def create_opf(self, input_path, pecha_id):
input_path = Path(input_path)
self._build_dirs(input_path, id_=pecha_id)
self.dirs["release_path"] = self.dirs["opf_path"].parent / "releases"
self.dirs["release_path"].mkdir(exist_ok=True, parents=True)
for i, vol_path in enumerate(sorted(input_path.iterdir())):
print(f"[INFO] Processing {input_path.name}-{vol_path.name} ...")
base_id = f"v{i+1:03}"
if (self.dirs["opf_path"] / "base" / f"{base_id}.txt").is_file():
continue
responses = self.get_input(vol_path)
layers = self.build_layers(responses, vol_path.name, base_id)
formatted_layers = self.format_layer(layers, base_id)
base_text = self.get_base_text()
# save base_text
(self.dirs["opf_path"] / "base" /
f"{base_id}.txt").write_text(base_text)
# save layers
vol_layer_path = self.dirs["layers_path"] / base_id
vol_layer_path.mkdir(exist_ok=True)
for layer, ann in formatted_layers.items():
layer_fn = vol_layer_path / f"{layer}.yml"
dump_yaml(ann, layer_fn)
self.set_vols_meta(vol_path.name, f"{base_id}.txt")
# create meta.yml
meta_fn = self.dirs["opf_path"] / "meta.yml"
dump_yaml(self.get_metadata(input_path.name), meta_fn)
return self.dirs["opf_path"].parent
def update_layers(self, vol_id, updater):
"""
Update all the layer annotations
"""
for layer_fn in (self.layer_path / vol_id).iterdir():
layer = load_yaml(layer_fn)
update_ann_layer(layer, updater)
dump_yaml(layer, layer_fn)
def update_index_layer(self):
layer = load_yaml(self.index_path)
for ann in layer["annotations"]:
# update text span
self.update_text_span(ann["span"])
# update sub-text span
for sub_text in ann["parts"]:
self.update_text_span(sub_text["span"])
dump_yaml(layer, self.index_path)
def create_opf(self, input_path):
input_path = Path(input_path)
self._build_dirs(input_path)
m_text = self.get_input(input_path)
layers = self.build_layers(m_text)
base_text = self.get_base_text(m_text)
# save layers
for layer, ann in layers.items():
layer_fn = self.dirs["layers_path"] / f"{layer}.yml"
dump_yaml(ann, layer_fn)
# save base_text
(self.dirs["opf_path"] / "base.txt").write_text(base_text)
def save_index(self):
try:
dump_yaml(json.loads(self.index.json()), self.index_fn)
except FileNotFoundError:
pass
def save_layer(self, base_name: str, layer_name: LayerEnum, layer: Layer):
layer_fn = self._mkdir(
self.layers_path / base_name) / f"{layer_name.value}.yml"
dump_yaml(json.loads(layer.json()), layer_fn)
def save_meta(self):
dump_yaml(json.loads(self.meta.json()), self.meta_fn)