Improve performance of sanitizeUTF8/readFileUTF8.

Uses std.encoding.sanitize, which returns the original string, if all code points are properly encoded. Note that the performance could still be improved considerably by iterating over multiple bytes at once, fast skipping over characters that don't have the most significant bit set.
This commit is contained in:
Sönke Ludwig 2019-10-22 11:33:57 +02:00
parent 81ba969fd6
commit 2e4bc6a316

View file

@ -25,24 +25,20 @@ import core.exception;
Takes a string with possibly invalid UTF8 sequences and outputs a valid UTF8 string as near to Takes a string with possibly invalid UTF8 sequences and outputs a valid UTF8 string as near to
the original as possible. the original as possible.
*/ */
string sanitizeUTF8(in ubyte[] str) string sanitizeUTF8(immutable(ubyte)[] str)
@safe pure { @safe pure {
import std.utf; import std.encoding : sanitize;
auto ret = appender!string(); auto ustr = cast(immutable(char)[])str;
ret.reserve(str.length); return () @trusted { return sanitize(ustr); } ();
size_t i = 0;
while (i < str.length) {
dchar ch = str[i];
try ch = std.utf.decode(cast(const(char[]))str, i);
catch( UTFException ){ i++; }
//catch( AssertError ){ i++; }
char[4] dst;
auto len = std.utf.encode(dst, ch);
ret.put(dst[0 .. len]);
} }
/// ditto
return ret.data; string sanitizeUTF8(in ubyte[] str)
@trusted pure {
import std.encoding : sanitize;
auto ustr = cast(immutable(char)[])str;
auto ret = sanitize(ustr);
if (ret.ptr is ustr.ptr) return ustr.idup;
else return ret;
} }
/** /**