bourbon

actes-princiers/conf/base/catalog.yml

@@ -10,32 +10,32 @@ bourbon_xmlcontent:
   housename: bourbon
   folderpath: data/02_intermediate/houses/bourbon/xml
 
-#bourbon_json:
-#  type: actesdataset.XMLDataSetCollection
-#  housename: bourbon
-#  folderpath: data/02_intermediate/houses/bourbon/json
+bourbon_json:
+  type: actesdataset.XMLDataSetCollection
+  housename: bourbon
+  folderpath: data/02_intermediate/houses/bourbon/json
 
-# ________________________________________________________________________
+## ________________________________________________________________________
 
-berry:
-  type: actesdataset.XMLDataSetCollection
-  housename: berry
-  folderpath: data/01_raw/houses/berry
+#berry:
+#  type: actesdataset.XMLDataSetCollection
+#  housename: berry
+#  folderpath: data/01_raw/houses/berry
 
-berry_xmlcontent:
-  type: actesdataset.XMLDataSetCollection
-  housename: berry
-  folderpath: data/02_intermediate/houses/berry/xml
+#berry_xmlcontent:
+#  type: actesdataset.XMLDataSetCollection
+#  housename: berry
+#  folderpath: data/02_intermediate/houses/berry/xml
 
-# ________________________________________________________________________
+## ________________________________________________________________________
 
-anjou:
-  type: actesdataset.XMLDataSetCollection
-  housename: berry
-  folderpath: data/01_raw/houses/anjou
+#anjou:
+#  type: actesdataset.XMLDataSetCollection
+#  housename: berry
+#  folderpath: data/01_raw/houses/anjou
 
-anjou_xmlcontent:
-  type: actesdataset.XMLDataSetCollection
-  housename: berry
-  folderpath: data/02_intermediate/houses/anjou/xml
+#anjou_xmlcontent:
+#  type: actesdataset.XMLDataSetCollection
+#  housename: berry
+#  folderpath: data/02_intermediate/houses/anjou/xml
 

actes-princiers/src/actes_princiers/pipelines/xml_processing/nodes.py

@@ -8,16 +8,17 @@ from actesdataset import EtreeXMLDataSet, XMLDataSetCollection
 
 logger = logging.getLogger(__name__)
 
-with KedroSession.create() as session:
-    context = session.load_context()
-    catalog = context.get_catalog()
+#with KedroSession.create() as session:
+#    context = session.load_context()
+#    catalog = context.get_catalog()
 
 def parse_xml_collection(datasetcol: XMLDataSetCollection) -> Dict[str, EtreeXMLDataSet]:
     "node function entry point, performs batch processing"
     datasets = datasetcol.datasets
-    housename = datasetcol.housename
-    output_catalog = catalog[housename + '_xmlcontent']
-    outputfolderpath = output_catalog['folderpath']
+    housename = datasetcol._housename
+    outputfolderpath = f"data/02_intermediate/houses/{housename}/xml"    
+#    output_catalog = catalog[housename + '_xmlcontent']
+#    outputfolderpath = output_catalog['folderpath']
     output_datasets = dict()
     for dataset_filenamestem, dataset in datasets.items():
         # a manual load is required here, because

actes-princiers/src/actes_princiers/pipelines/xml_processing/pipeline.py

@@ -9,22 +9,22 @@ def create_pipeline(**kwargs) -> Pipeline:
         [
             node(
                 func=parse_xml_collection,
-                inputs=["bourbon"],
-                outputs="bourbon_xmlcontent",
+                inputs="bourbon",
+                outputs=None, #"bourbon_xmlcontent",
                 name="bourbon_ds_collection",
             ),
-            node(
-                func=parse_xml_collection,
-                inputs=["berry"],
-                outputs="berry_xmlcontent",
-                name="berry_ds_collection",
-            ),
-            node(
-                func=parse_xml_collection,
-                inputs=["anjou"],
-                outputs="anjou_xmlcontent",
-                name="anjou_ds_collection",
-            ),
+#            node(
+#                func=parse_xml_collection,
+#                inputs="berry",
+#                outputs=None, #"berry_xmlcontent",
+#                name="berry_ds_collection",
+#            ),
+#            node(
+#                func=parse_xml_collection,
+#                inputs="anjou",
+#                outputs=None, # "anjou_xmlcontent",
+#                name="anjou_ds_collection",
+#            ),
 
         ]
     )

actes-princiers/src/actesdataset.py

@@ -122,7 +122,7 @@ class XMLDataSetCollection(AbstractDataSet):
     def __init__(self, 
         housename: str,
         folderpath: str) -> None:
-        self.housename = housename
+        self._housename = housename
         self._folderpath = Path(folderpath)
                     
     def _load(self) -> dict[str, EtreeXMLDataSet]:
@@ -145,7 +145,7 @@ class XMLDataSetCollection(AbstractDataSet):
         
     def _describe(self) -> dict[str, Any]:
         "kedro's API repr()"
-        return dict(name=self.housename, folderpath=self._folderpath)
+        return dict(name=self._housename, folderpath=self._folderpath)
 
 #class TextDataSet:
 #    """loads/saves data from/to a text file using an underlying filesystem