jiayuasu commented on code in PR #612: URL: https://github.com/apache/incubator-sedona/pull/612#discussion_r853657807
########## sql/src/main/scala/org/apache/spark/sql/sedona_sql/io/GeotiffFileFormat.scala: ########## @@ -101,4 +126,123 @@ private[spark] class GeotiffFileFormat extends FileFormat with DataSourceRegiste } } } + + private def isValidGeoTiffSchema(options: Map[String, String], dataSchema: StructType): Boolean = { + val colImage = options.getOrElse("key_image", "image") + val fields = dataSchema.fieldNames + if (fields.contains(colImage) ){ + val schemaFields = dataSchema.fields(dataSchema.fieldIndex(colImage)).dataType.asInstanceOf[StructType] + if (schemaFields.fieldNames.length != 6) return false + } + else { + if (fields.length != 6) return false + } + true + } + +} + +// class for writing geoTiff images +private class GeotiffFileWriter(savePath: String, + options: Map[String, String], + dataSchema: StructType, + context: TaskAttemptContext) extends OutputWriter { + + // set writing parameters + private val DEFAULT_WRITE_PARAMS: GeoTiffWriteParams = new GeoTiffWriteParams() + DEFAULT_WRITE_PARAMS.setCompressionMode(ImageWriteParam.MODE_EXPLICIT) + DEFAULT_WRITE_PARAMS.setCompressionType("LZW") + DEFAULT_WRITE_PARAMS.setCompressionQuality(0.75F) + DEFAULT_WRITE_PARAMS.setTilingMode(ImageWriteParam.MODE_EXPLICIT) + DEFAULT_WRITE_PARAMS.setTiling(512, 512) + + private val hfs = new Path(savePath).getFileSystem(context.getConfiguration) + + override def write(row: InternalRow): Unit = { + // retrieving the metadata of a geotiff image + var rowFields: InternalRow = row + var schemaFields: StructType = dataSchema + val fields = dataSchema.fieldNames + + val colImage = options.getOrElse("key_image", "image") + val colOrigin = options.getOrElse("key_origin", "origin") + val colBands = options.getOrElse("key_n_bands", "nBands") + val colWidth = options.getOrElse("key_width", "width") + val colHeight = options.getOrElse("key_height", "height") + val colWkt = options.getOrElse("key_wkt", "wkt") + val colData = options.getOrElse("key_data", "data") + + if (fields.contains(colImage)) { + schemaFields = dataSchema.fields(dataSchema.fieldIndex(colImage)).dataType.asInstanceOf[StructType] + rowFields = row.getStruct(dataSchema.fieldIndex(colImage), 6) + } + + val tiffOrigin = rowFields.getString(schemaFields.fieldIndex(colOrigin)) + val tiffBands = rowFields.getInt(schemaFields.fieldIndex(colBands)) + val tiffWidth = rowFields.getInt(schemaFields.fieldIndex(colWidth)) + val tiffHeight = rowFields.getInt(schemaFields.fieldIndex(colHeight)) + val tiffGeometry = rowFields.getString(schemaFields.fieldIndex(colWkt)) + val tiffData = rowFields.getArray(schemaFields.fieldIndex(colData)).toDoubleArray() + + // create a writable raster object + val raster = RasterFactory.createBandedRaster(DataBuffer.TYPE_DOUBLE, tiffWidth, tiffHeight, tiffBands, null) + + // extract the pixels of the geotiff image and write to the writable raster + val pixelVal = Array.ofDim[Double](tiffBands) + for (i <- 0 until tiffHeight) { + for (j <- 0 until tiffWidth) { + for (k <- 0 until tiffBands) { + pixelVal(k) = tiffData(tiffHeight*tiffWidth*k + i * tiffWidth + j) + } + raster.setPixel(j, i, pixelVal) + } + } + + // CRS is always fixed to EPSG:4326 + val crs = CRS.decode("EPSG:4326", true) Review Comment: CRS information might be stored in Geometry SRID attribute but this is NOT mandatory. So we cannot find the CRS by ourselves. It has to come from the user. So let me tweak my proposal for the "writer" a bit: For writer, "writeToCRS" is optional. Making it optional is kind of important as some GeoTiff users don't know what is called CRS. "writeToCRS" is provided: store CRS info in GeoTiff metadata. No CRS transformation is needed. "writeToCRS" is NOT provided: store WGS84 (aka epsg:4326) in the metadata. No CRS transformation is needed. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: dev-unsubscr...@sedona.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org