这是我的代码，基本上是我测试过的4台计算机，它们都可以完美地处理非常大的数据大小，例如大小达500mb的文本文件，但是当我在服务器上使用真实数据甚至文件运行它们时小到6mb似乎在某处溢出并将垃圾写入我的文件末尾。
以下是整个功能的来源，因此人们可以有更深入的了解
/** Reads values from tagname between start_time and end_time which are strings in the format
of 01/01/1970-12:00, a null string is treated as 01/01/1970, values are stored in
"tagname".csv */
int ReadValues(char * tagname, char * start_time, char * end_time, char * working_directory, char * new_tag_name)
{
long lRet;
int number_of_samples;
int loop;
IHU_DATA_SAMPLE * pSamples=NULL;
IHU_TIMESTAMP StartTime;
IHU_TIMESTAMP EndTime;
FILE *stream;
char outputFileName[200];
char szQuality[100];
char newTempTagName[100];
int Year;
int Month;
int Day;
int Hour;
int Minute;
int Second;
long Subsecond;
int date[10];
//if we want to change the tagname do it now
if(new_tag_name != 0){
strncpy(newTempTagName, new_tag_name, 100);
} else {
strncpy(newTempTagName, tagname, 100);
}
// if the tagname contains a character that is invalid for a filename then we have to make a name
if ( (strstr(newTempTagName, "/")) || (strstr(newTempTagName, "\\")) || (strstr(newTempTagName, ":")))
{
sprintf(outputFileName, "MadeUpTag%d", MadeUpTagCount++);
}
else
{
//If a working directory was passed in use it
if(!working_directory){
strncpy(outputFileName, newTempTagName,199);
} else {
strcpy(outputFileName, working_directory); //Copy the working directory to the filename
// Append a / on the end of the working directory if it doesn't already have one
if(outputFileName[strlen(outputFileName)-1] == '\"'){
outputFileName[strlen(outputFileName)-1] = 0;
}
if(outputFileName[strlen(outputFileName)-1] != '\\'){
strncat(outputFileName, "\\", 199);
}
strncat(outputFileName, newTempTagName, 199); //Copy the tagname to the end of the working directory
}
}
//Add the csv file extension
strcat(outputFileName, ".csv");
#ifdef DEBUG
printf("Output filename: %s\n", outputFileName);
#endif
stream = fopen(outputFileName, "w");
if( stream == NULL ) {
printf("The file %s can not be opened\n", outputFileName );
} else {
//If start_time == 0 we want to start at 1970
if(start_time == 0){
// we want all the data so use an old start time
struct tm local;
local.tm_year = date[2] - 1900;
local.tm_mon = date[1] - 1;
local.tm_mday = date[0];
local.tm_hour = date[3];
local.tm_min = date[4];
local.tm_sec = 0;
time_t utc_seconds = mktime(&local);
StartTime.Seconds = (long)utc_seconds;
StartTime.Subseconds = 0;
} else {
//we have been supplied a start time
#ifdef DEBUG
printf("Start Time: ");
#endif
lRet = convert_date(start_time, &StartTime);
#ifdef DEBUG
printf("Seconds %ld \n", StartTime.Seconds);
#endif
}
//if end_time == 0 we want to go all the way until now.
if(end_time == 0){
// end time of 0 means now
memset(&EndTime, 0, sizeof(EndTime));
} else {
//we have been supplied an end time
#ifdef DEBUG
printf("End Time: ");
#endif
lRet = convert_date(end_time, &EndTime);
#ifdef DEBUG
printf("Seconds %ld \n", EndTime.Seconds);
#endif
}
// API will determine actual samples that are in that time range
number_of_samples=0;
// API will allocate the memory
pSamples = NULL;
//timeTaken();
lRet = ihuReadRawDataByTime
(
serverhandle, // the handle returned from the connect
tagname, // the single tagname to retrieve
&StartTime, // start time for query
&EndTime, // end time for query
&number_of_samples, // will be set by API
&pSamples // will be allocated and populated in the user API
);
char temp[100];
char Header[100];
char lengthOfHeader = 0;
char SampleAndTag[100];
int lengthOfSampleAndTag = 0;
char ActualSample[100];
int lengthOfActualSample = 0;
char NumberOfSamples[100];
int lengthOfNumberOfSamples = 0;
int oldQualityStatus = 99999;
if (lRet == ihuSTATUS_OK)
{
// successful read
lengthOfHeader = snprintf(Header, 100, "[Data]\nTagname,TimeStamp,Value,DataQuality\n");
if(lengthOfHeader < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(Header, 1, lengthOfHeader, stream) < lengthOfHeader){
printf("ERROR WRITING TO FILE!\n");
}
}
for (loop = 0;loop < number_of_samples;loop++)
{
struct tm * local;
local = localtime(&pSamples[loop].TimeStamp.Seconds);
Month = local->tm_mon + 1;
Day = local->tm_mday;
Year = local->tm_year + 1900;
Hour = local->tm_hour;
Minute = local->tm_min;
Second = local->tm_sec;
Subsecond = pSamples[loop].TimeStamp.Subseconds;
//lengthOfSampleAndTag = snprintf(SampleAndTag, 100, "Sample %d, %s",loop, newTempTagName);
lengthOfSampleAndTag = snprintf(SampleAndTag, 100, "%s", newTempTagName);
if(lengthOfSampleAndTag < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(SampleAndTag, 1,lengthOfSampleAndTag, stream) < lengthOfSampleAndTag){
printf("ERROR WRITING TO FILE!\n");
}
}
//Doing the formatting ourselves manually per character saves about 20% of cpu time
//on large databases this can save hours
temp[0] = ',';
temp[1] = ' ';
temp[2] = Month/10 + 0x30; //Tens
temp[3] = Month%10 + 0x30; //units
temp[4] = '/';
temp[5] = Day/10 + 0x30;
temp[6] = Day%10 + 0x30;
temp[7] = '/';
temp[8] = Year/1000 + 0x30;
temp[9] = '0';
temp[10] = (Year%100)/10 + 0x30;
temp[11] = (Year%100)%10 + 0x30;
temp[12] = ' ';
temp[13] = Hour/10 + 0x30; //Tens
temp[14] = Hour%10 + 0x30; //units
temp[15] = ':';
temp[16] = Minute/10 + 0x30;
temp[17] = Minute%10 + 0x30;
temp[18] = ':';
temp[19] = Second/10 + 0x30;
temp[20] = Second%10 + 0x30;
temp[21] = '.';
temp[22] = '0';
temp[23] = '0';
temp[24] = '0';
temp[25] = ',';
temp[26] = ' ';
temp[27] = 0; //Null termination
//This is to save copying the string if it is the same
if(oldQualityStatus != pSamples[loop].Quality.QualityStatus){
oldQualityStatus = pSamples[loop].Quality.QualityStatus;
switch(pSamples[loop].Quality.QualityStatus)
{
case ihuOPCBad:
strncpy(szQuality,"Bad",99);
break;
case ihuOPCUncertain:
strncpy(szQuality,"Uncertain",99);
break;
case ihuOPCNA:
strncpy(szQuality,"NotAvailable",99);
break;
case ihuOPCGood:
strncpy(szQuality,"Good",99);
break;
default:
strncpy(szQuality,"Really unknown",99);
break;
}
}
if ( pSamples[loop].ValueDataType == ihuFloat )
{
//lengthOfActualSample = snprintf(ActualSample, 100,"%s%8.7f, %s, Type:Float\n",temp,pSamples[loop].Value.Float,szQuality);
lengthOfActualSample = snprintf(ActualSample, 100,"%s%8.7f, %s\n",temp,pSamples[loop].Value.Float,szQuality);
if(lengthOfActualSample < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1, lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE!\n");
}
}
}
else if ( pSamples[loop].ValueDataType == ihuDoubleFloat )
{
//lengthOfActualSample = snprintf(ActualSample, 100,"%s%8.15f, %s, Type: DoubleFloat\n ",temp,pSamples[loop].Value.DoubleFloat,szQuality);
lengthOfActualSample = snprintf(ActualSample, 100,"%s%8.15f, %s\n ",temp,pSamples[loop].Value.DoubleFloat,szQuality);
if(lengthOfActualSample < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1,lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE!\n");
}
}
}
else if ( pSamples[loop].ValueDataType == ihuShort )
{
//lengthOfActualSample = snprintf(ActualSample, 100,"%s%d, %s, Type:Short\n",temp,pSamples[loop].Value.Short,szQuality);
lengthOfActualSample = snprintf(ActualSample, 100,"%s%d, %s\n",temp,pSamples[loop].Value.Short,szQuality);
if(lengthOfActualSample <0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1, lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE!\n");
}
}
}
else if ( pSamples[loop].ValueDataType == ihuInteger )
{
//lengthOfActualSample = snprintf(ActualSample, 100,"%s%d, %s, Type:Integer\n",temp, pSamples[loop].Value.Integer, szQuality);
lengthOfActualSample = snprintf(ActualSample, 100,"%s%d, %s\n",temp, pSamples[loop].Value.Integer, szQuality);
if(lengthOfActualSample < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1, lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE! \n");
}
}
}
else if ( pSamples[loop].ValueDataType == ihuString )
{
//lengthOfActualSample = snprintf(ActualSample, 100,"%s%s, %s, Type:String\n",temp,pSamples[loop].Value.String,szQuality);
lengthOfActualSample = snprintf(ActualSample, 100,"%s%s, %s\n",temp,pSamples[loop].Value.String,szQuality);
if(lengthOfActualSample < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1, lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE! \n");
}
}
}
}
}
lengthOfNumberOfSamples = snprintf(NumberOfSamples, 100,"Number of samples: %d\n",number_of_samples);
if(lengthOfNumberOfSamples < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(NumberOfSamples, 1, lengthOfNumberOfSamples, stream) < lengthOfNumberOfSamples){
printf("ERROR WRITING TO FILE!\n");
}
}
//timeDifference = GetTickCount() - timeDifference;
//fprintf(stream, "Time taken: %d\n", timeDifference);
// success or not you should free this
FreeSamples(pSamples, number_of_samples);
ihuFreePtr (pSamples);
pSamples = NULL;
timeTaken();
fclose(stream);
}
return 0;
}
这是该功能的缩减版本，因此您没有完整的源代码，但是如果您想要它就在这里。我已经通过手动格式化字符串（temp []）来优化代码，该字符串运行速度提高了大约40倍，但它绝对是空终止的，所以这不应该是一个问题。
stream = fopen(outputFileName, "w");
for(...){
... lots of conditions
if ( pSamples[loop].ValueDataType == ihuFloat )
{
lengthOfActualSample = snprintf(ActualSample, 100,"%s%8.7f, %s\n",temp,pSamples[loop].Value.Float,szQuality);
if(lengthOfActualSample < 0){
printf("ERROR WRITING TO BUFFER!\n");
} else {
if(fwrite(ActualSample, 1, lengthOfActualSample, stream) < lengthOfActualSample){
printf("ERROR WRITING TO FILE!\n");
}
}
}
}
fclose(stream);
现在结果：
Sample 81035, debug, 09/13/2010 11:10:55.000, 0.8900000, Good, Type:Float
Sample 81036, debug, 09/13/2010 11:11:00.000, 0.9500000, Good, Type:Float
Number of samples: 81037
**:56:15.000, 0.2800000, Good, Type:Float
Sample 80164, debug, 07/22/2010 15:56:20.000, 0.3400000, Good, Type:Float
Sample 80165, debug, 07/22/2010 15:56:30.000, 0.4100000, Good, Type:Float
... lots more samples
Sample 80322, debug, 08/01/2010 00:04:35.000, 0.5800000, Good, Type:Float
Sample 80323, debug, 08/01/2010 00:04:45.000, 0.6700000, Good, Type:Float
Sample 80324**
file ends here halfway through sample 80324
以粗体显示文件溢出的位置，并在最后一次打印调用后继续将垃圾打印到文件中（“样本数量：....）
结果应该是这样的，并且在我尝试过的每一台计算机上都是ARE（4），除了服务器是它必须处理的唯一PC。
Sample 81035, debug, 09/13/2010 11:10:55.000, 0.8900000, Good, Type:Float
Sample 81036, debug, 09/13/2010 11:11:00.000, 0.9500000, Good, Type:Float
Number of samples: 81037
我迫不及待地想出这个，并且完全不知道为什么会这样，有没有人有任何想法？
编辑：我在gcc -Wall -Wformat上有这些选项，我在整个项目中都有ZERO警告。我还编译了-pedantic和
edit2：做了一个小得多的运行，而不是用数据超限，它超过了0或NUL
run1, 02/01/2010 08:50:50.000, 0.6000000, Good
run1, 02/01/2010 08:51:00.000, 0.6900000, Good
run1, 02/01/2010 08:51:10.000, 0.7600000, Good #Sample505
Number of samples: 505
NULNULNULNULNULNULNULNULNULNULNULNULNULNULNULNULNULNULNULNUL (0 in a hex editor)
一旦运行大于1000，它就会再次开始打印垃圾
run2, 03/31/2010 15:25:30.000, 0.8200000, Good
Number of samples: 1742
/22/2010 09:39:35.000, 0.5400000, Good
run2, 01/22/2010 09:39:45.000, 0.6400000, Good
run2, 01/22/2010 09:39:50.000, 0.6900000, Good
出于某种原因，2次运行一致，样本量为76660
run5, 12/31/2009 16:03:15.000, 0.6600000, Good
run5, 12/31/2009 16:03:25.000, 0.7600000, Good
Number of samples: 76660
因此结果得出结论：＆lt; 1000将nuls打印到文件末尾，在1000和76660之间打印垃圾，然后在样本大小80234也在文件末尾打印垃圾，非常奇怪，但多余的数据总是如果多次运行则一致且相同。
答案 0 :(得分：6)
以下行使用无效的fprintf格式字符串：
fprintf(stream,"%s%8.15, %s, Type: DoubleFloat\n",temp,pSamples[loop].Value.DoubleFloat,szQuality)
具体来说，%8.15,部分。你遗漏了类型说明符。如果您使用的是gcc，则可以使用-Wformat选项让编译器检查*printf调用此类错误。
另外，我同意pmg。您发布的代码似乎与您发布的输出不匹配。请仔细检查这是否是正确的代码。
答案 1 :(得分：3)
for (loop = 0;loop< number_of_samples;loop++){
/* ... */
if(fprintf(stream,"Sample %d, %s",loop, tagname) < 0){
/* ... */
if(fprintf(stream,"%s%8.7f, %s, Type:Float\n",temp,pSamples[loop].Value.Float,szQuality) < 0){
/* ... */
if(fprintf(stream,"%s%s, %s, Type:String\n",temp,pSamples[loop].Value.String,szQuality) <0){
/* ... */
if(fprintf(stream, "Number of samples: %d\n",number_of_samples) < 0){
/* ... */
“Type：”输出在哪里？你在看正确的源文件吗？
答案 2 :(得分：0)
我解决了自己的问题！
我改变了代码：
FreeSamples(pSamples, number_of_samples);
ihuFreePtr (pSamples);
pSamples = NULL;
timeTaken();
fclose(stream);
对此：
fclose(stream);
FreeSamples(pSamples, number_of_samples);
ihuFreePtr (pSamples);
pSamples = NULL;
timeTaken();
出于某种原因，我使用的用户API在释放内存时损坏了我的文件，我完全不知道如何或为什么，但它确实存在。