[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user asfgit closed the pull request at: https://github.com/apache/spark/pull/17142 --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104103858
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
+ //
+ // with the input tokens,
+ //
+ // input tokens - [1, 2, "A"]
+ //
+ // Each input token is placed in each output row's position by mapping
these. In this case,
+ //
+ // output row - ["A", 2, null, 1]
+ //
+ // In more details,
+ // - `valueConverters`, input tokens - CSV data schema
+ // `valueConverters` keeps the positions of input token indices (by
its index) to each
+ // value's converter (by its value) in an order of CSV data schema. In
this case,
+ // [string->int, string->int, string->string].
+ //
+ // - `tokenIndexArr`, input tokens - required CSV data schema
+ // `tokenIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // fields given the required CSV data schema (by its value). In this
case, [2, 1, 0].
+ //
+ // - `rowIndexArr`, input tokens - required schema
+ // `rowIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // field indices given the required schema (by its value). In this
case, [0, 1, 3].
+ private val valueConverters: Array[ValueConverter] =
+dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
+
+ // Only used to create both `tokenIndexArr` and `rowIndexArr`. This
variable means
+ // the fields that we should try to convert.
+ private val reorderedFields = if (options.dropMalformed) {
+// If `dropMalformed` is enabled, then it needs to parse all the values
+// so that we can decide which row is malformed.
+requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
+ } else {
+requiredSchema
+ }
+
+ private val tokenIndexArr: Array[Int] = {
+reorderedFields
+ .filter(_.name != options.columnNameOfCorruptRecord)
+ .map(f => dataSchema.indexOf(f)).toArray
+ }
+
+ private val rowIndexArr: Array[Int] = if (corruptFieldIndex.isDefined) {
+val corrFieldIndex = corruptFieldIndex.get
+reorderedFields.indices.filter(_ !=
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104105409
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
--- End diff --
ISTM it'd be better to map the names into these variables here,
`reuiqredSchema` and `dataSchema`?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104107046
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
+ //
+ // with the input tokens,
+ //
+ // input tokens - [1, 2, "A"]
+ //
+ // Each input token is placed in each output row's position by mapping
these. In this case,
+ //
+ // output row - ["A", 2, null, 1]
+ //
+ // In more details,
+ // - `valueConverters`, input tokens - CSV data schema
+ // `valueConverters` keeps the positions of input token indices (by
its index) to each
+ // value's converter (by its value) in an order of CSV data schema. In
this case,
+ // [string->int, string->int, string->string].
+ //
+ // - `tokenIndexArr`, input tokens - required CSV data schema
+ // `tokenIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // fields given the required CSV data schema (by its value). In this
case, [2, 1, 0].
+ //
+ // - `rowIndexArr`, input tokens - required schema
--- End diff --
ditto
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104106202
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
+ //
+ // with the input tokens,
+ //
+ // input tokens - [1, 2, "A"]
+ //
+ // Each input token is placed in each output row's position by mapping
these. In this case,
+ //
+ // output row - ["A", 2, null, 1]
+ //
+ // In more details,
+ // - `valueConverters`, input tokens - CSV data schema
+ // `valueConverters` keeps the positions of input token indices (by
its index) to each
+ // value's converter (by its value) in an order of CSV data schema. In
this case,
+ // [string->int, string->int, string->string].
+ //
+ // - `tokenIndexArr`, input tokens - required CSV data schema
+ // `tokenIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // fields given the required CSV data schema (by its value). In this
case, [2, 1, 0].
+ //
+ // - `rowIndexArr`, input tokens - required schema
+ // `rowIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // field indices given the required schema (by its value). In this
case, [0, 1, 3].
+ private val valueConverters: Array[ValueConverter] =
+dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
+
+ // Only used to create both `tokenIndexArr` and `rowIndexArr`. This
variable means
+ // the fields that we should try to convert.
+ private val reorderedFields = if (options.dropMalformed) {
--- End diff --
`requiredFields` is better?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104105767
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
--- End diff --
I feel "required CSV data schema" is a little ambiguous because there is no
schema variable along this name in this class. So, it seems we need to describe
more?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104107028
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
+ //
+ // with the input tokens,
+ //
+ // input tokens - [1, 2, "A"]
+ //
+ // Each input token is placed in each output row's position by mapping
these. In this case,
+ //
+ // output row - ["A", 2, null, 1]
+ //
+ // In more details,
+ // - `valueConverters`, input tokens - CSV data schema
+ // `valueConverters` keeps the positions of input token indices (by
its index) to each
+ // value's converter (by its value) in an order of CSV data schema. In
this case,
+ // [string->int, string->int, string->string].
+ //
+ // - `tokenIndexArr`, input tokens - required CSV data schema
--- End diff --
ditto; `tokenIndexArr` is an index array in tokens corresponding to
`requiredSchema`?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104104893
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
--- End diff --
How about "_c0,_c1,_c2" in the header line? I couldn't first tell this is a
header.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104104154
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
+ // This parser loads an `tokenIndexArr`-th position value in input
tokens,
+ // then put the value in `row(rowIndexArr)`.
+ //
+ // For example, let's say there is CSV data as below:
+ //
+ // a,b,c
+ // 1,2,A
+ //
+ // Also, let's say `columnNameOfCorruptRecord` is set to "_unparsed",
`header` is `true`
+ // by user and the user selects "c", "b", "_unparsed" and "a" fields. In
this case, we need
+ // to map those values below:
+ //
+ // required schema - ["c", "b", "_unparsed", "a"]
+ // CSV data schema - ["a", "b", "c"]
+ // required CSV data schema - ["c", "b", "a"]
+ //
+ // with the input tokens,
+ //
+ // input tokens - [1, 2, "A"]
+ //
+ // Each input token is placed in each output row's position by mapping
these. In this case,
+ //
+ // output row - ["A", 2, null, 1]
+ //
+ // In more details,
+ // - `valueConverters`, input tokens - CSV data schema
+ // `valueConverters` keeps the positions of input token indices (by
its index) to each
+ // value's converter (by its value) in an order of CSV data schema. In
this case,
+ // [string->int, string->int, string->string].
+ //
+ // - `tokenIndexArr`, input tokens - required CSV data schema
+ // `tokenIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // fields given the required CSV data schema (by its value). In this
case, [2, 1, 0].
+ //
+ // - `rowIndexArr`, input tokens - required schema
+ // `rowIndexArr` keeps the positions of input token indices (by its
index) to reordered
+ // field indices given the required schema (by its value). In this
case, [0, 1, 3].
+ private val valueConverters: Array[ValueConverter] =
+dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
+
+ // Only used to create both `tokenIndexArr` and `rowIndexArr`. This
variable means
+ // the fields that we should try to convert.
+ private val reorderedFields = if (options.dropMalformed) {
+// If `dropMalformed` is enabled, then it needs to parse all the values
+// so that we can decide which row is malformed.
+requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
+ } else {
+requiredSchema
+ }
+
+ private val tokenIndexArr: Array[Int] = {
--- End diff --
How about `fromIndexInTokens` instead of `tokenIndexArr ` for
self-describing more?
Along with his, `rowIndexArr` to `toIndexInRow`?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user HyukjinKwon commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104063545
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
--- End diff --
It is just a little bit for codes.. actually :). I hope the comment makes
reading this code easier and not look too verbose.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user maropu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r104062715
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
--- End diff --
Thanks for cc'ing and brushing-up code ;) I'll check in hours
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
Github user HyukjinKwon commented on a diff in the pull request:
https://github.com/apache/spark/pull/17142#discussion_r103986635
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/csv/UnivocityParser.scala
---
@@ -54,39 +54,77 @@ private[csv] class UnivocityParser(
private val dataSchema = StructType(schema.filter(_.name !=
options.columnNameOfCorruptRecord))
- private val valueConverters =
-dataSchema.map(f => makeConverter(f.name, f.dataType, f.nullable,
options)).toArray
-
private val tokenizer = new CsvParser(options.asParserSettings)
private var numMalformedRecords = 0
private val row = new GenericInternalRow(requiredSchema.length)
- // This gets the raw input that is parsed lately.
+ // In `PERMISSIVE` parse mode, we should be able to put the raw
malformed row into the field
+ // specified in `columnNameOfCorruptRecord`. The raw input is retrieved
by this method.
private def getCurrentInput(): String =
tokenizer.getContext.currentParsedContent().stripLineEnd
- // This parser loads an `indexArr._1`-th position value in input tokens,
- // then put the value in `row(indexArr._2)`.
- private val indexArr: Array[(Int, Int)] = {
-val fields = if (options.dropMalformed) {
- // If `dropMalformed` is enabled, then it needs to parse all the
values
- // so that we can decide which row is malformed.
- requiredSchema ++ schema.filterNot(requiredSchema.contains(_))
-} else {
- requiredSchema
-}
-// TODO: Revisit this; we need to clean up code here for readability.
-// See an URL below for related discussions:
-// https://github.com/apache/spark/pull/16928#discussion_r102636720
-val fieldsWithIndexes = fields.zipWithIndex
-corruptFieldIndex.map { case corrFieldIndex =>
- fieldsWithIndexes.filter { case (_, i) => i != corrFieldIndex }
-}.getOrElse {
- fieldsWithIndexes
-}.map { case (f, i) =>
- (dataSchema.indexOf(f), i)
-}.toArray
--- End diff --
cc @maropu and @cloud-fan, could you check if this comment look nicer to
you?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17142: [SPARK-18699][SQL][FOLLOWUP] Add explanation in C...
GitHub user HyukjinKwon opened a pull request: https://github.com/apache/spark/pull/17142 [SPARK-18699][SQL][FOLLOWUP] Add explanation in CSV parser and minor cleanup ## What changes were proposed in this pull request? This PR suggests adding some comments in `UnivocityParser` logics to explain what happens. Also, it proposes, IMHO, a little bit cleaner (at least easy for me to explain). ## How was this patch tested? Unit tests in `CSVSuite`. You can merge this pull request into a Git repository by running: $ git pull https://github.com/HyukjinKwon/spark SPARK-18699 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/spark/pull/17142.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #17142 --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
