Fixes #170 - remove dbg print instead of error

CHANGELOG.md

@@ -7,6 +7,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+- `DataParseError::CodebookAndDataRowsMismatch` variant for when the number of rows in the codebook and the number of rows in the data do not match.
+
+### Fixed
+- Initializing an engine with a codebook that has a different number of rows than the data will result in an error instead of printing a bunch on nonsense.
+
 ## [python-0.6.0] - 2024-01-23
 
 ### Added

lace/src/data/init.rs

@@ -287,17 +287,14 @@ pub fn df_to_col_models<R: rand::Rng>(
         })
         .collect::<Result<_, DataParseError>>()?;
 
-    if col_models
-        .iter()
-        .any(|cm| cm.len() != codebook.row_names.len())
-    {
-        dbg!(
-            col_models.iter().map(|cm| cm.len()).collect::<Vec<_>>(),
-            codebook.row_names.len()
-        );
-        // FIXME!
-        // return Err(CsvParseError::CodebookAndDataRowMismatch);
+    let n_codebook_rows = codebook.row_names.len();
+    if let Some(cm) = col_models.iter().find(|cm| cm.len() != n_codebook_rows) {
+        return Err(DataParseError::CodebookAndDataRowsMismatch {
+            n_codebook_rows,
+            n_data_rows: cm.len(),
+        });
     }
+
     Ok((codebook, col_models))
 }
 

lace/src/interface/engine/error.rs

@@ -18,9 +18,9 @@ pub enum DataParseError {
     /// The supplied data source is not currently supported for this operation
     #[error("Provided an unsupported data source")]
     UnsupportedDataSource,
-    /// The user supplied column_metdata in the codebook but provided an empty
+    /// The user supplied column_metadata in the codebook but provided an empty
     /// data source
-    #[error("non-empty column_metdata the codebook but empty DataSource")]
+    #[error("non-empty column_metadata the codebook but empty DataSource")]
     ColumnMetadataSuppliedForEmptyData,
     /// The user supplied row_names in the codebook but provided an empty
     /// data source
@@ -36,6 +36,12 @@ pub enum DataParseError {
     /// externally
     #[error("Column `{col_name}` has type `{col_type}`, which is unsupported for external data sources")]
     UnsupportedColumnType { col_name: String, col_type: String },
+    /// The codebook and the data have a different number of rows
+    #[error("The codebook contains {n_codebook_rows} rows, but the data contain {n_data_rows} rows")]
+    CodebookAndDataRowsMismatch {
+        n_codebook_rows: usize,
+        n_data_rows: usize,
+    },
 }
 
 /// Errors that can arise when creating a new engine