passage au wrapper _dom

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

@@ -25,7 +25,7 @@ def parse_xml_collection(datasets: Dict[str, EtreeXMLDataSet], param: str) -> Di
         dataset._load()
         output_source_doc = transform(dataset.get_source_doc(), param)
         # set dataset's output filepath
-        output_filepath = dataset.get_filepath().replace("01_raw", "02_intermediate")
+        output_filepath = dataset.filepath.replace("01_raw", "02_intermediate")
         output_xmldataset = EtreeXMLDataSet(output_filepath)
         output_xmldataset.set_source_doc(output_source_doc)
         output_datasets[dataset_filenamestem] = output_xmldataset

actes-princiers/src/actesdataset.py

@@ -12,17 +12,40 @@ from kedro.framework.session import KedroSession
 logger = logging.getLogger(__name__)
 
 
-# FIXME hériter de abc (cf dans le code de kedro)    
 class XMLDataSet:
     "Abstract base class for an XML dataset loader"
-
+        
     def __init__(self, filepath: str) -> None:
         self._filepath = filepath
+        # xml etree internal representation
+        self._dom = None
+        # xml as an str output
+        self._str = None 
 
-    def get_filepath(self) -> str:
+    def _load(self):
-        "xml file's long filename getters"
+        "kedro's API-like loader"
-        return self._filepath
+        pass
         
+    def _save(self, data:str) -> None:
+        "kedro's API-like saver"
+        pass
+    
+    @property
+    def tostring(self) -> str:
+        "XML source_doc (xml as a string) getter"
+        # FIXME : charger le _dom d'abord, puis génerer le str ici 
+        if getattr(self, '_str') is not None:
+            return self._str
+        else:
+            attr_error_msg = str(self._describe())
+            raise AttributeError(f"XMLDataSet dom object {attr_error_msg} has not been loaded yet")
+
+    @property
+    def filepath(self) -> str: 
+        "xml file's filename getters"
+        return self._filepath
+
+    # FIXME à supprimer
     def get_source_doc(self) -> str:
         "XML source_doc (xml as a string) getter"
         if hasattr(self, 'source_doc'):
@@ -31,6 +54,7 @@ class XMLDataSet:
             attr_error_msg = str(self._describe())
             raise AttributeError(f"XMLDataSet bject {attr_error_msg} has no attribute named : 'source_doc'")
 
+    # FIXME : À SUPPRIMER
     def set_source_doc(self, source_doc: str) -> None:
         "XML source_doc (xml as a string) setter"
         self.source_doc = source_doc
@@ -42,6 +66,9 @@ class XMLDataSet:
 
 class EtreeXMLDataSet(XMLDataSet):
     "XMLDataSet loader with lxml.etree (lxml.etree._ElementTree)"
+
+    def __str__(self):
+        return self.str 
     
     def _transform_source_doc(self) -> etree._ElementTree:
         "xml transformer (with element tree)"
@@ -52,12 +79,10 @@ class EtreeXMLDataSet(XMLDataSet):
             #replacing element name with its local name
             element.tag = etree.QName(element).localname
         etree.cleanup_namespaces(self.source_doc)
-        return self.source_doc
 
-    def _load(self) -> etree._ElementTree:
+    def _load(self):
         "kedro's API-like loader"
         self._transform_source_doc()
-        return self.source_doc
 
     def _save(self, data:str) -> None:
         "kedro's API-like saver"
@@ -67,17 +92,27 @@ class EtreeXMLDataSet(XMLDataSet):
 
 class BsXMLDataSet(XMLDataSet):
     "XMLDataSet loaded with BeautifulSoup"
-
+    def _load(self) -> str:
-    def _load(self) -> etree._ElementTree:
         "kedro's API-like loader"
-        self._transform_source_doc()
+        self.source_doc = self._load_soup()
         return self.source_doc
 
     def _load_soup(self):
         """open a xml file and return a BeautifulSoup object"""
         with open(self._filepath, 'r', encoding="utf-8") as opening:
             xml = BeautifulSoup(opening, 'xml')
-        return xml
+            self.internal_xml = xml
+        ## xml.prettify() -> str (source_doc)
+        return xml.prettify()
+
+    def get_internal_xml(self):
+        "beautiful soup internal DOM"
+        if hasattr(self, 'internal_xml'):
+            return self.internal_xml
+        else:
+            attr_error_msg = str(self._describe())
+            raise AttributeError(f"XMLDataSet bject {attr_error_msg} has no attribute named : 'internal_xml'")
+        return self.internal_xml
 
     def _save(self, data:str) -> None:
         "kedro's API-like saver"
@@ -142,7 +177,7 @@ class XMLDataSetCollection(AbstractDataSet):
             raise AttributeError(f"Object {attr_error_msg} has no attribute named : 'datasets'")
             
     def _load(self) -> dict[str, EtreeXMLDataSet]:
-        "kedro's API loader"
+        "kedro's API loader method"
         self.datasets = dict()
         for filepath in sorted(self._folderpath.glob("*.xml")):
             self.datasets[filepath.stem] = EtreeXMLDataSet(
@@ -150,7 +185,7 @@ class XMLDataSetCollection(AbstractDataSet):
         return self.datasets
 
     def _save(self, datasets: dict[str, EtreeXMLDataSet]) -> None:
-        "kedro's API saver"
+        "kedro's API saver method"
         for stemfilename, dataset in datasets.items():
             dataset._save(dataset.get_source_doc())