You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

445 lines
15 KiB

/*
**************************************************************************
* © 2016 and later: Unicode, Inc. and others.
* License & terms of use: http://www.unicode.org/copyright.html
**************************************************************************
**************************************************************************
* Copyright (C) 2014, International Business Machines
* Corporation and others. All Rights Reserved.
**************************************************************************
* file name: unisetperf.cpp
* encoding: UTF-8
* tab size: 8 (not used)
* indentation:4
*
* created on: 2007jan31
* created by: Markus Scherer
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "unicode/uperf.h"
#include "unicode/uniset.h"
#include "unicode/unistr.h"
#include "uoptions.h"
#include "cmemory.h" // for UPRV_LENGTHOF
// Command-line options specific to unisetperf.
// Options do not have abbreviations: Force readable command lines.
// (Using U+0001 for abbreviation characters.)
enum {
SET_PATTERN,
FAST_TYPE,
UNISETPERF_OPTIONS_COUNT
};
static UOption options[UNISETPERF_OPTIONS_COUNT]={
UOPTION_DEF("pattern", '\x01', UOPT_REQUIRES_ARG),
UOPTION_DEF("type", '\x01', UOPT_REQUIRES_ARG)
};
static const char *const unisetperf_usage =
"\t--pattern UnicodeSet pattern for instantiation.\n"
"\t Default: [:ID_Continue:]\n"
"\t--type Type of UnicodeSet: slow fast\n"
"\t Default: slow\n";
// Test object with setup data.
class UnicodeSetPerformanceTest : public UPerfTest {
public:
UnicodeSetPerformanceTest(int32_t argc, const char *argv[], UErrorCode &status)
: UPerfTest(argc, argv, options, UPRV_LENGTHOF(options), unisetperf_usage, status),
utf8(NULL), utf8Length(0), countInputCodePoints(0), spanCount(0) {
if (U_SUCCESS(status)) {
UnicodeString pattern=UnicodeString(options[SET_PATTERN].value, -1, US_INV).unescape();
set.applyPattern(pattern, status);
prefrozen=set;
if(0==strcmp(options[FAST_TYPE].value, "fast")) {
set.freeze();
}
int32_t inputLength;
UPerfTest::getBuffer(inputLength, status);
if(U_SUCCESS(status) && inputLength>0) {
countInputCodePoints = u_countChar32(buffer, bufferLen);
countSpans();
// Preflight the UTF-8 length and allocate utf8.
u_strToUTF8(NULL, 0, &utf8Length, buffer, bufferLen, &status);
if(status==U_BUFFER_OVERFLOW_ERROR) {
utf8=(char *)malloc(utf8Length);
if(utf8!=NULL) {
status=U_ZERO_ERROR;
u_strToUTF8(utf8, utf8Length, NULL, buffer, bufferLen, &status);
} else {
status=U_MEMORY_ALLOCATION_ERROR;
}
}
if(verbose) {
printf("code points:%ld len16:%ld len8:%ld spans:%ld "
"cp/span:%.3g UChar/span:%.3g B/span:%.3g B/cp:%.3g\n",
(long)countInputCodePoints, (long)bufferLen, (long)utf8Length, (long)spanCount,
(double)countInputCodePoints/spanCount, (double)bufferLen/spanCount, (double)utf8Length/spanCount,
(double)utf8Length/countInputCodePoints);
}
}
}
}
virtual UPerfFunction* runIndexedTest(int32_t index, UBool exec, const char* &name, char* par = NULL);
// Count spans of characters that are in the set,
// and spans of characters that are not in the set.
// If the very first character is in the set, then one additional
// not-span is counted.
void countSpans() {
const UChar *s=getBuffer();
int32_t length=getBufferLen();
int32_t i=0;
UBool tf=FALSE;
while(i<length) {
i=span(s, length, i, tf);
tf=(UBool)(!tf);
++spanCount;
}
}
int32_t span(const UChar *s, int32_t length, int32_t start, UBool tf) const {
UChar32 c;
int32_t prev;
while((prev=start)<length) {
U16_NEXT(s, start, length, c);
if(tf!=set.contains(c)) {
break;
}
}
return prev;
}
const UChar *getBuffer() const { return buffer; }
int32_t getBufferLen() const { return bufferLen; }
char *utf8;
int32_t utf8Length;
// Number of code points in the input text.
int32_t countInputCodePoints;
int32_t spanCount;
UnicodeSet set;
UnicodeSet prefrozen;
};
// Performance test function object.
class Command : public UPerfFunction {
protected:
Command(const UnicodeSetPerformanceTest &testcase) : testcase(testcase) {}
public:
virtual ~Command() {}
// virtual void call(UErrorCode* pErrorCode) { ... }
virtual long getOperationsPerIteration() {
// Number of code points tested:
// Input code points, plus one for the end of each span except the last span.
return testcase.countInputCodePoints+testcase.spanCount-1;
}
virtual long getEventsPerIteration() {
return testcase.spanCount;
}
const UnicodeSetPerformanceTest &testcase;
};
class Contains : public Command {
protected:
Contains(const UnicodeSetPerformanceTest &testcase) : Command(testcase) {
// Verify that the frozen set is equal to the unfrozen one.
UnicodeSet set;
UChar32 c;
for(c=0; c<=0x10ffff; ++c) {
if(testcase.set.contains(c)) {
set.add(c);
}
}
if(set!=testcase.set) {
fprintf(stderr, "error: frozen set != original!\n");
}
}
public:
static UPerfFunction* get(const UnicodeSetPerformanceTest &testcase) {
return new Contains(testcase);
}
virtual void call(UErrorCode* pErrorCode) {
const UnicodeSet &set=testcase.set;
const UChar *s=testcase.getBuffer();
int32_t length=testcase.getBufferLen();
int32_t count=0;
int32_t i=0;
UBool tf=FALSE;
while(i<length) {
i+=span(set, s+i, length-i, tf);
tf=(UBool)(!tf);
++count;
}
if(count!=testcase.spanCount) {
fprintf(stderr, "error: Contains() count=%ld != %ld=UnicodeSetPerformanceTest.spanCount\n",
(long)count, (long)testcase.spanCount);
}
}
static int32_t span(const UnicodeSet &set, const UChar *s, int32_t length, UBool tf) {
UChar32 c;
int32_t start=0, prev;
while((prev=start)<length) {
U16_NEXT(s, start, length, c);
if(tf!=set.contains(c)) {
break;
}
}
return prev;
}
};
class SpanUTF16 : public Command {
protected:
SpanUTF16(const UnicodeSetPerformanceTest &testcase) : Command(testcase) {
// Verify that the frozen set is equal to the unfrozen one.
UnicodeSet set;
UChar utf16[2];
UChar32 c, c2;
for(c=0; c<=0xffff; ++c) {
utf16[0]=(UChar)c;
if(testcase.set.span(utf16, 1, USET_SPAN_CONTAINED)>0) {
set.add(c);
}
}
for(c=0xd800; c<=0xdbff; ++c) {
utf16[0]=(UChar)c;
for(c2=0xdc00; c2<=0xdfff; ++c2) {
utf16[1]=(UChar)c2;
if(testcase.set.span(utf16, 2, USET_SPAN_CONTAINED)>0) {
set.add(U16_GET_SUPPLEMENTARY(c, c2));
}
}
}
if(set!=testcase.set) {
fprintf(stderr, "error: frozen set != original!\n");
}
}
public:
static UPerfFunction* get(const UnicodeSetPerformanceTest &testcase) {
return new SpanUTF16(testcase);
}
virtual void call(UErrorCode* pErrorCode) {
const UnicodeSet &set=testcase.set;
const UChar *s=testcase.getBuffer();
int32_t length=testcase.getBufferLen();
int32_t count=0;
int32_t i=0;
UBool tf=FALSE;
while(i<length) {
i+=set.span(s+i, length-i, (USetSpanCondition)tf);
tf=(UBool)(!tf);
++count;
}
if(count!=testcase.spanCount) {
fprintf(stderr, "error: SpanUTF16() count=%ld != %ld=UnicodeSetPerformanceTest.spanCount\n",
(long)count, (long)testcase.spanCount);
}
}
};
class SpanBackUTF16 : public Command {
protected:
SpanBackUTF16(const UnicodeSetPerformanceTest &testcase) : Command(testcase) {
// Verify that the frozen set is equal to the unfrozen one.
UnicodeSet set;
UChar utf16[2];
UChar32 c, c2;
for(c=0; c<=0xffff; ++c) {
utf16[0]=(UChar)c;
if(testcase.set.spanBack(utf16, 1, USET_SPAN_CONTAINED)==0) {
set.add(c);
}
}
for(c=0xd800; c<=0xdbff; ++c) {
utf16[0]=(UChar)c;
for(c2=0xdc00; c2<=0xdfff; ++c2) {
utf16[1]=(UChar)c2;
if(testcase.set.spanBack(utf16, 2, USET_SPAN_CONTAINED)==0) {
set.add(U16_GET_SUPPLEMENTARY(c, c2));
}
}
}
if(set!=testcase.set) {
fprintf(stderr, "error: frozen set != original!\n");
}
}
public:
static UPerfFunction* get(const UnicodeSetPerformanceTest &testcase) {
return new SpanBackUTF16(testcase);
}
virtual void call(UErrorCode* pErrorCode) {
const UnicodeSet &set=testcase.set;
const UChar *s=testcase.getBuffer();
int32_t length=testcase.getBufferLen();
int32_t count=0;
/*
* Get the same spans as with span() where we always start with a not-contained span.
* If testcase.spanCount is an odd number, then the last span() was not-contained.
* The last spanBack() must be not-contained to match the first span().
*/
UBool tf=(UBool)((testcase.spanCount&1)==0);
while(length>0 || !tf) {
length=set.spanBack(s, length, (USetSpanCondition)tf);
tf=(UBool)(!tf);
++count;
}
if(count!=testcase.spanCount) {
fprintf(stderr, "error: SpanBackUTF16() count=%ld != %ld=UnicodeSetPerformanceTest.spanCount\n",
(long)count, (long)testcase.spanCount);
}
}
};
class SpanUTF8 : public Command {
protected:
SpanUTF8(const UnicodeSetPerformanceTest &testcase) : Command(testcase) {
// Verify that the frozen set is equal to the unfrozen one.
UnicodeSet set;
char utf8[4];
UChar32 c;
int32_t length;
for(c=0; c<=0x10ffff; ++c) {
if(c==0xd800) {
c=0xe000;
}
length=0;
U8_APPEND_UNSAFE(utf8, length, c);
if(testcase.set.spanUTF8(utf8, length, USET_SPAN_CONTAINED)>0) {
set.add(c);
}
}
if(set!=testcase.set) {
fprintf(stderr, "error: frozen set != original!\n");
}
}
public:
static UPerfFunction* get(const UnicodeSetPerformanceTest &testcase) {
return new SpanUTF8(testcase);
}
virtual void call(UErrorCode* pErrorCode) {
const UnicodeSet &set=testcase.set;
const char *s=testcase.utf8;
int32_t length=testcase.utf8Length;
int32_t count=0;
int32_t i=0;
UBool tf=FALSE;
while(i<length) {
i+=set.spanUTF8(s+i, length-i, (USetSpanCondition)tf);
tf=(UBool)(!tf);
++count;
}
if(count!=testcase.spanCount) {
fprintf(stderr, "error: SpanUTF8() count=%ld != %ld=UnicodeSetPerformanceTest.spanCount\n",
(long)count, (long)testcase.spanCount);
}
}
};
class SpanBackUTF8 : public Command {
protected:
SpanBackUTF8(const UnicodeSetPerformanceTest &testcase) : Command(testcase) {
// Verify that the frozen set is equal to the unfrozen one.
UnicodeSet set;
char utf8[4];
UChar32 c;
int32_t length;
for(c=0; c<=0x10ffff; ++c) {
if(c==0xd800) {
c=0xe000;
}
length=0;
U8_APPEND_UNSAFE(utf8, length, c);
if(testcase.set.spanBackUTF8(utf8, length, USET_SPAN_CONTAINED)==0) {
set.add(c);
}
}
if(set!=testcase.set) {
fprintf(stderr, "error: frozen set != original!\n");
}
}
public:
static UPerfFunction* get(const UnicodeSetPerformanceTest &testcase) {
return new SpanBackUTF8(testcase);
}
virtual void call(UErrorCode* pErrorCode) {
const UnicodeSet &set=testcase.set;
const char *s=testcase.utf8;
int32_t length=testcase.utf8Length;
int32_t count=0;
/*
* Get the same spans as with span() where we always start with a not-contained span.
* If testcase.spanCount is an odd number, then the last span() was not-contained.
* The last spanBack() must be not-contained to match the first span().
*/
UBool tf=(UBool)((testcase.spanCount&1)==0);
while(length>0 || !tf) {
length=set.spanBackUTF8(s, length, (USetSpanCondition)tf);
tf=(UBool)(!tf);
++count;
}
if(count!=testcase.spanCount) {
fprintf(stderr, "error: SpanBackUTF8() count=%ld != %ld=UnicodeSetPerformanceTest.spanCount\n",
(long)count, (long)testcase.spanCount);
}
}
};
UPerfFunction* UnicodeSetPerformanceTest::runIndexedTest(int32_t index, UBool exec, const char* &name, char* par) {
switch (index) {
case 0: name = "Contains"; if (exec) return Contains::get(*this); break;
case 1: name = "SpanUTF16"; if (exec) return SpanUTF16::get(*this); break;
case 2: name = "SpanBackUTF16";if (exec) return SpanBackUTF16::get(*this); break;
case 3: name = "SpanUTF8"; if (exec) return SpanUTF8::get(*this); break;
case 4: name = "SpanBackUTF8"; if (exec) return SpanBackUTF8::get(*this); break;
default: name = ""; break;
}
return NULL;
}
int main(int argc, const char *argv[])
{
// Default values for command-line options.
options[SET_PATTERN].value = "[:ID_Continue:]";
options[FAST_TYPE].value = "slow";
UErrorCode status = U_ZERO_ERROR;
UnicodeSetPerformanceTest test(argc, argv, status);
if (U_FAILURE(status)){
printf("The error is %s\n", u_errorName(status));
test.usage();
return status;
}
if (test.run() == FALSE){
fprintf(stderr, "FAILED: Tests could not be run, please check the "
"arguments.\n");
return 1;
}
return 0;
}