[GitHub] [carbondata] akashrn5 commented on a change in pull request #3896: [CARBONDATA-3955] Fix load failures due to daylight saving time changes

2020-08-25 Thread GitBox


akashrn5 commented on a change in pull request #3896:
URL: https://github.com/apache/carbondata/pull/3896#discussion_r476487539



##
File path: 
core/src/main/java/org/apache/carbondata/core/constants/CarbonCommonConstants.java
##
@@ -1592,6 +1592,15 @@ private CarbonCommonConstants() {
 
   public static final String CARBON_LUCENE_INDEX_STOP_WORDS_DEFAULT = "false";
 
+  // Property to enable parsing the timestamp/date data with setLenient = true 
in load
+  // flow if it fails with parse invalid timestamp data. (example: 1941-03-15 
00:00:00
+  // is valid time in Asia/Calcutta zone and is invalid and will fail to parse 
in Asia/Shanghai
+  // zone as DST is observed and clocks were turned forward 1 hour to 
1941-03-15 01:00:00)
+  @CarbonProperty(dynamicConfigurable = true) public static final String

Review comment:
   `move public static final String` to next line

##
File path: core/src/main/java/org/apache/carbondata/core/util/DataTypeUtil.java
##
@@ -444,9 +446,38 @@ private static Object parseTimestamp(String 
dimensionValue, String dateFormat) {
 dateFormatter = timestampFormatter.get();
   }
   dateToStr = dateFormatter.parse(dimensionValue);
-  return dateToStr.getTime();
+  timeValue = dateToStr.getTime();
+  validateTimeStampRange(timeValue);
+  return timeValue;
 } catch (ParseException e) {
-  throw new NumberFormatException(e.getMessage());
+  // If the parsing fails, try to parse again with setLenient to true if 
the property is set
+  if (CarbonProperties.getInstance().isSetLenientEnabled()) {
+try {
+  dateFormatter.setLenient(true);
+  dateToStr = dateFormatter.parse(dimensionValue);
+  dateFormatter.setLenient(false);
+  timeValue = dateToStr.getTime();
+  validateTimeStampRange(timeValue);
+  LOGGER.info("Parsed data with lenience as true, setting back to 
default mode");
+  return timeValue;
+} catch (ParseException ex) {
+  dateFormatter.setLenient(false);
+  LOGGER.info("Failed to parse data with lenience as true, setting 
back to default mode");
+  throw new NumberFormatException(ex.getMessage());
+}
+  } else {
+throw new NumberFormatException(e.getMessage());
+  }
+}
+  }
+
+  private static void validateTimeStampRange(Long timeValue) {
+if (timeValue < DateDirectDictionaryGenerator.MIN_VALUE
+|| timeValue > DateDirectDictionaryGenerator.MAX_VALUE) {
+  throw new NumberFormatException(
+  "timestamp column data value: " + timeValue + "is not in valid " + 
"range of: "

Review comment:
   `"is not in valid " + "range of: "` correct the formatting here, there 
are unnecessary concatenations

##
File path: 
integration/spark/src/test/scala/org/apache/carbondata/spark/util/BadRecordUtil.scala
##
@@ -68,4 +71,32 @@ object BadRecordUtil {
 badRecordLocation = badRecordLocation + "/" + dbName + "/" + tableName
 
FileFactory.deleteAllCarbonFilesOfDir(FileFactory.getCarbonFile(badRecordLocation))
   }
+
+  def createCSV(rows: ListBuffer[Array[String]], csvPath: String): Unit = {
+val out = new BufferedWriter(new FileWriter(csvPath))
+val writer: CSVWriter = new CSVWriter(out)
+try {
+  for (row <- rows) {
+writer.writeNext(row)
+  }
+}
+catch {
+  case e: Exception =>
+Assert.fail(e.getMessage)
+}
+finally {
+  out.close()
+  writer.close()
+}
+  }
+
+  def deleteCSVFile(csvPath: String): Unit = {

Review comment:
   i don't think this method is required, just call 
`FileUtils.forceDelete(new File(csvPath))` in each test case itself and no need 
of any Assert here, as its not doing any functional validation

##
File path: 
integration/spark/src/test/scala/org/apache/carbondata/spark/util/BadRecordUtil.scala
##
@@ -68,4 +71,32 @@ object BadRecordUtil {
 badRecordLocation = badRecordLocation + "/" + dbName + "/" + tableName
 
FileFactory.deleteAllCarbonFilesOfDir(FileFactory.getCarbonFile(badRecordLocation))
   }
+
+  def createCSV(rows: ListBuffer[Array[String]], csvPath: String): Unit = {
+val out = new BufferedWriter(new FileWriter(csvPath))

Review comment:
   correct the formatting of this method and may be you can use 
try-with-resource here, you can check once





This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org




[GitHub] [carbondata] akashrn5 commented on a change in pull request #3896: [CARBONDATA-3955] Fix load failures due to daylight saving time changes

2020-08-24 Thread GitBox


akashrn5 commented on a change in pull request #3896:
URL: https://github.com/apache/carbondata/pull/3896#discussion_r475716479



##
File path: core/src/main/java/org/apache/carbondata/core/util/DataTypeUtil.java
##
@@ -435,18 +436,48 @@ public static Object 
getDataDataTypeForNoDictionaryColumn(String dimensionValue,
 
   private static Object parseTimestamp(String dimensionValue, String 
dateFormat) {
 Date dateToStr;
-DateFormat dateFormatter;
+DateFormat dateFormatter = null;
+long timeValue;
 try {
   if (null != dateFormat && !dateFormat.trim().isEmpty()) {
 dateFormatter = new SimpleDateFormat(dateFormat);
-dateFormatter.setLenient(false);
   } else {
 dateFormatter = timestampFormatter.get();
   }
+  dateFormatter.setLenient(false);
   dateToStr = dateFormatter.parse(dimensionValue);
-  return dateToStr.getTime();
+  timeValue = dateToStr.getTime();
+  validateTimeStampRange(timeValue);
+  return timeValue;
 } catch (ParseException e) {
-  throw new NumberFormatException(e.getMessage());
+  // If the parsing fails, try to parse again with setLenient to true if 
the property is set
+  if (CarbonProperties.getInstance().isSetLenientEnabled()) {
+try {
+  LOGGER.info("Changing setLenient to true");
+  dateFormatter.setLenient(true);
+  dateToStr = dateFormatter.parse(dimensionValue);
+  dateFormatter.setLenient(false);
+  LOGGER.info("Changing setLenient back to false");

Review comment:
   i think these logs not required here, may be you can add only one log 
say after line 462, saying you had set true as parsing failed for invalid data 
and parsing finished, something meaningful like this and in line 466, you can 
say `"failed to parse data with lenience as true, setting back to default"` 
like this , it will look clean and good

##
File path: 
integration/spark/src/test/scala/org/apache/carbondata/spark/testsuite/badrecordloger/BadRecordActionTest.scala
##
@@ -270,6 +274,53 @@ class BadRecordActionTest extends QueryTest {
 }
   }
 
+  test("test bad record FAIL with invalid timestamp range") {
+val csvPath = s"$resourcesPath/badrecords/invalidTimeStampRange.csv"
+val rows1 = new ListBuffer[Array[String]]
+rows1 += Array("ID", "date", "time")
+rows1 += Array("1", "2016-7-24", "342016-7-24 01:02:30")
+createCSV(rows1, csvPath)
+sql("DROP TABLE IF EXISTS test_time")
+sql("""
+   CREATE TABLE IF NOT EXISTS test_time
+   (ID Int, date Date, time Timestamp)
+   STORED AS carbondata TBLPROPERTIES('dateformat'='-MM-dd',
+   'timestampformat'='-MM-dd HH:mm:ss')
+""")
+val exception = intercept[Exception] {
+  sql(s" LOAD DATA LOCAL INPATH 
'$resourcesPath/badrecords/invalidTimeStampRange.csv' " +
+  s"into table test_time options ('bad_records_action'='fail')")
+}
+assert(exception.getMessage
+  .contains(
+"Data load failed due to bad record: The value with column name time 
and column data" +
+" type TIMESTAMP is not a valid TIMESTAMP type.Please enable bad 
record logger to know" +
+" the detail reason"))
+sql("DROP TABLE IF EXISTS test_time")
+deleteCSVFile(csvPath)
+  }
+
+  def createCSV(rows: ListBuffer[Array[String]], csvPath: String): Unit = {
+val out = new BufferedWriter(new FileWriter(csvPath))
+val writer: CSVWriter = new CSVWriter(out)
+
+for (row <- rows) {
+  writer.writeNext(row)
+}
+out.close()
+writer.close()
+  }
+
+  def deleteCSVFile(csvPath: String): Unit = {

Review comment:
   no need to create new method only for delete, call the file delete API 
directly in test case as not used in many test case.

##
File path: 
integration/spark/src/test/scala/org/apache/carbondata/spark/testsuite/dataload/TestLoadDataWithDiffTimestampFormat.scala
##
@@ -306,7 +315,107 @@ class TestLoadDataWithDiffTimestampFormat extends 
QueryTest with BeforeAndAfterA
 }
   }
 
+  test("test load, update data with setlenient carbon property for daylight " +
+   "saving time from different timezone") {
+CarbonProperties.getInstance().addProperty(
+  CarbonCommonConstants.CARBON_LOAD_SETLENIENT_ENABLE, "true")
+val defaultTimeZone = TimeZone.getDefault
+TimeZone.setDefault(TimeZone.getTimeZone("Asia/Shanghai"))
+sql("DROP TABLE IF EXISTS test_time")
+sql(
+  """
+   CREATE TABLE IF NOT EXISTS test_time
+   (ID Int, date Date, time Timestamp)
+   STORED AS carbondata TBLPROPERTIES('dateformat'='-MM-dd',
+   'timestampformat'='-MM-dd HH:mm')
+""")
+sql(s" LOAD DATA LOCAL INPATH '$resourcesPath/differentZoneTimeStamp.csv' 
into table test_time")
+sql(s"insert into test_time select 11, '2016-7-24', '1941-3-15 00:00:00' ")
+sql("update test_time set (time) = 

[GitHub] [carbondata] akashrn5 commented on a change in pull request #3896: [CARBONDATA-3955] Fix load failures due to daylight saving time changes

2020-08-24 Thread GitBox


akashrn5 commented on a change in pull request #3896:
URL: https://github.com/apache/carbondata/pull/3896#discussion_r475346485



##
File path: 
core/src/main/java/org/apache/carbondata/core/constants/CarbonCommonConstants.java
##
@@ -1592,6 +1592,13 @@ private CarbonCommonConstants() {
 
   public static final String CARBON_LUCENE_INDEX_STOP_WORDS_DEFAULT = "false";
 
+  // Property to enable parsing the timestamp/date data with setLenient = true 
in load
+  // flow if it fails with parse invalid timestamp data.

Review comment:
   can give the timestamp example and mention DST, so that we will know in 
future why this property was added.





This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org




[GitHub] [carbondata] akashrn5 commented on a change in pull request #3896: [CARBONDATA-3955] Fix load failures due to daylight saving time changes

2020-08-21 Thread GitBox


akashrn5 commented on a change in pull request #3896:
URL: https://github.com/apache/carbondata/pull/3896#discussion_r474608789



##
File path: core/src/main/java/org/apache/carbondata/core/util/DataTypeUtil.java
##
@@ -435,19 +436,48 @@ public static Object 
getDataDataTypeForNoDictionaryColumn(String dimensionValue,
 
   private static Object parseTimestamp(String dimensionValue, String 
dateFormat) {
 Date dateToStr;
-DateFormat dateFormatter;
+DateFormat dateFormatter = null;
 try {
   if (null != dateFormat && !dateFormat.trim().isEmpty()) {
 dateFormatter = new SimpleDateFormat(dateFormat);
-dateFormatter.setLenient(false);
   } else {
 dateFormatter = timestampFormatter.get();
   }
+  dateFormatter.setLenient(false);
   dateToStr = dateFormatter.parse(dimensionValue);
-  return dateToStr.getTime();
+  return validateTimeStampRange(dateToStr.getTime());
 } catch (ParseException e) {
-  throw new NumberFormatException(e.getMessage());
+  // If the parsing fails, try to parse again with setLenient to true if 
the property is set
+  if (CarbonProperties.getInstance().isSetLenientEnabled()) {
+try {
+  LOGGER.info("Changing setLenient to true for TimeStamp: " + 
dimensionValue);
+  dateFormatter.setLenient(true);
+  dateToStr = dateFormatter.parse(dimensionValue);
+  LOGGER.info("Changing " + dimensionValue + " to " + dateToStr);
+  dateFormatter.setLenient(false);
+  LOGGER.info("Changing setLenient back to false");
+  return validateTimeStampRange(dateToStr.getTime());
+} catch (ParseException ex) {
+  dateFormatter.setLenient(false);
+  LOGGER.info("Changing setLenient back to false");
+  throw new NumberFormatException(ex.getMessage());
+}
+  } else {
+throw new NumberFormatException(e.getMessage());
+  }
+}
+  }
+
+  private static Long validateTimeStampRange(Long timeValue) {
+long minValue = DateDirectDictionaryGenerator.MIN_VALUE;
+long maxValue = DateDirectDictionaryGenerator.MAX_VALUE;
+if (timeValue < minValue || timeValue > maxValue) {
+  if (LOGGER.isDebugEnabled()) {
+LOGGER.debug("Value for timestamp type column is not in valid range.");
+  }
+  throw new NumberFormatException("Value for timestamp type column is not 
in valid range.");

Review comment:
   ```suggestion
 throw new NumberFormatException("timestamp column data is not in valid 
range of: " + DateDirectDictionaryGenerator.MIN_VALUE + " and " + 
DateDirectDictionaryGenerator.MAX_VALUE );
   ```

##
File path: core/src/main/java/org/apache/carbondata/core/util/DataTypeUtil.java
##
@@ -435,19 +436,48 @@ public static Object 
getDataDataTypeForNoDictionaryColumn(String dimensionValue,
 
   private static Object parseTimestamp(String dimensionValue, String 
dateFormat) {
 Date dateToStr;
-DateFormat dateFormatter;
+DateFormat dateFormatter = null;
 try {
   if (null != dateFormat && !dateFormat.trim().isEmpty()) {
 dateFormatter = new SimpleDateFormat(dateFormat);
-dateFormatter.setLenient(false);
   } else {
 dateFormatter = timestampFormatter.get();
   }
+  dateFormatter.setLenient(false);
   dateToStr = dateFormatter.parse(dimensionValue);
-  return dateToStr.getTime();
+  return validateTimeStampRange(dateToStr.getTime());
 } catch (ParseException e) {
-  throw new NumberFormatException(e.getMessage());
+  // If the parsing fails, try to parse again with setLenient to true if 
the property is set
+  if (CarbonProperties.getInstance().isSetLenientEnabled()) {
+try {
+  LOGGER.info("Changing setLenient to true for TimeStamp: " + 
dimensionValue);
+  dateFormatter.setLenient(true);
+  dateToStr = dateFormatter.parse(dimensionValue);
+  LOGGER.info("Changing " + dimensionValue + " to " + dateToStr);
+  dateFormatter.setLenient(false);
+  LOGGER.info("Changing setLenient back to false");
+  return validateTimeStampRange(dateToStr.getTime());
+} catch (ParseException ex) {
+  dateFormatter.setLenient(false);
+  LOGGER.info("Changing setLenient back to false");
+  throw new NumberFormatException(ex.getMessage());
+}
+  } else {
+throw new NumberFormatException(e.getMessage());
+  }
+}
+  }
+
+  private static Long validateTimeStampRange(Long timeValue) {
+long minValue = DateDirectDictionaryGenerator.MIN_VALUE;
+long maxValue = DateDirectDictionaryGenerator.MAX_VALUE;
+if (timeValue < minValue || timeValue > maxValue) {
+  if (LOGGER.isDebugEnabled()) {

Review comment:
   here debug log is not required, because always the exception is thrown 
here.

##
File path: