From f97350fcf1d992d34b29add468f9104a4fc7d7aa Mon Sep 17 00:00:00 2001
From: jens <jens@caret.be>
Date: Wed, 23 Sep 2020 19:23:38 +0000
Subject: [PATCH] Add 'devdocs/vrtdownload.py'

---
 devdocs/vrtdownload.py | 49 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100644 devdocs/vrtdownload.py

diff --git a/devdocs/vrtdownload.py b/devdocs/vrtdownload.py
new file mode 100644
index 0000000..4fed1f6
--- /dev/null
+++ b/devdocs/vrtdownload.py
@@ -0,0 +1,49 @@
+# curl https://www.vrt.be/vrtnu/a-z/ | grep -o /a-z/.*relevant | sort -u | cut -d '.' -f 1  > all_vrt
+# 1 can be a lot of numbers, sometimes the first seasons are not available
+
+import re
+import os
+import requests
+import youtube_dl
+
+
+startdir = os.getcwd()
+
+ydl_opts = {
+        'username': '<vrt nu username here>', 
+        'password': '<vrt nu password here>',
+        'no-overwrites': True,
+}
+ydl = youtube_dl.YoutubeDL(ydl_opts)
+
+#all_vrt = open('all_vrt').readlines()
+#all_vrt = set(re.findall('/a-z/(.*).relevant', requests.get('https://www.vrt.be/vrtnu/a-z/').text))
+all_vrt = ['bevergem', 'de-collega-s', 'dox', 'duts', 'fc-de-kampioenen', 'het-peulengaleis', 'kamp-waes', 'kulderzipken', 'pano', 'postbus-x', 'reizen-waes', 'rick-stein-s-secret-france', 'stille-waters', 'team-scheire', 'the-handmaid-s-tale', 'tomtesterom', 'w817', 'wauters-vs--waes', 'witse', 'in-de-ban-van-tsjernobyl', 'factcheckers', 'flikken-maastricht', 'gentbrugge', 'geub', 'de-smurfen', 'deadwind', 'clinch','100-dagen', 'belpop-classics', 'de-blacklist', 'de-val--10-jaar-na-de-crisis','gevoel-voor-tumor' ]
+for series in all_vrt:
+    for season in range(20):
+        # curl -v  https://www.vrt.be/vrtnu/a-z/het-peulengaleis/3.lists.all-episodes/ | grep -o vrtnu[^\"]*
+        url = 'https://www.vrt.be/vrtnu/a-z/%s/%d.lists.all-episodes/' % (series.strip(), season)
+        #print(url)
+        all_eps = requests.get(url).text
+        episodes = set(re.findall('vrtnu/a-z/([^"]*)',all_eps))
+        print(episodes)
+        print([x.split('/')[-2] for x in episodes])
+        if episodes:
+            path = os.path.join(startdir,series,'Season_%d' % season)
+            try:
+                os.makedirs(path)
+            except:
+                pass
+
+        for episode in episodes:
+            print(episode)
+            path = os.path.join(startdir,series,'Season_%d' % season, episode.split('/')[-2])
+            print(path)
+            try:
+                os.mkdir(path)
+                os.chdir(path)
+                ydl.download(['https://www.vrt.be/vrtnu/a-z/' + episode])
+            except Exception as e:
+                print('skipping existing folder %s' % episode)
+                print(e)
+