/**** Global functions definitions. ****/
TA_RetCode TA_GetHistoryDataFromWeb( TA_DataSourceHandle *handle,
TA_CategoryHandle *categoryHandle,
TA_SymbolHandle *symbolHandle,
TA_Period period,
const TA_Timestamp *start,
const TA_Timestamp *end,
TA_Field fieldToAlloc,
TA_ParamForAddData *paramForAddData )
{
TA_PROLOG
TA_RetCode retCode;
TA_StringCache *stringCache;
TA_String *yahooName;
TA_WebPage *webPage;
TA_PrivateYahooHandle *yahooHandle;
TA_DecodingParam localDecodingParam;
TA_DecodingParam directYahooDecodingParam;
const TA_DecodingParam *decodingParam;
TA_FileHandle *fileHandle;
TA_ReadOpInfo *readOpInfo;
TA_UIRSuffixParsing suffixParsing;
TA_Timestamp firstBarTimestamp, lastBarTimestamp, prevEndDate;
TA_InfoFromAddedData infoFromAddedData;
TA_DayOfWeek dayOfWeek;
TA_Timestamp curAdjustYear, lastAdjustYear;
const char *overideServerAddr;
int nbEstimateBar;
int nbField;
unsigned int nbBarAdded, nbTotalBarAdded;
int again, firstTime, nbBatch;
int zeroBarAddedAttempt;
int doAdjustment;
TA_TRACE_BEGIN( TA_GetHistoryDataFromWeb );
/* Initialize some local variables. */
stringCache = TA_GetGlobalStringCache();
yahooHandle = (TA_PrivateYahooHandle *)handle->opaqueData;
readOpInfo = NULL;
nbEstimateBar = 0;
TA_ASSERT( categoryHandle != NULL );
TA_ASSERT( symbolHandle != NULL );
TA_ASSERT( categoryHandle->string != NULL );
TA_ASSERT( symbolHandle->string != NULL );
retCode = TA_BAD_PARAM;
/* Set the initial first/last timestamp */
if( start )
TA_TimestampCopy( &firstBarTimestamp, start );
else
TA_SetDate( 1950, 1, 1, &firstBarTimestamp );
if( end )
TA_TimestampCopy( &lastBarTimestamp, end );
else
TA_SetDateNow( &lastBarTimestamp );
/* Time component is not important for Yahoo! but all end of day
* price bar use 00:00:00, so do the same here.
*/
TA_SetTime( 0, 0, 0, &firstBarTimestamp );
TA_SetTime( 0, 0, 0, &lastBarTimestamp );
/* Make sure that lastBarTimestamp is a week-day. */
dayOfWeek = TA_GetDayOfTheWeek( &lastBarTimestamp );
if( (dayOfWeek == TA_SUNDAY) || (dayOfWeek == TA_SATURDAY) )
TA_PrevWeekday( &lastBarTimestamp );
TA_ASSERT( yahooHandle != NULL );
if( yahooHandle->param->id == TA_YAHOO_ONE_SYMBOL )
{
/* User specified Yahoo! name. */
yahooName = TA_StringDup(stringCache,yahooHandle->webSiteSymbol);
TA_ASSERT( yahooName != NULL );
}
else
{
/* Map the TA-Lib name into the Yahoo! name. */
retCode = TA_AllocStringFromLibName( categoryHandle->string,
symbolHandle->string,
&yahooName );
if( retCode != TA_SUCCESS )
{
TA_TRACE_RETURN( retCode );
}
TA_ASSERT( yahooName != NULL );
}
/* Check if the user did overide the server address in the location parameter.
* (as needed, convert from TA_String to char *)
*/
if( yahooHandle->userSpecifiedServer )
overideServerAddr = TA_StringToChar(yahooHandle->userSpecifiedServer);
else
overideServerAddr = NULL;
/* Get the decoding parameter for the CSV web page. */
if( yahooHandle->param->id == TA_YAHOO_ONE_SYMBOL )
{
retCode = TA_YahooIdxDataDecoding( yahooHandle->webSiteCountry,
TA_YAHOOIDX_CSV_PAGE,
&directYahooDecodingParam );
if( retCode != TA_SUCCESS )
{
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( retCode );
}
decodingParam = &directYahooDecodingParam;
}
else
{
decodingParam = TA_YahooIdxDecodingParam( yahooHandle->index, TA_YAHOOIDX_CSV_PAGE );
if( !decodingParam )
{
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_INTERNAL_ERROR(103) );
}
}
localDecodingParam = *decodingParam;
/* Check if split/value adjustment are necessary. */
if( (yahooHandle->param->flags & TA_DO_NOT_SPLIT_ADJUST) &&
(yahooHandle->param->flags & TA_DO_NOT_VALUE_ADJUST) )
{
doAdjustment = 0;
}
else
doAdjustment = 1;
/* Parse the uirSuffix so the start/end date can be changed. */
if( !setUIRSuffixParsing( decodingParam->uirSuffix, &suffixParsing ) )
{
/* This should never happen unless the
* Yahoo! index protocol has been broken.
*/
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_INTERNAL_ERROR(104) );
}
/* Use a local copy of the decoding param.
* This is because the uirSuffix is replaced with
* an allocated buffer (so the date field can be
* manipulated).
*/
/* Replace the uirSuffix with a large local buffer. */
localDecodingParam.uirSuffix = TA_Malloc( suffixParsing.maxTotalLength );
if( !localDecodingParam.uirSuffix )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_ALLOC_ERR );
}
/* Change the dates in the uirSuffix. */
buildUIRSuffix( &suffixParsing,
&firstBarTimestamp, &lastBarTimestamp,
(char *)localDecodingParam.uirSuffix );
/* nbBatch is a safety net to make sure that
* TA-Lib won't stay forever in the while loop
* in case Yahoo! changes their protocol.
*/
nbBatch = 0;
/* Sometime Yahoo! return an empty csv file. Make
* multiple attempts in that case.
*/
zeroBarAddedAttempt = 0;
again = 1;
firstTime = 1;
nbTotalBarAdded = 0;
while( again && (++nbBatch < 100) && (zeroBarAddedAttempt < 10) )
{
retCode = TA_DriverShouldContinue(paramForAddData);
if( retCode != TA_SUCCESS )
{
TA_StringFree( stringCache, yahooName );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( retCode );
}
if( TA_TimestampLess(&lastBarTimestamp,&firstBarTimestamp) )
{
/* Get out of this loop if all the requested data
* has been retreived already.
*/
again = 0;
break;
}
retCode = TA_WebPageAllocFromYahooName( &localDecodingParam,
TA_StringToChar(yahooName),
overideServerAddr,
&webPage, paramForAddData );
if( retCode != TA_SUCCESS )
{
TA_StringFree( stringCache, yahooName );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( retCode );
}
/* Disguise the webPage stream into a "file". That way the speed
* optimized ASCII decoder can be re-used (TA_ReadOp stuff).
*/
retCode = TA_FileSeqOpenFromStream( webPage->content, &fileHandle );
if( retCode != TA_SUCCESS )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_WebPageFree( webPage );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( retCode );
}
if( firstTime )
{
/* Make assumption of the data provided
* base on the number of fields in the CSV file.
*/
nbField = nbCommaField( webPage->content );
switch( nbField )
{
case 2:
readOpInfo = yahooHandle->readOp2Fields;
break;
case 5:
readOpInfo = yahooHandle->readOp5Fields;
break;
default:
readOpInfo = yahooHandle->readOp6Fields;
}
#if !defined( TA_SINGLE_THREAD )
/* Must use a local copy if multi-threaded */
retCode = TA_ReadOpInfoClone(&readOpInfo, readOpInfo);
if( retCode != TA_SUCCESS )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_WebPageFree( webPage );
TA_Free( (char *)localDecodingParam.uirSuffix );
TA_TRACE_RETURN( retCode );
}
#endif
/* User asking for all the fields? */
if( fieldToAlloc == TA_ALL )
{
switch( nbField )
{
case 2:
fieldToAlloc = TA_CLOSE|TA_TIMESTAMP;
break;
case 5:
fieldToAlloc = TA_OPEN|TA_HIGH|TA_LOW|TA_CLOSE|TA_TIMESTAMP;
break;
default:
fieldToAlloc = TA_OPEN|TA_HIGH|TA_LOW|TA_CLOSE|TA_VOLUME|TA_TIMESTAMP;
}
}
/* Optimize the read op for the requested data. */
retCode = TA_ReadOp_Optimize( readOpInfo,
period,
fieldToAlloc );
if( retCode != TA_SUCCESS )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_WebPageFree( webPage );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( retCode );
}
/* Make an estimation of the number of price bar. */
nbEstimateBar = TA_StreamCountChar( webPage->content, '\n' ) + 1;
if( nbEstimateBar < 100 )
nbEstimateBar = 100;
}
/* Interpret the CSV data. */
retCode = TA_ReadOp_Do( fileHandle,
readOpInfo,
period, &firstBarTimestamp, &lastBarTimestamp,
nbEstimateBar, fieldToAlloc,
paramForAddData,
&nbBarAdded );
TA_FileSeqClose( fileHandle );
TA_WebPageFree( webPage );
nbTotalBarAdded += nbBarAdded;
if( retCode != TA_SUCCESS )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( retCode );
}
/* Yahoo! does not always return all the data it could, up to
* the requested end date. It is important to detect these occurence
* and cancel the usage of all data accumulated up to now.
*/
TA_GetInfoFromAddedData( paramForAddData, &infoFromAddedData );
if( infoFromAddedData.barAddedSinceLastCall )
{
/* Do some more checking by considering holidays, week-end etc... */
if( !isGapAcceptable(&infoFromAddedData.highestTimestampAddedSinceLastCall, &lastBarTimestamp) )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( TA_DATA_GAP );
}
TA_TimestampCopy( &lastBarTimestamp, &infoFromAddedData.lowestTimestamp );
}
#if DEBUG_PRINTF
printf( "NB BAR ADDED=%d, TOTAL=%d\n", nbBarAdded, nbTotalBarAdded );
#endif
/* Verify if more data should be processed.
* Yahoo! sometimes slice their data, in
* batch of 200 price bars.
*/
if( firstTime && (nbBarAdded > 200) )
{
again = 0; /* Assume all the data extracted... exit the loop. */
}
else if( nbBarAdded == 0 )
{
/* Make multiple attempts when retreiving data succeed,
* but somehow there is zero bar returned.
*
* Sometimes this might be correct when there is truly no
* more data available, so choosing an algorithm before
* giving up is a comprimise between reliability and
* usability. The data source is free... and you get
* what you pay for after all ;)
*/
if( (nbTotalBarAdded < 1000) && (zeroBarAddedAttempt >= 1) && (zeroBarAddedAttempt < 7) )
{
/* I did choose to add a delay when insufficient total data is returned. When
* there is already ~5 years of data, most likely there is "Zero" returned
* because there is NO more data available, so just do the retry without delay.
*/
TA_Sleep(zeroBarAddedAttempt*2);
}
#if DEBUG_PRINTF
printf( "Retry %d", zeroBarAddedAttempt );
#endif
zeroBarAddedAttempt++;
}
else
{
zeroBarAddedAttempt = 0;
if( TA_TimestampEqual( &lastBarTimestamp, &prevEndDate ) )
{
/* prevEndDate is a "safety net" to
* exit the loop early in case Yahoo! starts
* to return always the same batch of data.
* Just ignore the repetitive data and exit.
*/
TA_Free( (char *)localDecodingParam.uirSuffix );
TA_StringFree( stringCache, yahooName );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
TA_TRACE_RETURN( TA_SUCCESS );
}
TA_TimestampCopy( &prevEndDate, &lastBarTimestamp );
/* Request the data up to the day BEFORE
* the last batch of data received.
*/
TA_PrevDay( &lastBarTimestamp );
/* Make sure that lastBarTimestamp is a week-day. */
dayOfWeek = TA_GetDayOfTheWeek( &lastBarTimestamp );
if( (dayOfWeek == TA_SUNDAY) || (dayOfWeek == TA_SATURDAY) )
TA_PrevWeekday( &lastBarTimestamp );
/* Change the dates in the uirSuffix. */
buildUIRSuffix( &suffixParsing,
&firstBarTimestamp, &lastBarTimestamp,
(char *)localDecodingParam.uirSuffix );
/* From that point, data is expected to be most likely
* sent in batch of 200.
*/
nbEstimateBar = 200;
}
firstTime = 0;
}
/* Get rid of some memory not used anymore. */
TA_Free( (char *)localDecodingParam.uirSuffix );
#if !defined( TA_SINGLE_THREAD )
if( readOpInfo )
TA_ReadOpInfoFree( readOpInfo );
#endif
/* If adjusted data is requested, use splits and dividend info from Yahoo!. */
if( doAdjustment && (nbTotalBarAdded >= 1) )
{
/* Get the decoding parameter for the adjustment page. */
if( yahooHandle->param->id == TA_YAHOO_ONE_SYMBOL )
{
retCode = TA_YahooIdxDataDecoding( yahooHandle->webSiteCountry,
TA_YAHOOIDX_ADJUSTMENT,
&directYahooDecodingParam );
if( retCode != TA_SUCCESS )
{
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( retCode );
}
decodingParam = &directYahooDecodingParam;
}
else
{
decodingParam = TA_YahooIdxDecodingParam( yahooHandle->index, TA_YAHOOIDX_ADJUSTMENT );
if( !decodingParam )
{
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_INTERNAL_ERROR(140) );
}
}
localDecodingParam = *decodingParam;
if( !setUIRSuffixParsing( decodingParam->uirSuffix, &suffixParsing ) )
{
/* This should never happen unless the
* Yahoo! index protocol has been broken.
*/
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_INTERNAL_ERROR(141) );
}
/* Use a local copy of the decoding param.
* This is because the uirSuffix is replaced with
* an allocated buffer (so the date field can be
* manipulated).
*/
/* Replace the uirSuffix with a large local buffer. */
localDecodingParam.uirSuffix = TA_Malloc( suffixParsing.maxTotalLength );
if( !localDecodingParam.uirSuffix )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( TA_ALLOC_ERR );
}
/* curAdjustYear indicates for which year the download is
* taking place.
*/
TA_SetDefault( &curAdjustYear );
TA_SetDateNow( &curAdjustYear );
/* Identify the oldest year for which data was downloaded */
TA_GetInfoFromAddedData( paramForAddData, &infoFromAddedData );
TA_TimestampCopy( &lastAdjustYear, &infoFromAddedData.lowestTimestamp );
TA_PrevYear( &lastAdjustYear ); /* Get one more year to be on the safe side. */
while( TA_TimestampLess( &lastAdjustYear, &curAdjustYear ) )
{
/* Set prevEndDate to two years earlier. */
TA_TimestampCopy( &prevEndDate, &curAdjustYear );
TA_PrevYear( &prevEndDate );
TA_PrevYear( &prevEndDate );
/* Change the dates in the uirSuffix. */
TA_SetDate( TA_GetYear(&curAdjustYear), 12, 31, &curAdjustYear );
TA_SetDate( TA_GetYear(&prevEndDate), 1, 1, &prevEndDate );
buildUIRSuffix( &suffixParsing,
&prevEndDate, &curAdjustYear,
(char *)localDecodingParam.uirSuffix );
retCode = doAdjustments( &localDecodingParam, yahooName,
yahooHandle, NULL, paramForAddData );
if( retCode != TA_SUCCESS )
{
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_Free( (char *)localDecodingParam.uirSuffix );
TA_TRACE_RETURN( retCode );
}
/* Move 3 years earlier. */
TA_PrevYear( &prevEndDate );
TA_TimestampCopy( &curAdjustYear, &prevEndDate );
}
/* Clean-up what was allocated for the adjustment logic. */
TA_Free( (char *)localDecodingParam.uirSuffix );
}
/* Clean-up and exit */
TA_StringFree( stringCache, yahooName );
TA_TRACE_RETURN( retCode );
}
TA_RetCode TA_YAHOO_GetHistoryData( TA_DataSourceHandle *handle,
TA_CategoryHandle *categoryHandle,
TA_SymbolHandle *symbolHandle,
TA_Period period,
const TA_Timestamp *start,
const TA_Timestamp *end,
TA_Field fieldToAlloc,
TA_ParamForAddData *paramForAddData )
{
TA_PROLOG
TA_RetCode tempRetCode, retCode;
TA_PrivateYahooHandle *yahooHandle;
int again, j;
TA_TRACE_BEGIN( TA_YAHOO_GetHistoryData );
TA_ASSERT( handle != NULL );
TA_ASSERT( paramForAddData != NULL );
TA_ASSERT( categoryHandle != NULL );
TA_ASSERT( symbolHandle != NULL );
yahooHandle = (TA_PrivateYahooHandle *)handle->opaqueData;
TA_ASSERT( yahooHandle != NULL );
/* If the requested period is too precise for the
* period that can be provided by this data source,
* simply return without error.
* Since no data has been added, the TA-LIB will ignore
* this data source.
*/
if( period < TA_DAILY )
{
TA_TRACE_RETURN( TA_SUCCESS );
}
/* Get the data from the WEB.
*
* Yahoo! sometimes have "gaps" in its data (like one
* week missing), when this is being detected, we throw
* away all the data up to now and start over (up to
* 5 times before giving up).
*/
again = 5;
do
{
retCode = TA_GetHistoryDataFromWeb( handle,
categoryHandle, symbolHandle,
TA_DAILY, start, end,
fieldToAlloc, paramForAddData );
if( retCode == TA_DATA_GAP )
{
retCode = TA_HistoryAddDataReset( paramForAddData );
if( retCode != TA_SUCCESS )
again = 0; /* Give up */
else
{
--again; /* Try again */
/* Sometimes giving Yahoo! a break helps. */
tempRetCode = TA_DriverShouldContinue(paramForAddData);
j = 0;
while( (j++ < 5) && (tempRetCode != TA_DATA_RETREIVE_TIMEOUT) )
{
TA_Sleep(1);
tempRetCode = TA_DriverShouldContinue(paramForAddData);
}
if( tempRetCode == TA_DATA_RETREIVE_TIMEOUT )
{
retCode = tempRetCode;
again = 0;
}
}
}
else
{
again = 0; /* Exit the loop */
}
} while( again > 0 );
TA_TRACE_RETURN( retCode );
}
/**** Global functions definitions. ****/
TA_RetCode TA_WebPageAlloc( const char *webSiteAddr,
const char *webSitePage,
const char *proxyName,
const char *proxyPort,
TA_WebPage **webPageAllocated,
unsigned int nbAttempt,
void *paramForAddData )
{
TA_RetCode retCode;
unsigned int i,j, k;
TA_ParamForAddData *paramForAddDataPtr;
paramForAddDataPtr = (TA_ParamForAddData *)paramForAddData;
/* Make sure there is at least one attempt. */
if( nbAttempt == 0 )
nbAttempt = 1;
/* Do not attempt more than 50 times (realistic upper limit). */
if( nbAttempt > 50 )
nbAttempt = 50;
retCode = TA_SUCCESS;
for( i=0; i < nbAttempt; i++ )
{
if( i > 0 )
{
/* Some delay before a new attempt.
* We do not want to irritate the server.
* Wait up to 20 seconds.
*/
j = i*2;
if( j > 20 ) j = 20;
retCode = TA_DriverShouldContinue(paramForAddDataPtr);
k = 0;
while( (retCode == TA_SUCCESS) && (k++<j) )
{
TA_Sleep( 1 );
retCode = TA_DriverShouldContinue(paramForAddDataPtr);
}
if( retCode != TA_SUCCESS )
return retCode;
}
/* On data retreival problems, keep retrying
* many times. TA_INTERNET_READ_DATA_FAILED means
* that the server can be reached, but somehow the
* transmission of the data was interupted. So it is
* worth to give multiple re-try immediatly.
*
* All other type of failure will retry
* only "nbAttempt" times.
*/
retCode = TA_INTERNET_READ_DATA_FAILED;
for( j=0; (retCode == TA_INTERNET_READ_DATA_FAILED) && (j < 100); j++ )
{
retCode = internalWebPageAlloc( webSiteAddr,
webSitePage,
proxyName,
proxyPort,
webPageAllocated );
if( retCode == TA_INTERNET_READ_DATA_FAILED )
TA_Sleep( 1 ); /* 1 second */
}
if( retCode == TA_SUCCESS )
return TA_SUCCESS;
}
return retCode;
}
