Merge pull request #60 from OpenPecha/fix-pecha_db_serializer

updated BaseSerializer and added source_type flag

src/openpecha/pecha/serializers/__init__.py

@@ -11,5 +11,6 @@ def name(self):
  def serialize(
  self,
  pecha_path: Path,
+ source_type: str,
  ):
  pass

src/openpecha/pecha/serializers/pecha_db.py

@@ -15,7 +15,7 @@ def __init__(self, output_path: Path = PECHAS_PATH):
  self.mappings: Dict[str, List[int]] = {}
  self.chapter: List[str] = []
 
- def create_content_list(self, pecha):
+ def create_dharmanexus_content_list(self, pecha):
  for _, layers in pecha.layers.items():
  chapter_num = 1
  for ann_store in layers[LayerEnum.meaning_segment]:
@@ -29,20 +29,23 @@ def create_content_list(self, pecha):
  self.contents.append(self.chapter)
  self.chapter = []
 
- def serialize(self, pecha_path: Path):
+ def serialize(self, pecha_path: Path, source_type: str):
  pecha_id = pecha_path.stem
  pecha = Pecha(pecha_id=pecha_id, pecha_path=pecha_path)
- self.create_content_list(pecha)
- pecha_json = {
- "title": pecha.metadata.title,
- "language": pecha.metadata.language.value,
- "versionSource": " ",
- "completestatus": "done",
- "content": self.contents,
- "direction": "ltr",
- }
- pecha_db_json_path = f"{self.output_path}/{pecha_id}.json"
- mapping_path = f"{self.output_path}/{pecha_id}_mapping.json"
- write_json(pecha_db_json_path, pecha_json)
- write_json(mapping_path, self.mappings)
- return pecha_db_json_path, mapping_path
+ if source_type == "dharmanexus":
+ self.create_dharmanexus_content_list(pecha)
+ pecha_json = {
+ "title": pecha.metadata.title,
+ "language": pecha.metadata.language.value,
+ "versionSource": " ",
+ "completestatus": "done",
+ "content": self.contents,
+ "direction": "ltr",
+ }
+ pecha_db_json_path = f"{self.output_path}/{pecha_id}.json"
+ mapping_path = f"{self.output_path}/{pecha_id}_mapping.json"
+ write_json(pecha_db_json_path, pecha_json)
+ write_json(mapping_path, self.mappings)
+ return pecha_db_json_path, mapping_path
+ elif source_type == "pedurma":
+ pass

tests/pecha/serializers/pecha_db/test_pecha_db_serializer.py

@@ -10,8 +10,10 @@ def test_pecha_db_serializer():
  output_path = Path(tmpdirname)
  serializer = PechaDBSerializer(output_path=output_path)
  pecha_path = Path("tests/pecha/serializers/pecha_db/data/IB1A4B480")
+ source_type = "dharmanexus"
  pecha_db_json_path, mapping_json_path = serializer.serialize(
- pecha_path=pecha_path
+ pecha_path=pecha_path,
+ source_type=source_type,
  )
  assert read_json(pecha_db_json_path) == read_json(
  "tests/pecha/serializers/pecha_db/data/expected_pecha_db.json"