add_data에서 dict도 허용

ko_lm_dataformat/archive.py

@@ -55,16 +55,17 @@ def set_chunk_name(self):
 
     def add_data(
         self,
-        data: Union[str, List[str]],
+        data: Union[str, List, dict],
         meta: Optional[Dict] = None,
         split_sent: bool = False,
         clean_sent: bool = False,
     ):
         """
         Args:
-            data (Union[str, List[str]]):
+            data (Union[str, List, dict]):
                 - Simple text
-                - List of text (multiple sentences)
+                - List of text (multiple sentences), List of dict
+                - Json style data
             meta (Dict, optional): metadata. Defaults to None.
             split_sent (bool): Whether to split text into sentences
             clean_sent (bool): Whether to clean text (NFC, remove control char etc.)
@@ -73,7 +74,7 @@ def add_data(
             meta = {}
         if split_sent:
             assert self.sentence_splitter
-            assert type(data) is not list  # Shouldn't be List[str]
+            assert type(data) is str
             data = self.sentence_splitter.split(data, clean_sent=clean_sent)
 
         if clean_sent and type(data) is str:
@@ -133,10 +134,10 @@ def __init__(self, out_dir: str, sentence_splitter: Optional[SentenceSplitterBas
 
         self.sentence_splitter = sentence_splitter
 
-    def add_data(self, data: Union[str, List[str]], split_sent: bool = False, clean_sent: bool = False):
+    def add_data(self, data: Union[str, List, dict], split_sent: bool = False, clean_sent: bool = False):
         if split_sent:
             assert self.sentence_splitter
-            assert type(data) is str  # Shouldn't be List[str]
+            assert type(data) is str
             data = self.sentence_splitter.split(data, clean_sent=clean_sent)
 
         self.data.append(data)
@@ -191,7 +192,7 @@ def __init__(self, out_dir: str, sentence_splitter: Optional[SentenceSplitterBas
 
         self.sentence_splitter = sentence_splitter
 
-    def add_data(self, data: Union[str, List[str]], split_sent: bool = False, clean_sent: bool = False):
+    def add_data(self, data: Union[str, List, dict], split_sent: bool = False, clean_sent: bool = False):
         if split_sent:
             assert self.sentence_splitter
             assert type(data) is str  # Shouldn't be List[str]

ko_lm_dataformat/version.txt

@@ -1 +1 @@
-0.2.0
+0.3.0

tests/assets/sample.json

@@ -0,0 +1,4 @@
+{
+  "sentence_a": "안녕하세요",
+  "sentence_b": "반가워요"
+}

tests/test_archive.py

@@ -1,3 +1,4 @@
+import json
 import shutil
 
 import pytest
@@ -16,6 +17,28 @@ def test_kowiki_archive():
     shutil.rmtree(TMP_DIR_NAME)
 
 
+def test_archive_json_data():
+    remove_tmp_dir()
+    archive = kldf.Archive(TMP_DIR_NAME)
+    with open(get_tests_dir(append_path="assets/sample.json"), "r", encoding="utf-8") as f:
+        data = json.load(f)
+    archive.add_data(data)
+
+    shutil.rmtree(TMP_DIR_NAME)
+
+
+def test_archive_json_data_is_same():
+    remove_tmp_dir()
+    archive = kldf.Archive(TMP_DIR_NAME)
+    with open(get_tests_dir(append_path="assets/sample.json"), "r", encoding="utf-8") as f:
+        orig_data = json.load(f)
+    archive.add_data(orig_data)
+    archive.commit()
+    reader = kldf.Reader(TMP_DIR_NAME)
+    kldf_data = list(reader.stream_data())
+    assert kldf_data[0] == orig_data
+
+
 def test_kor_str_is_same():
     remove_tmp_dir()
     archive = kldf.Archive(TMP_DIR_NAME)