| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
/* |
|
2
|
|
|
|
|
|
|
* Copyright 1999-2009, Gisle Aas. |
|
3
|
|
|
|
|
|
|
* |
|
4
|
|
|
|
|
|
|
* This library is free software; you can redistribute it and/or |
|
5
|
|
|
|
|
|
|
* modify it under the same terms as Perl itself. |
|
6
|
|
|
|
|
|
|
*/ |
|
7
|
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
#ifndef EXTERN |
|
9
|
|
|
|
|
|
|
#define EXTERN extern |
|
10
|
|
|
|
|
|
|
#endif |
|
11
|
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
EXTERN SV* |
|
14
|
47246
|
|
|
|
|
|
sv_lower(pTHX_ SV* sv) |
|
15
|
|
|
|
|
|
|
{ |
|
16
|
|
|
|
|
|
|
STRLEN len; |
|
17
|
47246
|
50
|
|
|
|
|
char *s = SvPV_force(sv, len); |
|
18
|
1281714
|
100
|
|
|
|
|
for (; len--; s++) |
|
19
|
1234468
|
100
|
|
|
|
|
*s = toLOWER(*s); |
|
20
|
47246
|
|
|
|
|
|
return sv; |
|
21
|
|
|
|
|
|
|
} |
|
22
|
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
EXTERN int |
|
24
|
270
|
|
|
|
|
|
strnEQx(const char* s1, const char* s2, STRLEN n, int ignore_case) |
|
25
|
|
|
|
|
|
|
{ |
|
26
|
2149
|
100
|
|
|
|
|
while (n--) { |
|
27
|
1880
|
100
|
|
|
|
|
if (ignore_case) { |
|
28
|
1874
|
100
|
|
|
|
|
if (toLOWER(*s1) != toLOWER(*s2)) |
|
|
|
50
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
29
|
1
|
|
|
|
|
|
return 0; |
|
30
|
|
|
|
|
|
|
} |
|
31
|
|
|
|
|
|
|
else { |
|
32
|
6
|
50
|
|
|
|
|
if (*s1 != *s2) |
|
33
|
0
|
|
|
|
|
|
return 0; |
|
34
|
|
|
|
|
|
|
} |
|
35
|
1879
|
|
|
|
|
|
s1++; |
|
36
|
1879
|
|
|
|
|
|
s2++; |
|
37
|
|
|
|
|
|
|
} |
|
38
|
269
|
|
|
|
|
|
return 1; |
|
39
|
|
|
|
|
|
|
} |
|
40
|
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
static void |
|
42
|
26
|
|
|
|
|
|
grow_gap(pTHX_ SV* sv, STRLEN grow, char** t, char** s, char** e) |
|
43
|
|
|
|
|
|
|
{ |
|
44
|
|
|
|
|
|
|
/* |
|
45
|
|
|
|
|
|
|
SvPVX ---> AAAAAA...BBBBBB |
|
46
|
|
|
|
|
|
|
^ ^ ^ |
|
47
|
|
|
|
|
|
|
t s e |
|
48
|
|
|
|
|
|
|
*/ |
|
49
|
26
|
|
|
|
|
|
STRLEN t_offset = *t - SvPVX(sv); |
|
50
|
26
|
|
|
|
|
|
STRLEN s_offset = *s - SvPVX(sv); |
|
51
|
26
|
|
|
|
|
|
STRLEN e_offset = *e - SvPVX(sv); |
|
52
|
|
|
|
|
|
|
|
|
53
|
26
|
50
|
|
|
|
|
SvGROW(sv, e_offset + grow + 1); |
|
|
|
100
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
|
|
55
|
26
|
|
|
|
|
|
*t = SvPVX(sv) + t_offset; |
|
56
|
26
|
|
|
|
|
|
*s = SvPVX(sv) + s_offset; |
|
57
|
26
|
|
|
|
|
|
*e = SvPVX(sv) + e_offset; |
|
58
|
|
|
|
|
|
|
|
|
59
|
26
|
|
|
|
|
|
Move(*s, *s+grow, *e - *s, char); |
|
60
|
26
|
|
|
|
|
|
*s += grow; |
|
61
|
26
|
|
|
|
|
|
*e += grow; |
|
62
|
26
|
|
|
|
|
|
} |
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
EXTERN SV* |
|
65
|
4577
|
|
|
|
|
|
decode_entities(pTHX_ SV* sv, HV* entity2char, bool expand_prefix) |
|
66
|
|
|
|
|
|
|
{ |
|
67
|
|
|
|
|
|
|
STRLEN len; |
|
68
|
4577
|
50
|
|
|
|
|
char *s = SvPV_force(sv, len); |
|
69
|
4577
|
|
|
|
|
|
char *t = s; |
|
70
|
4577
|
|
|
|
|
|
char *end = s + len; |
|
71
|
|
|
|
|
|
|
char *ent_start; |
|
72
|
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
char *repl; |
|
74
|
|
|
|
|
|
|
STRLEN repl_len; |
|
75
|
|
|
|
|
|
|
char buf[UTF8_MAXLEN]; |
|
76
|
|
|
|
|
|
|
int repl_utf8; |
|
77
|
4577
|
|
|
|
|
|
int high_surrogate = 0; |
|
78
|
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
#if defined(__GNUC__) |
|
80
|
|
|
|
|
|
|
/* gcc -Wall reports this variable as possibly used uninitialized */ |
|
81
|
4577
|
|
|
|
|
|
repl_utf8 = 0; |
|
82
|
|
|
|
|
|
|
#endif |
|
83
|
|
|
|
|
|
|
|
|
84
|
95962
|
100
|
|
|
|
|
while (s < end) { |
|
85
|
|
|
|
|
|
|
assert(t <= s); |
|
86
|
|
|
|
|
|
|
|
|
87
|
91385
|
100
|
|
|
|
|
if ((*t++ = *s++) != '&') |
|
88
|
90560
|
|
|
|
|
|
continue; |
|
89
|
|
|
|
|
|
|
|
|
90
|
825
|
|
|
|
|
|
ent_start = s; |
|
91
|
825
|
|
|
|
|
|
repl = 0; |
|
92
|
|
|
|
|
|
|
|
|
93
|
936
|
100
|
|
|
|
|
if (s < end && *s == '#') { |
|
|
|
100
|
|
|
|
|
|
|
94
|
111
|
|
|
|
|
|
UV num = 0; |
|
95
|
111
|
|
|
|
|
|
int ok = 0; |
|
96
|
111
|
|
|
|
|
|
s++; |
|
97
|
152
|
50
|
|
|
|
|
if (s < end && (*s == 'x' || *s == 'X')) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
98
|
41
|
|
|
|
|
|
s++; |
|
99
|
174
|
100
|
|
|
|
|
while (s < end) { |
|
100
|
162
|
|
|
|
|
|
char *tmp = strchr(PL_hexdigit, *s); |
|
101
|
162
|
100
|
|
|
|
|
if (!tmp) |
|
102
|
25
|
|
|
|
|
|
break; |
|
103
|
137
|
|
|
|
|
|
num = num << 4 | ((tmp - PL_hexdigit) & 15); |
|
104
|
137
|
100
|
|
|
|
|
if (num > 0x10FFFF) { |
|
105
|
|
|
|
|
|
|
/* overflow */ |
|
106
|
4
|
|
|
|
|
|
ok = 0; |
|
107
|
4
|
|
|
|
|
|
break; |
|
108
|
|
|
|
|
|
|
} |
|
109
|
133
|
|
|
|
|
|
s++; |
|
110
|
133
|
|
|
|
|
|
ok = 1; |
|
111
|
|
|
|
|
|
|
} |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
else { |
|
114
|
282
|
100
|
|
|
|
|
while (s < end && isDIGIT(*s)) { |
|
|
|
100
|
|
|
|
|
|
|
115
|
214
|
|
|
|
|
|
num = num * 10 + (*s - '0'); |
|
116
|
214
|
100
|
|
|
|
|
if (num > 0x10FFFF) { |
|
117
|
|
|
|
|
|
|
/* overflow */ |
|
118
|
2
|
|
|
|
|
|
ok = 0; |
|
119
|
2
|
|
|
|
|
|
break; |
|
120
|
|
|
|
|
|
|
} |
|
121
|
212
|
|
|
|
|
|
s++; |
|
122
|
212
|
|
|
|
|
|
ok = 1; |
|
123
|
|
|
|
|
|
|
} |
|
124
|
|
|
|
|
|
|
} |
|
125
|
111
|
100
|
|
|
|
|
if (num && ok) { |
|
|
|
100
|
|
|
|
|
|
|
126
|
96
|
100
|
|
|
|
|
if (!SvUTF8(sv) && num <= 255) { |
|
|
|
100
|
|
|
|
|
|
|
127
|
41
|
|
|
|
|
|
buf[0] = (char) num; |
|
128
|
41
|
|
|
|
|
|
repl = buf; |
|
129
|
41
|
|
|
|
|
|
repl_len = 1; |
|
130
|
41
|
|
|
|
|
|
repl_utf8 = 0; |
|
131
|
|
|
|
|
|
|
} |
|
132
|
55
|
50
|
|
|
|
|
else if (num == 0xFFFE || num == 0xFFFF) { |
|
|
|
100
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
/* illegal */ |
|
134
|
|
|
|
|
|
|
} |
|
135
|
|
|
|
|
|
|
else { |
|
136
|
|
|
|
|
|
|
char *tmp; |
|
137
|
54
|
100
|
|
|
|
|
if ((num & 0xFFFFFC00) == 0xDC00) { /* low-surrogate */ |
|
138
|
1
|
50
|
|
|
|
|
if (high_surrogate != 0) { |
|
139
|
1
|
|
|
|
|
|
t -= 3; /* Back up past 0xFFFD */ |
|
140
|
1
|
|
|
|
|
|
num = ((high_surrogate - 0xD800) << 10) + |
|
141
|
|
|
|
|
|
|
(num - 0xDC00) + 0x10000; |
|
142
|
1
|
|
|
|
|
|
high_surrogate = 0; |
|
143
|
|
|
|
|
|
|
} else { |
|
144
|
1
|
|
|
|
|
|
num = 0xFFFD; |
|
145
|
|
|
|
|
|
|
} |
|
146
|
|
|
|
|
|
|
} |
|
147
|
53
|
100
|
|
|
|
|
else if ((num & 0xFFFFFC00) == 0xD800) { /* high-surrogate */ |
|
148
|
2
|
|
|
|
|
|
high_surrogate = num; |
|
149
|
2
|
|
|
|
|
|
num = 0xFFFD; |
|
150
|
|
|
|
|
|
|
} |
|
151
|
|
|
|
|
|
|
else { |
|
152
|
51
|
|
|
|
|
|
high_surrogate = 0; |
|
153
|
|
|
|
|
|
|
/* otherwise invalid? */ |
|
154
|
51
|
100
|
|
|
|
|
if ((num >= 0xFDD0 && num <= 0xFDEF) || |
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
155
|
46
|
50
|
|
|
|
|
((num & 0xFFFE) == 0xFFFE) || |
|
156
|
|
|
|
|
|
|
num > 0x10FFFF) |
|
157
|
|
|
|
|
|
|
{ |
|
158
|
5
|
|
|
|
|
|
num = 0xFFFD; |
|
159
|
|
|
|
|
|
|
} |
|
160
|
|
|
|
|
|
|
} |
|
161
|
|
|
|
|
|
|
|
|
162
|
54
|
|
|
|
|
|
tmp = (char*)uvuni_to_utf8((U8*)buf, num); |
|
163
|
54
|
|
|
|
|
|
repl = buf; |
|
164
|
54
|
|
|
|
|
|
repl_len = tmp - buf; |
|
165
|
54
|
|
|
|
|
|
repl_utf8 = 1; |
|
166
|
|
|
|
|
|
|
} |
|
167
|
|
|
|
|
|
|
} |
|
168
|
|
|
|
|
|
|
} |
|
169
|
|
|
|
|
|
|
else { |
|
170
|
714
|
|
|
|
|
|
char *ent_name = s; |
|
171
|
2325
|
100
|
|
|
|
|
while (s < end && isALNUM(*s)) |
|
|
|
100
|
|
|
|
|
|
|
172
|
1611
|
|
|
|
|
|
s++; |
|
173
|
714
|
100
|
|
|
|
|
if (ent_name != s && entity2char) { |
|
|
|
100
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
SV** svp; |
|
175
|
417
|
100
|
|
|
|
|
if ( (svp = hv_fetch(entity2char, ent_name, s - ent_name, 0)) || |
|
|
|
100
|
|
|
|
|
|
|
176
|
16
|
100
|
|
|
|
|
(*s == ';' && (svp = hv_fetch(entity2char, ent_name, s - ent_name + 1, 0))) |
|
177
|
|
|
|
|
|
|
) |
|
178
|
|
|
|
|
|
|
{ |
|
179
|
293
|
50
|
|
|
|
|
repl = SvPV(*svp, repl_len); |
|
180
|
293
|
|
|
|
|
|
repl_utf8 = SvUTF8(*svp); |
|
181
|
|
|
|
|
|
|
} |
|
182
|
124
|
100
|
|
|
|
|
else if (expand_prefix) { |
|
183
|
4
|
|
|
|
|
|
char *ss = s - 1; |
|
184
|
16
|
100
|
|
|
|
|
while (ss > ent_name) { |
|
185
|
14
|
|
|
|
|
|
svp = hv_fetch(entity2char, ent_name, ss - ent_name, 0); |
|
186
|
14
|
100
|
|
|
|
|
if (svp) { |
|
187
|
2
|
50
|
|
|
|
|
repl = SvPV(*svp, repl_len); |
|
188
|
2
|
|
|
|
|
|
repl_utf8 = SvUTF8(*svp); |
|
189
|
2
|
|
|
|
|
|
s = ss; |
|
190
|
2
|
|
|
|
|
|
break; |
|
191
|
|
|
|
|
|
|
} |
|
192
|
12
|
|
|
|
|
|
ss--; |
|
193
|
|
|
|
|
|
|
} |
|
194
|
|
|
|
|
|
|
} |
|
195
|
|
|
|
|
|
|
} |
|
196
|
714
|
|
|
|
|
|
high_surrogate = 0; |
|
197
|
|
|
|
|
|
|
} |
|
198
|
|
|
|
|
|
|
|
|
199
|
825
|
100
|
|
|
|
|
if (repl) { |
|
200
|
390
|
|
|
|
|
|
char *repl_allocated = 0; |
|
201
|
390
|
100
|
|
|
|
|
if (s < end && *s == ';') |
|
|
|
100
|
|
|
|
|
|
|
202
|
245
|
|
|
|
|
|
s++; |
|
203
|
390
|
|
|
|
|
|
t--; /* '&' already copied, undo it */ |
|
204
|
|
|
|
|
|
|
|
|
205
|
390
|
100
|
|
|
|
|
if (*s != '&') { |
|
206
|
101
|
|
|
|
|
|
high_surrogate = 0; |
|
207
|
|
|
|
|
|
|
} |
|
208
|
|
|
|
|
|
|
|
|
209
|
408
|
100
|
|
|
|
|
if (!SvUTF8(sv) && repl_utf8) { |
|
|
|
100
|
|
|
|
|
|
|
210
|
|
|
|
|
|
|
/* need to upgrade sv before we continue */ |
|
211
|
18
|
|
|
|
|
|
STRLEN before_gap_len = t - SvPVX(sv); |
|
212
|
18
|
|
|
|
|
|
char *before_gap = (char*)bytes_to_utf8((U8*)SvPVX(sv), &before_gap_len); |
|
213
|
18
|
|
|
|
|
|
STRLEN after_gap_len = end - s; |
|
214
|
18
|
|
|
|
|
|
char *after_gap = (char*)bytes_to_utf8((U8*)s, &after_gap_len); |
|
215
|
|
|
|
|
|
|
|
|
216
|
18
|
|
|
|
|
|
sv_setpvn(sv, before_gap, before_gap_len); |
|
217
|
18
|
|
|
|
|
|
sv_catpvn(sv, after_gap, after_gap_len); |
|
218
|
18
|
|
|
|
|
|
SvUTF8_on(sv); |
|
219
|
|
|
|
|
|
|
|
|
220
|
18
|
|
|
|
|
|
Safefree(before_gap); |
|
221
|
18
|
|
|
|
|
|
Safefree(after_gap); |
|
222
|
|
|
|
|
|
|
|
|
223
|
18
|
|
|
|
|
|
s = t = SvPVX(sv) + before_gap_len; |
|
224
|
18
|
|
|
|
|
|
end = SvPVX(sv) + before_gap_len + after_gap_len; |
|
225
|
|
|
|
|
|
|
} |
|
226
|
372
|
100
|
|
|
|
|
else if (SvUTF8(sv) && !repl_utf8) { |
|
|
|
100
|
|
|
|
|
|
|
227
|
212
|
|
|
|
|
|
repl = (char*)bytes_to_utf8((U8*)repl, &repl_len); |
|
228
|
212
|
|
|
|
|
|
repl_allocated = repl; |
|
229
|
|
|
|
|
|
|
} |
|
230
|
|
|
|
|
|
|
|
|
231
|
390
|
100
|
|
|
|
|
if (t + repl_len > s) { |
|
232
|
|
|
|
|
|
|
/* need to grow the string */ |
|
233
|
26
|
|
|
|
|
|
grow_gap(aTHX_ sv, repl_len - (s - t), &t, &s, &end); |
|
234
|
|
|
|
|
|
|
} |
|
235
|
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
/* copy replacement string into string */ |
|
237
|
1141
|
100
|
|
|
|
|
while (repl_len--) |
|
238
|
751
|
|
|
|
|
|
*t++ = *repl++; |
|
239
|
|
|
|
|
|
|
|
|
240
|
390
|
100
|
|
|
|
|
if (repl_allocated) |
|
241
|
390
|
|
|
|
|
|
Safefree(repl_allocated); |
|
242
|
|
|
|
|
|
|
} |
|
243
|
|
|
|
|
|
|
else { |
|
244
|
697
|
100
|
|
|
|
|
while (ent_start < s) |
|
245
|
262
|
|
|
|
|
|
*t++ = *ent_start++; |
|
246
|
|
|
|
|
|
|
} |
|
247
|
|
|
|
|
|
|
} |
|
248
|
|
|
|
|
|
|
|
|
249
|
4577
|
|
|
|
|
|
*t = '\0'; |
|
250
|
4577
|
|
|
|
|
|
SvCUR_set(sv, t - SvPVX(sv)); |
|
251
|
|
|
|
|
|
|
|
|
252
|
4577
|
|
|
|
|
|
return sv; |
|
253
|
|
|
|
|
|
|
} |
|
254
|
|
|
|
|
|
|
|
|
255
|
|
|
|
|
|
|
static bool |
|
256
|
13
|
|
|
|
|
|
has_hibit(char *s, char *e) |
|
257
|
|
|
|
|
|
|
{ |
|
258
|
29
|
100
|
|
|
|
|
while (s < e) { |
|
259
|
25
|
|
|
|
|
|
U8 ch = *s++; |
|
260
|
25
|
100
|
|
|
|
|
if (!UTF8_IS_INVARIANT(ch)) { |
|
261
|
9
|
|
|
|
|
|
return 1; |
|
262
|
|
|
|
|
|
|
} |
|
263
|
|
|
|
|
|
|
} |
|
264
|
4
|
|
|
|
|
|
return 0; |
|
265
|
|
|
|
|
|
|
} |
|
266
|
|
|
|
|
|
|
|
|
267
|
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
EXTERN bool |
|
269
|
13
|
|
|
|
|
|
probably_utf8_chunk(pTHX_ char *s, STRLEN len) |
|
270
|
|
|
|
|
|
|
{ |
|
271
|
13
|
|
|
|
|
|
char *e = s + len; |
|
272
|
|
|
|
|
|
|
STRLEN clen; |
|
273
|
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
/* ignore partial utf8 char at end of buffer */ |
|
275
|
18
|
100
|
|
|
|
|
while (s < e && UTF8_IS_CONTINUATION((U8)*(e - 1))) |
|
|
|
100
|
|
|
|
|
|
|
276
|
5
|
|
|
|
|
|
e--; |
|
277
|
13
|
100
|
|
|
|
|
if (s < e && UTF8_IS_START((U8)*(e - 1))) |
|
|
|
100
|
|
|
|
|
|
|
278
|
6
|
|
|
|
|
|
e--; |
|
279
|
13
|
|
|
|
|
|
clen = len - (e - s); |
|
280
|
13
|
100
|
|
|
|
|
if (clen && UTF8SKIP(e) == clen) { |
|
|
|
100
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
/* all promised continuation bytes are present */ |
|
282
|
2
|
|
|
|
|
|
e = s + len; |
|
283
|
|
|
|
|
|
|
} |
|
284
|
|
|
|
|
|
|
|
|
285
|
13
|
100
|
|
|
|
|
if (!has_hibit(s, e)) |
|
286
|
4
|
|
|
|
|
|
return 0; |
|
287
|
|
|
|
|
|
|
|
|
288
|
9
|
|
|
|
|
|
return is_utf8_string((U8*)s, e - s); |
|
289
|
|
|
|
|
|
|
} |