d14960da08
Merge in RR/cv-analysis from RED-5527 to master
Squashed commit of the following:
commit 1748095ce45f4a76e9d3d8547d9fc70f7deae2fb
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Nov 8 15:15:16 2022 +0100
RED-5527 update pdf2img with skip_pages_without_images flag, add ENV for this setting for table parsing
62 lines
2.2 KiB
Python
62 lines
2.2 KiB
Python
from dataclasses import asdict
|
|
from operator import truth
|
|
|
|
from funcy import lmap, flatten
|
|
|
|
from cv_analysis.figure_detection.figure_detection import detect_figures
|
|
from cv_analysis.table_parsing import parse_tables
|
|
from cv_analysis.utils.structures import Rectangle
|
|
from pdf2img.conversion import convert_pages_to_images
|
|
from pdf2img.default_objects.image import ImagePlus, ImageInfo
|
|
from pdf2img.default_objects.rectangle import RectanglePlus
|
|
|
|
|
|
def get_analysis_pipeline(operation, table_parsing_skip_pages_without_images):
|
|
if operation == "table":
|
|
return make_analysis_pipeline(
|
|
parse_tables,
|
|
table_parsing_formatter,
|
|
dpi=200,
|
|
skip_pages_without_images=table_parsing_skip_pages_without_images,
|
|
)
|
|
elif operation == "figure":
|
|
return make_analysis_pipeline(detect_figures, figure_detection_formatter, dpi=200)
|
|
else:
|
|
raise
|
|
|
|
|
|
def make_analysis_pipeline(analysis_fn, formatter, dpi, skip_pages_without_images=False):
|
|
def analyse_pipeline(pdf: bytes, index=None):
|
|
def parse_page(page: ImagePlus):
|
|
image = page.asarray()
|
|
rects = analysis_fn(image)
|
|
if not rects:
|
|
return
|
|
infos = formatter(rects, page, dpi)
|
|
return infos
|
|
|
|
pages = convert_pages_to_images(pdf, index=index, dpi=dpi, skip_pages_without_images=skip_pages_without_images)
|
|
results = map(parse_page, pages)
|
|
|
|
yield from flatten(filter(truth, results))
|
|
|
|
return analyse_pipeline
|
|
|
|
|
|
def table_parsing_formatter(rects, page: ImagePlus, dpi):
|
|
def format_rect(rect: Rectangle):
|
|
rect_plus = RectanglePlus.from_pixels(*rect.xyxy(), page.info, alpha=False, dpi=dpi)
|
|
return rect_plus.asdict(derotate=True)
|
|
|
|
bboxes = lmap(format_rect, rects)
|
|
|
|
return {"pageInfo": page.asdict(natural_index=True), "tableCells": bboxes}
|
|
|
|
|
|
def figure_detection_formatter(rects, page, dpi):
|
|
def format_rect(rect: Rectangle):
|
|
rect_plus = RectanglePlus.from_pixels(*rect.xyxy(), page.info, alpha=False, dpi=dpi)
|
|
return asdict(ImageInfo(page.info, rect_plus.asbbox(derotate=False), rect_plus.alpha))
|
|
|
|
return lmap(format_rect, rects)
|