fix(glob): avoid subcapture nesting too deep error (#29520)

Use Cmt to evaluate Cond and Elem during match to avoid building the
nested capture structure later.
This commit is contained in:
Zoltán Nyikos 2024-07-06 11:40:08 +02:00 committed by GitHub
parent 0abaccb2a7
commit b109b1abce
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 28 additions and 6 deletions

View File

@ -1,6 +1,6 @@
local lpeg = vim.lpeg
local P, S, V, R, B = lpeg.P, lpeg.S, lpeg.V, lpeg.R, lpeg.B
local C, Cc, Ct, Cf = lpeg.C, lpeg.Cc, lpeg.Ct, lpeg.Cf
local C, Cc, Ct, Cf, Cmt = lpeg.C, lpeg.Cc, lpeg.Ct, lpeg.Cf, lpeg.Cmt
local M = {}
@ -47,13 +47,22 @@ function M.to_lpeg(pattern)
return (-after * P(1)) ^ 0 * after
end
-- luacheck: push ignore s
local function cut(s, idx, match)
return idx, match
end
-- luacheck: pop
local p = P({
'Pattern',
Pattern = V('Elem') ^ -1 * V('End'),
Elem = Cf(
(V('DStar') + V('Star') + V('Ques') + V('Class') + V('CondList') + V('Literal'))
* (V('Elem') + V('End')),
mul
Elem = Cmt(
Cf(
(V('DStar') + V('Star') + V('Ques') + V('Class') + V('CondList') + V('Literal'))
* (V('Elem') + V('End')),
mul
),
cut
),
DStar = (B(pathsep) + -B(P(1)))
* P('**')
@ -72,7 +81,7 @@ function M.to_lpeg(pattern)
-- pattern" which in all other cases is the entire succeeding part of the pattern, but at the end of a {}
-- condition means "everything after the {}" where several other options separated by ',' may
-- exist in between that should not be matched by '*'.
Cond = Cf((V('Ques') + V('Class') + V('Literal') - S(',}')) ^ 1, mul) + Cc(P(0)),
Cond = Cmt(Cf((V('Ques') + V('Class') + V('Literal') - S(',}')) ^ 1, mul), cut) + Cc(P(0)),
Literal = P(1) / P,
End = P(-1) * Cc(P(-1)),
})

View File

@ -205,6 +205,19 @@ describe('glob', function()
eq(true, match('[!a-zA-Z0-9]', '!'))
end)
it('should handle long patterns', function()
-- lpeg has a recursion limit of 200 by default, make sure the grammar does trigger it on
-- strings longer than that
local fill_200 =
'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa'
eq(200, fill_200:len())
local long_lit = fill_200 .. 'a'
eq(false, match(long_lit, 'b'))
eq(true, match(long_lit, long_lit))
local long_pat = fill_200 .. 'a/**/*.c'
eq(true, match(long_pat, fill_200 .. 'a/b/c/d.c'))
end)
it('should match complex patterns', function()
eq(false, match('**/*.{c,h}', ''))
eq(false, match('**/*.{c,h}', 'c'))