org.oewntk.sql.out.Synsets.kt Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of tosql Show documentation
Export Wordnet model to SQL files.
The newest version!
/*
 * Copyright (c) 2024. Bernard Bou.
 */
package org.oewntk.sql.out

import org.oewntk.model.Relation
import org.oewntk.model.Synset
import org.oewntk.model.SynsetId
import org.oewntk.sql.out.Printers.printInsert
import org.oewntk.sql.out.Printers.printInsertWithComment
import org.oewntk.sql.out.Printers.printInserts
import org.oewntk.sql.out.Printers.printInsertsWithComment
import org.oewntk.sql.out.Utils.escape
import java.io.PrintStream

/**
 * Process synsets
 */
object Synsets {

    /**
     * Make synset id-to-nid map
     *
     * @param synsets synsets
     * @return id-to-nid map
     */
    fun makeSynsetNIDs(synsets: Collection): Map {
        return synsets
            .asSequence()
            .map { s: Synset -> s.synsetId }
            .sorted()
            .withIndex()
            .associate { it.value to it.index + 1 }
    }

    /**
     * Generate synsets table
     *
     * @param ps      print stream
     * @param synsets synsets
     * @return synsets id-to-nid map
     */
    fun generateSynsets(ps: PrintStream, synsets: Collection): Map {
        // make synsetId-to-nid map
        val synsetIdToNID = makeSynsetNIDs(synsets)

        // insert map
        val columns = listOf(
            Names.SYNSETS.synsetid,
            Names.SYNSETS.posid,
            Names.SYNSETS.domainid,
            Names.SYNSETS.definition
        ).joinToString(",")

        val toSqlRow = { synset: Synset ->
            val type = synset.type
            val definition = synset.definition
            val domain = synset.lexfile
            val lexdomainId = BuiltIn.LEXFILE_NIDS[domain]!!
            "'$type',$lexdomainId,'${escape(definition!!)}'"
        }
        if (!Printers.WITH_COMMENT) {
            printInsert(ps, Names.SYNSETS.TABLE, columns, synsets, { it.synsetId }, synsetIdToNID, toSqlRow)
        } else {
            val toSqlRowWithComment = { synset: Synset -> toSqlRow.invoke(synset) to synset.synsetId }
            printInsertWithComment(ps, Names.SYNSETS.TABLE, columns, synsets, { it.synsetId }, synsetIdToNID, toSqlRowWithComment)
        }
        return synsetIdToNID
    }

    /**
     * Generate synset relations table
     *
     * @param ps               print stream
     * @param synsets          synsets
     * @param synsetIdToNIDMap id-to-nid map
     */
    fun generateSynsetRelations(ps: PrintStream, synsets: Collection, synsetIdToNIDMap: Map) {

        // synset sequence
        val synsetSeq = synsets
            .asSequence()
            .filter { !it.relations.isNullOrEmpty() }
            .sortedBy(Synset::synsetId)

        // insert
        val columns = listOf(
            Names.SEMRELATIONS.synset1id,
            Names.SEMRELATIONS.synset2id,
            Names.SEMRELATIONS.relationid
        ).joinToString(",")

        val toTargetData = { synset: Synset ->
            synset.relations!!.keys
                .asSequence()
                .onEach { require(BuiltIn.OEWN_RELATION_TYPES.containsKey(it)) { it } } // relation type
                .flatMap {
                    val relation: Relation = it
                    val relationNID: Int = BuiltIn.OEWN_RELATION_TYPES[it]!! // relation NID
                    synset.relations!![it]!!
                        .asSequence() // sequence of synset2 ids
                        .map { synset2Id -> (relation to relationNID) to synset2Id }
                } // sequence of ((relation, relationNID), synset2Id_1) ((relation, relationNID, synset2Id_2) ...
                .sortedWith(
                    Comparator
                        .comparingInt { data: Pair, SynsetId> -> data.first.second } // relationNID
                        .thenComparing { data -> data.second } //  synset2Id
                )
        }

        val toSqlRows = { synset: Synset ->
            val synset1NID = NIDMaps.lookup(synsetIdToNIDMap, synset.synsetId)
            toTargetData(synset) // sequence of ((relation, relationNID), synset2Id_1) ((relation, relationNID, synset2Id_2) ...
                .map {
                    val relationNID: Int = BuiltIn.OEWN_RELATION_TYPES[it.first.first]!! // relation
                    val synset2NID = NIDMaps.lookup(synsetIdToNIDMap, it.second)
                    "$synset1NID,$synset2NID,$relationNID"
                }
                .toList()
        }

        if (!Printers.WITH_COMMENT) {
            printInserts(ps, Names.SEMRELATIONS.TABLE, columns, synsetSeq, toSqlRows, false)
        } else {
            val toSqlRowsWithComments = { synset: Synset ->
                val rows = toSqlRows.invoke(synset)
                val comments = toTargetData(synset) // sequence of ((relation, relationNID), synset2Id_1) ((relation, relationNID, synset2Id_2) ...
                    .map { "${synset.synsetId} -${it.first.first}-> ${it.second}" }
                rows
                    .asSequence()
                    .zip(comments)
            }
            printInsertsWithComment(ps, Names.SEMRELATIONS.TABLE, columns, synsetSeq, toSqlRowsWithComments, false)
        }
    }

    /**
     * Generate samples table
     *
     * @param ps               print stream
     * @param synsets          synsets
     * @param synsetIdToNIDMap id-to-nid map
     */
    fun generateSamples(ps: PrintStream, synsets: Collection, synsetIdToNIDMap: Map) {

        // sequence of synsets
        val synsetSeq = synsets
            .asSequence()
            .filter { !it.examples.isNullOrEmpty() }
            .sortedBy(Synset::synsetId)

        // insert
        val columns = listOf(
            Names.SAMPLES.sampleid,
            Names.SAMPLES.synsetid,
            Names.SAMPLES.sample
        ).joinToString(",")
        val toSqlRows = { synset: Synset ->
            val synsetNID1 = NIDMaps.lookup(synsetIdToNIDMap, synset.synsetId)
            synset.examples!!
                .map { "$synsetNID1,'${escape(it)}'" }
                .toList()
        }
        printInserts(ps, Names.SAMPLES.TABLE, columns, synsetSeq, toSqlRows, true)
    }
}