Legacy Office Reader

The LegacyOfficeReader is the reader for Word-97(.doc) files. Under the hood, it uses Apache Tika to parse the file.

Get Started

If you’re opening this Notebook on colab, you will probably need to install LlamaIndex 🦙 and the legacy office reader.

Note: Apache Tika is a dependency of the legacy office reader and it requires Java to be installed and call-able via java --version.

For instance, on colab, you can install it with !apt-get install default-jdk. or on macOS, you can install it with brew install openjdk.

%pip install llama-index-readers-legacy-office

Prepare Data

So we need to prepare a .doc file for testing. Supposedly it’s in test_dir/harry_potter_lagacy.doc

from llama_index.readers.legacy_office import LegacyOfficeReader

Option 1: Load the file with LegacyOfficeReader

file_path = "./test_dir/harry_potter_lagacy.doc"
reader = LegacyOfficeReader(
    excluded_embed_metadata_keys=["file_path", "file_name"],
    excluded_llm_metadata_keys=["file_type"],
)

docs = reader.load_data(file=file_path)
print(f"Loaded {len(docs)} docs")

Loaded 1 docs

Option 2: Load the file with SimpleDirectoryReader

This is the path where we have .doc files together with other files in the same directory.

from llama_index.core import SimpleDirectoryReader

reader = SimpleDirectoryReader(
    input_dir="./test_dir/",
    file_extractor={
        ".doc": LegacyOfficeReader(),
        }
)