19241805651

Committed 10 Nov 2025 06:22PM UTC coverage: 88.832%. First build

Build # 19241805651

Build Type

Pull #1083

github

Committed by

web-flow

Commit Message

Merge dac631b93 into 113de40ca

Pull Request Pull Request #1083: feat: new gdal source workflow optimization

Coverage Stats

6213 of 7052 new or added lines in 71 files covered. (88.1%)

116189 of 130797 relevant lines covered (88.83%)

496404.71 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

93.38

/operators/src/source/csv.rs

use crate::engine::{
    CanonicOperatorName, InitializedVectorOperator, OperatorData, OperatorName, QueryContext,
    SourceOperator, TypedVectorQueryProcessor, VectorOperator, VectorQueryProcessor,
    VectorResultDescriptor,
};
use crate::engine::{QueryProcessor, WorkflowOperatorPath};
use crate::error;
use crate::optimization::OptimizationError;
use crate::util::{Result, safe_lock_mutex};
use async_trait::async_trait;
use csv::{Position, Reader, StringRecord};
use futures::stream::BoxStream;
use futures::task::{Context, Poll};
use futures::{Stream, StreamExt};
use geoengine_datatypes::collections::{
    BuilderProvider, GeoFeatureCollectionRowBuilder, MultiPointCollection, VectorDataType,
};
use geoengine_datatypes::dataset::NamedData;
use geoengine_datatypes::primitives::SpatialResolution;
use geoengine_datatypes::{
    primitives::{
        BoundingBox2D, ColumnSelection, Coordinate2D, TimeInterval, VectorQueryRectangle,
    },
    spatial_reference::SpatialReference,
};
use serde::{Deserialize, Serialize};
use snafu::{OptionExt, ResultExt, ensure};
use std::path::PathBuf;
use std::pin::Pin;
use std::sync::atomic::Ordering;
use std::sync::{Arc, Mutex};
use std::{fs::File, sync::atomic::AtomicBool};

/// Parameters for the CSV Source Operator
#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone)]
#[serde(rename_all = "camelCase")]
pub struct CsvSourceParameters {
    pub file_path: PathBuf,
    pub field_separator: char,
    pub geometry: CsvGeometrySpecification,
    #[serde(default)]
    pub time: CsvTimeSpecification,
}

#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone)]
#[serde(tag = "type", rename_all = "lowercase")]
pub enum CsvGeometrySpecification {
    #[allow(clippy::upper_case_acronyms)]
    XY { x: String, y: String },
}

#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone, Default)]
pub enum CsvTimeSpecification {
    #[default]
    None,
}

enum ReaderState {
    Untouched(Reader<File>),
    OnGoing {
        header: ParsedHeader,
        records: csv::StringRecordsIntoIter<File>,
    },
    Error,
}

impl ReaderState {
    pub fn setup_once(&mut self, geometry_specification: CsvGeometrySpecification) -> Result<()> {
        if let ReaderState::Untouched(..) = self {
            // pass
        } else {
            return Ok(());
        }

        let old_state = std::mem::replace(self, ReaderState::Error);

        if let ReaderState::Untouched(mut csv_reader) = old_state {
            let header = CsvSourceStream::setup_read(geometry_specification, &mut csv_reader)?;

            let mut records = csv_reader.into_records();

            // consume the first row, which is the header
            if header.has_header {
                // TODO: throw error
                records.next();
            }

            *self = ReaderState::OnGoing { header, records }
        }

        Ok(())
    }
}

pub struct CsvSourceStream {
    parameters: CsvSourceParameters,
    bbox: BoundingBox2D,
    chunk_size: usize,
    reader_state: Arc<Mutex<ReaderState>>,
    thread_is_computing: Arc<AtomicBool>,
    #[allow(clippy::option_option)]
    poll_result: Arc<Mutex<Option<Option<Result<MultiPointCollection>>>>>,
}

pub type CsvSource = SourceOperator<CsvSourceParameters>;

impl OperatorName for CsvSource {
    const TYPE_NAME: &'static str = "CsvSource";
}

impl OperatorData for CsvSourceParameters {
    fn data_names_collect(&self, _data_names: &mut Vec<NamedData>) {}
}

#[typetag::serde]
#[async_trait]
impl VectorOperator for CsvSource {
    async fn _initialize(
        self: Box<Self>,
        path: WorkflowOperatorPath,
        _context: &dyn crate::engine::ExecutionContext,
    ) -> Result<Box<dyn InitializedVectorOperator>> {
        let initialized_source = InitializedCsvSource {
            name: CanonicOperatorName::from(&self),
            path,
            result_descriptor: VectorResultDescriptor {
                data_type: VectorDataType::MultiPoint, // TODO: get as user input
                spatial_reference: SpatialReference::epsg_4326().into(), // TODO: get as user input
                columns: Default::default(), // TODO: get when source allows loading other columns
                time: None,
                bbox: None,
            },
            state: self.params,
        };

        Ok(initialized_source.boxed())
    }

    span_fn!(CsvSource);
}

pub struct InitializedCsvSource {
    name: CanonicOperatorName,
    path: WorkflowOperatorPath,
    result_descriptor: VectorResultDescriptor,
    state: CsvSourceParameters,
}

impl InitializedVectorOperator for InitializedCsvSource {
    fn query_processor(&self) -> Result<crate::engine::TypedVectorQueryProcessor> {
        Ok(TypedVectorQueryProcessor::MultiPoint(
            CsvSourceProcessor {
                params: self.state.clone(),
                result_descriptor: self.result_descriptor.clone(),
            }
            .boxed(),
        ))
    }

    fn result_descriptor(&self) -> &VectorResultDescriptor {
        &self.result_descriptor
    }

    fn canonic_name(&self) -> CanonicOperatorName {
        self.name.clone()
    }

    fn name(&self) -> &'static str {
        CsvSource::TYPE_NAME
    }

    fn path(&self) -> WorkflowOperatorPath {
        self.path.clone()
    }

    fn optimize(
        &self,
        _target_resolution: SpatialResolution,
    ) -> Result<Box<dyn VectorOperator>, OptimizationError> {
        Ok(CsvSource {
            params: self.state.clone(),
        }
        .boxed())
    }
}

impl CsvSourceStream {
    /// Creates a new `CsvSource`
    ///
    /// # Errors
    ///
    /// This constructor fails if the delimiter is not an ASCII character.
    /// Furthermore, there are IO errors from the reader.
    ///
    // TODO: include time interval, e.g. QueryRectangle parameter
    pub fn new(
        parameters: CsvSourceParameters,
        bbox: BoundingBox2D,
        chunk_size: usize,
    ) -> Result<Self> {
        ensure!(
            parameters.field_separator.is_ascii(),
            error::CsvSource {
                details: "Delimiter must be ASCII character"
            }
        );

        Ok(Self {
            reader_state: Arc::new(Mutex::new(ReaderState::Untouched(
                csv::ReaderBuilder::new()
                    .delimiter(parameters.field_separator as u8)
                    .has_headers(true)
                    .from_path(parameters.file_path.as_path())
                    .context(error::CsvSourceReader {})?,
            ))),
            thread_is_computing: Arc::new(AtomicBool::new(false)),
            poll_result: Arc::new(Mutex::new(None)),
            parameters,
            bbox,
            chunk_size,
        })
    }

    fn setup_read(
        geometry_specification: CsvGeometrySpecification,
        csv_reader: &mut Reader<File>,
    ) -> Result<ParsedHeader> {
        csv_reader
            .seek(Position::new())
            .context(error::CsvSourceReader {})?; // start at beginning

        ensure!(
            csv_reader.has_headers(),
            error::CsvSource {
                details: "CSV file must contain header",
            }
        );

        let header = csv_reader.headers().context(error::CsvSourceReader)?;

        let CsvGeometrySpecification::XY { x, y } = geometry_specification;
        let x_index = header
            .iter()
            .position(|v| v == x)
            .context(error::CsvSource {
                details: "Cannot find x index in csv header",
            })?;
        let y_index = header
            .iter()
            .position(|v| v == y)
            .context(error::CsvSource {
                details: "Cannot find y index in csv header",
            })?;

        Ok(ParsedHeader {
            has_header: true,
            x_index,
            y_index,
        })
    }

    /// Parse a single CSV row
    fn parse_row(header: &ParsedHeader, row: &StringRecord) -> Result<ParsedRow> {
        let x: f64 = row
            .get(header.x_index)
            .context(error::CsvSource {
                details: "Cannot find x index key",
            })?
            .parse()
            .map_err(|_error| error::Error::CsvSource {
                details: "Cannot parse x coordinate".to_string(),
            })?;
        let y: f64 = row
            .get(header.y_index)
            .context(error::CsvSource {
                details: "Cannot find y index key",
            })?
            .parse()
            .map_err(|_error| error::Error::CsvSource {
                details: "Cannot parse y coordinate".to_string(),
            })?;

        Ok(ParsedRow {
            coordinate: (x, y).into(),
            time_interval: TimeInterval::default(),
        })
    }
}

impl Stream for CsvSourceStream {
    type Item = Result<MultiPointCollection>;

    fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
        // TODO: handle lock poisoning on multiple occasions

        if self.thread_is_computing.load(Ordering::Relaxed) {
            return Poll::Pending;
        }

        let mut poll_result = safe_lock_mutex(&self.poll_result);
        if let Some(x) = poll_result.take() {
            return Poll::Ready(x);
        }

        self.thread_is_computing.store(true, Ordering::Relaxed);

        let is_working = self.thread_is_computing.clone();
        let reader_state = self.reader_state.clone();
        let poll_result = self.poll_result.clone();

        let bbox = self.bbox;
        let chunk_size = self.chunk_size;
        let parameters = self.parameters.clone();
        let waker = cx.waker().clone();

        crate::util::spawn_blocking(move || {
            let mut csv_reader = safe_lock_mutex(&reader_state);
            let computation_result = || -> Result<Option<MultiPointCollection>> {
                // TODO: is clone necessary?
                let geometry_specification = parameters.geometry.clone();
                csv_reader.setup_once(geometry_specification)?;

                let (header, records) = match &mut *csv_reader {
                    ReaderState::OnGoing { header, records } => (header, records),
                    ReaderState::Error => return Ok(None),
                    ReaderState::Untouched(_) => unreachable!(),
                };

                let mut builder = MultiPointCollection::builder().finish_header();
                let mut number_of_entries = 0; // TODO: add size/len to builder

                while number_of_entries < chunk_size {
                    let Some(record) = records.next() else {
                        break;
                    };

                    let row = record.context(error::CsvSourceReader)?;
                    let parsed_row = CsvSourceStream::parse_row(header, &row)?;

                    // TODO: filter time
                    if bbox.contains_coordinate(&parsed_row.coordinate) {
                        builder.push_geometry(parsed_row.coordinate.into());
                        builder.push_time_interval(parsed_row.time_interval);
                        builder.finish_row();

                        number_of_entries += 1;
                    }
                }

                // TODO: is this the correct cancellation criterion?
                if number_of_entries > 0 {
                    let collection = builder.build()?;
                    Ok(Some(collection))
                } else {
                    Ok(None)
                }
            }();

            *safe_lock_mutex(&poll_result) = Some(match computation_result {
                Ok(Some(collection)) => Some(Ok(collection)),
                Ok(None) => None,
                Err(e) => Some(Err(e)),
            });
            is_working.store(false, Ordering::Relaxed);

            waker.wake();
        });

        Poll::Pending
    }
}

#[derive(Debug)]
struct CsvSourceProcessor {
    params: CsvSourceParameters,
    result_descriptor: VectorResultDescriptor,
}

#[async_trait]
impl QueryProcessor for CsvSourceProcessor {
    type Output = MultiPointCollection;
    type SpatialBounds = BoundingBox2D;
    type Selection = ColumnSelection;
    type ResultDescription = VectorResultDescriptor;

    async fn _query<'a>(
        &'a self,
        query: VectorQueryRectangle,
        _ctx: &'a dyn QueryContext,
    ) -> Result<BoxStream<'a, Result<Self::Output>>> {
        // TODO: properly handle chunk_size
        Ok(CsvSourceStream::new(self.params.clone(), query.spatial_bounds(), 10)?.boxed())
    }

    fn result_descriptor(&self) -> &VectorResultDescriptor {
        &self.result_descriptor
    }
}

#[derive(Clone, Copy, Debug)]
struct ParsedHeader {
    pub has_header: bool,
    pub x_index: usize,
    pub y_index: usize,
}

struct ParsedRow {
    pub coordinate: Coordinate2D,
    pub time_interval: TimeInterval,
    // TODO: fields
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::engine::MockExecutionContext;
    use geoengine_datatypes::{
        collections::{FeatureCollectionInfos, ToGeoJson},
        util::test::TestDefault,
    };
    use std::io::{Seek, SeekFrom, Write};

    #[test]
    fn it_deserializes() {
        let json_string = r#"
            {
                "type": "CsvSource",
                "params": {
                    "filePath": "/foo/bar.csv",
                    "fieldSeparator": ",",
                    "geometry": {
                        "type": "xy",
                        "x": "x",
                        "y": "y"
                    }
                }
            }"#;

        let operator: CsvSource = serde_json::from_str(json_string).unwrap();

        assert_eq!(
            operator,
            CsvSource {
                params: CsvSourceParameters {
                    file_path: "/foo/bar.csv".into(),
                    field_separator: ',',
                    geometry: CsvGeometrySpecification::XY {
                        x: "x".into(),
                        y: "y".into()
                    },
                    time: CsvTimeSpecification::None,
                },
            }
        );
    }

    #[tokio::test]
    async fn read_points() {
        let mut fake_file = tempfile::NamedTempFile::new().unwrap();
        write!(
            fake_file,
            "\
x,y
0,1
2,3
4,5
"
        )
        .unwrap();
        fake_file.seek(SeekFrom::Start(0)).unwrap();

        let mut csv_source = CsvSourceStream::new(
            CsvSourceParameters {
                file_path: fake_file.path().into(),
                field_separator: ',',
                geometry: CsvGeometrySpecification::XY {
                    x: "x".into(),
                    y: "y".into(),
                },
                time: CsvTimeSpecification::None,
            },
            BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),
            2,
        )
        .unwrap();

        assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 2);
        assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);
        assert!(csv_source.next().await.is_none());
    }

    #[tokio::test]
    async fn erroneous_point_rows() {
        let mut fake_file = tempfile::NamedTempFile::new().unwrap();
        write!(
            fake_file,
            "\
x,y
0,1
CORRUPT
4,5
"
        )
        .unwrap();
        fake_file.seek(SeekFrom::Start(0)).unwrap();

        let mut csv_source = CsvSourceStream::new(
            CsvSourceParameters {
                file_path: fake_file.path().into(),
                field_separator: ',',
                geometry: CsvGeometrySpecification::XY {
                    x: "x".into(),
                    y: "y".into(),
                },
                time: CsvTimeSpecification::None,
            },
            BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),
            1,
        )
        .unwrap();

        assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);
        assert!(csv_source.next().await.unwrap().is_err());
        assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);
        assert!(csv_source.next().await.is_none());
    }

    #[tokio::test]
    async fn corrupt_point_header() {
        let mut fake_file = tempfile::NamedTempFile::new().unwrap();
        write!(
            fake_file,
            "\
x,z
0,1
2,3
4,5
"
        )
        .unwrap();
        fake_file.seek(SeekFrom::Start(0)).unwrap();

        let mut csv_source = CsvSourceStream::new(
            CsvSourceParameters {
                file_path: fake_file.path().into(),
                field_separator: ',',
                geometry: CsvGeometrySpecification::XY {
                    x: "x".into(),
                    y: "y".into(),
                },
                time: CsvTimeSpecification::None,
            },
            BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),
            1,
        )
        .unwrap();

        assert!(csv_source.next().await.unwrap().is_err());
        assert!(csv_source.next().await.is_none());
    }

    #[tokio::test]
    async fn processor() {
        let mut fake_file = tempfile::NamedTempFile::new().unwrap();
        write!(
            fake_file,
            "\
x,y
0,1
2,3
4,5
"
        )
        .unwrap();
        fake_file.seek(SeekFrom::Start(0)).unwrap();

        let params = CsvSourceParameters {
            file_path: fake_file.path().into(),
            field_separator: ',',
            geometry: CsvGeometrySpecification::XY {
                x: "x".into(),
                y: "y".into(),
            },
            time: CsvTimeSpecification::None,
        };

        let p = CsvSourceProcessor {
            params,
            result_descriptor: VectorResultDescriptor {
                data_type: VectorDataType::MultiPoint,
                spatial_reference: SpatialReference::epsg_4326().into(),
                columns: Default::default(),
                time: None,
                bbox: None,
            },
        };

        let query = VectorQueryRectangle::new(
            BoundingBox2D::new_unchecked(Coordinate2D::new(0., 0.), Coordinate2D::new(3., 3.)),
            TimeInterval::new_unchecked(0, 1),
            ColumnSelection::all(),
        );
        let ecx = MockExecutionContext::test_default();
        let ctx = ecx.mock_query_context((10 * 8 * 2).into());

        let r: Vec<Result<MultiPointCollection>> =
            p.query(query, &ctx).await.unwrap().collect().await;

        assert_eq!(r.len(), 1);

        assert_eq!(
            serde_json::from_str::<serde_json::Value>(&r[0].as_ref().unwrap().to_geo_json())
                .unwrap(),
            serde_json::json!({
                "type": "FeatureCollection",
                "features": [{
                    "type": "Feature",
                    "geometry": {
                        "type": "Point",
                        "coordinates": [0.0, 1.0]
                    },
                    "properties": {},
                    "when": {
                        "start": "-262143-01-01T00:00:00+00:00",
                        "end": "+262142-12-31T23:59:59.999+00:00",
                        "type": "Interval"
                    }
                }, {
                    "type": "Feature",
                    "geometry": {
                        "type": "Point",
                        "coordinates": [2.0, 3.0]
                    },
                    "properties": {},
                    "when": {
                        "start": "-262143-01-01T00:00:00+00:00",
                        "end": "+262142-12-31T23:59:59.999+00:00",
                        "type": "Interval"
                    }
                }]
            })
        );
    }

    #[test]
    fn operator() {
        let mut temp_file = tempfile::NamedTempFile::new().unwrap();
        write!(
            temp_file,
            "\
x;y
0;1
2;3
4;5
"
        )
        .unwrap();
        temp_file.seek(SeekFrom::Start(0)).unwrap();

        let params = CsvSourceParameters {
            file_path: temp_file.path().into(),
            field_separator: ';',
            geometry: CsvGeometrySpecification::XY {
                x: "x".into(),
                y: "y".into(),
            },
            time: CsvTimeSpecification::None,
        };

        let operator = CsvSource { params }.boxed();

        let operator_json = serde_json::to_value(&operator).unwrap();

        assert_eq!(
            operator_json,
            serde_json::json!({
                "type": "CsvSource",
                "params": {
                    "filePath": temp_file.path(),
                    "fieldSeparator": ";",
                    "geometry": {
                        "type": "xy",
                        "x": "x",
                        "y": "y"
                    },
                    "time": "None"
                }
            })
        );

        let _operator: Box<dyn VectorOperator> = serde_json::from_value(operator_json).unwrap();
    }
}

1	use crate::engine::{
2	CanonicOperatorName, InitializedVectorOperator, OperatorData, OperatorName, QueryContext,
3	SourceOperator, TypedVectorQueryProcessor, VectorOperator, VectorQueryProcessor,
4	VectorResultDescriptor,
5	};
6	use crate::engine::{QueryProcessor, WorkflowOperatorPath};
7	use crate::error;
8	use crate::optimization::OptimizationError;
9	use crate::util::{Result, safe_lock_mutex};
10	use async_trait::async_trait;
11	use csv::{Position, Reader, StringRecord};
12	use futures::stream::BoxStream;
13	use futures::task::{Context, Poll};
14	use futures::{Stream, StreamExt};
15	use geoengine_datatypes::collections::{
16	BuilderProvider, GeoFeatureCollectionRowBuilder, MultiPointCollection, VectorDataType,
17	};
18	use geoengine_datatypes::dataset::NamedData;
19	use geoengine_datatypes::primitives::SpatialResolution;
20	use geoengine_datatypes::{
21	primitives::{
22	BoundingBox2D, ColumnSelection, Coordinate2D, TimeInterval, VectorQueryRectangle,
23	},
24	spatial_reference::SpatialReference,
25	};
26	use serde::{Deserialize, Serialize};
27	use snafu::{OptionExt, ResultExt, ensure};
28	use std::path::PathBuf;
29	use std::pin::Pin;
30	use std::sync::atomic::Ordering;
31	use std::sync::{Arc, Mutex};
32	use std::{fs::File, sync::atomic::AtomicBool};
33
34	/// Parameters for the CSV Source Operator
35	#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone)]
36	#[serde(rename_all = "camelCase")]
37	pub struct CsvSourceParameters {
38	pub file_path: PathBuf,
39	pub field_separator: char,
40	pub geometry: CsvGeometrySpecification,
41	#[serde(default)]
42	pub time: CsvTimeSpecification,
43	}
44
45	#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone)]
46	#[serde(tag = "type", rename_all = "lowercase")]
47	pub enum CsvGeometrySpecification {
48	#[allow(clippy::upper_case_acronyms)]
49	XY { x: String, y: String },
50	}
51
52	#[derive(Debug, PartialEq, Eq, Serialize, Deserialize, Clone, Default)]
53	pub enum CsvTimeSpecification {
54	#[default]
55	None,
56	}
57
58	enum ReaderState {
59	Untouched(Reader<File>),
60	OnGoing {
61	header: ParsedHeader,
62	records: csv::StringRecordsIntoIter<File>,
63	},
64	Error,
65	}
66
67	impl ReaderState {
68	pub fn setup_once(&mut self, geometry_specification: CsvGeometrySpecification) -> Result<()> {	17✔
69	if let ReaderState::Untouched(..) = self {	17✔
70	// pass	7✔
71	} else {	7✔
72	return Ok(());	10✔
73	}
74
75	let old_state = std::mem::replace(self, ReaderState::Error);	7✔
76
77	if let ReaderState::Untouched(mut csv_reader) = old_state {	7✔
78	let header = CsvSourceStream::setup_read(geometry_specification, &mut csv_reader)?;	7✔
79
80	let mut records = csv_reader.into_records();	6✔
81
82	// consume the first row, which is the header
83	if header.has_header {	6✔
84	// TODO: throw error	6✔
85	records.next();	6✔
86	}	6✔
87
88	*self = ReaderState::OnGoing { header, records }	6✔
89	}	×
90
91	Ok(())	6✔
92	}	17✔
93	}
94
95	pub struct CsvSourceStream {
96	parameters: CsvSourceParameters,
97	bbox: BoundingBox2D,
98	chunk_size: usize,
99	reader_state: Arc<Mutex<ReaderState>>,
100	thread_is_computing: Arc<AtomicBool>,
101	#[allow(clippy::option_option)]
102	poll_result: Arc<Mutex<Option<Option<Result<MultiPointCollection>>>>>,
103	}
104
105	pub type CsvSource = SourceOperator<CsvSourceParameters>;
106
107	impl OperatorName for CsvSource {
108	const TYPE_NAME: &'static str = "CsvSource";
109	}
110
111	impl OperatorData for CsvSourceParameters {
112	fn data_names_collect(&self, _data_names: &mut Vec<NamedData>) {}	×
113	}
114
115	#[typetag::serde]	×
116	#[async_trait]
117	impl VectorOperator for CsvSource {
118	async fn _initialize(
119	self: Box<Self>,
120	path: WorkflowOperatorPath,
121	_context: &dyn crate::engine::ExecutionContext,
122	) -> Result<Box<dyn InitializedVectorOperator>> {	4✔
123	let initialized_source = InitializedCsvSource {
124	name: CanonicOperatorName::from(&self),
125	path,
126	result_descriptor: VectorResultDescriptor {
127	data_type: VectorDataType::MultiPoint, // TODO: get as user input
128	spatial_reference: SpatialReference::epsg_4326().into(), // TODO: get as user input
129	columns: Default::default(), // TODO: get when source allows loading other columns
130	time: None,
131	bbox: None,
132	},
133	state: self.params,
134	};
135
136	Ok(initialized_source.boxed())
137	}	4✔
138
139	span_fn!(CsvSource);
140	}
141
142	pub struct InitializedCsvSource {
143	name: CanonicOperatorName,
144	path: WorkflowOperatorPath,
145	result_descriptor: VectorResultDescriptor,
146	state: CsvSourceParameters,
147	}
148
149	impl InitializedVectorOperator for InitializedCsvSource {
150	fn query_processor(&self) -> Result<crate::engine::TypedVectorQueryProcessor> {	3✔
151	Ok(TypedVectorQueryProcessor::MultiPoint(	3✔
152	CsvSourceProcessor {	3✔
153	params: self.state.clone(),	3✔
154	result_descriptor: self.result_descriptor.clone(),	3✔
155	}	3✔
156	.boxed(),	3✔
157	))	3✔
158	}	3✔
159
160	fn result_descriptor(&self) -> &VectorResultDescriptor {	4✔
161	&self.result_descriptor	4✔
162	}	4✔
163
164	fn canonic_name(&self) -> CanonicOperatorName {	×
165	self.name.clone()	×
166	}	×
167
168	fn name(&self) -> &'static str {	3✔
169	CsvSource::TYPE_NAME	3✔
170	}	3✔
171
172	fn path(&self) -> WorkflowOperatorPath {	3✔
173	self.path.clone()	3✔
174	}	3✔
175
NEW 176	fn optimize(	×
NEW 177	&self,	×
NEW 178	_target_resolution: SpatialResolution,	×
NEW 179	) -> Result<Box<dyn VectorOperator>, OptimizationError> {	×
NEW 180	Ok(CsvSource {	×
NEW 181	params: self.state.clone(),	×
NEW 182	}	×
NEW 183	.boxed())	×
NEW 184	}	×
185	}
186
187	impl CsvSourceStream {
188	/// Creates a new `CsvSource`
189	///
190	/// # Errors
191	///
192	/// This constructor fails if the delimiter is not an ASCII character.
193	/// Furthermore, there are IO errors from the reader.
194	///
195	// TODO: include time interval, e.g. QueryRectangle parameter
196	pub fn new(	7✔
197	parameters: CsvSourceParameters,	7✔
198	bbox: BoundingBox2D,	7✔
199	chunk_size: usize,	7✔
200	) -> Result<Self> {	7✔
201	ensure!(	7✔
202	parameters.field_separator.is_ascii(),	7✔
203	error::CsvSource {	×
204	details: "Delimiter must be ASCII character"	×
205	}	×
206	);
207
208	Ok(Self {
209	reader_state: Arc::new(Mutex::new(ReaderState::Untouched(	7✔
210	csv::ReaderBuilder::new()	7✔
211	.delimiter(parameters.field_separator as u8)	7✔
212	.has_headers(true)	7✔
213	.from_path(parameters.file_path.as_path())	7✔
214	.context(error::CsvSourceReader {})?,	7✔
215	))),
216	thread_is_computing: Arc::new(AtomicBool::new(false)),	7✔
217	poll_result: Arc::new(Mutex::new(None)),	7✔
218	parameters,	7✔
219	bbox,	7✔
220	chunk_size,	7✔
221	})
222	}	7✔
223
224	fn setup_read(	7✔
225	geometry_specification: CsvGeometrySpecification,	7✔
226	csv_reader: &mut Reader<File>,	7✔
227	) -> Result<ParsedHeader> {	7✔
228	csv_reader	7✔
229	.seek(Position::new())	7✔
230	.context(error::CsvSourceReader {})?; // start at beginning	7✔
231
232	ensure!(	7✔
233	csv_reader.has_headers(),	7✔
234	error::CsvSource {	×
235	details: "CSV file must contain header",	×
236	}	×
237	);
238
239	let header = csv_reader.headers().context(error::CsvSourceReader)?;	7✔
240
241	let CsvGeometrySpecification::XY { x, y } = geometry_specification;	7✔
242	let x_index = header	7✔
243	.iter()	7✔
244	.position(\|v\| v == x)	7✔
245	.context(error::CsvSource {	7✔
246	details: "Cannot find x index in csv header",	7✔
247	})?;	7✔
248	let y_index = header	7✔
249	.iter()	7✔
250	.position(\|v\| v == y)	14✔
251	.context(error::CsvSource {	7✔
252	details: "Cannot find y index in csv header",	7✔
253	})?;	7✔
254
255	Ok(ParsedHeader {	6✔
256	has_header: true,	6✔
257	x_index,	6✔
258	y_index,	6✔
259	})	6✔
260	}	7✔
261
262	/// Parse a single CSV row
263	fn parse_row(header: &ParsedHeader, row: &StringRecord) -> Result<ParsedRow> {	17✔
264	let x: f64 = row	17✔
265	.get(header.x_index)	17✔
266	.context(error::CsvSource {	17✔
267	details: "Cannot find x index key",	17✔
268	})?	17✔
269	.parse()	17✔
270	.map_err(\|_error\| error::Error::CsvSource {	17✔
271	details: "Cannot parse x coordinate".to_string(),	×
272	})?;	×
273	let y: f64 = row	17✔
274	.get(header.y_index)	17✔
275	.context(error::CsvSource {	17✔
276	details: "Cannot find y index key",	17✔
277	})?	17✔
278	.parse()	17✔
279	.map_err(\|_error\| error::Error::CsvSource {	17✔
280	details: "Cannot parse y coordinate".to_string(),	×
281	})?;	×
282
283	Ok(ParsedRow {	17✔
284	coordinate: (x, y).into(),	17✔
285	time_interval: TimeInterval::default(),	17✔
286	})	17✔
287	}	17✔
288	}
289
290	impl Stream for CsvSourceStream {
291	type Item = Result<MultiPointCollection>;
292
293	fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {	34✔
294	// TODO: handle lock poisoning on multiple occasions
295
296	if self.thread_is_computing.load(Ordering::Relaxed) {	34✔
297	return Poll::Pending;	×
298	}	34✔
299
300	let mut poll_result = safe_lock_mutex(&self.poll_result);	34✔
301	if let Some(x) = poll_result.take() {	34✔
302	return Poll::Ready(x);	17✔
303	}	17✔
304
305	self.thread_is_computing.store(true, Ordering::Relaxed);	17✔
306
307	let is_working = self.thread_is_computing.clone();	17✔
308	let reader_state = self.reader_state.clone();	17✔
309	let poll_result = self.poll_result.clone();	17✔
310
311	let bbox = self.bbox;	17✔
312	let chunk_size = self.chunk_size;	17✔
313	let parameters = self.parameters.clone();	17✔
314	let waker = cx.waker().clone();	17✔
315
316	crate::util::spawn_blocking(move \|\| {	17✔
317	let mut csv_reader = safe_lock_mutex(&reader_state);	17✔
318	let computation_result = \|\| -> Result<Option<MultiPointCollection>> {	17✔
319	// TODO: is clone necessary?
320	let geometry_specification = parameters.geometry.clone();	17✔
321	csv_reader.setup_once(geometry_specification)?;	17✔
322
323	let (header, records) = match &mut *csv_reader {	16✔
324	ReaderState::OnGoing { header, records } => (header, records),	15✔
325	ReaderState::Error => return Ok(None),	1✔
326	ReaderState::Untouched(_) => unreachable!(),	×
327	};
328
329	let mut builder = MultiPointCollection::builder().finish_header();	15✔
330	let mut number_of_entries = 0; // TODO: add size/len to builder	15✔
331
332	while number_of_entries < chunk_size {	32✔
333	let Some(record) = records.next() else {	29✔
334	break;	11✔
335	};
336
337	let row = record.context(error::CsvSourceReader)?;	18✔
338	let parsed_row = CsvSourceStream::parse_row(header, &row)?;	17✔
339
340	// TODO: filter time
341	if bbox.contains_coordinate(&parsed_row.coordinate) {	17✔
342	builder.push_geometry(parsed_row.coordinate.into());	16✔
343	builder.push_time_interval(parsed_row.time_interval);	16✔
344	builder.finish_row();	16✔
345		16✔
346	number_of_entries += 1;	16✔
347	}	16✔
348	}
349
350	// TODO: is this the correct cancellation criterion?
351	if number_of_entries > 0 {	14✔
352	let collection = builder.build()?;	8✔
353	Ok(Some(collection))	8✔
354	} else {
355	Ok(None)	6✔
356	}
357	}();	17✔
358
359	*safe_lock_mutex(&poll_result) = Some(match computation_result {	17✔
360	Ok(Some(collection)) => Some(Ok(collection)),	8✔
361	Ok(None) => None,	7✔
362	Err(e) => Some(Err(e)),	2✔
363	});
364	is_working.store(false, Ordering::Relaxed);	17✔
365
366	waker.wake();	17✔
367	});	17✔
368
369	Poll::Pending	17✔
370	}	34✔
371	}
372
373	#[derive(Debug)]
374	struct CsvSourceProcessor {
375	params: CsvSourceParameters,
376	result_descriptor: VectorResultDescriptor,
377	}
378
379	#[async_trait]
380	impl QueryProcessor for CsvSourceProcessor {
381	type Output = MultiPointCollection;
382	type SpatialBounds = BoundingBox2D;
383	type Selection = ColumnSelection;
384	type ResultDescription = VectorResultDescriptor;
385
386	async fn _query<'a>(
387	&'a self,
388	query: VectorQueryRectangle,
389	_ctx: &'a dyn QueryContext,
390	) -> Result<BoxStream<'a, Result<Self::Output>>> {	4✔
391	// TODO: properly handle chunk_size
392	Ok(CsvSourceStream::new(self.params.clone(), query.spatial_bounds(), 10)?.boxed())
393	}	4✔
394
395	fn result_descriptor(&self) -> &VectorResultDescriptor {	13✔
396	&self.result_descriptor	13✔
397	}	13✔
398	}
399
400	#[derive(Clone, Copy, Debug)]
401	struct ParsedHeader {
402	pub has_header: bool,
403	pub x_index: usize,
404	pub y_index: usize,
405	}
406
407	struct ParsedRow {
408	pub coordinate: Coordinate2D,
409	pub time_interval: TimeInterval,
410	// TODO: fields
411	}
412
413	#[cfg(test)]
414	mod tests {
415	use super::*;
416	use crate::engine::MockExecutionContext;
417	use geoengine_datatypes::{
418	collections::{FeatureCollectionInfos, ToGeoJson},
419	util::test::TestDefault,
420	};
421	use std::io::{Seek, SeekFrom, Write};
422
423	#[test]
424	fn it_deserializes() {	1✔
425	let json_string = r#"	1✔
426	{	1✔
427	"type": "CsvSource",	1✔
428	"params": {	1✔
429	"filePath": "/foo/bar.csv",	1✔
430	"fieldSeparator": ",",	1✔
431	"geometry": {	1✔
432	"type": "xy",	1✔
433	"x": "x",	1✔
434	"y": "y"	1✔
435	}	1✔
436	}	1✔
437	}"#;	1✔
438
439	let operator: CsvSource = serde_json::from_str(json_string).unwrap();	1✔
440
441	assert_eq!(	1✔
442	operator,
443	CsvSource {	1✔
444	params: CsvSourceParameters {	1✔
445	file_path: "/foo/bar.csv".into(),	1✔
446	field_separator: ',',	1✔
447	geometry: CsvGeometrySpecification::XY {	1✔
448	x: "x".into(),	1✔
449	y: "y".into()	1✔
450	},	1✔
451	time: CsvTimeSpecification::None,	1✔
452	},	1✔
453	}	1✔
454	);
455	}	1✔
456
457	#[tokio::test]
458	async fn read_points() {	1✔
459	let mut fake_file = tempfile::NamedTempFile::new().unwrap();	1✔
460	write!(	1✔
461	fake_file,	1✔
462	"\	1✔
463	x,y	1✔
464	0,1	1✔
465	2,3	1✔
466	4,5	1✔
467	"	1✔
468	)
469	.unwrap();	1✔
470	fake_file.seek(SeekFrom::Start(0)).unwrap();	1✔
471
472	let mut csv_source = CsvSourceStream::new(	1✔
473	CsvSourceParameters {	1✔
474	file_path: fake_file.path().into(),	1✔
475	field_separator: ',',	1✔
476	geometry: CsvGeometrySpecification::XY {	1✔
477	x: "x".into(),	1✔
478	y: "y".into(),	1✔
479	},	1✔
480	time: CsvTimeSpecification::None,	1✔
481	},	1✔
482	BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),	1✔
483	2,
484	)
485	.unwrap();	1✔
486
487	assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 2);	1✔
488	assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);	1✔
489	assert!(csv_source.next().await.is_none());	1✔
490	}	1✔
491
492	#[tokio::test]
493	async fn erroneous_point_rows() {	1✔
494	let mut fake_file = tempfile::NamedTempFile::new().unwrap();	1✔
495	write!(	1✔
496	fake_file,	1✔
497	"\	1✔
498	x,y	1✔
499	0,1	1✔
500	CORRUPT	1✔
501	4,5	1✔
502	"	1✔
503	)
504	.unwrap();	1✔
505	fake_file.seek(SeekFrom::Start(0)).unwrap();	1✔
506
507	let mut csv_source = CsvSourceStream::new(	1✔
508	CsvSourceParameters {	1✔
509	file_path: fake_file.path().into(),	1✔
510	field_separator: ',',	1✔
511	geometry: CsvGeometrySpecification::XY {	1✔
512	x: "x".into(),	1✔
513	y: "y".into(),	1✔
514	},	1✔
515	time: CsvTimeSpecification::None,	1✔
516	},	1✔
517	BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),	1✔
518	1,
519	)
520	.unwrap();	1✔
521
522	assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);	1✔
523	assert!(csv_source.next().await.unwrap().is_err());	1✔
524	assert_eq!(csv_source.next().await.unwrap().unwrap().len(), 1);	1✔
525	assert!(csv_source.next().await.is_none());	1✔
526	}	1✔
527
528	#[tokio::test]
529	async fn corrupt_point_header() {	1✔
530	let mut fake_file = tempfile::NamedTempFile::new().unwrap();	1✔
531	write!(	1✔
532	fake_file,	1✔
533	"\	1✔
534	x,z	1✔
535	0,1	1✔
536	2,3	1✔
537	4,5	1✔
538	"	1✔
539	)
540	.unwrap();	1✔
541	fake_file.seek(SeekFrom::Start(0)).unwrap();	1✔
542
543	let mut csv_source = CsvSourceStream::new(	1✔
544	CsvSourceParameters {	1✔
545	file_path: fake_file.path().into(),	1✔
546	field_separator: ',',	1✔
547	geometry: CsvGeometrySpecification::XY {	1✔
548	x: "x".into(),	1✔
549	y: "y".into(),	1✔
550	},	1✔
551	time: CsvTimeSpecification::None,	1✔
552	},	1✔
553	BoundingBox2D::new_unchecked((0., 0.).into(), (5., 5.).into()),	1✔
554	1,
555	)
556	.unwrap();	1✔
557
558	assert!(csv_source.next().await.unwrap().is_err());	1✔
559	assert!(csv_source.next().await.is_none());	1✔
560	}	1✔
561
562	#[tokio::test]
563	async fn processor() {	1✔
564	let mut fake_file = tempfile::NamedTempFile::new().unwrap();	1✔
565	write!(	1✔
566	fake_file,	1✔
567	"\	1✔
568	x,y	1✔
569	0,1	1✔
570	2,3	1✔
571	4,5	1✔
572	"	1✔
573	)
574	.unwrap();	1✔
575	fake_file.seek(SeekFrom::Start(0)).unwrap();	1✔
576
577	let params = CsvSourceParameters {	1✔
578	file_path: fake_file.path().into(),	1✔
579	field_separator: ',',	1✔
580	geometry: CsvGeometrySpecification::XY {	1✔
581	x: "x".into(),	1✔
582	y: "y".into(),	1✔
583	},	1✔
584	time: CsvTimeSpecification::None,	1✔
585	};	1✔
586
587	let p = CsvSourceProcessor {	1✔
588	params,	1✔
589	result_descriptor: VectorResultDescriptor {	1✔
590	data_type: VectorDataType::MultiPoint,	1✔
591	spatial_reference: SpatialReference::epsg_4326().into(),	1✔
592	columns: Default::default(),	1✔
593	time: None,	1✔
594	bbox: None,	1✔
595	},	1✔
596	};	1✔
597
598	let query = VectorQueryRectangle::new(	1✔
599	BoundingBox2D::new_unchecked(Coordinate2D::new(0., 0.), Coordinate2D::new(3., 3.)),	1✔
600	TimeInterval::new_unchecked(0, 1),	1✔
601	ColumnSelection::all(),	1✔
602	);
603	let ecx = MockExecutionContext::test_default();	1✔
604	let ctx = ecx.mock_query_context((10 * 8 * 2).into());	1✔
605
606	let r: Vec<Result<MultiPointCollection>> =	1✔
607	p.query(query, &ctx).await.unwrap().collect().await;	1✔
608
609	assert_eq!(r.len(), 1);	1✔
610
611	assert_eq!(	1✔
612	serde_json::from_str::<serde_json::Value>(&r[0].as_ref().unwrap().to_geo_json())	1✔
613	.unwrap(),	1✔
614	serde_json::json!({	1✔
615	"type": "FeatureCollection",	1✔
616	"features": [{	1✔
617	"type": "Feature",	1✔
618	"geometry": {	1✔
619	"type": "Point",	1✔
620	"coordinates": [0.0, 1.0]	1✔
621	},	1✔
622	"properties": {},	1✔
623	"when": {	1✔
624	"start": "-262143-01-01T00:00:00+00:00",	1✔
625	"end": "+262142-12-31T23:59:59.999+00:00",	1✔
626	"type": "Interval"	1✔
627	}	1✔
628	}, {	1✔
629	"type": "Feature",	1✔
630	"geometry": {	1✔
631	"type": "Point",	1✔
632	"coordinates": [2.0, 3.0]	1✔
633	},	1✔
634	"properties": {},	1✔
635	"when": {	1✔
636	"start": "-262143-01-01T00:00:00+00:00",	1✔
637	"end": "+262142-12-31T23:59:59.999+00:00",	1✔
638	"type": "Interval"	1✔
639	}	1✔
640	}]	1✔
641	})	1✔
642	);	1✔
643	}	1✔
644
645	#[test]
646	fn operator() {	1✔
647	let mut temp_file = tempfile::NamedTempFile::new().unwrap();	1✔
648	write!(	1✔
649	temp_file,	1✔
650	"\	1✔
651	x;y	1✔
652	0;1	1✔
653	2;3	1✔
654	4;5	1✔
655	"	1✔
656	)
657	.unwrap();	1✔
658	temp_file.seek(SeekFrom::Start(0)).unwrap();	1✔
659
660	let params = CsvSourceParameters {	1✔
661	file_path: temp_file.path().into(),	1✔
662	field_separator: ';',	1✔
663	geometry: CsvGeometrySpecification::XY {	1✔
664	x: "x".into(),	1✔
665	y: "y".into(),	1✔
666	},	1✔
667	time: CsvTimeSpecification::None,	1✔
668	};	1✔
669
670	let operator = CsvSource { params }.boxed();	1✔
671
672	let operator_json = serde_json::to_value(&operator).unwrap();	1✔
673
674	assert_eq!(	1✔
675	operator_json,
676	serde_json::json!({	1✔
677	"type": "CsvSource",	1✔
678	"params": {	1✔
679	"filePath": temp_file.path(),	1✔
680	"fieldSeparator": ";",	1✔
681	"geometry": {	1✔
682	"type": "xy",	1✔
683	"x": "x",	1✔
684	"y": "y"	1✔
685	},
686	"time": "None"	1✔
687	}
688	})
689	);
690
691	let _operator: Box<dyn VectorOperator> = serde_json::from_value(operator_json).unwrap();	1✔
692	}	1✔
693	}

geo-engine / geoengine / 19241805651

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous