rewriting IDF stuff

develop
Timo Bryant 2023-12-17 17:46:51 +01:00
parent ca51b50306
commit d995b26459
18 changed files with 314 additions and 39 deletions

View File

@ -6,10 +6,12 @@ import com.github.ajalt.clikt.parameters.options.option
import com.github.ajalt.clikt.parameters.options.required
import com.github.ajalt.clikt.parameters.types.enum
import com.github.ajalt.clikt.parameters.types.file
import de.itkl.textprocessing.TextFile
import de.itkl.tfidf.Language
import de.itkl.tfidf.TfIdf
import de.itkl.tfidf.TfIdfPipeline
import kotlinx.coroutines.flow.take
import kotlinx.coroutines.runBlocking
import java.io.File
class ComputeTf : CliktCommand() {
private val corpus by option(help = "corpus")
@ -20,13 +22,17 @@ class ComputeTf : CliktCommand() {
.required()
override fun run() = runBlocking {
val tfIdf = TfIdf()
val histogram = tfIdf.computeTf(
corpus,
language
)
tfIdf.normalizeTf(histogram, corpus.toPath().parent.resolve("${corpus.nameWithoutExtension}-tf.csv").toFile())
TfIdfPipeline(language = Language.DE)
.input(corpus)
// TextFile(corpus).splitByEmptyLines()
// .take(10)
// .collect { println(it) }
// val tfIdf = TfIdf()
// val histogram = tfIdf.computeTf(
// corpus,
// language
// )
// val tf = tfIdf.normalizeTf(histogram, corpus.toPath().parent.resolve("${corpus.nameWithoutExtension}-tf.csv").toFile())
}
}

View File

@ -0,0 +1,7 @@
plugins {
id("docthor.kotlin-library-conventions")
}
dependencies {
implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core:1.7.3")
}

View File

@ -0,0 +1,10 @@
package de.itkl.fileprocessing
import java.io.File
import java.io.InputStream
import java.nio.file.Path
interface FileProcessor {
fun willProduce(path: Path): Path
suspend fun process(resource: Resource): File
}

View File

@ -0,0 +1,25 @@
package de.itkl.fileprocessing
import io.github.oshai.kotlinlogging.KotlinLogging
import java.io.File
import kotlin.io.path.exists
private val Log = KotlinLogging.logger { }
abstract class FileProcessingPipeline {
protected abstract val fileProcessor: List<FileProcessor>
suspend fun input(file: File) {
var currentFile = file
fileProcessor.forEach { processor ->
val target = processor.willProduce(currentFile.toPath())
if(target.exists()) {
Log.info { "$target exists. Skipping" }
} else {
Log.info { "$target does not exists. Creating" }
processor.process(FileResource(currentFile))
}
currentFile = target.toFile()
}
}
}

View File

@ -0,0 +1,39 @@
package de.itkl.fileprocessing
import java.io.InputStream
/**
* Represents an input stream that tracks the progress of reading from an underlying input stream.
*
* @property inputStream The underlying input stream to read from.
* @property updateOp The operation to be executed when the number of bytes read changes.
* @property bytesRead The number of bytes read from the input stream.
*/
class ProgressInputStream(
private val inputStream: InputStream,
private val updateOp: (Long) -> Unit) : InputStream() {
@Volatile
var bytesRead: Long = 0
private set(value) {
field = value
updateOp(value)
}
override fun read(): Int {
val byte = inputStream.read()
if (byte != -1) {
bytesRead++
}
return byte
}
override fun read(b: ByteArray, off: Int, len: Int): Int {
val bytesRead = inputStream.read(b, off, len)
if (bytesRead != -1) {
this.bytesRead += bytesRead
}
return bytesRead
}
override fun read(b: ByteArray): Int {
return this.read(b, 0, b.size)
}
}

View File

@ -0,0 +1,34 @@
package de.itkl.fileprocessing
import java.io.File
import java.io.InputStream
import java.nio.file.Files
import java.nio.file.Path
interface Resource {
val path: Path
val size: Long
fun toFile(): File = path.toFile()
fun read(): InputStream
}
class ProgressResource(
private val resource: Resource,
private val progressOpSupplier: () -> (Long) -> Unit
) : Resource by resource
{
override fun read(): InputStream {
return ProgressInputStream(
read(),
progressOpSupplier()
)
}
}
class FileResource(override val path: Path) : Resource {
constructor(file: File): this(file.toPath())
override val size: Long by lazy { path.toFile().length() }
override fun read(): InputStream {
return Files.newInputStream(path)
}
}

View File

@ -0,0 +1,25 @@
package de.itkl.processing
import kotlinx.coroutines.async
import kotlinx.coroutines.channels.Channel
import kotlinx.coroutines.channels.consumeEach
import kotlinx.coroutines.coroutineScope
import kotlinx.coroutines.flow.*
import kotlinx.coroutines.launch
import kotlinx.coroutines.sync.Semaphore
import kotlinx.coroutines.sync.withPermit
import kotlin.coroutines.coroutineContext
class ParallelFlowProcessor<T,U>(
private val mapperFn: (T) -> U,
private val concurrencyLimit: Int) {
suspend fun process(flow: Flow<T>): Flow<U> = coroutineScope {
val gate = Semaphore(concurrencyLimit)
flow.map { item ->
async {
gate.withPermit { mapperFn(item) }
}
}
.map { it.await() }
}
}

View File

@ -0,0 +1,33 @@
package de.itkl.textprocessing
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.toSet
class BagOfWords(private val data: MutableSet<String> = mutableSetOf()) : Iterable<String> {
companion object {
suspend fun from(flow: Flow<String>): BagOfWords {
return BagOfWords(flow.toSet().toMutableSet())
}
}
fun add(word: String) {
data.add(word)
}
fun join(bagOfWords: BagOfWords): BagOfWords {
return BagOfWords(data.toMutableSet().apply { addAll(bagOfWords.data) })
}
override fun iterator(): Iterator<String> {
return iterator {
data.forEach { yield(it) }
}
}
override fun toString(): String {
return data.joinToString(",")
}
}

View File

@ -0,0 +1,4 @@
package de.itkl.textprocessing
class DocumentFrequency {
}

View File

@ -1,6 +1,6 @@
package de.itkl.textprocessing
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.*
class Histogram(private val histo: MutableMap<String,UInt> = mutableMapOf()) : Iterable<Pair<String, UInt>>{
@ -11,6 +11,14 @@ class Histogram(private val histo: MutableMap<String,UInt> = mutableMapOf()) : I
}
}
suspend fun fromBagOfWords(flow: Flow<BagOfWords>): Histogram {
val result = Histogram()
flow.collect { bagOfWords ->
bagOfWords.forEach(result::add)
}
return result
}
fun from(sequence: Sequence<Map<String, String>>): Histogram {
val histo = sequence.associate { map -> map["word"]!! to map["count"]!!.toUInt() }
.toMutableMap()

View File

@ -11,6 +11,23 @@ import java.io.InputStreamReader
class TextFile(val file: File) {
fun splitByEmptyLines(progressOp: (read: Long) -> Unit = {}): Flow<List<String>> {
val reader = InputStreamReader(ProgressInputStream(file.inputStream(), progressOp))
var list = mutableListOf<String>()
return flow {
reader.useLines { lines ->
lines.forEach { line ->
if(line.isEmpty()) {
emit(list)
list = mutableListOf()
} else {
list.add(line)
}
}
}
}
}
fun words(progressOp: (read: Long) -> Unit = {}): Flow<String> {
val factory = AttributeFactory.DEFAULT_ATTRIBUTE_FACTORY
val tokenizer = StandardTokenizer(factory)

View File

@ -0,0 +1,44 @@
package de.itkl.textprocessing
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.flow
import kotlinx.coroutines.flow.onCompletion
import org.apache.lucene.analysis.standard.StandardTokenizer
import org.apache.lucene.analysis.tokenattributes.CharTermAttribute
import org.apache.lucene.util.AttributeFactory
import java.io.StringReader
class Tokenizer : AutoCloseable{
private val tokenizer by lazy {
val factory = AttributeFactory.DEFAULT_ATTRIBUTE_FACTORY
val tokenizer = StandardTokenizer(factory)
tokenizer
// val reader = ProgressInputStream(file.inputStream(), progressOp)
// tokenizer.setReader(InputStreamReader(reader))
// tokenizer.reset()
// val attr = tokenizer.addAttribute(CharTermAttribute::class.java)
// return flow {
// while (kotlin.runCatching { tokenizer.incrementToken() }.getOrElse { true } ) {
// emit(attr.toString())
// }
// }.onCompletion {
// tokenizer.close()
// }
}
fun tokenize(input: String): Flow<String> {
val reader = StringReader(input)
tokenizer.setReader(reader)
tokenizer.reset()
val attr = tokenizer.addAttribute(CharTermAttribute::class.java)
return flow {
while (kotlin.runCatching { tokenizer.incrementToken() }.getOrElse { true }) {
emit(attr.toString())
}
}.onCompletion { tokenizer.close() }
}
override fun close() {
tokenizer.close()
}
}

View File

@ -4,6 +4,7 @@ plugins {
dependencies {
api(project(":libraries:textprocessing"))
api(project(":libraries:fileprocessing"))
implementation("com.github.ajalt.mordant:mordant:2.2.0")
implementation("com.github.doyaaaaaken:kotlin-csv-jvm:1.9.2")
}

View File

@ -0,0 +1,40 @@
package de.itkl.tfidf
import de.itkl.fileprocessing.FileProcessor
import de.itkl.fileprocessing.Resource
import de.itkl.textprocessing.*
import io.github.oshai.kotlinlogging.KotlinLogging
import kotlinx.coroutines.flow.map
import kotlinx.coroutines.flow.reduce
import kotlinx.coroutines.flow.take
import java.io.File
import java.nio.file.Path
import kotlin.io.path.nameWithoutExtension
private val Log = KotlinLogging.logger { }
class Idf : FileProcessor {
override fun willProduce(path: Path): Path {
return path.parent.resolve(path.nameWithoutExtension + "-idf.csv")
}
override suspend fun process(resource: Resource): File {
Log.info { "Would produce: ${willProduce(resource.path)}" }
val resultFile = willProduce(resource.path).toFile()
val textFile = TextFile(resource.toFile())
val bagOfWords = textFile.splitByEmptyLines { }
.take(10)
.map { document ->
val tokenizer = Tokenizer()
document.map { line ->
val tokens = tokenizer.tokenize(line)
BagOfWords.from(tokens)
}
.reduce { acc, bagOfWords -> acc.join(bagOfWords) }
}
val histogram = Histogram.fromBagOfWords(bagOfWords)
HistogramCsvStorage().save(histogram, resultFile)
return resultFile
}
}

View File

@ -1,31 +1,9 @@
package de.itkl.tfidf
import com.github.doyaaaaaken.kotlincsv.dsl.csvReader
import com.github.doyaaaaaken.kotlincsv.dsl.csvWriter
import de.itkl.textprocessing.Histogram
import io.github.oshai.kotlinlogging.KotlinLogging
import java.io.File
import kotlin.math.max
private val Log = KotlinLogging.logger { }
class Tf {
private val data: MutableMap<String, Double> = mutableMapOf()
fun update(histogram: Histogram): Tf {
val max = histogram.maxOf { (_, count) -> count }
.toDouble()
histogram.forEach { (word, count) ->
val tf = count.toDouble() / max
data[word] = tf
}
return this
}
suspend fun saveToCsv(file: File) {
csvWriter {}
.openAsync(file, append = false) {
writeRow("term", "frequency")
data.forEach { (t, u) ->
writeRow(t, u)
}
}
}
}

View File

@ -41,13 +41,6 @@ class TfIdf {
return histogram
}
suspend fun normalizeTf(histogram: Histogram, destination: File) {
Log.info { "Write tf to $destination" }
Tf()
.update(histogram)
.saveToCsv(destination)
}
private fun stemmer(language: Language): SnowballStemmer {
return when(language) {
Language.DE -> GermanStemmer()

View File

@ -0,0 +1,10 @@
package de.itkl.tfidf
import de.itkl.fileprocessing.FileProcessingPipeline
import de.itkl.fileprocessing.FileProcessor
class TfIdfPipeline(private val language: Language) : FileProcessingPipeline() {
override val fileProcessor = listOf(
Idf()
)
}

View File

@ -11,4 +11,5 @@ include(
"app",
"libraries:tfidf",
"libraries:textprocessing",
"libraries:fileprocessing",
)