From d1d41de61c50ccdcc06e3f4bee8d26fe81992974 Mon Sep 17 00:00:00 2001
From: Peva Blanchard <pblanchard@kleis.ch>
Date: Sat, 11 Nov 2023 12:29:21 +0100
Subject: [PATCH] plugin: basic data source csv loader

---
 .../reducer/DataExpressionReducer.kt          |  2 +-
 .../datasources/LcaDataSourceOperations.kt    | 42 +++++++++++++++++--
 2 files changed, 39 insertions(+), 5 deletions(-)

diff --git a/core/src/main/kotlin/ch/kleis/lcaac/core/lang/evaluator/reducer/DataExpressionReducer.kt b/core/src/main/kotlin/ch/kleis/lcaac/core/lang/evaluator/reducer/DataExpressionReducer.kt
index 6173e7a45..12afeca72 100644
--- a/core/src/main/kotlin/ch/kleis/lcaac/core/lang/evaluator/reducer/DataExpressionReducer.kt
+++ b/core/src/main/kotlin/ch/kleis/lcaac/core/lang/evaluator/reducer/DataExpressionReducer.kt
@@ -46,7 +46,7 @@ class DataExpressionReducer<Q>(
     private fun reduceDataFrom(expression: EDataFrom<Q>): DataExpression<Q> {
         val source = dataSourceRegister[DataSourceKey(expression.source)]
             ?: throw EvaluatorException("unknown data source ${expression.source}")
-        return sourceOps.read(source, expression.row, expression.column)
+        return reduce(sourceOps.read(source, expression.row, expression.column))
     }
 
     private fun reduceUnitOf(unitOf: EUnitOf<Q>): DataExpression<Q> {
diff --git a/plugin/src/main/kotlin/ch/kleis/lcaac/plugin/datasources/LcaDataSourceOperations.kt b/plugin/src/main/kotlin/ch/kleis/lcaac/plugin/datasources/LcaDataSourceOperations.kt
index e566c4bfc..dae9d9934 100644
--- a/plugin/src/main/kotlin/ch/kleis/lcaac/plugin/datasources/LcaDataSourceOperations.kt
+++ b/plugin/src/main/kotlin/ch/kleis/lcaac/plugin/datasources/LcaDataSourceOperations.kt
@@ -1,15 +1,49 @@
 package ch.kleis.lcaac.plugin.datasources
 
 import ch.kleis.lcaac.core.datasource.DataSourceOperations
-import ch.kleis.lcaac.core.lang.expression.DataExpression
-import ch.kleis.lcaac.core.lang.expression.DataSourceExpression
-import ch.kleis.lcaac.core.lang.expression.SliceIndex
+import ch.kleis.lcaac.core.lang.evaluator.EvaluatorException
+import ch.kleis.lcaac.core.lang.expression.*
 import ch.kleis.lcaac.core.math.QuantityOperations
+import org.apache.commons.csv.CSVFormat
+import org.apache.commons.csv.CSVParser
+import java.nio.file.Path
 
 class LcaDataSourceOperations<Q>(
     private val ops: QuantityOperations<Q>,
 ) : DataSourceOperations<Q> {
     override fun read(source: DataSourceExpression<Q>, row: SliceIndex, column: SliceIndex): DataExpression<Q> {
-        TODO("Not yet implemented")
+        if (row !is StrIndex || column !is StrIndex) {
+            throw EvaluatorException("Not supported yet")
+        }
+        return when (source) {
+            is ECsvSource -> readCsv(source, row, column)
+        }
+    }
+
+    private val format = CSVFormat.DEFAULT.builder()
+        .setHeader()
+        .setSkipHeaderRecord(true)
+        .build()
+
+    // TODO: Very inefficient, csv is scanned on every call
+    private fun readCsv(source: ECsvSource<Q>, row: StrIndex, column: StrIndex): DataExpression<Q> {
+        val schema = source.schema
+        val columnType = schema[column.s] ?: throw EvaluatorException("${column.s} not found in schema")
+
+        val file = Path.of(source.location).toFile()
+        val index = source.index ?: throw EvaluatorException("Requires index")
+        val parser = CSVParser(file.inputStream().reader(), format)
+        val rawValue = parser.records.find {
+            it[index] == row.s
+        }?.get(column.s)
+            ?: throw EvaluatorException("value not found (row=${row.s}, col=${column.s})")
+
+        return when(columnType) {
+            is CQuantity -> {
+                val amount = ops.pure(rawValue.toDouble())
+                return EQuantityScale(amount, columnType.unit)
+            }
+            is CText -> EStringLiteral(rawValue)
+        }
     }
 }