Feature: respect newline characters in strings

src/geometamaker/geometamaker.py

@@ -261,7 +261,7 @@ def describe_raster(source_dataset_path, scheme):
         b = i + 1
         bands.append(models.BandSchema(
             index=b,
-            gdal_type=info['datatype'],
+            gdal_type=gdal.GetDataTypeName(info['datatype']),
             numpy_type=numpy.dtype(info['numpy_type']).name,
             nodata=info['nodata'][i]))
     description['schema'] = models.RasterSchema(

src/geometamaker/models.py

@@ -8,23 +8,12 @@
 import yaml
 
 import geometamaker
+from . import utils
 
 
 LOGGER = logging.getLogger(__name__)
 
 
-# https://stackoverflow.com/questions/13518819/avoid-references-in-pyyaml
-class _NoAliasDumper(yaml.SafeDumper):
-    """Keep the yaml human-readable by avoiding anchors and aliases."""
-
-    def ignore_aliases(self, data):
-        return True
-
-
-def _yaml_dump(data):
-    return yaml.dump(data, allow_unicode=True, Dumper=_NoAliasDumper)
-
-
 @dataclass(frozen=True)
 class BoundingBox:
     """Class for a spatial bounding box."""
@@ -107,7 +96,7 @@ class BandSchema:
     """Class for metadata for a raster band."""
 
     index: int
-    gdal_type: int
+    gdal_type: str
     numpy_type: str
     nodata: int | float
     description: str = ''
@@ -275,7 +264,7 @@ def write(self, target_path):
 
         """
         with open(target_path, 'w') as file:
-            file.write(_yaml_dump(dataclasses.asdict(self)))
+            file.write(utils.yaml_dump(dataclasses.asdict(self)))
 
 
 @dataclass()
@@ -520,7 +509,7 @@ def write(self, workspace=None):
                 workspace, os.path.basename(self.metadata_path))
 
         with open(target_path, 'w') as file:
-            file.write(_yaml_dump(dataclasses.asdict(self)))
+            file.write(utils.yaml_dump(dataclasses.asdict(self)))
 
     def to_string(self):
         pass

src/geometamaker/utils.py

@@ -0,0 +1,29 @@
+import yaml
+
+
+def _represent_str(dumper, data):
+    scalar = yaml.representer.SafeRepresenter.represent_str(dumper, data)
+    if len(data.splitlines()) > 1:
+        scalar.style = '|'  # literal style, newline chars will be new lines
+    return scalar
+
+
+class _SafeDumper(yaml.SafeDumper):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        # Patch the default string representer to use a literal block
+        # style when the data contain newline characters
+        self.add_representer(str, _represent_str)
+
+    # https://stackoverflow.com/questions/13518819/avoid-references-in-pyyaml
+    def ignore_aliases(self, data):
+        """Keep the yaml human-readable by avoiding anchors and aliases."""
+        return True
+
+
+def yaml_dump(data):
+    return yaml.dump(
+        data,
+        allow_unicode=True,
+        Dumper=_SafeDumper)

tests/test_geometamaker.py

@@ -255,7 +255,8 @@ def test_raster_attributes(self):
         band = resource.schema.bands[band_idx]
         self.assertEqual(band.title, title)
         self.assertEqual(band.description, description)
-        self.assertEqual(band.gdal_type, raster_info['datatype'])
+        self.assertEqual(
+            band.gdal_type, gdal.GetDataTypeName(raster_info['datatype']))
         self.assertEqual(band.numpy_type, numpy.dtype(numpy_type).name)
         self.assertEqual(band.nodata, raster_info['nodata'][band_idx])
         self.assertEqual(band.units, units)
@@ -395,11 +396,28 @@ def test_set_and_get_lineage(self):
         import geometamaker
 
         resource = geometamaker.models.Resource()
-        statement = 'a lineage statment'
+        statement = 'a lineage statment\n is long and has\n many lines.'
 
         resource.set_lineage(statement)
         self.assertEqual(resource.get_lineage(), statement)
 
+    def test_lineage_roundtrip(self):
+        """Test writing and loading yaml with block indicator."""
+        import geometamaker
+
+        datasource_path = os.path.join(self.workspace_dir, 'raster.tif')
+        numpy_type = numpy.int16
+        create_raster(numpy_type, datasource_path)
+
+        resource = geometamaker.describe(datasource_path)
+
+        statement = 'a lineage statment\n is long and has\n many lines.'
+        resource.set_lineage(statement)
+        resource.write()
+
+        new_resource = geometamaker.describe(datasource_path)
+        self.assertEqual(new_resource.get_lineage(), statement)
+
     def test_set_and_get_purpose(self):
         """Test set and get purpose of resource."""
         import geometamaker