From 74d7d2e85a8b3b2de3111ff7d832b96511269d4a Mon Sep 17 00:00:00 2001
From: Max Voit <max.voit+gtdv@with-eyes.net>
Date: Sat, 8 Jan 2022 10:24:52 +0100
Subject: [PATCH] new feature: preserve document time as file mtime

---
 docdl/cli.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/docdl/cli.py b/docdl/cli.py
index fceac33..54faaa5 100644
--- a/docdl/cli.py
+++ b/docdl/cli.py
@@ -1,5 +1,6 @@
 """download documents from web portals"""
 
+import os
 import pkg_resources
 import click
 import click_plugins
@@ -85,6 +86,16 @@
     help="webdriver to use for selenium based plugins",
     show_default=True
 )
+@click.option(
+    "-T",
+    "--time-preserve",
+    type=bool,
+    default=True,
+    envvar="DOCDL_TIME_PRESERVE",
+    show_envvar=True,
+    help="preserve the document creation time for the downloaded file",
+    show_default=True
+)
 @click.option(
     "-t",
     "--timeout",
@@ -131,7 +142,7 @@
 # pylint: disable=W0613,C0103,R0913
 def documentdl(
     ctx, username, password, string_matches, regex_matches, jq_matches,
-    headless, browser, timeout, image_loading, action, output_format
+    headless, browser, time_preserve, timeout, image_loading, action, output_format
 ):
     """download documents from web portals"""
     # set browser that SeleniumWebPortal plugins should use
@@ -179,7 +190,10 @@ def run(ctx, plugin_class):
                 continue
             # download ?
             if root_params['action'] == "download":
-                portal.download(document)
+                filename = portal.download(document)
+                if root_params['time_preserve']:
+                    timest = document.attributes['date'].timestamp()
+                    os.utime(filename, (timest, timest))
             # line buffered dict output?
             if root_params['output_format'] == "dicts":
                 # always output as json dict