feat: create a docling reader for docs as JSON. (#10)

New class DoclingJSONReader from BaseDoclingReader to read docs as JSON.
Unit test for this class.

Signed-off-by: Cesar Berrospi Ramis <[email protected]>

quackling/llama_index/readers/__init__.py

@@ -3,4 +3,5 @@
 # SPDX-License-Identifier: MIT
 #
 
+from quackling.llama_index.readers.docling_json_reader import DoclingJSONReader  # noqa
 from quackling.llama_index.readers.docling_pdf_reader import DoclingPDFReader  # noqa

quackling/llama_index/readers/docling_json_reader.py

@@ -0,0 +1,25 @@
+#
+# Copyright IBM Corp. 2024 - 2024
+# SPDX-License-Identifier: MIT
+#
+
+import json
+from typing import Iterable
+
+from docling_core.types import Document as DLDocument
+from llama_index.core.schema import Document as LIDocument
+
+from quackling.llama_index.readers.base import BaseDoclingReader
+
+
+class DoclingJSONReader(BaseDoclingReader):
+    def lazy_load_data(self, file_path: str | list[str]) -> Iterable[LIDocument]:
+
+        file_paths = file_path if isinstance(file_path, list) else [file_path]
+
+        for source in file_paths:
+            with open(source, encoding="utf-8") as file_obj:
+                data = json.load(file_obj)
+            dl_doc: DLDocument = DLDocument.model_validate(data)
+            li_doc: LIDocument = self._create_li_doc_from_dl_doc(dl_doc=dl_doc)
+            yield li_doc

tests/unit/test_li_docling_json_reader.py

@@ -0,0 +1,14 @@
+#
+# Copyright IBM Corp. 2024 - 2024
+# SPDX-License-Identifier: MIT
+#
+
+from quackling.llama_index.readers import DoclingJSONReader
+
+
+def test_lazy_load_data():
+    reader = DoclingJSONReader(parse_type=DoclingJSONReader.ParseType.JSON)
+
+    file_path = "tests/unit/data/0_inp_dl_doc.json"
+    li_docs = list(reader.lazy_load_data(file_path))
+    assert len(li_docs) == 1