof') == 1:
            proof_match = re.search('Proof(?: using[^\.]*)?\.', proof_part)
            if proof_match:
                if proof_match.group() == suggestion:
                    return FOUND_BUT_UNCHANGED # already correct
                elif proof_match.group() == 'Proof.':
                    return (before_match + match.group() + after_match[:proof_match.start()] +
                            suggestion +
                            after_match[proof_match.end():])
                else:
                    env['log']('Warning: Mismatch between existing Proof using and suggested Proof using:', level=0)
                    env['log']('In %s, id %s, found %s, expected %s' % (filename, rest_id, proof_match.group(), suggestion), level=0)
                    return FOUND_BUT_UNCHANGED
            else:
                env['log']('Warning: Mismatched Proofs found in %s for %s' % (filename, rest_id), level=0)
                return FOUND_BUT_UNCHANGED
        elif proof_part.count('Proof') == 0:
            loc = get_end_of_first_line_location(proof_part, **env)
            indent = get_indent(match.group())
            env['log']('Before proof: %s\nAfter proof: %s' % (proof_part[:loc], proof_part[loc:]), level=3)
            return '%s%s%s\n%s%s\n%s%s' % (before_match, match.group(), proof_part[:loc], indent, suggestion, proof_part[loc:], after_match[ending.start():])
        else:
            env['log']('Warning: Too many Proofs found in %s for %s' % (filename, rest_id), level=0)
    else:
        env['log']('Warning: No %s found in %s for %s' % (ALL_ENDINGS, filename, rest_id), level=0)
    return None

def unsafe_update_definitions(name, contents, filename, rest_id, suggestion, **env):
    match = re.search(ALL_DEFINITIONS_STR % name, contents, re.MULTILINE)
    if match:
        return update_proof(name, contents[:match.start()], match, contents[match.end():], filename, rest_id, suggestion, **env)
    elif name[-len('_Proper'):] == '_Proper':
        for match in re.finditer(ALL_DEFINITIONS_LESS_PROPER_STR, contents, re.MULTILINE | re.DOTALL):
            if match.group().strip('\n\t .').split(' ')[-1] + '_Proper' == name:
                return update_proof(name, contents[:match.start()], match, contents[match.end():], filename, rest_id, suggestion, **env)
    env['log']('Warning: No %s found in %s' % (rest_id, filename),
               level=(2 if re.search('|'.join(env['hide_reg']), rest_id) else 0))
    return None

def update_definitions(contents, filename, rest_id, suggestion, **env):
    name = rest_id.split('#')[-1]
    if len(re.findall(ALL_DEFINITIONS_STR % name, contents, re.MULTILINE)) <= 1:
        ret = unsafe_update_definitions(name, contents, filename, rest_id, suggestion, **env)
        if ret is None:
            return None
        elif ret is FOUND_BUT_UNCHANGED:
            return contents
        else:
            return ret
    else:
        modules = rest_id.split('#')[0].split('.')
        pre, mod_body, post = '', contents, ''
        while len(modules) > 0:
            mod_name = modules.pop(0)
            cur_mod = 'Module ' + mod_name
            cur_end = 'End ' + mod_name + '.'
            if mod_body.count(cur_mod) == 1:
                pre += mod_body[:mod_body.index(cur_mod) + len(cur_mod)]
                mod_body = mod_body[mod_body.index(cur_mod) + len(cur_mod):]
                if mod_body.count(cur_end) == 1:
                    post = mod_body[mod_body.index(cur_end):] + post
                    mod_body = mod_body[:mod_body.index(cur_end)]
                    if len(re.findall(ALL_DEFINITIONS_STR % name, mod_body, re.MULTILINE)) <= 1:
                        ret = unsafe_update_definitions(name, mod_body, filename, rest_id, suggestion, **env)
                        if ret is None:
                            return None
                        elif ret is FOUND_BUT_UNCHANGED:
                            return contents
                        else:
                            return pre + ret + post
                else:
                    env['log']('Warning: Too many %s found for %s in %s' % (cur_end, rest_id, filename), level=0)
                    break
            else:
                env['log']('Warning: Too many %s found for %s in %s' % (cur_mod, rest_id, filename), level=0)
                break
        env['log']('Warning: Module disambiguation was insufficient to uniqueize %s in %s' % (rest_id, filename), level=0)
        env['log']('Found: %s' % repr(re.findall(ALL_DEFINITIONS_STR % name, contents, re.MULTILINE)), level=2)
    return contents

if __name__ == '__main__':
    args = process_logging_arguments(parser.parse_args())
    source = args.source
    env = {
        'lib_to_dir': lib_to_dir_map(args.libnames + args.non_recursive_libnames),
        'log': args.log,
        'hide_reg': args.hide_reg
        }
    update_env_with_libnames(env, args)
    for theorem_id, suggestions in REG_PROOF_USING.findall('\n'.join(source)):
        all_suggestions = REG_SUB_PROOF_USING.findall(suggestions)
        suggestion = pick_suggestion(all_suggestions)
        filenames = list(reversed(split_to_file_and_rest(theorem_id, **env)))
        if filenames:
            is_first = True
            for filename, rest_id in filenames:
                orig = read_from_file(filename)
                updated = update_definitions(orig, filename, rest_id, suggestion, **env)
                if updated is not None:
                    if updated != orig:
                        env['log']('Updating %s in %s' % (rest_id, filename), level=1)
                        mwrite_to_file(filename, updated, do_backup=True)
                    elif len(filenames) > 1 and not is_first:
                        env['log']('Found %s in %s' % (rest_id, filename), level=1)
                    break
                is_first = False
        else:
            env['log']('Warning: Could not find theorem %s' % theorem_id, level=LOG_ALWAYS)
back to top