Extracted figures are overwritten

[eryk-mazus]

Hi,

I'm new to unstructured. When I run the sample code to perform the partitioning of several pdf/doc files, the extracted images are saved to the separate folder called figures. The naming convention seems to be: figure-{page_number}-{#}, as a result, the images extracted from different documents appearing on the same page number, will overwrite themselves - the metadata in the resulting json will point out to wrong file.

For instance the link to figures/figure-3-2.jpg is included in three separate json files.

Here is my code:

    Pipeline.from_configs(
        context=ProcessorConfig(),
        indexer_config=LocalIndexerConfig(input_path=str(INPUT_DIR)),
        downloader_config=LocalDownloaderConfig(),
        source_connection_config=LocalConnectionConfig(),
        partitioner_config=PartitionerConfig(
            ocr_languages=["eng"],
            strategy="hi_res",
            partition_by_api=False,
            additional_partition_args={"extract_image_block_types": ["Image", "Table"]},
        ),
        # chunker_config=ChunkerConfig(
        #     chunking_strategy="by_title",
        #     chunk_max_characters=512,
        #     chunk_combine_text_under_n_chars=200,
        # ),
        # embedder_config=EmbedderConfig(embedding_provider="huggingface"),
        uploader_config=LocalUploaderConfig(output_dir=str(OUTPUT_DIR)),
    ).run()

Is this a bug or am I doing something wrong ?