Gestion des définitions vides…

wikstraktor.py

@@ -117,9 +117,15 @@ class Definition(SubInfo):
 		if text != "":
 			self.lang = lang
 			self.text = text
+			self.metadata = {}
 		else:
 			raise ValueError(f"Definition.__init__: “{text}” empty definition.")
 
+	def add_metadata(self, key, value):
+		if key in self.metadata.keys():
+			self.log.add_log("Definition.add_metadata", f"for {self.text} replaced {key}:“{self.metadata['key']}” by {key}:“{value}”")
+		self.metadata["key"]=value
+
 	def __eq__(self, other):
 		return isinstance(other, self.__class__) and self.lang == other.lang and self.text == other.text
 
@@ -127,6 +133,8 @@ class Definition(SubInfo):
 		res = super().serializable(prefix)
 		res["lang"] = self.lang
 		res[self.__class__.key] = self.text
+		if len(self.metadata.keys()) > 0 :
+			res["metadata"] = self.metadata
 		return res
 
 class Translation(Definition):
@@ -587,34 +595,44 @@ class Wikstraktor:
 			res = self.wtp.parse(example_wiki_text).plain_text().strip()
 		return res
 
+	#can be overloaded
+	def parse_definition(self, definition):
+		return self.wtp.parse(definition).plain_text().strip()
+
 	#can be overloaded
 	def process_definition(self, definition, sub_items, def_level = True):
-		if def_level:
-			newSense = Sense(self.entry_language, self.wtp.parse(definition).plain_text().strip(), self.wiki_language)
-			pattern_ex = self.constants['sense_pattern'][0]["ex"]
-			pattern_subdef = self.constants['sense_pattern'][0]["add_subdef"] + self.constants['sense_pattern'][0]["def"]
-		else:
-			newSense = SubSense(self.entry_language, self.wtp.parse(definition).plain_text().strip(), self.wiki_language)
-			pattern_subdef = None
-			pattern_ex = self.constants['sense_pattern'][0]["add_subdef"] + self.constants['sense_pattern'][0]["ex"]
-		#Process examples
-		a = 0
-		for item_list in sub_items:
-			if item_list.pattern == pattern_ex:
-				for item in item_list.items:
-					newSense.add_example(self.process_example(item))
-					#Si on veut traiter les sous items (ex traductions), on peut utiliser
-			if def_level and item_list.pattern == pattern_subdef:
-				b = 0
-				for item in item_list.items:
-					try:
-						sub_sub = item_list.sublists(b)
-					except IndexError as err:
-						sub_sub = []
-						print(f"There is an error in the selection of subitems:\n\t{b}th item of\n\t{itm_list.sublists()}\ntriggered {err}")
-					newSense.add_subsense(self.process_definition(item, sub_sub, False))
-					b += 1
-			a += 1
+		#does not process wk_en quotations
+		try:
+			if def_level:
+				newSense = Sense(self.entry_language, self.parse_definition(definition), self.wiki_language)
+				pattern_ex = self.constants['sense_pattern'][0]["ex"]
+				pattern_subdef = self.constants['sense_pattern'][0]["add_subdef"] + self.constants['sense_pattern'][0]["def"]
+			else:
+				newSense = SubSense(self.entry_language, self.parse_definition(definition), self.wiki_language)
+				pattern_subdef = None
+				pattern_ex = self.constants['sense_pattern'][0]["add_subdef"] + self.constants['sense_pattern'][0]["ex"]
+			#Process examples
+			a = 0
+			for item_list in sub_items:
+				if item_list.pattern == pattern_ex:
+					for item in item_list.items:
+						newSense.add_example(self.process_example(item))
+						#Si on veut traiter les sous items (ex traductions), on peut utiliser
+				if def_level and item_list.pattern == pattern_subdef:
+					b = 0
+					for item in item_list.items:
+						try:
+							sub_sub = item_list.sublists(b)
+						except IndexError as err:
+							sub_sub = []
+							print(f"There is an error in the selection of subitems:\n\t{b}th item of\n\t{itm_list.sublists()}\ntriggered {err}")
+						newSense.add_subsense(self.process_definition(item, sub_sub, False))
+						b += 1
+				a += 1
+		except ValueError as err:
+			self.log.add_log("Wikstraktor.process_definition", f"“{definition}” processed as empty")
+			raise ValueError(f"Wikstraktor.process_definition with empty definition\n{err}")
+
 		return newSense
 
 	def process_senses(self, sensesContent):
@@ -626,8 +644,11 @@ class Wikstraktor:
 		if l.pattern == self.constants['sense_pattern'][0]["def"]:
 			i = 0
 			for item in l.items:
-				senses.append(self.process_definition(item, l.sublists(i)))
-				i += 1
+				try:
+					senses.append(self.process_definition(item, l.sublists(i)))
+					i += 1
+				except ValueError as err:
+					print("Skipped empty definition")
 		return senses
 
 	def __str__(self):