ENH: add "driver" property to read_info result (#224)

Author: theroggy

CHANGES.md

@@ -1,5 +1,11 @@
 # CHANGELOG
 
+## 0.6.0 (???)
+
+### Improvements
+
+- ENH: add "driver" property to read_info result (#224)
+
 ## 0.5.1 (2023-01-26)
 
 ### Bug fixes

docs/source/introduction.md

@@ -76,14 +76,16 @@ first layer.
 >>> from pyogrio import read_info
 >>> read_info('ne_10m_admin_0_countries.shp')
 
-# Outputs a dictionary with `crs`, `encoding`, `fields`, `geometry_type`, and `features`
+# Outputs a dictionary with `crs`, `driver`, `encoding`, `fields`, `geometry_type`, and
+# `features`
 {
   'crs': 'EPSG:4326',
   'encoding': 'UTF-8',
   'fields': array(['featurecla', 'scalerank', 'LABELRANK', ...], dtype=object),
   'dtypes': array(['int64', 'object', 'object', 'object', 'float64'], dtype=object),
   'geometry_type': 'Polygon',
-  'features': 255
+  'features': 255,
+  'driver': 'ESRI Shapefile',
 }
 ```
 

pyogrio/_io.pyx

@@ -264,35 +264,49 @@ cdef str get_crs(OGRLayerH ogr_layer):
         return wkt
 
 
-cdef detect_encoding(OGRDataSourceH ogr_dataset, OGRLayerH ogr_layer):
-    """Attempt to detect the encoding of the layer.
-    If it supports UTF-8, use that.
-    If it is a shapefile, it must otherwise be ISO-8859-1.
-
+cdef get_driver(OGRDataSourceH ogr_dataset):
+    """Get the driver for a dataset.
+    
     Parameters
     ----------
     ogr_dataset : pointer to open OGR dataset
-    ogr_layer : pointer to open OGR layer
-
     Returns
     -------
     str or None
     """
     cdef void *ogr_driver
 
-    if OGR_L_TestCapability(ogr_layer, OLCStringsAsUTF8):
-        return 'UTF-8'
-
     try:
         ogr_driver = exc_wrap_pointer(GDALGetDatasetDriver(ogr_dataset))
 
     except NullPointerError:
-        raise DataLayerError(f"Could not detect encoding of layer") from None
+        raise DataLayerError(f"Could not detect driver of dataset") from None
 
     except CPLE_BaseError as exc:
         raise DataLayerError(str(exc))
 
     driver = OGR_Dr_GetName(ogr_driver).decode("UTF-8")
+    return driver
+
+
+cdef detect_encoding(OGRDataSourceH ogr_dataset, OGRLayerH ogr_layer):
+    """Attempt to detect the encoding of the layer.
+    If it supports UTF-8, use that.
+    If it is a shapefile, it must otherwise be ISO-8859-1.
+
+    Parameters
+    ----------
+    ogr_dataset : pointer to open OGR dataset
+    ogr_layer : pointer to open OGR layer
+
+    Returns
+    -------
+    str or None
+    """
+    if OGR_L_TestCapability(ogr_layer, OLCStringsAsUTF8):
+        return 'UTF-8'
+
+    driver = get_driver(ogr_dataset)
     if driver == 'ESRI Shapefile':
         return 'ISO-8859-1'
 
@@ -973,7 +987,7 @@ def ogr_read(
             'crs': crs,
             'encoding': encoding,
             'fields': fields[:,2], # return only names
-            'geometry_type': geometry_type
+            'geometry_type': geometry_type,
         }
 
     finally:
@@ -1205,6 +1219,7 @@ def ogr_read_info(str path, object layer=None, object encoding=None, **kwargs):
         'dtypes': fields[:,3],
         'geometry_type': get_geometry_type(ogr_layer),
         'features': OGR_L_GetFeatureCount(ogr_layer, 1),
+        'driver': get_driver(ogr_dataset),
         "capabilities": {
             "random_read": OGR_L_TestCapability(ogr_layer, OLCRandomRead),
             "fast_set_next_by_index": OGR_L_TestCapability(ogr_layer, OLCFastSetNextByIndex),

pyogrio/core.py

@@ -172,7 +172,8 @@ def read_info(path_or_buffer, /, layer=None, encoding=None):
                 "dtypes": <ndarray of field dtypes>,
                 "encoding": "<encoding>",
                 "geometry": "<geometry type>",
-                "features": <feature count>
+                "features": <feature count>,
+                "driver": "<driver>",
             }
     """
     path, buffer = get_vsi_path(path_or_buffer)

pyogrio/raw.py

@@ -118,7 +118,7 @@ def read(
             "crs": "<crs>",
             "fields": <ndarray of field names>,
             "encoding": "<encoding>",
-            "geometry": "<geometry type>"
+            "geometry_type": "<geometry type>"
         }
     """
     path, buffer = get_vsi_path(path_or_buffer)

pyogrio/tests/test_core.py

@@ -227,6 +227,7 @@ def test_read_info(naturalearth_lowres):
     assert meta["fields"].shape == (5,)
     assert meta["dtypes"].tolist() == ["int64", "object", "object", "object", "float64"]
     assert meta["features"] == 177
+    assert meta["driver"] == "ESRI Shapefile"
 
 
 @pytest.mark.parametrize(