add xslt stylesheet in the catalog collection

actes-princiers/conf/base/catalog.yml

@@ -5,12 +5,14 @@ bourbon:
   type: actesdataset.XMLDataSetCollection
   housename: bourbon
   folderpath: data/01_raw/xml/Bourbon
+  xsltstylesheet: templates/xsl/actes_princiers.xsl
   
 # output (write) **pseudo xml** dataset
 bourbon_xmlcontent:
   type: actesdataset.XMLDataSetCollection
   housename: bourbon
   folderpath: data/02_intermediate/xml/Bourbon/xml
+  xsltstylesheet: templates/xsl/actes_princiers.xsl
   
 # input (read) **pseudo xml** dataset
 bourbon_pseudoxmlcontent:

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

@@ -12,10 +12,9 @@ logger = logging.getLogger(__name__)
 
 with KedroSession.create() as session:
     context = session.load_context()
-#    catalog = context.get_catalog() # FIXME : ça porte à confusion de renvoyer un dict
+#    catalog = context.get_catalog()
 
-
+def parse_xml_collection(datasetcol: XMLDataSetCollection, params: str) -> XMLDataSetCollection:
-def parse_xml_collection(datasetcol: XMLDataSetCollection) -> XMLDataSetCollection:
     "node function entry point, performs batch processing"
     datasets = datasetcol.datasets
     housename = datasetcol._housename
@@ -28,7 +27,7 @@ def parse_xml_collection(datasetcol: XMLDataSetCollection) -> XMLDataSetCollecti
         output_source_doc = dataset.transform() 
         # set dataset's output filepath
         output_filepath = outputfolderpath / Path(dataset_filenamestem).with_suffix(".pseudoxml")
-        output_xmldataset = EtreeXMLDataSet(str(output_filepath))
+        output_xmldataset = EtreeXMLDataSet(str(output_filepath), params)
         # let's create subfolders, if they don't exist
         output_xmldataset_dir = output_filepath.parent
         output_xmldataset_dir.mkdir(parents=True, exist_ok=True)

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

@@ -10,7 +10,7 @@ def create_pipeline(**kwargs) -> Pipeline:
         [
             node(
                 func=parse_xml_collection,
-                inputs="bourbon",
+                inputs=["bourbon", "params:xsltstylesheet"],
                 outputs="bourbon_xmlcontent",
                 name="bourbon_ds_collection",
             ),

actes-princiers/src/actesdataset.py

@@ -13,20 +13,6 @@ from kedro.framework.session import KedroSession
 
 logger = logging.getLogger(__name__)
 
-with KedroSession.create() as session:
-    context = session.load_context()
-    xlststylesheet = context.params['xsltstylesheet']
-        
-#xlststylesheet = "templates/xsl/actes_princiers.xsl"
-# XXX is it usefull to make this bunch of code a classmethod ?
-def _xslt(xsltstylesheet):
-    "performs XML transformation on each dataset"
-    xslt_doc = etree.parse(xlststylesheet)
-    xslt_transformer = etree.XSLT(xslt_doc)
-    return xslt_transformer
-
-xslt_transformer = _xslt(xlststylesheet)
-
 class XMLDataSet(ABC):
     "Abstract base class for an XML dataset loader"
         
@@ -52,6 +38,10 @@ class XMLDataSet(ABC):
 class EtreeXMLDataSet(XMLDataSet):
     "XMLDataSet loader with lxml.etree (lxml.etree._ElementTree)"
 
+    def __init__(self, filepath, params):
+        self._filepath = filepath
+        self.xsltstylesheet = params
+        
     def _load(self):
         "from the xml file loads a internal xml repr (with element tree)"
         # self.source_doc is an etree internal xml repr document
@@ -68,7 +58,15 @@ class EtreeXMLDataSet(XMLDataSet):
         with open(self._filepath, 'w') as fhandle:
             fhandle.write(data)
 
+    @staticmethod
+    def _xslt(xsltstylesheet):
+        "performs XML transformation on each dataset"
+        xslt_doc = etree.parse(xsltstylesheet)
+        xslt_transformer = etree.XSLT(xslt_doc)
+        return xslt_transformer
+
     def transform(self):
+        xslt_transformer = self._xslt(self.xsltstylesheet)
         return str(xslt_transformer(self.source_doc))
 
 class BsXMLDataSet(XMLDataSet):
@@ -149,11 +147,15 @@ class DataSetCollection(AbstractDataSet):
 
 
 class XMLDataSetCollection(DataSetCollection):
+    def __init__(self, housename: str,
+        folderpath: str, xsltstylesheet: str) -> None:
+        super().__init__(housename, folderpath)
+        self.xsltstylesheet = xsltstylesheet
+
     def _load(self) -> dict[str, EtreeXMLDataSet]:
         "kedro's API loader method"
         for filepath in sorted(self._folderpath.glob("*.xml")):
-            self.datasets[filepath.stem] = EtreeXMLDataSet(
+            self.datasets[filepath.stem] = EtreeXMLDataSet(str(filepath), self.xsltstylesheet)
-                filepath=str(filepath))
         return self