def generate_ocr_insights(
source: str,
insight_type: str,
output: Path = Option(
...,
help="File to write output to, stdout if not specified",
dir_okay=False,
writable=True,
),
keep_empty: bool = Argument(..., help="Keep documents with empty insight"),
) -> None:
"""Generate OCR insights of the requested type.
SOURCE can be either:
* the path to a JSON file, a (gzipped-)JSONL file, or a directory
containing JSON files
* a barcode
* the '-' character: input is read from stdin and assumed to be JSONL
Output is JSONL, each line containing the insights for one document.
"""
from typing import TextIO, Union
from robotoff.cli import insights
from robotoff.insights._enum import InsightType
from robotoff.utils import get_logger
input_: Union[str, TextIO] = sys.stdin if source == "-" else source
get_logger()
insights.run_from_ocr_archive(input_, InsightType[insight_type], output,
keep_empty)
def generate_ocr_insights(source: str, insight_type: str, output: str,
keep_empty: bool):
from typing import TextIO, Union
from robotoff.cli import insights
from robotoff.insights._enum import InsightType
from robotoff.utils import get_logger
input_: Union[str, TextIO] = sys.stdin if source == "-" else source
get_logger()
insights.run_from_ocr_archive(input_, InsightType[insight_type],
output, keep_empty)
def generate_ocr_insights(input_: str, insight_type: str, output: str):
from robotoff.cli import insights
insights.run_from_ocr_archive(input_, insight_type, output)
def generate_ocr_insights(input_: str, insight_type: str, output: str):
from robotoff.cli import insights
from robotoff.utils import get_logger
get_logger()
insights.run_from_ocr_archive(input_, insight_type, output)