mirror of
https://github.com/VectorCamp/vectorscan.git
synced 2025-06-28 16:41:01 +03:00
Corpus generator: fix random char value of UTF-8.
fixes github issue #184
This commit is contained in:
parent
4d4940dfbe
commit
a9ca0e4de3
@ -476,14 +476,14 @@ void CorpusGeneratorUtf8::generateCorpus(vector<string> &data) {
|
|||||||
* that we've been asked for. */
|
* that we've been asked for. */
|
||||||
unichar CorpusGeneratorUtf8::getRandomChar() {
|
unichar CorpusGeneratorUtf8::getRandomChar() {
|
||||||
u32 range = MAX_UNICODE + 1
|
u32 range = MAX_UNICODE + 1
|
||||||
- (UNICODE_SURROGATE_MAX + UNICODE_SURROGATE_MIN + 1);
|
- (UNICODE_SURROGATE_MAX - UNICODE_SURROGATE_MIN + 1);
|
||||||
range = min(cProps.alphabetSize, range);
|
range = min(cProps.alphabetSize, range);
|
||||||
assert(range);
|
assert(range);
|
||||||
|
|
||||||
unichar c = 'a' + cProps.rand(0, range - 1);
|
unichar c = 'a' + cProps.rand(0, range - 1);
|
||||||
|
|
||||||
if (c >= UNICODE_SURROGATE_MIN) {
|
if (c >= UNICODE_SURROGATE_MIN) {
|
||||||
c =+ UNICODE_SURROGATE_MAX + 1;
|
c += UNICODE_SURROGATE_MAX - UNICODE_SURROGATE_MIN + 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
return c % (MAX_UNICODE + 1);
|
return c % (MAX_UNICODE + 1);
|
||||||
|
Loading…
x
Reference in New Issue
Block a user