summaryrefslogtreecommitdiff
path: root/prolog/util.py
blob: 48e3345a07e2b3e6c3bd620894092943e505fbb4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
#!/usr/bin/python3

from collections import namedtuple

from .lexer import lexer, operators

# Stores a token's type and value, and optionally the position of the first
# character in the lexed stream.
class Token(namedtuple('Token', ['type', 'val', 'pos'])):
    __slots__ = ()

    # Custom constructor to support default parameters.
    def __new__(cls, type, val='', pos=None):
        return super(Token, cls).__new__(cls, type, val, pos)

    def __str__(self):
        return self.val

    # Ignore position when comparing tokens. There is probably a cleaner way of
    # doing these.
    __eq__ = lambda x, y: x[0] == y[0] and x[1] == y[1]
    __ne__ = lambda x, y: x[0] != y[0] or x[1] != y[1]
    __lt__ = lambda x, y: tuple.__lt__(x[0:2], y[0:2])
    __le__ = lambda x, y: tuple.__le__(x[0:2], y[0:2])
    __ge__ = lambda x, y: tuple.__ge__(x[0:2], y[0:2])
    __gt__ = lambda x, y: tuple.__gt__(x[0:2], y[0:2])

    # Only hash token's value (we don't care about position, and types are
    # determined by values).
    def __hash__(self):
        return hash(self[1])

# Return a list of tokens in [text].
def tokenize(text):
    lexer.input(text)
    return [Token(t.type, t.value, t.lexpos) for t in lexer]

# Return a one-line string representation of [tokens].
def stringify(tokens):
    def token_str(t):
        if t.type in ('PERIOD', 'COMMA'):
            return str(t) + ' '
        if t.type in operators.values():
            return ' ' + str(t) + ' '
        return str(t)
    return ''.join(map(token_str, tokens))

# Yield the sequence of rules in [code].
def split(code):
    tokens = tokenize(code)
    start = 0
    for idx, token in enumerate(tokens):
        if token.type == 'PERIOD' and idx - start > 1:
            yield stringify(tokens[start:idx])
            start = idx + 1

# Return a list of lines in [code] and a list of rule ranges.
def decompose(code):
    lines = []
    rules = []

    rule_start = 0     # lowest line number in the current rule
    line = []          # tokens in the current line
    break_line = True  # for each comma, go to a new line
    parens = []        # stack of currently open parens/brackets/braces

    for t in tokenize(code) + [Token('EOF')]:
        # Always break the line on a semicolon, even inside parens.
        if t.type == 'SEMI':
            if line:
                lines.append(tuple(line))
                line = []
            lines.append((t,))
            continue

        # Break the line on these tokens if we are not inside parens. Don't
        # append the final token unless it is the :- operator.
        if break_line and t.type in ('PERIOD', 'FROM', 'COMMA', 'EOF'):
            # Only append :- at the end of the line, ignore commas and periods.
            if t.type == 'FROM':
                line.append(t)

            # Append nonempty lines to the output list.
            if line:
                lines.append(tuple(line))
                line = []

            # Commit a new rule if it contains some lines.
            if t.type in ('PERIOD', 'EOF') and rule_start < len(lines):
                rules.append((rule_start, len(lines)))
                rule_start = len(lines)
            continue

        # Handle parens.
        if t.type == 'LPAREN':
            # Disallow breaking lines inside "name( )" (e.g. member(X, L)) but
            # not other ( ).
            if line and line[-1].type == 'NAME':
                parens.append('paren')
                break_line = False
            else:
                parens.append('ignore')
        elif t.type in ('LBRACKET', 'LBRACE'):
            # Disallow breaking lines inside "[ ]" and "{ }".
            parens.append('paren')
            break_line = False
        elif parens:
            if t.type in ('RPAREN', 'RBRACE', 'RBRACKET'):
                parens.pop()
            break_line = 'paren' not in parens

        # Append this token to the current line.
        line.append(t)

    return lines, rules

# Format a list of [lines] according to [rules] (as returned by decompose).
def compose(lines, rules):
    code = ''
    for start, end in rules:
        for i in range(start, end):
            line = lines[i]
            if i > start:
                code += '  '
            code += stringify(line)
            if i == end-1:
                code += '.\n'
            elif i == start:
                code += '\n'
            else:
                if line and line[-1].type != 'SEMI' and lines[i+1][-1].type != 'SEMI':
                    code += ','
                code += '\n'
    return code.strip()

# Rename variables in [tokens] to A0, A1, A2,… in order of appearance.
def rename_vars(tokens, names=None):
    if names is None:
        names = {}
    next_id = len(names)

    # Return a new list.
    tokens = list(tokens)
    for i in range(len(tokens)):
        if tokens[i].type == 'PERIOD':
            names.clear()
            next_id = 0
        elif tokens[i] == Token('VARIABLE', '_'):
            tokens[i] = Token('VARIABLE', 'A{}'.format(next_id))
            next_id += 1
        elif tokens[i].type == 'VARIABLE':
            cur_name = tokens[i].val
            if cur_name not in names:
                names[cur_name] = 'A{}'.format(next_id)
                next_id += 1
            tokens[i] = Token('VARIABLE', names[cur_name])
    return tokens

# Helper function to remove trailing punctuation from lines and rename
# variables to A1,A2,A3,… (potentially using [var_names]). Return a tuple.
def normalized(line, var_names=None):
    # Remove trailing punctuation.
    i = len(line)
    while i > 0:
        if line[i-1].type not in ('COMMA', 'PERIOD', 'SEMI'):
            break
        i -= 1
    return tuple(rename_vars(line[:i], var_names))

# transformation = before → after; applied on line which is part of rule
# return mapping from formal vars in before+after to actual vars in rule
# line and rule should of course not be normalized
def map_vars(before, after, line, rule):
    mapping = {}
    new_index = 0
    for i in range(len(before)):
        if line[i].type == 'VARIABLE':
            formal_name = before[i].val
            if line[i].val != '_':
                actual_name = line[i].val
            else:
                actual_name = 'New'+str(new_index)
                new_index += 1
            mapping[formal_name] = actual_name

    remaining_formal = [t.val for t in after if t.type == 'VARIABLE' and t.val not in mapping.keys()]
    remaining_actual = [t.val for t in rule if t.type == 'VARIABLE' and t.val != '_' and t.val not in mapping.values()]

    while len(remaining_actual) < len(remaining_formal):
        remaining_actual.append('New'+str(new_index))
        new_index += 1

    for i, formal_name in enumerate(remaining_formal):
        mapping[formal_name] = remaining_actual[i]

    return mapping

# Basic sanity check.
if __name__ == '__main__':
    code = 'dup([H|T], [H1|T1]) :- dup(T1, T2). '
    lines, rules = decompose(code)
    print(compose(lines, rules))

    var_names = {}
    before = rename_vars(tokenize("dup([A0|A1], [A2|A3])"), var_names)
    after = rename_vars(tokenize("dup([A0|A1], [A5, A4|A3])"), var_names)

    line = lines[0]
    rule = tokenize(code)

    mapping = map_vars(before, after, line, rule)
    print(mapping)