diff --git a/config.yaml b/config.yaml index 1399cee..e57728c 100755 --- a/config.yaml +++ b/config.yaml @@ -9,6 +9,7 @@ service: operations: conversion: input: + multi: False subdir: "" extension: ORIGIN.pdf.gz output: @@ -16,6 +17,7 @@ service: extension: json.gz extraction: input: + multi: False subdir: "" extension: ORIGIN.pdf.gz output: @@ -23,6 +25,7 @@ service: extension: json.gz table_parsing: input: + multi: True subdir: "pages_as_images" extension: json.gz output: @@ -30,6 +33,7 @@ service: extension: json.gz image_classification: input: + multi: True subdir: "extracted_images" extension: json.gz output: @@ -37,6 +41,7 @@ service: extension: IMAGE_INFO.json.gz default: input: + multi: False subdir: "" extension: in.gz output: diff --git a/scripts/show_compressed_json.py b/scripts/show_compressed_json.py index 86ea339..57976dc 100644 --- a/scripts/show_compressed_json.py +++ b/scripts/show_compressed_json.py @@ -14,7 +14,10 @@ def parse_args(): def interpret(parsed): - return {**parsed, "data": str(string_to_bytes(parsed["data"]))} + try: + return {**parsed, "data": str(string_to_bytes(parsed["data"]))} + except KeyError: + return parsed def main(fp):