-
Notifications
You must be signed in to change notification settings - Fork 0
/
lib.py
343 lines (298 loc) · 10 KB
/
lib.py
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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
from dataclasses import dataclass
from pathlib import Path
from typing import Iterator
import math
import re
import sys
M_VAR_MAX = 6
def solo(items, default=None):
if len(items) == 1:
return next(iter(items))
return default or items
class Bnf:
"""Automatic parse rule based on bnf rule text
Spec 4.1. Production Syntax
Represents different entries using python expressions:
- Atomic terms:
"abc" Text string is str "abc" (no backslash escaping)
'c' Text character is str "c" (no backslash escaping)
x30 Escaped character is str "0"
[xA0-xD7FF] Character range is exclusive range(0xA0, 0xD800)
l-empty(n,c) Production is tuple ("rule", "l-empty", "n", "c")
- Lookarounds produce regex
[ lookahead = 'c' ] Production is ("?=", "c")
[ lookahead ≠ 'c' ] Production is ("?!", "c")
[ lookbehind = 'c' ] Production is ("?<=", "c")
- Special productions also produce regex (different DOTALL)
<start-of-line> Start of line is ("^",)
<end-of-input> End of whole text stream is ("$",)
<empty> Empty string is redundant -- would already be ("concat",)
"a" "b" Concatenation is tuple ("concat", "a", "b")
"a" | "b" Alternation is frozenset({"a", "b"})
"a"? Option is tuple ("repeat", 0, 1, "a")
"a"* Repeat is tuple ("repeat", 0, inf, "a")
"a"+ Repeat is tuple ("repeat", 1, inf, "a")
"a"{4} Repeat is tuple ("repeat", 4, 4, "a")
dig - "0" - "1" Difference is tuple ("diff", ("rule", "dig"), "0", "1")
"""
def __init__(self, text: str):
# '# Comments' shouldn't have semantics
self.text = re.sub(r'# .*', '', text).strip()
#TODO some comments have semantics!
# Not in spec...
self.text = re.sub(r'/\*.*?\*/', '', self.text, flags=re.DOTALL)
self.i = 0
self.expr = self.parse()
if self.i < len(self.text):
raise ValueError('remaining', self.text[self.i:self.i + 10], 'got',
self.expr)
def parse(self):
return self.parseOr()
def parseOr(self):
items = set()
while True:
items.add(self.parseConcat())
if not self.try_take(r'\|'):
return solo(frozenset(items))
def parseConcat(self):
items = []
while True:
item = self.parseDiff()
if item:
items.append(item)
else:
return solo(items, ('concat', *items))
def parseDiff(self):
subtrahend = self.parseRepeat()
minuends = []
while self.try_take('-'):
minuends.append(self.parseRepeat())
if minuends:
return ('diff', subtrahend, *minuends)
return subtrahend
def parseRepeat(self):
e = self.parseSingle()
if c := self.try_take('[+?*{]'):
lo, hi = 0, math.inf
if c == '+':
lo = 1
elif c == '?':
hi = 1
elif c == '{':
lo = hi = int(self.take(r'\d+'))
self.take('}')
return ('repeat', lo, hi, e)
return e
# Rule names can contain '+' so if followed by a letter it's part of the name not a regex repeat.
ident_reg = r'^((?:[\w-]|\+\w)+)(\([\w(),<≤/\+-]+\))?'
def parseSingle(self):
if self.try_take('"'):
return self.parseString()
if self.try_take("'"):
c = self.take() # \ isn't used as an escape
self.take("'")
return c
elif self.try_take('x'):
return chr(int(self.take(r'[0-9A-F]{1,6}'), 16))
elif self.try_take(r'\[x'):
begin = int(self.take(r'[0-9A-F]{1,6}'), 16)
self.take('-x')
end = int(self.take(r'[0-9A-F]{1,6}'), 16) + 1
self.take(r'\]')
return range(begin, end)
elif name := self.try_take(Bnf.ident_reg):
match = re.match(Bnf.ident_reg, name)
name, args = match.groups()
if '(' in name:
raise ValueError(name)
args = args.strip('()').split(',') if args else ()
return "rule", name, *args
elif self.try_take(r'\[ look'):
return self.parseLookaround()
elif self.try_take('<'):
return self.parseSpecial()
elif self.try_take(r'\('):
parens = self.parse()
self.take(r'\)')
return parens
else:
return None
def parseLookaround(self):
if self.try_take('ahead'):
pos = bool(self.try_take('='))
if not pos:
self.take('≠')
e = self.parseSingle()
self.take(']')
return ("?=" if pos else "?!", e)
self.take('behind =')
e = self.parseSingle()
self.take(']')
return ("?<=", e)
def parseSpecial(self):
if self.try_take('start-of-line>'): return ("^",)
if self.try_take('end-of-input>'): return ("$",)
self.take('empty>')
return ("concat",)
def parseString(self):
cs = []
while not self.try_take('"'):
cs.append(self.take())
return ''.join(cs)
def try_take(self, pattern='.') -> str:
m = re.match(pattern, self.text[self.i:])
if not m:
return None
s = m.group()
self.i += len(s)
self.try_take(r'\s+')
return s
def take(self, pattern='.'):
s = self.try_take(pattern)
if not s:
raise ValueError('expected', pattern, 'at', self.text[self.i:self.i + 10])
return s
def str_concat(head, tail):
if isinstance(head, str) and isinstance(tail, str):
return head + tail
if tail is None:
return head
try:
comb = (head, *tail)
except:
return (head, tail)
return solo(comb)
def split_defs(bnf_text):
lines = bnf_text.split('\n')
def_lines = [i for i, line in enumerate(lines) if '::=' in line] + [len(lines)]
for b, e in zip(def_lines, def_lines[1:]):
def_str = '\n'.join(lines[b:e]).strip()
name, text = (s.strip() for s in def_str.split('::='))
yield name, text
@dataclass(frozen=True)
class ParseResult:
name: str
start: int
end: int
expr: object
def find_vars(expr):
match expr:
case ('rule', _, *args):
return set(a for arg in args for a in arg.split('+') if a.isalpha() and a.islower() and len(a) == 1)
case (_, *es):
return set(v for e in es for v in find_vars(e))
case set() | frozenset():
return set(v for e in expr for v in find_vars(e))
case _:
return set()
def define_unbound(vars):
if not vars:
yield {}
return
var, *vars = vars
match var:
case 'm':
for f in define_unbound(vars):
for m in range(0, M_VAR_MAX):
yield f | {'m': m}
case 't':
for f in define_unbound(vars):
for t in 'CLIP KEEP STRIP'.split():
yield f | {'t': t}
case _:
raise ValueError(var)
def automagically_define_unbound(expr: any, frame: dict[str, str]):
vars = find_vars(expr) - set(frame)
for f in define_unbound(vars):
yield f | frame
class Lib:
def __init__(self, *, show_parse=False):
self.bnf = {}
self.load_defs()
self.show_parse = show_parse
def load_defs(self):
productions_path = (Path(__file__).parent / 'productions.bnf').resolve()
with open(productions_path, 'r', encoding="utf-8") as f:
productions = f.read()
for name, text in split_defs(productions):
_, name, *params = Bnf(name).expr
try:
rule = Bnf(text)
except Exception as e:
raise type(e)(f"{name}: {str(e)}").with_traceback(sys.exc_info()[2])
self.bnf.setdefault(name, []).append((params, rule.expr))
def parse(self, text, expr):
self.text = text
results = set()
for result, lastI in self.resolve(0, expr, {}):
if lastI == len(text):
results.add(result)
if not results:
raise ValueError('no results')
return solo(results)
enums = set('BLOCK-IN BLOCK-KEY BLOCK-OUT CLIP FLOW-IN FLOW-KEY FLOW-OUT KEEP STRIP'.split())
def new_frame(self, params, args, old_frame):
frame = {}
for param, arg in zip(params, (old_frame.get(arg, arg) for arg in args)):
if param.isdigit() or param in self.enums:
if param != arg: return None
elif param == 'n+1':
if arg == '0': return None
frame['n'] = str(int(arg)-1)
else:
if not param.isalpha():
raise NotImplementedError(param)
frame[param] = arg
return frame
def resolve(self, i: int, expr: any, frame: dict[str, str]) -> Iterator[tuple[object, int]]:
match expr:
case str(s):
if i < len(self.text) and self.text[i] == s:
yield s, i + 1
case range():
if i < len(self.text) and ord(self.text[i]) in expr:
yield self.text[i], i + 1
case set() | frozenset():
for e in expr:
yield from self.resolve(i, e, frame)
case ('concat',):
yield None, i
case ('concat', e, *exprs):
for vv, ii in self.resolve(i, e, frame):
for vvv, iii in self.resolve(ii, ('concat', *exprs), frame):
yield str_concat(vv, vvv), iii
case ('repeat', lo, hi, e):
if not lo:
yield None, i
if hi:
dec = ('repeat', max(lo - 1, 0), hi - 1, e)
for vv, ii in self.resolve(i, e, frame):
for vvv, iii in self.resolve(ii, dec, frame):
yield str_concat(vv, vvv), iii
case ('rule', name, *args):
for params, expr in self.bnf[name]:
if len(params) != len(args):
raise ValueError("arity mismatch")
new_frame = self.new_frame(params, args, frame)
if new_frame is None: continue
for bound_frame in automagically_define_unbound(expr, new_frame):
rec = self.resolve(i, expr, bound_frame)
if self.show_parse:
for e, ii in rec:
yield ParseResult(name, i, ii, e), ii
else:
yield from rec
case ('diff', e, *subtrahends):
for s in subtrahends:
for o in self.resolve(i, s, frame):
return
if not any(any(self.resolve(i, s, frame)) for s in subtrahends):
yield from self.resolve(i, e, frame)
case ('^',):
if i == 0 or self.text[i - 1] == '\n':
yield '', i
case ('$',):
if i == len(self.text):
yield '', i
case _:
raise ValueError('unknown type:', expr)