use of edu.harvard.iq.dataverse.ingest.tabulardata.InvalidData in project dataverse by IQSS.
the class SAVFileReader method decodeRecordType999.
void decodeRecordType999(BufferedInputStream stream) throws IOException {
dbgLog.fine("decodeRecordType999(): start");
try {
if (stream == null) {
throw new IllegalArgumentException("RT999: stream == null!");
}
// first check the 4-byte header value
// if (stream.markSupported()){
stream.mark(1000);
// }
// 999.0 check the first 4 bytes
byte[] headerCodeRt999 = new byte[LENGTH_RECORD_TYPE999_CODE];
// dbgLog.fine("RT999: stream position="+stream.pos);
int nbytes_rt999 = stream.read(headerCodeRt999, 0, LENGTH_RECORD_TYPE999_CODE);
// to-do check against nbytes
// printHexDump(headerCodeRt999, "RT999 header test");
ByteBuffer bb_header_code_rt999 = ByteBuffer.wrap(headerCodeRt999, 0, LENGTH_RECORD_TYPE999_CODE);
if (isLittleEndian) {
bb_header_code_rt999.order(ByteOrder.LITTLE_ENDIAN);
}
int intRT999test = bb_header_code_rt999.getInt();
dbgLog.fine("header test value: RT999=" + intRT999test);
if (intRT999test != 999) {
// if (stream.markSupported()){
dbgLog.fine("intRT999test failed=" + intRT999test);
stream.reset();
throw new IOException("RT999:Header value(999) was not correctly detected:" + intRT999test);
// }
}
// 999.1 check 4-byte integer Filler block
byte[] length_filler = new byte[LENGTH_RT999_FILLER];
int nbytes_rt999_1 = stream.read(length_filler, 0, LENGTH_RT999_FILLER);
// to-do check against nbytes
// printHexDump(length_how_many_line_bytes, "RT999 how_many_line_bytes");
ByteBuffer bb_filler = ByteBuffer.wrap(length_filler, 0, LENGTH_RT999_FILLER);
if (isLittleEndian) {
bb_filler.order(ByteOrder.LITTLE_ENDIAN);
}
int rt999filler = bb_filler.getInt();
dbgLog.fine("rt999filler=" + rt999filler);
if (rt999filler == 0) {
dbgLog.fine("the end of the dictionary section");
} else {
throw new IOException("RT999: failed to detect the end mark(0): value=" + rt999filler);
}
// missing value processing concerning HIGHEST/LOWEST values
Set<Map.Entry<String, InvalidData>> msvlc = invalidDataTable.entrySet();
for (Iterator<Map.Entry<String, InvalidData>> itc = msvlc.iterator(); itc.hasNext(); ) {
Map.Entry<String, InvalidData> et = itc.next();
String variable = et.getKey();
dbgLog.fine("variable=" + variable);
InvalidData invalidDataInfo = et.getValue();
if (invalidDataInfo.getInvalidRange() != null && !invalidDataInfo.getInvalidRange().isEmpty()) {
if (invalidDataInfo.getInvalidRange().get(0).equals(OBSTypeHexValue.get("LOWEST"))) {
dbgLog.fine("1st value is LOWEST");
invalidDataInfo.getInvalidRange().set(0, "LOWEST");
} else if (invalidDataInfo.getInvalidRange().get(1).equals(OBSTypeHexValue.get("HIGHEST"))) {
dbgLog.fine("2nd value is HIGHEST");
invalidDataInfo.getInvalidRange().set(1, "HIGHEST");
}
}
}
dbgLog.fine("invalidDataTable:\n" + invalidDataTable);
// TODO: take care of the invalid data! - add the appropriate
// value labels (?)
// should it be done here, or at the end of ingest?
// -- L.A. 4.0 alpha
// /smd.setInvalidDataTable(invalidDataTable);
} catch (IOException ex) {
// exit(1);
throw ex;
}
dbgLog.fine("decodeRecordType999(): end");
}
use of edu.harvard.iq.dataverse.ingest.tabulardata.InvalidData in project dataverse by IQSS.
the class SAVFileReader method decodeRecordType2.
void decodeRecordType2(BufferedInputStream stream) throws IOException {
dbgLog.fine("decodeRecordType2(): start");
if (stream == null) {
throw new IllegalArgumentException("stream == null!");
}
Map<String, String> printFormatNameTable = new LinkedHashMap<String, String>();
Map<String, String> variableLabelMap = new LinkedHashMap<String, String>();
Map<String, List<String>> missingValueTable = new LinkedHashMap<String, List<String>>();
List<Integer> printFormatList = new ArrayList<Integer>();
String caseWeightVariableName = null;
int caseWeightVariableIndex = 0;
boolean lastVariableIsExtendable = false;
boolean extendedVariableMode = false;
boolean obs255 = false;
String lastVariableName = null;
String lastExtendedVariable = null;
// this field repeats as many as the number of variables in
// this sav file
// (note that the above statement is not technically correct, this
// record repeats not just for every variable in the file, but for
// every OBS (8 byte unit); i.e., if a string is split into multiple
// OBS units, each one will have its own RT2 record -- L.A.).
// Each field constists of a fixed (32-byte) segment and
// then a few variable segments:
// if the variable has a label (3rd INT4 set to 1), then there's 4 more
// bytes specifying the length of the label, and then that many bytes
// holding the label itself (no more than 256).
// Then if there are optional missing value units (4th INT4 set to 1)
// there will be 3 more OBS units attached = 24 extra bytes.
int variableCounter = 0;
int obsSeqNumber = 0;
int j;
dbgLog.fine("RT2: Reading " + OBSUnitsPerCase + " OBS units.");
for (j = 0; j < OBSUnitsPerCase; j++) {
dbgLog.fine("RT2: " + j + "-th RT2 unit is being decoded.");
// 2.0: read the fixed[=non-optional] 32-byte segment
byte[] recordType2Fixed = new byte[LENGTH_RECORDTYPE2_FIXED];
try {
int nbytes = stream.read(recordType2Fixed, 0, LENGTH_RECORDTYPE2_FIXED);
if (nbytes == 0) {
throw new IOException("reading recordType2: no bytes read!");
}
int offset = 0;
// 2.1: create int-view of the bytebuffer for the first 16-byte segment
int rt2_1st_4_units = 4;
ByteBuffer[] bb_record_type2_fixed_part1 = new ByteBuffer[rt2_1st_4_units];
int[] recordType2FixedPart1 = new int[rt2_1st_4_units];
for (int i = 0; i < rt2_1st_4_units; i++) {
bb_record_type2_fixed_part1[i] = ByteBuffer.wrap(recordType2Fixed, offset, LENGTH_SAV_INT_BLOCK);
offset += LENGTH_SAV_INT_BLOCK;
if (isLittleEndian) {
bb_record_type2_fixed_part1[i].order(ByteOrder.LITTLE_ENDIAN);
}
recordType2FixedPart1[i] = bb_record_type2_fixed_part1[i].getInt();
}
// 1st ([0]) element must be 2 otherwise no longer Record Type 2
if (recordType2FixedPart1[0] != 2) {
dbgLog.warning(j + "-th RT header value is no longet RT2! " + recordType2FixedPart1[0]);
break;
}
dbgLog.fine("variable type[must be 2]=" + recordType2FixedPart1[0]);
// 2.3 variable name: 8 byte(space[x20]-padded)
// This field is located at the very end of the 32 byte
// fixed-size RT2 header (bytes 24-31).
// We are processing it now, so that
// we can make the decision on whether this variable is part
// of a compound variable:
String RawVariableName = getNullStrippedString(new String(Arrays.copyOfRange(recordType2Fixed, 24, (24 + LENGTH_VARIABLE_NAME)), defaultCharSet));
// offset +=LENGTH_VARIABLE_NAME;
String variableName = null;
if (RawVariableName.indexOf(' ') >= 0) {
variableName = RawVariableName.substring(0, RawVariableName.indexOf(' '));
} else {
variableName = RawVariableName;
}
// 2nd ([1]) element: numeric variable = 0 :for string variable
// this block indicates its datum-length, i.e, >0 ;
// if -1, this RT2 unit is a non-1st RT2 unit for a string variable
// whose value is longer than 8 character.
boolean isNumericVariable = false;
dbgLog.fine("variable type(0: numeric; > 0: String;-1 continue )=" + recordType2FixedPart1[1]);
// OBSwiseTypelList.add(recordType2FixedPart1[1]);
int HowManyRt2Units = 1;
if (recordType2FixedPart1[1] == -1) {
dbgLog.fine("this RT2 is an 8 bit continuation chunk of an earlier string variable");
if (obs255) {
if (obsSeqNumber < 30) {
OBSwiseTypelList.add(recordType2FixedPart1[1]);
obsSeqNumber++;
} else {
OBSwiseTypelList.add(-2);
obs255 = false;
obsSeqNumber = 0;
}
} else {
OBSwiseTypelList.add(recordType2FixedPart1[1]);
}
obsNonVariableBlockSet.add(j);
continue;
} else if (recordType2FixedPart1[1] == 0) {
// This is a numeric variable
extendedVariableMode = false;
// And as such, it cannot be an extension of a
// previous, long string variable.
OBSwiseTypelList.add(recordType2FixedPart1[1]);
variableCounter++;
isNumericVariable = true;
variableTypelList.add(recordType2FixedPart1[1]);
} else if (recordType2FixedPart1[1] > 0) {
if (recordType2FixedPart1[1] == 255) {
obs255 = true;
}
if (lastVariableIsExtendable) {
String varNameBase = null;
if (lastVariableName.length() > 5) {
varNameBase = lastVariableName.substring(0, 5);
} else {
varNameBase = lastVariableName;
}
if (extendedVariableMode) {
if (variableNameIsAnIncrement(varNameBase, lastExtendedVariable, variableName)) {
OBSwiseTypelList.add(-1);
lastExtendedVariable = variableName;
// OK, we stay in the "extended variable" mode;
// but we can't move on to the next OBS (hence the commented out
// "continue" below:
// continue;
// see the next comment below for the explanation.
//
// Should we also set "extendable" flag to false at this point
// if it's shorter than 255 bytes, i.e. the last extended chunk?
} else {
extendedVariableMode = false;
}
} else {
if (variableNameIsAnIncrement(varNameBase, variableName)) {
OBSwiseTypelList.add(-1);
extendedVariableMode = true;
dbgLog.fine("RT2: in extended variable mode; variable " + variableName);
lastExtendedVariable = variableName;
// Before we move on to the next OBS unit, we need to check
// if this current extended variable has its own label specified;
// If so, we need to determine its length, then read and skip
// that many bytes.
// Hence the commented out "continue" below:
// continue;
}
}
}
if (!extendedVariableMode) {
// OK, this is a "real"
// string variable, and not a continuation chunk of a compound
// string.
OBSwiseTypelList.add(recordType2FixedPart1[1]);
variableCounter++;
if (recordType2FixedPart1[1] == 255) {
// This variable is 255 bytes long, i.e. this is
// either the single "atomic" variable of the
// max allowed size, or it's a 255 byte segment
// of a compound variable. So we will check
// the next variable and see if it is the continuation
// of this one.
lastVariableIsExtendable = true;
} else {
lastVariableIsExtendable = false;
}
if (recordType2FixedPart1[1] % LENGTH_SAV_OBS_BLOCK == 0) {
HowManyRt2Units = recordType2FixedPart1[1] / LENGTH_SAV_OBS_BLOCK;
} else {
HowManyRt2Units = recordType2FixedPart1[1] / LENGTH_SAV_OBS_BLOCK + 1;
}
variableTypelList.add(recordType2FixedPart1[1]);
}
}
if (!extendedVariableMode) {
// Again, we only want to do the following steps for the "real"
// variables, not the chunks of split mega-variables:
dbgLog.fine("RT2: HowManyRt2Units for this variable=" + HowManyRt2Units);
lastVariableName = variableName;
// caseWeightVariableOBSIndex starts from 1: 0 is used for does-not-exist cases
if (j == (caseWeightVariableOBSIndex - 1)) {
caseWeightVariableName = variableName;
// TODO: do we need this "index"? -- 4.0 alpha
caseWeightVariableIndex = variableCounter;
// /smd.setCaseWeightVariableName(caseWeightVariableName);
// /smd.getFileInformation().put("caseWeightVariableIndex", caseWeightVariableIndex);
}
OBSIndexToVariableName.put(j, variableName);
// dbgLog.fine("\nvariable name="+variableName+"<-");
dbgLog.fine("RT2: " + j + "-th variable name=" + variableName + "<-");
dbgLog.fine("RT2: raw variable: " + RawVariableName);
variableNameList.add(variableName);
}
// 3rd ([2]) element: = 1 variable-label block follows; 0 = no label
//
dbgLog.fine("RT: variable label follows?(1:yes; 0: no)=" + recordType2FixedPart1[2]);
boolean hasVariableLabel = recordType2FixedPart1[2] == 1 ? true : false;
if ((recordType2FixedPart1[2] != 0) && (recordType2FixedPart1[2] != 1)) {
throw new IOException("RT2: reading error: value is neither 0 or 1" + recordType2FixedPart1[2]);
}
if (hasVariableLabel) {
byte[] length_variable_label = new byte[4];
int nbytes_2_4 = stream.read(length_variable_label);
if (nbytes_2_4 == 0) {
throw new IOException("RT 2: error reading recordType2.4: no bytes read!");
} else {
dbgLog.fine("nbytes_2_4=" + nbytes_2_4);
}
ByteBuffer bb_length_variable_label = ByteBuffer.wrap(length_variable_label, 0, LENGTH_VARIABLE_LABEL);
if (isLittleEndian) {
bb_length_variable_label.order(ByteOrder.LITTLE_ENDIAN);
}
int rawVariableLabelLength = bb_length_variable_label.getInt();
dbgLog.fine("rawVariableLabelLength=" + rawVariableLabelLength);
int variableLabelLength = getSAVintAdjustedBlockLength(rawVariableLabelLength);
dbgLog.fine("RT2: variableLabelLength=" + variableLabelLength);
// 2.5 [optional]variable label whose length is found at 2.4
String variableLabel = "";
if (rawVariableLabelLength > 0) {
byte[] variable_label = new byte[variableLabelLength];
int nbytes_2_5 = stream.read(variable_label);
if (nbytes_2_5 == 0) {
throw new IOException("RT 2: error reading recordType2.5: " + variableLabelLength + " bytes requested, no bytes read!");
} else {
dbgLog.fine("nbytes_2_5=" + nbytes_2_5);
}
variableLabel = getNullStrippedString(new String(Arrays.copyOfRange(variable_label, 0, rawVariableLabelLength), defaultCharSet));
dbgLog.fine("RT2: variableLabel=" + variableLabel + "<-");
dbgLog.fine(variableName + " => " + variableLabel);
} else {
dbgLog.fine("RT2: defaulting to empty variable label.");
}
if (!extendedVariableMode) {
// We only have any use for this label if it's a "real" variable.
// Thinking about it, it doesn't make much sense for the "fake"
// variables that are actually chunks of large strings to store
// their own labels. But in some files they do. Then failing to read
// the bytes would result in getting out of sync with the RT record
// borders. So we always read the bytes, but only use them for
// the real variable entries.
/*String variableLabel = new String(Arrays.copyOfRange(variable_label,
0, rawVariableLabelLength),"US-ASCII");*/
variableLabelMap.put(variableName, variableLabel);
}
}
if (extendedVariableMode) {
// that's how SPSS stores them.
continue;
}
// 4th ([3]) element: Missing value type code
// 0[none], 1, 2, 3 [point-type],-2[range], -3 [range type+ point]
dbgLog.fine("RT: missing value unit follows?(if 0, none)=" + recordType2FixedPart1[3]);
boolean hasMissingValues = (validMissingValueCodeSet.contains(recordType2FixedPart1[3]) && (recordType2FixedPart1[3] != 0)) ? true : false;
InvalidData invalidDataInfo = null;
if (recordType2FixedPart1[3] != 0) {
invalidDataInfo = new InvalidData(recordType2FixedPart1[3]);
dbgLog.fine("RT: missing value type=" + invalidDataInfo.getType());
}
// 2.2: print/write formats: 4-byte each = 8 bytes
byte[] printFormt = Arrays.copyOfRange(recordType2Fixed, offset, offset + LENGTH_PRINT_FORMAT_CODE);
dbgLog.fine("printFrmt=" + new String(Hex.encodeHex(printFormt)));
offset += LENGTH_PRINT_FORMAT_CODE;
int formatCode = isLittleEndian ? printFormt[2] : printFormt[1];
int formatWidth = isLittleEndian ? printFormt[1] : printFormt[2];
// TODO:
// What should we be doing with these "format decimal positions"
// in 4.0?
// -- L.A. 4.0 alpha
int formatDecimalPointPosition = isLittleEndian ? printFormt[0] : printFormt[3];
dbgLog.fine("RT2: format code{5=F, 1=A[String]}=" + formatCode);
formatDecimalPointPositionList.add(formatDecimalPointPosition);
if (!SPSSConstants.FORMAT_CODE_TABLE_SAV.containsKey(formatCode)) {
throw new IOException("Unknown format code was found = " + formatCode);
} else {
printFormatList.add(formatCode);
}
byte[] writeFormt = Arrays.copyOfRange(recordType2Fixed, offset, offset + LENGTH_WRITE_FORMAT_CODE);
dbgLog.fine("RT2: writeFrmt=" + new String(Hex.encodeHex(writeFormt)));
if (writeFormt[3] != 0x00) {
dbgLog.fine("byte-order(write format): reversal required");
}
offset += LENGTH_WRITE_FORMAT_CODE;
if (!SPSSConstants.ORDINARY_FORMAT_CODE_SET.contains(formatCode)) {
StringBuilder sb = new StringBuilder(SPSSConstants.FORMAT_CODE_TABLE_SAV.get(formatCode) + formatWidth);
if (formatDecimalPointPosition > 0) {
sb.append("." + formatDecimalPointPosition);
}
dbgLog.fine("formattable[i] = " + variableName + " -> " + sb.toString());
printFormatNameTable.put(variableName, sb.toString());
}
printFormatTable.put(variableName, SPSSConstants.FORMAT_CODE_TABLE_SAV.get(formatCode));
if (hasMissingValues) {
dbgLog.fine("RT2: decoding missing value: type=" + recordType2FixedPart1[3]);
int howManyMissingValueUnits = missingValueCodeUnits.get(recordType2FixedPart1[3]);
// int howManyMissingValueUnits = recordType2FixedPart1[3] > 0 ? recordType2FixedPart1[3] : 0;
dbgLog.fine("RT2: howManyMissingValueUnits=" + howManyMissingValueUnits);
byte[] missing_value_code_units = new byte[LENGTH_SAV_OBS_BLOCK * howManyMissingValueUnits];
int nbytes_2_6 = stream.read(missing_value_code_units);
if (nbytes_2_6 == 0) {
throw new IOException("RT 2: reading recordType2.6: no byte was read");
} else {
dbgLog.fine("nbytes_2_6=" + nbytes_2_6);
}
if (isNumericVariable) {
double[] missingValues = new double[howManyMissingValueUnits];
// List<String> mvp = new ArrayList<String>();
List<String> mv = new ArrayList<String>();
ByteBuffer[] bb_missig_value_code = new ByteBuffer[howManyMissingValueUnits];
int offset_start = 0;
for (int i = 0; i < howManyMissingValueUnits; i++) {
bb_missig_value_code[i] = ByteBuffer.wrap(missing_value_code_units, offset_start, LENGTH_SAV_OBS_BLOCK);
offset_start += LENGTH_SAV_OBS_BLOCK;
if (isLittleEndian) {
bb_missig_value_code[i].order(ByteOrder.LITTLE_ENDIAN);
}
ByteBuffer temp = bb_missig_value_code[i].duplicate();
missingValues[i] = bb_missig_value_code[i].getDouble();
if (Double.toHexString(missingValues[i]).equals("-0x1.ffffffffffffep1023")) {
dbgLog.fine("1st value is LOWEST");
mv.add(Double.toHexString(missingValues[i]));
} else if (Double.valueOf(missingValues[i]).equals(Double.MAX_VALUE)) {
dbgLog.fine("2nd value is HIGHEST");
mv.add(Double.toHexString(missingValues[i]));
} else {
mv.add(doubleNumberFormatter.format(missingValues[i]));
}
dbgLog.fine(i + "-th missing value=" + Double.toHexString(missingValues[i]));
}
dbgLog.fine("variableName=" + variableName);
if (recordType2FixedPart1[3] > 0) {
// point cases only
dbgLog.fine("mv(>0)=" + mv);
missingValueTable.put(variableName, mv);
invalidDataInfo.setInvalidValues(mv);
} else if (recordType2FixedPart1[3] == -2) {
dbgLog.fine("mv(-2)=" + mv);
// range
invalidDataInfo.setInvalidRange(mv);
} else if (recordType2FixedPart1[3] == -3) {
// mixed case
dbgLog.fine("mv(-3)=" + mv);
invalidDataInfo.setInvalidRange(mv.subList(0, 2));
invalidDataInfo.setInvalidValues(mv.subList(2, 3));
missingValueTable.put(variableName, mv.subList(2, 3));
}
dbgLog.fine("missing value=" + StringUtils.join(missingValueTable.get(variableName), "|"));
dbgLog.fine("invalidDataInfo(Numeric):\n" + invalidDataInfo);
invalidDataTable.put(variableName, invalidDataInfo);
} else {
// string variable case
String[] missingValues = new String[howManyMissingValueUnits];
List<String> mv = new ArrayList<String>();
int offset_start = 0;
int offset_end = LENGTH_SAV_OBS_BLOCK;
for (int i = 0; i < howManyMissingValueUnits; i++) {
missingValues[i] = StringUtils.stripEnd(new String(Arrays.copyOfRange(missing_value_code_units, offset_start, offset_end), defaultCharSet), " ");
dbgLog.fine("missing value=" + missingValues[i] + "<-");
offset_start = offset_end;
offset_end += LENGTH_SAV_OBS_BLOCK;
mv.add(missingValues[i]);
}
invalidDataInfo.setInvalidValues(mv);
missingValueTable.put(variableName, mv);
invalidDataTable.put(variableName, invalidDataInfo);
dbgLog.fine("missing value(str)=" + StringUtils.join(missingValueTable.get(variableName), "|"));
dbgLog.fine("invalidDataInfo(String):\n" + invalidDataInfo);
}
// string case
dbgLog.fine("invalidDataTable:\n" + invalidDataTable);
}
// if msv
} catch (IOException ex) {
// ex.printStackTrace();
throw ex;
} catch (Exception ex) {
ex.printStackTrace();
// should we be throwing some exception here?
}
}
if (j != OBSUnitsPerCase) {
dbgLog.fine("RT2: attention! didn't reach the end of the OBS list!");
throw new IOException("RT2: didn't reach the end of the OBS list!");
}
dbgLog.fine("RT2 metadata-related exit-chores");
// /smd.getFileInformation().put("varQnty", variableCounter);
dataTable.setVarQuantity(new Long(variableCounter));
dbgLog.fine("RT2: varQnty=" + variableCounter);
// 4.0 Initialize variables:
List<DataVariable> variableList = new ArrayList<DataVariable>();
for (int i = 0; i < variableCounter; i++) {
DataVariable dv = new DataVariable();
String varName = variableNameList.get(i);
dbgLog.fine("name: " + varName);
dv.setName(varName);
String varLabel = variableLabelMap.get(varName);
if (varLabel != null && varLabel.length() > 255) {
// TODO:
// variable labels will be changed into type 'TEXT' in the
// database - this will eliminate the 255 char. limit.
// -- L.A. 4.0 beta11
dbgLog.fine("Have to truncate label: " + varLabel);
varLabel = varLabel.substring(0, 255);
}
dbgLog.fine("label: " + varLabel);
dv.setLabel(varLabel);
dv.setInvalidRanges(new ArrayList<VariableRange>());
dv.setSummaryStatistics(new ArrayList<SummaryStatistic>());
dv.setUnf("UNF:6:");
dv.setCategories(new ArrayList<VariableCategory>());
variableList.add(dv);
dv.setFileOrder(i);
dv.setDataTable(dataTable);
}
dataTable.setDataVariables(variableList);
// /smd.setVariableName(variableNameList.toArray(new String[variableNameList.size()]));
// /smd.setVariableLabel(variableLabelMap);
// TODO:
// figure out what to do with the missing value table!
// -- 4.0 alpha
// well, they were used to generate merged summary statistics for
// the variable. So need to verify what the DDI import was doing
// with them and replicate the same in 4.0.
// (add appropriate value labels?)
// /TODO: 4.0 smd.setMissingValueTable(missingValueTable);
// /smd.getFileInformation().put("caseWeightVariableName", caseWeightVariableName);
dbgLog.fine("sumstat:long case=" + Arrays.deepToString(variableTypelList.toArray()));
dbgLog.fine("RT2: OBSwiseTypelList=" + OBSwiseTypelList);
dbgLog.fine("decodeRecordType2(): end");
}
use of edu.harvard.iq.dataverse.ingest.tabulardata.InvalidData in project dataverse by IQSS.
the class PORFileReader method processMissingValueData.
private void processMissingValueData() {
/*
POR's missing-value storage differs form the counterpart of SAV;
this method transforms the POR-native storage to the SAV-type
after this process, missingValueTable contains point-type
missing values for later catStat/sumStat processing;
range and mixed type cases are stored in invalidDataTable
missingValueCodeTable=
{VAR1=[9], VAR2=[A], VAR3=[9, 8], VAR4=[A, 8],
VAR5=[8, 8, 8], VAR6=[B], VAR7=[B, 8]}
missingValueTable=
{VAR1=[-1], VAR2=[-1], VAR3=[-2, -1], VAR4=[-1, -2],
VAR5=[-1, -2, -3], VAR6=[-2, -1], VAR7=[-3, -2, -1]}
missingValueTable={VAR1=[], VAR2=[], VAR3=[-1], VAR4=[-2],
VAR5=[-1, -2, -3], VAR6=[], VAR7=[-2]}
*/
dbgLog.fine("missingValueCodeTable=" + missingValueCodeTable);
Set<Map.Entry<String, List<String>>> msvlc = missingValueCodeTable.entrySet();
for (Map.Entry<String, List<String>> et : msvlc) {
String variable = et.getKey();
dbgLog.fine("variable=" + variable);
List<String> codeList = et.getValue();
List<String> valueList = missingValueTable.get(variable);
dbgLog.fine("codeList=" + codeList);
dbgLog.fine("valueList=" + valueList);
int type;
InvalidData invalidDataInfo = null;
if (valueList.size() == 3) {
if (codeList.get(0).equals("8") && codeList.get(1).equals("8") && codeList.get(2).equals("8")) {
type = 3;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList);
} else if (codeList.get(0).equals("9") && codeList.get(1).equals("8")) {
type = -3;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList.subList(2, 3));
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else if (codeList.get(0).equals("A") && codeList.get(1).equals("8")) {
type = -3;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList.subList(2, 3));
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else if (codeList.get(0).equals("B") && codeList.get(1).equals("8")) {
type = -3;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList.subList(2, 3));
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else {
dbgLog.severe("unkown missing-value combination(3 values)");
}
} else if (valueList.size() == 2) {
if (codeList.get(0).equals("8") && codeList.get(1).equals("8")) {
type = 2;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList);
} else if (codeList.get(0).equals("9")) {
type = -2;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else if (codeList.get(0).equals("A")) {
type = -2;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else if (codeList.get(0).equals("B")) {
type = -2;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidRange(valueList.subList(0, 2));
} else {
dbgLog.severe("unknown missing value combination(2 values)");
}
} else if (valueList.size() == 1) {
if (codeList.get(0).equals("8")) {
type = 1;
invalidDataInfo = new InvalidData(type);
invalidDataInfo.setInvalidValues(valueList);
} else {
dbgLog.severe("unknown missing value combination(2 values)");
}
}
invalidDataTable.put(variable, invalidDataInfo);
}
dbgLog.fine("invalidDataTable=" + invalidDataTable);
Set<Map.Entry<String, List<String>>> msvl = missingValueTable.entrySet();
for (Map.Entry<String, List<String>> et : msvl) {
String variable = et.getKey();
List<String> valueList = et.getValue();
List<String> codeList = missingValueCodeTable.get(variable);
dbgLog.finer("var=" + variable + "\tvalue=" + valueList + "\t code" + codeList);
List<String> temp = new ArrayList<>();
for (int j = 0; j < codeList.size(); j++) {
if (codeList.get(j).equals("8")) {
temp.add(valueList.get(j));
}
}
missingValueTable.put(variable, temp);
}
dbgLog.fine("missingValueTable=" + missingValueTable);
}
Aggregations