json_entry_added_etc (#500)

run-llama · Sep 12, 2023 · b078914 · b078914
1 parent 33155c7
commit b078914
Show file tree

Hide file tree

Showing 7 changed files with 15 additions and 137 deletions.
diff --git a/llama_hub/hangeul/README.md b/llama_hub/hangeul/README.md
diff --git a/llama_hub/hangeul/__init__.py b/llama_hub/hangeul/__init__.py
diff --git a/llama_hub/hangeul/base.py b/llama_hub/hangeul/base.py
diff --git a/llama_hub/hangeul/requirements.txt b/llama_hub/hangeul/requirements.txt
diff --git a/llama_hub/hwp/base.py b/llama_hub/hwp/base.py
@@ -8,7 +8,7 @@
 from llama_index.readers.schema.base import Document
 
 class HWPReader(BaseReader):
-    """Hangeul Reader. Reads contents from Hangeul file.
+    """Hwp Reader. Reads contents from Hwp file.
     Args: None
     """
     def __init__(
@@ -26,14 +26,13 @@ def __init__(
     def load_data(
             self,
             file: Path,
+            extra_info: Optional[Dict] = None
     ) -> Document:
-        """Load data and extract table from PDF file.
-
+        """Load data and extract table from Hwp file.
         Args:
-            file (Path): Path for the PDF file.
-
+            file (Path): Path for the Hwp file.
         Returns:
-            List[Document]: List of documents.
+            Document
         """
         import olefile
 
@@ -44,7 +43,7 @@ def load_data(
             raise Exception("Not Valid HwpFile")
 
         result_text = self._get_text(load_file, file_dir)
-        result = self._text_to_document(text=result_text)
+        result = self._text_to_document(text=result_text, extra_info=extra_info)
         return result
 
     def is_valid(self, dirs):

diff --git a/llama_hub/library.json b/llama_hub/library.json
@@ -939,5 +939,12 @@
     "keywords": [
       "linear"
     ]
+  },
+  "HWPReader": {
+    "id": "hwp",
+    "author": "sangwongenip",
+    "keywords": [
+      "hwp"
+    ]
   }
 }
diff --git a/test_requirements.txt b/test_requirements.txt
@@ -11,6 +11,8 @@ typing_extensions==4.5.0
 llama-index>=0.6.9
 atlassian-python-api
 html2text
+olefile
+
 # hotfix
 psutil