markup: improve path routines

markup/img.py

@@ -11,7 +11,7 @@ from pathlib import Path
 
 from pdfminer.psparser import LIT
 
-from .utils import cv2_rect, ensure_dir, set_file_perms, WORKDIR
+from .utils import cv2_rect, ensure_dir, set_file_perms
 
 # https://www.pyimagesearch.com/2014/10/20/finding-shapes-images-using-python-opencv/
 

markup/spreadsheet.py

@@ -4,7 +4,7 @@ from itertools import zip_longest
 from openpyxl import Workbook
 from openpyxl.styles import PatternFill, Border, Side, Alignment, Protection, Font
 
-from .utils import ensure_dir, set_file_perms, WORKDIR
+from .utils import ensure_dir, set_file_perms
 
 
 def format_season(s):
@@ -19,7 +19,7 @@ def format_name(name, gender):
     return '{}-{}'.format(name, gender[:1])
 
 
-def write_spreadsheet(matches, workdir, file_base):
+def write_spreadsheet(matches, xls_path):
     if not matches:
         print('write_spreadsheet: no matches.  skipping.')
         return None
@@ -78,9 +78,5 @@ def write_spreadsheet(matches, workdir, file_base):
             cell.border = border
 
     # save
-    ensure_dir(workdir)
+    wb.save(xls_path)
-    path = os.path.join(workdir, f"{file_base}.xlsx")
+    set_file_perms(xls_path)
-    wb.save(path)
-    set_file_perms(path)
-
-    return path

markup/tasks.py

@@ -24,7 +24,7 @@ django.setup()
 
 from django.contrib.auth.models import User
 
-from .utils import clean_path, ensure_dir, set_file_perms, WORKDIR
+from .utils import clean_path, ensure_dir, set_file_perms, MARKUP_WORK_DIR
 from .email import reply, reply_missing, reply_no_matches, send_error_email
 from .matching import find_marked_products
 from .spreadsheet import write_spreadsheet
@@ -47,6 +47,25 @@ def on_fail_handler(self, exc, task_id, args, kwargs, einfo):
 #     raise KeyError()
 
 
+def workdir(username, pdf_path):
+    stem = clean_path(Path(pdf_path).stem)
+    return os.path.join(MARKUP_WORK_DIR, clean_path(username), stem)
+
+
+def dest_spreadsheet_path(dest_pdf_path):
+    path = Path(dest_pdf_path)
+    return os.path.join(path.parent, f"{path.stem}.xlsx")
+
+
+def work_pdf_path(workdir, pdf_path):
+    return os.path.join(workdir, Path(pdf_path).name)
+
+
+def work_spreadsheet_path(username, pdf_path):
+    stem = Path(pdf_path).stem
+    return os.path.join(workdir(username, pdf_path), f"{stem}.xlsx")
+
+
 @shared_task(on_failure=on_fail_handler)
 def process_markup_pdf(pdf_path, username):
     if not Path(pdf_path).is_file():
@@ -55,22 +74,21 @@ def process_markup_pdf(pdf_path, username):
 
     user = User.objects.get(username=username)
 
-    pdf_stem = Path(pdf_path).stem
+    work_dir = workdir(user.username, pdf_path)
-    workdir = os.path.join(WORKDIR, clean_path(user.username), clean_path(pdf_stem))
+    ensure_dir(work_dir)
-    ensure_dir(workdir)
+
+    work_pdf = work_pdf_path(work_dir, pdf_path)
 
-    pdf_name = Path(pdf_path).name
-    dest_path = os.path.join(workdir, pdf_name)
     print(f'copying pdf from {pdf_path}')
-    print(f'copying pdf to {dest_path}')
+    print(f'copying pdf to {work_pdf}')
-    shutil.copy(pdf_path, dest_path)
+    shutil.copy(pdf_path, work_pdf)
-    set_file_perms(dest_path)
+    set_file_perms(work_pdf)
 
     frm = str(make_header(decode_header(f'{user.get_full_name()} <{user.email}>')))
-    subject = str(make_header(decode_header(pdf_stem)))
+    subject = str(make_header(decode_header(Path(pdf_path).stem)))
 
     # find matches
-    matches = find_marked_products(dest_path, workdir, debug=0)
+    matches = find_marked_products(work_pdf, work_dir, debug=0)
     if not matches:
         print('no product matches')
         # reply_no_matches(frm, subject)
@@ -79,21 +97,20 @@ def process_markup_pdf(pdf_path, username):
     print(f'{len(matches)} product matches')
 
     # write spreadsheet
-    xls_path = write_spreadsheet(matches, workdir, pdf_stem)
+    work_xls_path = work_spreadsheet_path(username, pdf_path)
+    write_spreadsheet(matches, work_xls_path)
 
-    if not xls_path:
+    if not work_xls_path:
         # TODO send error
         print(f'error creating spreadsheet')
         return
 
-    webdav_dir = Path(pdf_path).parent
+    dest_xls_path = dest_spreadsheet_path(pdf_path)
-    xls_name = Path(xls_path).name
-    xls_webdav_path = os.path.join(webdav_dir, xls_name)
 
-    print(f'wrote spreadsheet: {xls_path}')
+    print(f'wrote spreadsheet: {work_xls_path}')
-    print(f'copying xls to {xls_webdav_path}')
+    print(f'copying xls to {dest_xls_path}')
 
-    shutil.copy(xls_path, xls_webdav_path)
+    shutil.copy(work_xls_path, dest_xls_path)
-    set_file_perms(xls_webdav_path)
+    set_file_perms(dest_xls_path)
 
-    reply(frm, subject, xls_webdav_path, pdf_path)
+    reply(frm, subject, dest_xls_path, pdf_path)

markup/utils.py

@@ -4,7 +4,7 @@ import shutil
 
 from django.conf import settings
 
-WORKDIR = os.path.join(settings.ASSET_DIR, 'markup', 'work')
+MARKUP_WORK_DIR = os.path.join(settings.ASSET_DIR, 'markup', 'work')
 
 
 def pdf_rect(rect, container_height):

markup/webdav.py

@@ -27,7 +27,7 @@ from djangodav.views import DavView
 from procat2.models import Catalog
 from procat2.settings import BASE_DIR, ASSET_DIR
 
-from .utils import clean_path, ensure_dir, set_file_perms, WORKDIR
+from .utils import clean_path, ensure_dir, set_file_perms
 from .tasks import process_markup_pdf
 
 log = logging.getLogger(__name__)