add entries in acte database

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

@@ -52,6 +52,8 @@ class Acte(Document):
     xmlcontent = StringField(required=True) # no max_length
     place = DictField()
     folium = StringField(required=False) # no max_length
+    diplo_state = StringField(required=False) # sample: "diplo_state": "Original",
+    diplo_type = StringField(required=False) # sample: "diplo_type": "Lettres_patentes",
 
 
 def db_connect(storage_ip, db_name, mongodb_admin, mongodb_password):

actes-princiers/src/actesdataset.py

@@ -104,7 +104,7 @@ class BsXMLDataSet(XMLDataSet):
         prince_name = ps.get_text()
         return prince_name
 
-    def make_prince_code_from_filestem(self, filestem):
+    def extract_prince_code_from_filestem(self, filestem):
         """
         builds prince code
         
@@ -137,9 +137,8 @@ class BsXMLDataSet(XMLDataSet):
         # //sourceDesc//msIdentifier/idno[@n='1'] is always the
         # archive box or manuscript collection id
         #doc = self.soup.msIdentifier.find_all("idno", {"n": "1"})[0]
-        #type_diplo = self.soup.body.div["subtype"]
+        type_diplo = self.soup.body.div["subtype"]
-        #diplo_state = self.soup.body.div["type"]
+        diplo_state = self.soup.body.div["type"]
-
         # geolocalisation
         place = self.soup.find("place")
         place_name = place.find("placeName")
@@ -174,11 +173,10 @@ class BsXMLDataSet(XMLDataSet):
                      latitude = latitude,
                      longitude = longitude
                      )
-
         return {
-#            "num_acte": counter,
+#           "num_acte": counter,
             "prince_name": self.find_prince_name(),
-            "prince_code": self.make_prince_code_from_filestem(numb),
+            "prince_code": self.extract_prince_code_from_filestem(numb),
             "filename": numb,
             "date_time": date_time,
             "date": date,
@@ -187,9 +185,11 @@ class BsXMLDataSet(XMLDataSet):
 #            "doc_acte": doc_query[0],
             "ref_acte": ref_acte,
             "transcribers": self.find_transcribers(),
-            "place": place
+            "place": place,
-#            "state_doc": state_query[0],
+            "diplo_type": type_diplo,
-#            "diplo_type_acte": diplo_query[0]
+            "diplo_state": diplo_state
+#           "state_doc": state_query[0],
+#           "diplo_type_acte": diplo_query[0]
             }
 
 class DataSetCollection(AbstractDataSet):