Merge pull request #52 from weso/issue51

Add DCTap support

Author: labra

dctap/Cargo.toml

@@ -11,8 +11,8 @@ repository.workspace = true
 
 [dependencies]
 csv = "1.3.0"
+indexmap = { version = "2"}
 serde = "1.0"
 serde_derive = "1.0"
-
 thiserror = "1"
 tracing = "0.1"

dctap/src/dctap.rs

@@ -1,14 +1,16 @@
-use crate::{tap_config::TapConfig, tap_error::TapError};
+use crate::{tap_config::TapConfig, tap_error::TapError, ShapeId, TapReaderBuilder, TapShape};
+use indexmap::IndexMap;
 use serde_derive::{Deserialize, Serialize};
-use std::path::Path;
+use std::{io, path::Path};
 use tracing::debug;
 
 #[derive(Debug, Serialize, Deserialize)]
 struct TapShapeId(String);
 
-#[derive(Debug, Serialize)]
+#[derive(Debug, Serialize, PartialEq)]
 pub struct DCTap {
     version: String,
+    shapes: IndexMap<Option<ShapeId>, TapShape>,
 }
 
 impl Default for DCTap {
@@ -21,12 +23,58 @@ impl DCTap {
     pub fn new() -> DCTap {
         DCTap {
             version: "0.1".to_string(),
+            shapes: IndexMap::new(),
         }
     }
 
-    pub fn read_buf(_path: &Path, _config: TapConfig) -> Result<DCTap, TapError> {
-        let dctap = DCTap::new();
+    pub fn add_shape(&mut self, shape: &TapShape) {
+        self.shapes.insert(shape.shape_id(), shape.clone());
+    }
+
+    pub fn from_path(path: &Path, _config: TapConfig) -> Result<DCTap, TapError> {
+        let mut dctap = DCTap::new();
+        debug!("DCTap parsed: {:?}", dctap);
+        let mut tap_reader = TapReaderBuilder::new().flexible(true).from_path(path)?;
+        for maybe_shape in tap_reader.shapes() {
+            let shape = maybe_shape?;
+            println!("Shape read: {shape:?}");
+            dctap.add_shape(&shape)
+        }
+        Ok(dctap)
+    }
+
+    pub fn from_reader<R: io::Read>(reader: R) -> Result<DCTap, TapError> {
+        let mut dctap = DCTap::new();
         debug!("DCTap parsed: {:?}", dctap);
+        let mut tap_reader = TapReaderBuilder::new().flexible(true).from_reader(reader)?;
+        for maybe_shape in tap_reader.shapes() {
+            let shape = maybe_shape?;
+            println!("Shape read: {shape:?}");
+            dctap.add_shape(&shape)
+        }
         Ok(dctap)
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use crate::{PropertyId, TapShape, TapStatement};
+
+    use super::*;
+
+    #[test]
+    fn test_simple() {
+        let data = "\
+shapeId,shapeLabel,propertyId,propertyLabel
+Person,PersonLabel,knows,KnowsLabel
+";
+        let dctap = DCTap::from_reader(data.as_bytes()).unwrap();
+        let mut expected_shape = TapShape::new();
+        expected_shape.set_shape_id(&ShapeId::new("Person"));
+        expected_shape.add_statement(TapStatement::new(PropertyId::new("knows")));
+
+        let mut expected_dctap = DCTap::new();
+        expected_dctap.add_shape(&expected_shape);
+        assert_eq!(dctap, expected_dctap);
+    }
+}

dctap/src/lib.rs

@@ -8,11 +8,41 @@
 pub mod dctap;
 pub mod tap_config;
 pub mod tap_error;
+pub mod tap_headers;
+pub mod tap_reader;
 pub mod tap_shape;
 pub mod tap_statement;
 
 pub use crate::tap_config::*;
 pub use crate::tap_error::*;
+pub use crate::tap_reader::*;
 pub use crate::tap_shape::*;
 pub use crate::tap_statement::*;
 pub use dctap::*;
+use serde_derive::{Deserialize, Serialize};
+
+#[derive(Deserialize, Serialize, Debug, PartialEq, Default, Clone)]
+pub struct PropertyId {
+    str: String,
+}
+
+impl PropertyId {
+    pub fn new(str: &str) -> PropertyId {
+        PropertyId {
+            str: str.to_string(),
+        }
+    }
+}
+
+#[derive(Deserialize, Serialize, Debug, Hash, PartialEq, Eq, Clone)]
+pub struct ShapeId {
+    str: String,
+}
+
+impl ShapeId {
+    pub fn new(str: &str) -> ShapeId {
+        ShapeId {
+            str: str.to_string(),
+        }
+    }
+}

dctap/src/tap_error.rs

@@ -1,4 +1,21 @@
+use std::result;
+
+use csv::StringRecord;
 use thiserror::Error;
 
+pub type Result<T> = result::Result<T, TapError>;
+
 #[derive(Error, Debug)]
-pub enum TapError {}
+pub enum TapError {
+    #[error("CSV Error: {err}")]
+    RDFParseError {
+        #[from]
+        err: csv::Error,
+    },
+
+    #[error("Cannot obtain shape id with index {shape_id} from record {record:?}")]
+    NoShapeId {
+        shape_id: usize,
+        record: StringRecord,
+    },
+}

dctap/src/tap_headers.rs

@@ -0,0 +1,46 @@
+use crate::tap_error::Result;
+use csv::StringRecord;
+use tracing::debug;
+
+#[derive(Debug, Default)]
+pub(crate) struct TapHeaders {
+    shape_id: Option<usize>,
+    property_id: Option<usize>,
+}
+
+impl TapHeaders {
+    pub(crate) fn new() -> TapHeaders {
+        TapHeaders::default()
+    }
+
+    pub(crate) fn from_record(record: &StringRecord) -> Result<TapHeaders> {
+        let mut shape_id = None;
+        let mut property_id = None;
+
+        for (idx, field) in record.iter().enumerate() {
+            match clean(field).as_str() {
+                "SHAPEID" => shape_id = Some(idx),
+                "PROPERTYID" => property_id = Some(idx),
+                _ => {
+                    debug!("Unknown field reading headers: {field}")
+                }
+            }
+        }
+        Ok(TapHeaders {
+            shape_id,
+            property_id,
+        })
+    }
+
+    pub fn shape_id(&self) -> Option<usize> {
+        self.shape_id
+    }
+
+    pub fn property_id(&self) -> Option<usize> {
+        self.property_id
+    }
+}
+
+fn clean(str: &str) -> String {
+    str.to_uppercase()
+}