vim-patch:9.0.0105: illegal memory access when pattern starts with illegal byte

Problem:    Illegal memory access when pattern starts with illegal byte.
Solution:   Do not match a character with an illegal byte.

f50940531d

Co-authored-by: Bram Moolenaar <Bram@vim.org>
This commit is contained in:
zeertzjq 2024-07-31 06:37:58 +08:00
parent 63cd2adf3d
commit 17f95fe79b
2 changed files with 18 additions and 1 deletions

View File

@ -1804,7 +1804,9 @@ static inline char *cstrchr(const char *const s, const int c)
if (c > 0x80) {
const int folded_c = utf_fold(c);
for (const char *p = s; *p != NUL; p += utfc_ptr2len(p)) {
if (utf_fold(utf_ptr2char(p)) == folded_c) {
const int uc = utf_ptr2char(p);
// Do not match an illegal byte. E.g. 0xff matches 0xc3 0xbf, not 0xff.
if ((uc < 0x80 || uc != (uint8_t)(*p)) && utf_fold(uc) == folded_c) {
return (char *)p;
}
}

View File

@ -1,5 +1,7 @@
" Tests for regexp in utf8 encoding
source shared.vim
func s:equivalence_test()
let str = "AÀÁÂÃÄÅĀĂĄǍǞǠǺȂȦȺḀẠẢẤẦẨẪẬẮẰẲẴẶ BƁɃḂḄḆ CÇĆĈĊČƇȻḈꞒ DĎĐƊḊḌḎḐḒ EÈÉÊËĒĔĖĘĚȄȆȨɆḔḖḘḚḜẸẺẼẾỀỂỄỆ FƑḞ GĜĞĠĢƓǤǦǴḠꞠ HĤĦȞḢḤḦḨḪⱧ IÌÍÎÏĨĪĬĮİƗǏȈȊḬḮỈỊ JĴɈ KĶƘǨḰḲḴⱩꝀ LĹĻĽĿŁȽḶḸḺḼⱠ MḾṀṂ NÑŃŅŇǸṄṆṈṊꞤ OÒÓÔÕÖØŌŎŐƟƠǑǪǬǾȌȎȪȬȮȰṌṎṐṒỌỎỐỒỔỖỘỚỜỞỠỢ PƤṔṖⱣ QɊ RŔŖŘȐȒɌṘṚṜṞⱤꞦ SŚŜŞŠȘṠṢṤṦṨⱾꞨ TŢŤŦƬƮȚȾṪṬṮṰ UÙÚÛÜŨŪŬŮŰƯǕǙǛǓǗȔȖɄṲṴṶṸṺỤỦỨỪỬỮỰ VƲṼṾ WŴẀẂẄẆẈ XẊẌ YÝŶŸƳȲɎẎỲỴỶỸ ZŹŻŽƵẐẒẔⱫ aàáâãäåāăąǎǟǡǻȃȧᶏḁẚạảấầẩẫậắằẳẵặⱥ bƀɓᵬᶀḃḅḇ cçćĉċčƈȼḉꞓꞔ dďđɗᵭᶁᶑḋḍḏḑḓ eèéêëēĕėęěȅȇȩɇᶒḕḗḙḛḝẹẻẽếềểễệ fƒᵮᶂḟ gĝğġģǥǧǵɠḡꞡ hĥħȟḣḥḧḩḫẖⱨꞕ iìíîïĩīĭįǐȉȋɨᶖḭḯỉị jĵǰɉ kķƙǩᶄḱḳḵⱪꝁ lĺļľŀłƚḷḹḻḽⱡ mᵯḿṁṃ nñńņňʼnǹᵰᶇṅṇṉṋꞥ oòóôõöøōŏőơǒǫǭǿȍȏȫȭȯȱɵṍṏṑṓọỏốồổỗộớờởỡợ pƥᵱᵽᶈṕṗ qɋʠ rŕŗřȑȓɍɽᵲᵳᶉṛṝṟꞧ sśŝşšșȿᵴᶊṡṣṥṧṩꞩ tţťŧƫƭțʈᵵṫṭṯṱẗⱦ uùúûüũūŭůűųǚǖưǔǘǜȕȗʉᵾᶙṳṵṷṹṻụủứừửữự vʋṽṿ wŵẁẃẅẇẉẘ xẋẍ yýÿŷƴȳɏẏẙỳỵỷỹ zźżžƶᵶᶎẑẓẕⱬ"
let groups = split(str)
@ -559,6 +561,19 @@ func Test_match_invalid_byte()
call delete('Xinvalid')
endfunc
func Test_match_illegal_byte()
let lines =<< trim END
silent! buffer ÿ\c
next ÿ
0scriptnames
source
END
call writefile(lines, 'Xregexp')
call system(GetVimCommand() .. ' -X -Z -e -s -S Xregexp -c qa!')
call delete('Xregexp')
endfunc
func Test_match_too_complicated()
set regexpengine=1
exe "noswapfile vsplit \xeb\xdb\x99"