From f819ac65b58a666d5d7391e3b35f47ff8670e4f2 Mon Sep 17 00:00:00 2001 From: Seth Ladygo Date: Fri, 18 Oct 2019 16:39:01 -0700 Subject: [PATCH] markup: move work files to a per-cat subdir --- markup/img.py | 6 ++++-- markup/matching.py | 6 +++--- markup/pdf.py | 8 ++++---- 3 files changed, 11 insertions(+), 9 deletions(-) diff --git a/markup/img.py b/markup/img.py index 9d01143..1ae5cbc 100644 --- a/markup/img.py +++ b/markup/img.py @@ -76,9 +76,11 @@ def find_shapes(image_path): return img.width, img.height, bboxes -def write_debug_image(subdir, cat_name, page_num, prods, scribbles): +def write_debug_image(subdir, name, page_num, prods, scribbles): """Draw an image with boxes for products, images, and shapes.""" - path = os.path.join(WORKDIR, subdir, f"{cat_name}-debug-page{page_num:03d}.png") + dir = os.path.join(WORKDIR, subdir, name) + ensure_dir(dir) + path = os.path.join(dir, f"debug-page{page_num:03d}.png") pagew = int(11*72) pageh = int(8.5*72) diff --git a/markup/matching.py b/markup/matching.py index 2aba05b..b5269de 100644 --- a/markup/matching.py +++ b/markup/matching.py @@ -1,6 +1,6 @@ -from markup.img import find_shapes, write_debug_image -from markup.pdf import parse_pdf -from markup.utils import overlaps +from .img import find_shapes, write_debug_image +from .pdf import parse_pdf +from .utils import overlaps def find_marked_products(pdf, subdir, catname, debug=0): diff --git a/markup/pdf.py b/markup/pdf.py index a5a9c8a..460375c 100644 --- a/markup/pdf.py +++ b/markup/pdf.py @@ -59,10 +59,10 @@ def make_scribble(obj, pagenum, mediabox, subdir, name): def export_jp2(obj, subdir, name, pagenum): oid = obj.objid - jp2_path = os.path.join(WORKDIR, subdir, f"{name}-export-page{pagenum:03d}-{oid}.jp2") - png_path = os.path.join(WORKDIR, subdir, f"{name}-export-page{pagenum:03d}-{oid}.png") - - ensure_dir(os.path.join(WORKDIR, subdir)) + dir = os.path.join(WORKDIR, subdir, name) + ensure_dir(dir) + jp2_path = os.path.join(dir, f"export-page{pagenum:03d}-obj{oid:05d}.jp2") + png_path = os.path.join(dir, f"export-page{pagenum:03d}-obj{oid:05d}.png") data = obj.get_rawdata() print('extracting jp2: {}'.format(jp2_path))