xref: /aosp_15_r20/external/pcre/testdata/testinput14 (revision 22dc650d8ae982c6770746019a6f94af92b0f024)
1*22dc650dSSadaf Ebrahimi# These test special UTF and UCP features of DFA matching. The output is
2*22dc650dSSadaf Ebrahimi# different for the different widths.
3*22dc650dSSadaf Ebrahimi
4*22dc650dSSadaf Ebrahimi#subject dfa
5*22dc650dSSadaf Ebrahimi
6*22dc650dSSadaf Ebrahimi# ----------------------------------------------------
7*22dc650dSSadaf Ebrahimi# These are a selection of the more comprehensive tests that are run for
8*22dc650dSSadaf Ebrahimi# non-DFA matching.
9*22dc650dSSadaf Ebrahimi
10*22dc650dSSadaf Ebrahimi/X/utf
11*22dc650dSSadaf Ebrahimi    XX\x{d800}
12*22dc650dSSadaf Ebrahimi    XX\x{d800}\=offset=3
13*22dc650dSSadaf Ebrahimi    XX\x{d800}\=no_utf_check
14*22dc650dSSadaf Ebrahimi    XX\x{da00}
15*22dc650dSSadaf Ebrahimi    XX\x{da00}\=no_utf_check
16*22dc650dSSadaf Ebrahimi    XX\x{dc00}
17*22dc650dSSadaf Ebrahimi    XX\x{dc00}\=no_utf_check
18*22dc650dSSadaf Ebrahimi    XX\x{de00}
19*22dc650dSSadaf Ebrahimi    XX\x{de00}\=no_utf_check
20*22dc650dSSadaf Ebrahimi    XX\x{dfff}
21*22dc650dSSadaf Ebrahimi    XX\x{dfff}\=no_utf_check
22*22dc650dSSadaf Ebrahimi    XX\x{110000}
23*22dc650dSSadaf Ebrahimi    XX\x{d800}\x{1234}
24*22dc650dSSadaf Ebrahimi
25*22dc650dSSadaf Ebrahimi/badutf/utf
26*22dc650dSSadaf Ebrahimi    X\xdf
27*22dc650dSSadaf Ebrahimi    XX\xef
28*22dc650dSSadaf Ebrahimi    XXX\xef\x80
29*22dc650dSSadaf Ebrahimi    X\xf7
30*22dc650dSSadaf Ebrahimi    XX\xf7\x80
31*22dc650dSSadaf Ebrahimi    XXX\xf7\x80\x80
32*22dc650dSSadaf Ebrahimi
33*22dc650dSSadaf Ebrahimi/shortutf/utf
34*22dc650dSSadaf Ebrahimi    XX\xdf\=ph
35*22dc650dSSadaf Ebrahimi    XX\xef\=ph
36*22dc650dSSadaf Ebrahimi    XX\xef\x80\=ph
37*22dc650dSSadaf Ebrahimi    \xf7\=ph
38*22dc650dSSadaf Ebrahimi    \xf7\x80\=ph
39*22dc650dSSadaf Ebrahimi
40*22dc650dSSadaf Ebrahimi# ----------------------------------------------------
41*22dc650dSSadaf Ebrahimi# UCP and casing tests - except for the first two, these will all fail in 8-bit
42*22dc650dSSadaf Ebrahimi# mode because they are testing UCP without UTF and use characters > 255.
43*22dc650dSSadaf Ebrahimi
44*22dc650dSSadaf Ebrahimi/\x{c1}/i,no_start_optimize
45*22dc650dSSadaf Ebrahimi\= Expect no match
46*22dc650dSSadaf Ebrahimi    \x{e1}
47*22dc650dSSadaf Ebrahimi
48*22dc650dSSadaf Ebrahimi/\x{c1}+\x{e1}/iB,ucp
49*22dc650dSSadaf Ebrahimi    \x{c1}\x{c1}\x{c1}
50*22dc650dSSadaf Ebrahimi    \x{e1}\x{e1}\x{e1}
51*22dc650dSSadaf Ebrahimi
52*22dc650dSSadaf Ebrahimi/\x{120}\x{c1}/i,ucp,no_start_optimize
53*22dc650dSSadaf Ebrahimi    \x{121}\x{e1}
54*22dc650dSSadaf Ebrahimi
55*22dc650dSSadaf Ebrahimi/\x{120}\x{c1}/i,ucp
56*22dc650dSSadaf Ebrahimi    \x{121}\x{e1}
57*22dc650dSSadaf Ebrahimi
58*22dc650dSSadaf Ebrahimi/[^\x{120}]/i,no_start_optimize
59*22dc650dSSadaf Ebrahimi    \x{121}
60*22dc650dSSadaf Ebrahimi
61*22dc650dSSadaf Ebrahimi/[^\x{120}]/i,ucp,no_start_optimize
62*22dc650dSSadaf Ebrahimi\= Expect no match
63*22dc650dSSadaf Ebrahimi    \x{121}
64*22dc650dSSadaf Ebrahimi
65*22dc650dSSadaf Ebrahimi/[^\x{120}]/i
66*22dc650dSSadaf Ebrahimi    \x{121}
67*22dc650dSSadaf Ebrahimi
68*22dc650dSSadaf Ebrahimi/[^\x{120}]/i,ucp
69*22dc650dSSadaf Ebrahimi\= Expect no match
70*22dc650dSSadaf Ebrahimi    \x{121}
71*22dc650dSSadaf Ebrahimi
72*22dc650dSSadaf Ebrahimi/\x{120}{2}/i,ucp
73*22dc650dSSadaf Ebrahimi    \x{121}\x{121}
74*22dc650dSSadaf Ebrahimi
75*22dc650dSSadaf Ebrahimi/[^\x{120}]{2}/i,ucp
76*22dc650dSSadaf Ebrahimi\= Expect no match
77*22dc650dSSadaf Ebrahimi    \x{121}\x{121}
78*22dc650dSSadaf Ebrahimi
79*22dc650dSSadaf Ebrahimi# ----------------------------------------------------
80*22dc650dSSadaf Ebrahimi
81*22dc650dSSadaf Ebrahimi# ----------------------------------------------------
82*22dc650dSSadaf Ebrahimi# Tests for handling 0xffffffff in caseless UCP mode. They only apply to 32-bit
83*22dc650dSSadaf Ebrahimi# mode; for the other widths they will fail.
84*22dc650dSSadaf Ebrahimi
85*22dc650dSSadaf Ebrahimi/k*\x{ffffffff}/caseless,ucp
86*22dc650dSSadaf Ebrahimi    \x{ffffffff}
87*22dc650dSSadaf Ebrahimi
88*22dc650dSSadaf Ebrahimi/k+\x{ffffffff}/caseless,ucp,no_start_optimize
89*22dc650dSSadaf Ebrahimi    K\x{ffffffff}
90*22dc650dSSadaf Ebrahimi\= Expect no match
91*22dc650dSSadaf Ebrahimi    \x{ffffffff}\x{ffffffff}
92*22dc650dSSadaf Ebrahimi
93*22dc650dSSadaf Ebrahimi/k{2}\x{ffffffff}/caseless,ucp,no_start_optimize
94*22dc650dSSadaf Ebrahimi\= Expect no match
95*22dc650dSSadaf Ebrahimi    \x{ffffffff}\x{ffffffff}\x{ffffffff}
96*22dc650dSSadaf Ebrahimi
97*22dc650dSSadaf Ebrahimi/k\x{ffffffff}/caseless,ucp,no_start_optimize
98*22dc650dSSadaf Ebrahimi    K\x{ffffffff}
99*22dc650dSSadaf Ebrahimi\= Expect no match
100*22dc650dSSadaf Ebrahimi    \x{ffffffff}\x{ffffffff}\x{ffffffff}
101*22dc650dSSadaf Ebrahimi
102*22dc650dSSadaf Ebrahimi/k{2,}?Z/caseless,ucp,no_start_optimize,no_auto_possess
103*22dc650dSSadaf Ebrahimi\= Expect no match
104*22dc650dSSadaf Ebrahimi    Kk\x{ffffffff}\x{ffffffff}\x{ffffffff}Z
105*22dc650dSSadaf Ebrahimi
106*22dc650dSSadaf Ebrahimi# ----------------------------------------------------
107*22dc650dSSadaf Ebrahimi
108*22dc650dSSadaf Ebrahimi# End of testinput14
109