kscience/dataforge-core
8
0
Fork 0
Code Issues Pull Requests 1 Packages Projects Releases 12 Wiki Activity

Text tables test working

Browse Source
This commit is contained in:
Alexander Nozik 2020-02-16 15:18:09 +03:00
parent cd4f07267b
commit dfea68b65c
15 changed files with 243 additions and 90 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
dataforge-io/src/commonMain/kotlin/hep/dataforge/io/Envelope.kt
View File

@ -28,7 +28,7 @@ interface Envelope {
/**
* Build a static envelope using provided builder
*/
operator fun invoke(block: EnvelopeBuilder.() -> Unit) = EnvelopeBuilder().apply(block).build()
inline operator fun invoke(block: EnvelopeBuilder.() -> Unit) = EnvelopeBuilder().apply(block).build()
}
}

18
dataforge-io/src/commonMain/kotlin/hep/dataforge/io/EnvelopeBuilder.kt
View File

@ -18,8 +18,15 @@ class EnvelopeBuilder {
metaBuilder.update(meta)
}
/**
* The general purpose of the envelope
*/
var type by metaBuilder.string(key = Envelope.ENVELOPE_TYPE_KEY)
var dataType by metaBuilder.string(key = Envelope.ENVELOPE_DATA_TYPE_KEY)
/**
* Data unique identifier to bypass identity checks
*/
var dataID by metaBuilder.string(key = Envelope.ENVELOPE_DATA_ID_KEY)
var description by metaBuilder.string(key = Envelope.ENVELOPE_DESCRIPTION_KEY)
var name by metaBuilder.string(key = Envelope.ENVELOPE_NAME_KEY)
@ -32,5 +39,14 @@ class EnvelopeBuilder {
data = ArrayBinary.write(builder = block)
}
internal fun build() = SimpleEnvelope(metaBuilder.seal(), data)
fun build() = SimpleEnvelope(metaBuilder.seal(), data)
}
//@ExperimentalContracts
//suspend fun EnvelopeBuilder.buildData(block: suspend Output.() -> Unit): Binary{
// contract {
// callsInPlace(block, InvocationKind.EXACTLY_ONCE)
// }
// val scope = CoroutineScope(coroutineContext)
//}

6
dataforge-meta/src/commonMain/kotlin/hep/dataforge/meta/MutableMeta.kt
View File

@ -115,6 +115,8 @@ operator fun MutableMeta<*>.set(name: NameToken, value: Any?) = set(name.asName(
operator fun MutableMeta<*>.set(key: String, value: Any?) = set(key.toName(), value)
operator fun MutableMeta<*>.set(key: String, index: String, value: Any?) = set(key.toName().withIndex(index), value)
/**
* Update existing mutable node with another node. The rules are following:
* * value replaces anything
@ -161,7 +163,7 @@ operator fun MutableMeta<*>.set(name: String, metas: Iterable<Meta>): Unit = set
/**
* Append the node with a same-name-sibling, automatically generating numerical index
*/
fun <M: MutableMeta<M>> M.append(name: Name, value: Any?) {
fun <M : MutableMeta<M>> M.append(name: Name, value: Any?) {
require(!name.isEmpty()) { "Name could not be empty for append operation" }
val newIndex = name.last()!!.index
if (newIndex.isNotEmpty()) {
@ -172,4 +174,4 @@ fun <M: MutableMeta<M>> M.append(name: Name, value: Any?) {
}
}
fun <M: MutableMeta<M>> M.append(name: String, value: Any?) = append(name.toName(), value)
fun <M : MutableMeta<M>> M.append(name: String, value: Any?) = append(name.toName(), value)

36
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/ColumnHeader.kt Normal file
View File

@ -0,0 +1,36 @@
package hep.dataforge.tables
import hep.dataforge.meta.Meta
import hep.dataforge.meta.get
import hep.dataforge.meta.int
import hep.dataforge.meta.string
import hep.dataforge.values.Value
import hep.dataforge.values.ValueType
import kotlin.reflect.KClass
typealias TableHeader<C> = List<ColumnHeader<C>>
typealias ValueTableHeader = List<ColumnHeader<Value>>
interface ColumnHeader<out T : Any> {
val name: String
val type: KClass<out T>
val meta: Meta
}
data class SimpleColumnHeader<T : Any>(
override val name: String,
override val type: KClass<out T>,
override val meta: Meta
) : ColumnHeader<T>
val ColumnHeader<Value>.valueType: ValueType? get() = meta["valueType"].string?.let { ValueType.valueOf(it) }
val ColumnHeader<Value>.textWidth: Int
get() = meta["columnWidth"].int ?: when (valueType) {
ValueType.NUMBER -> 8
ValueType.STRING -> 16
ValueType.BOOLEAN -> 5
ValueType.NULL -> 5
null -> 16
}

12
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/ColumnTable.kt
View File

@ -12,16 +12,20 @@ class ColumnTable<C : Any>(override val columns: Collection<Column<C>>) : Table<
require(columns.all { it.size == rowsNum }) { "All columns must be of the same size" }
}
override val rows: List<Row>
override val rows: List<Row<C>>
get() = (0 until rowsNum).map { VirtualRow(this, it) }
override fun <T : Any> getValue(row: Int, column: String, type: KClass<out T>): T? {
override fun <T : C> getValue(row: Int, column: String, type: KClass<out T>): T? {
val value = columns[column]?.get(row)
return type.cast(value)
}
}
internal class VirtualRow<C : Any>(val table: Table<C>, val index: Int) : Row {
override fun <T : Any> getValue(column: String, type: KClass<out T>): T? = table.getValue(index, column, type)
internal class VirtualRow<C : Any>(val table: Table<C>, val index: Int) : Row<C> {
override fun <T : C> getValue(column: String, type: KClass<out T>): T? = table.getValue(index, column, type)
// override fun <T : C> get(columnHeader: ColumnHeader<T>): T? {
// return table.co[columnHeader][index]
// }
}

4
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/MapRow.kt
View File

@ -2,8 +2,8 @@ package hep.dataforge.tables
import kotlin.reflect.KClass
inline class MapRow(val values: Map<String, Any?>) : Row {
override fun <T : Any> getValue(column: String, type: KClass<out T>): T? {
inline class MapRow<C: Any>(val values: Map<String, C?>) : Row<C> {
override fun <T : C> getValue(column: String, type: KClass<out T>): T? {
val value = values[column]
return type.cast(value)
}

4
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/MutableColumnTable.kt
View File

@ -10,11 +10,11 @@ class MutableColumnTable<C: Any>(val size: Int) : Table<C> {
private val _columns = ArrayList<Column<C>>()
override val columns: List<Column<C>> get() = _columns
override val rows: List<Row> get() = (0 until size).map {
override val rows: List<Row<C>> get() = (0 until size).map {
VirtualRow(this, it)
}
override fun <T : Any> getValue(row: Int, column: String, type: KClass<out T>): T? {
override fun <T : C> getValue(row: Int, column: String, type: KClass<out T>): T? {
val value = columns[column]?.get(row)
return type.cast(value)
}

25
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/MutableTable.kt
View File

@ -1,18 +1,14 @@
package hep.dataforge.tables
import hep.dataforge.meta.Meta
import hep.dataforge.values.Value
import kotlin.reflect.KClass
class SimpleColumnHeader<T : Any>(
override val name: String,
override val type: KClass<out T>,
override val meta: Meta
) : ColumnHeader<T>
class MutableTable<C : Any>(
override val rows: MutableList<Row>,
override val rows: MutableList<Row<C>>,
override val header: MutableList<ColumnHeader<C>>
) : RowTable<C, Row>(rows, header) {
) : RowTable<C>(rows, header) {
fun <T : C> column(name: String, type: KClass<out T>, meta: Meta): ColumnHeader<T> {
val column = SimpleColumnHeader(name, type, meta)
header.add(column)
@ -21,23 +17,24 @@ class MutableTable<C : Any>(
inline fun <reified T : C> column(
name: String,
noinline columnMetaBuilder: ColumnScheme.() -> Unit
noinline columnMetaBuilder: ColumnScheme.() -> Unit = {}
): ColumnHeader<T> {
return column(name, T::class, ColumnScheme(columnMetaBuilder).toMeta())
}
fun row(block: MutableMap<String, Any?>.() -> Unit): Row {
val map = HashMap<String, Any?>().apply(block)
fun row(map: Map<String, C?>): Row<C> {
val row = MapRow(map)
rows.add(row)
return row
}
operator fun <T : Any> MutableMap<String, Any?>.set(header: ColumnHeader<T>, value: T?) {
set(header.name, value)
}
fun <T : C> row(vararg pairs: Pair<ColumnHeader<T>, T>): Row<C> =
row(pairs.associate { it.first.name to it.second })
}
fun MutableTable<Value>.row(vararg pairs: Pair<ColumnHeader<Value>, Any?>): Row<Value> =
row(pairs.associate { it.first.name to Value.of(it.second) })
fun <C : Any> Table<C>.edit(block: MutableTable<C>.() -> Unit): Table<C> {
return MutableTable(rows.toMutableList(), header.toMutableList()).apply(block)
}

7
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/RowTable.kt
View File

@ -13,12 +13,11 @@ internal class RowTableColumn<C : Any, T : C>(val table: Table<C>, val header: C
override fun get(index: Int): T? = table.rows[index].getValue(name, type)
}
open class RowTable<C : Any, R : Row>(override val rows: List<R>, override val header: List<ColumnHeader<C>>) :
Table<C> {
override fun <T : Any> getValue(row: Int, column: String, type: KClass<out T>): T? =
open class RowTable<C : Any>(override val rows: List<Row<C>>, override val header: List<ColumnHeader<C>>) : Table<C> {
override fun <T : C> getValue(row: Int, column: String, type: KClass<out T>): T? =
rows[row].getValue(column, type)
override val columns: List<Column<C>> get() = header.map { RowTableColumn(this, it) }
}
suspend fun Rows.collect(): Table<*> = this as? Table<*> ?: RowTable(rowFlow().toList(), header)
suspend fun <C : Any> Rows<C>.collect(): Table<C> = this as? Table<C> ?: RowTable(rowFlow().toList(), header)

37
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/Table.kt
View File

@ -1,6 +1,5 @@
package hep.dataforge.tables
import hep.dataforge.meta.Meta
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.asFlow
import kotlin.reflect.KClass
@ -15,28 +14,30 @@ internal fun <T : Any> KClass<T>.cast(value: Any?): T? {
}
}
typealias TableHeader<C> = List<ColumnHeader<C>>
/**
* Finite or infinite row set. Rows are produced in a lazy suspendable [Flow].
* Each row must contain at least all the fields mentioned in [header].
*/
interface Rows {
val header: TableHeader<*>
fun rowFlow(): Flow<Row>
interface Rows<C : Any> {
val header: TableHeader<C>
fun rowFlow(): Flow<Row<C>>
}
interface Table<out C : Any> : Rows {
fun <T : Any> getValue(row: Int, column: String, type: KClass<out T>): T?
interface Table<C : Any> : Rows<C> {
fun <T : C> getValue(row: Int, column: String, type: KClass<out T>): T?
val columns: Collection<Column<C>>
override val header: TableHeader<C> get() = columns.toList()
val rows: List<Row>
override fun rowFlow(): Flow<Row> = rows.asFlow()
val rows: List<Row<C>>
override fun rowFlow(): Flow<Row<C>> = rows.asFlow()
/**
* Apply typed query to this table and return lazy [Flow] of resulting rows. The flow could be empty.
*/
//fun select(query: Any): Flow<Row> = error("Query of type ${query::class} is not supported by this table")
companion object {
inline operator fun <T : Any> invoke(block: MutableTable<T>.() -> Unit): Table<T> =
MutableTable<T>(arrayListOf(), arrayListOf()).apply(block)
}
}
operator fun Collection<Column<*>>.get(name: String): Column<*>? = find { it.name == name }
@ -44,15 +45,9 @@ operator fun Collection<Column<*>>.get(name: String): Column<*>? = find { it.nam
inline operator fun <C : Any, reified T : C> Table<C>.get(row: Int, column: String): T? =
getValue(row, column, T::class)
interface ColumnHeader<out T : Any> {
val name: String
val type: KClass<out T>
val meta: Meta
}
operator fun <C : Any, T : C> Table<C>.get(row: Int, column: Column<T>): T? = getValue(row, column.name, column.type)
interface Column<out T : Any> : ColumnHeader<T> {
interface Column<T : Any> : ColumnHeader<T> {
val size: Int
operator fun get(index: Int): T?
}
@ -65,9 +60,9 @@ operator fun <T : Any> Column<T>.iterator() = iterator {
}
}
interface Row {
fun <T : Any> getValue(column: String, type: KClass<out T>): T?
interface Row<C: Any> {
fun <T : C> getValue(column: String, type: KClass<out T>): T?
}
inline operator fun <reified T : Any> Row.get(column: String): T? = getValue(column, T::class)
operator fun <T : Any> Row.get(column: ColumnHeader<T>): T? = getValue(column.name, column.type)
inline operator fun <C : Any, reified T : C> Row<C>.get(column: String): T? = getValue(column, T::class)
operator fun <C : Any, T : C> Row<C>.get(column: ColumnHeader<T>): T? = getValue(column.name, column.type)

87
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/io/TextRows.kt
View File

@ -1,38 +1,56 @@
package hep.dataforge.tables.io
import hep.dataforge.meta.get
import hep.dataforge.meta.int
import hep.dataforge.meta.string
import hep.dataforge.tables.*
import hep.dataforge.values.*
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.collect
import kotlinx.coroutines.flow.flow
import kotlinx.coroutines.flow.toList
import kotlinx.io.Binary
import kotlinx.io.ExperimentalIoApi
import kotlinx.io.Output
import kotlinx.io.RandomAccessBinary
import kotlinx.io.text.forEachUtf8Line
import kotlinx.io.text.readUtf8Line
import kotlinx.io.text.readUtf8StringUntilDelimiter
import kotlinx.io.text.writeUtf8String
import kotlin.reflect.KClass
private fun readLine(header: List<ColumnHeader<Value>>, line: String): Row {
val values = line.split("\\s+".toRegex()).map { it.parseValue() }
/**
* Read a lin as a fixed width [Row]
*/
private fun readLine(header: ValueTableHeader, line: String): Row<Value> {
val values = line.trim().split("\\s+".toRegex()).map { it.lazyParseValue() }
if (values.size == header.size) {
val map = header.map { it.name }.zip(values).toMap()
return MapRow(map)
} else {
error("Can't read line $line. Expected ${header.size} values in a line, but found ${values.size}")
error("Can't read line \"$line\". Expected ${header.size} values in a line, but found ${values.size}")
}
}
/**
* Finite or infinite [Rows] created from a fixed width text binary
*/
@ExperimentalIoApi
class TextRows(override val header: List<ColumnHeader<Value>>, val binary: Binary) : Rows {
class TextRows(override val header: ValueTableHeader, val binary: Binary) : Rows<Value> {
override fun rowFlow(): Flow<Row> = binary.read {
/**
* A flow of indexes of string start offsets ignoring empty strings
*/
fun indexFlow(): Flow<Int> = binary.read {
var counter: Int = 0
flow {
val string = readUtf8StringUntilDelimiter('\n')
counter += string.length
if (!string.isBlank()) {
emit(counter)
}
}
}
override fun rowFlow(): Flow<Row<Value>> = binary.read {
flow {
forEachUtf8Line { line ->
if (line.isNotBlank()) {
@ -42,35 +60,57 @@ class TextRows(override val header: List<ColumnHeader<Value>>, val binary: Binar
}
}
}
companion object
}
/**
* Create a row offset index for [TextRows]
*/
@ExperimentalIoApi
suspend fun TextRows.buildRowIndex(): List<Int> = indexFlow().toList()
/**
* Finite table created from [RandomAccessBinary] with fixed width text table
*/
@ExperimentalIoApi
class TextTable(
override val header: List<ColumnHeader<Value>>,
override val header: ValueTableHeader,
val binary: RandomAccessBinary,
val index: List<Int>
) : Table<Value> {
override val columns: Collection<Column<Value>> get() = header.map { RowTableColumn(this, it) }
override val rows: List<Row> get() = index.map { readAt(it) }
override val rows: List<Row<Value>> get() = index.map { readAt(it) }
override fun rowFlow(): Flow<Row> = TextRows(header, binary).rowFlow()
override fun rowFlow(): Flow<Row<Value>> = TextRows(header, binary).rowFlow()
private fun readAt(offset: Int): Row {
private fun readAt(offset: Int): Row<Value> {
return binary.read(offset) {
val line = readUtf8Line()
return@read readLine(header, line)
}
}
override fun <T : Any> getValue(row: Int, column: String, type: KClass<out T>): T? {
override fun <T : Value> getValue(row: Int, column: String, type: KClass<out T>): T? {
val offset = index[row]
return type.cast(readAt(offset)[column])
}
companion object {
suspend operator fun invoke(header: ValueTableHeader, binary: RandomAccessBinary): TextTable {
val index = TextRows(header, binary).buildRowIndex()
return TextTable(header, binary, index)
}
}
}
fun Output.writeValue(value: Value, width: Int, left: Boolean = true) {
/**
* Write a fixed width value to the output
*/
private fun Output.writeValue(value: Value, width: Int, left: Boolean = true) {
require(width > 5) { "Width could not be less than 5" }
val str: String = when (value.type) {
ValueType.NUMBER -> value.number.toString() //TODO apply decimal format
@ -90,31 +130,20 @@ fun Output.writeValue(value: Value, width: Int, left: Boolean = true) {
writeUtf8String(padded)
}
val ColumnHeader<Value>.valueType: ValueType? get() = meta["valueType"].string?.let { ValueType.valueOf(it) }
private val ColumnHeader<Value>.width: Int
get() = meta["columnWidth"].int ?: when (valueType) {
ValueType.NUMBER -> 8
ValueType.STRING -> 16
ValueType.BOOLEAN -> 5
ValueType.NULL -> 5
null -> 16
}
/**
* Write rows without header to the output
*/
suspend fun Output.writeRows(rows: Rows) {
suspend fun Output.writeRows(rows: Rows<Value>) {
@Suppress("UNCHECKED_CAST") val header = rows.header.map {
if (it.type != Value::class) error("Expected Value column, but found ${it.type}") else (it as ColumnHeader<Value>)
}
val widths: List<Int> = header.map {
it.width
it.textWidth
}
rows.rowFlow().collect { row ->
header.forEachIndexed { index, columnHeader ->
writeValue(row[columnHeader] ?: Null, widths[index])
}
writeUtf8String("\r\n")
}
}

42
dataforge-tables/src/commonMain/kotlin/hep/dataforge/tables/io/textTableEnvelope.kt Normal file
View File

@ -0,0 +1,42 @@
package hep.dataforge.tables.io
import hep.dataforge.io.Envelope
import hep.dataforge.meta.*
import hep.dataforge.tables.SimpleColumnHeader
import hep.dataforge.tables.Table
import hep.dataforge.values.Value
import kotlinx.io.ByteArrayOutput
import kotlinx.io.EmptyBinary
import kotlinx.io.ExperimentalIoApi
import kotlinx.io.asBinary
@ExperimentalIoApi
suspend fun Table<Value>.wrap(): Envelope = Envelope {
meta {
header.forEachIndexed { index, columnHeader ->
set("column", index.toString(), buildMeta {
"name" put columnHeader.name
if (!columnHeader.meta.isEmpty()) {
"meta" put columnHeader.meta
}
})
}
}
type = "table.value"
dataID = "valueTable[${this@wrap.hashCode()}]"
data = ByteArrayOutput().apply { writeRows(this@wrap) }.toByteArray().asBinary()
}
@DFExperimental
@ExperimentalIoApi
fun TextRows.Companion.readEnvelope(envelope: Envelope): TextRows {
val header = envelope.meta.getIndexed("column")
.entries.sortedBy { it.key.toInt() }
.map { (_, item) ->
SimpleColumnHeader(item.node["name"].string!!, Value::class, item.node["meta"].node ?: Meta.EMPTY)
}
return TextRows(header, envelope.data ?: EmptyBinary)
}

6
dataforge-tables/src/commonTest/kotlin/hep/dataforge/tables/io/TextRowsTest.kt
View File

@ -1,6 +0,0 @@
package hep.dataforge.tables.io
class TextRowsTest{
}

6
dataforge-tables/src/jvmMain/kotlin/hep/dataforge/tables/CastColumn.kt
View File

@ -8,7 +8,7 @@ import kotlin.reflect.full.cast
import kotlin.reflect.full.isSubclassOf
@Suppress("UNCHECKED_CAST")
fun <T : Any> Column<*>.cast(type: KClass<T>): Column<T> {
fun <T : Any> Column<*>.cast(type: KClass<out T>): Column<T> {
return if (type.isSubclassOf(this.type)) {
this as Column<T>
} else {
@ -16,7 +16,7 @@ fun <T : Any> Column<*>.cast(type: KClass<T>): Column<T> {
}
}
class CastColumn<T : Any>(val origin: Column<*>, override val type: KClass<T>) : Column<T> {
class CastColumn<T : Any>(val origin: Column<*>, override val type: KClass<out T>) : Column<T> {
override val name: String get() = origin.name
override val meta: Meta get() = origin.meta
override val size: Int get() = origin.size
@ -32,5 +32,5 @@ class ColumnProperty<C: Any, T : C>(val table: Table<C>, val type: KClass<T>) :
}
}
operator fun <T : Any> Collection<Column<*>>.get(header: ColumnHeader<T>): Column<T>? =
operator fun <C: Any, T : C> Collection<Column<C>>.get(header: ColumnHeader<T>): Column<T>? =
find { it.name == header.name }?.cast(header.type)

39
dataforge-tables/src/jvmTest/kotlin/hep/dataforge/tables/io/TextRowsTest.kt Normal file
View File

@ -0,0 +1,39 @@
package hep.dataforge.tables.io
import hep.dataforge.meta.DFExperimental
import hep.dataforge.tables.Table
import hep.dataforge.tables.get
import hep.dataforge.tables.row
import hep.dataforge.values.Value
import hep.dataforge.values.int
import kotlinx.coroutines.flow.toList
import kotlinx.coroutines.runBlocking
import kotlinx.io.ExperimentalIoApi
import kotlinx.io.toByteArray
import kotlin.test.Test
import kotlin.test.assertEquals
@DFExperimental
@ExperimentalIoApi
class TextRowsTest {
val table = Table<Value> {
val a = column<Value>("a")
val b = column<Value>("b")
row(a to 1, b to "b1")
row(a to 2, b to "b2")
}
@Test
fun testTableWriteRead() {
runBlocking {
val envelope = table.wrap()
val string = envelope.data!!.toByteArray().decodeToString()
println(string)
val table = TextRows.readEnvelope(envelope)
val rows = table.rowFlow().toList()
assertEquals(1, rows[0]["a"]?.int)
assertEquals("b2", rows[1]["b"]?.string)
}
}
}