This commit is contained in:
j 2018-11-15 11:55:44 +00:00
parent a09fcaaa86
commit 19d4367ec7
2 changed files with 15 additions and 6 deletions

View file

@ -30,7 +30,7 @@ if __name__ == '__main__':
edit = api.getEdit(id=i['id'])['data'] edit = api.getEdit(id=i['id'])['data']
if edit['status'] == 'private': if edit['status'] == 'private':
continue continue
path = i['id'].replace(':', '_').replace('/', '_') + '.json' path = i['id'].replace(':', '\uf022').replace('/', '_').replace('|', '_') + '.json'
path = ox.decode_html(path) path = ox.decode_html(path)
path = os.path.join(target, path) path = os.path.join(target, path)
ox.makedirs(os.path.dirname(path)) ox.makedirs(os.path.dirname(path))

View file

@ -12,9 +12,12 @@ def get_extension(api, oshash):
r = api.findMedia({'query': { r = api.findMedia({'query': {
'conditions': [{'key': 'oshash', 'value': oshash}] 'conditions': [{'key': 'oshash', 'value': oshash}]
}, 'keys': ['extension']})['data'] }, 'keys': ['extension']})['data']
return r['items'][0]['extension'] return r['items'][0]['extension'].lower()
COLON = '\uf022'
COLON = ':'
if __name__ == '__main__': if __name__ == '__main__':
if len(sys.argv) > 1: if len(sys.argv) > 1:
target = sys.argv[1] target = sys.argv[1]
@ -28,7 +31,7 @@ if __name__ == '__main__':
api = ox.api.signin('https://%s/api/' % site) api = ox.api.signin('https://%s/api/' % site)
keep = [] keep = []
r = api.find({'range': [0, 1000], 'keys': ['id']}) r = api.find({'range': [0, 10000], 'keys': ['id']})
for i in r['data']['items']: for i in r['data']['items']:
item = api.get(id=i['id'], keys=['id', 'streams', 'title', 'director', 'year', 'instances'])['data'] item = api.get(id=i['id'], keys=['id', 'streams', 'title', 'director', 'year', 'instances'])['data']
director = item.get('director', ['Unknown Director']) director = item.get('director', ['Unknown Director'])
@ -38,7 +41,10 @@ if __name__ == '__main__':
path = ox.decode_html(path) path = ox.decode_html(path)
prefix = ox.decode_html(item['title']) prefix = ox.decode_html(item['title'])
keep.append(unicodedata.normalize('NFD', os.path.join(target, path, 'annotations.json'))) path = path.replace(':', COLON)
prefix = prefix.replace(':', COLON)
keep.append(unicodedata.normalize('NFD', os.path.join(target, path, 'annotations.json')).lower())
parts = [] parts = []
if len(item['streams']) == 1: if len(item['streams']) == 1:
@ -46,6 +52,7 @@ if __name__ == '__main__':
ext = get_extension(api, id) ext = get_extension(api, id)
name = '%s.%s' % (item['title'], ext) name = '%s.%s' % (item['title'], ext)
name = ox.decode_html(name) name = ox.decode_html(name)
name = name.replace(':', COLON)
part = 1 part = 1
parts.append([os.path.join(path, name), item['id'], id, part]) parts.append([os.path.join(path, name), item['id'], id, part])
else: else:
@ -54,12 +61,14 @@ if __name__ == '__main__':
ext = get_extension(api, id) ext = get_extension(api, id)
name = '%s.Part %d.%s' % (prefix, part, ext) name = '%s.Part %d.%s' % (prefix, part, ext)
name = ox.decode_html(name) name = ox.decode_html(name)
name = name.replace(':', COLON)
parts.append([os.path.join(path, name), item['id'], id, part]) parts.append([os.path.join(path, name), item['id'], id, part])
part += 1 part += 1
for path, id, oshash, part in parts: for path, id, oshash, part in parts:
abspath = os.path.join(target, path) abspath = os.path.join(target, path)
if os.path.exists(abspath) and ox.oshash(abspath) != oshash: if os.path.exists(abspath) and ox.oshash(abspath) != oshash:
print('file changed', abspath[len(target):])
os.unlink(abspath) os.unlink(abspath)
if not os.path.exists(abspath): if not os.path.exists(abspath):
url = 'https://%s/%s/download/source/%s' % (site, id, part) url = 'https://%s/%s/download/source/%s' % (site, id, part)
@ -68,11 +77,11 @@ if __name__ == '__main__':
api.save_url(url, abspath) api.save_url(url, abspath)
except: except:
print('failed to download', url) print('failed to download', url)
keep.append(unicodedata.normalize('NFD', abspath)) keep.append(unicodedata.normalize('NFD', abspath).lower())
for root, folders, files in os.walk(target): for root, folders, files in os.walk(target):
for f in files: for f in files:
path = os.path.join(root, f) path = os.path.join(root, f)
if unicodedata.normalize('NFD', path) not in keep: if unicodedata.normalize('NFD', path).lower() not in keep:
print('deleting', path) print('deleting', path)
os.unlink(path) os.unlink(path)