server/posts/upload: limit filesize for uploads through youtube-dl

This will be controlled by the config parameter 'max_dl_filesize'.

TODO: In a future commit, the regular downloader should also respect
this parameter.

server/config.yaml.dist

@@ -5,8 +5,6 @@
 name: szurubooru
 # full url to the homepage of this szurubooru site, with no trailing slash
 domain: # example: http://example.com
-# user agent name used to download files from the web on behalf of the api users
-user_agent:
 # used to salt the users' password hashes and generate filenames for static content
 secret: change
 
@@ -21,6 +19,10 @@ thumbnails:
     post_width: 300
     post_height: 300
 
+# settings used to download files from the web on behalf of the api users
+user_agent:
+max_dl_filesize: 25.0E+6 # maximum filesize limit in bytes
+
 # automatically convert animated GIF uploads to video formats
 convert:
    gif:

server/szurubooru/errors.py

@@ -56,3 +56,6 @@ class InvalidParameterError(ValidationError):
 
 class ThirdPartyError(BaseError):
     pass
+
+class DownloadTooLargeError(ProcessingError):
+    pass

server/szurubooru/func/net.py

@@ -31,9 +31,12 @@ def download(url: str, use_video_downloader: bool = False) -> bytes:
 def _youtube_dl_wrapper(url: str) -> bytes:
     options = {
         'quiet': True,
+        'ignoreerrors': False,
         'format': 'webm/mp4',
         'logger': logger,
         'noplaylist': True,
+        'max_filesize': config.config['max_dl_filesize'],
+        'max_downloads': 1,
         'outtmpl': os.path.join(
             config.config['data_dir'],
             'temporary-uploads',
@@ -42,9 +45,19 @@ def _youtube_dl_wrapper(url: str) -> bytes:
     with YoutubeDL(options) as ydl:
         try:
             ydl_info = ydl.extract_info(url, download=True)
+            # need to confirm if download was skipped due to size
+            if ydl_info['filesize'] > config.config['max_dl_filesize']:
+                raise errors.DownloadTooLargeError(
+                    'Requested video too large (%d MB > %d MB)' % (
+                        ydl_info['filesize'] / 1.0e6,
+                        config.config['max_dl_filesize'] / 1.0e6))
             ydl_filename = ydl.prepare_filename(ydl_info)
         except YoutubeDLError as ex:
             raise errors.ThirdPartyError(
                 'Error downloading video %s (%s)' % (url, ex))
-    with open(ydl_filename, 'rb') as f:
-        return f.read()
+    try:
+        with open(ydl_filename, 'rb') as f:
+            return f.read()
+    except FileNotFoundError as ex:
+        raise errors.ThirdPartyError(
+            'Error downloading video %s' % (url))

server/szurubooru/tests/func/test_net.py

@@ -1,13 +1,19 @@
 import pytest
-from szurubooru.errors import ThirdPartyError
+from szurubooru import errors
 from szurubooru.func import net
 from szurubooru.func.util import get_sha1
 
 
-def test_download(config_injector):
+@pytest.fixture(autouse=True)
+def inject_config(tmpdir, config_injector):
     config_injector({
-        'user_agent': None
+        'user_agent': None,
+        'max_dl_filesize': 1.0E+6,
+        'data_dir': str(tmpdir.mkdir('data')),
     })
+
+
+def test_download():
     url = 'http://info.cern.ch/hypertext/WWW/TheProject.html'
 
     expected_content = (
@@ -52,11 +58,15 @@ def test_download(config_injector):
     assert actual_content == expected_content
 
 
-def test_video_download(tmpdir, config_injector):
-    config_injector({
-        'user_agent': None,
-        'data_dir': str(tmpdir.mkdir('data'))
-    })
+def test_too_large_download():
+    pytest.xfail('Download limit not implemented yet')
+    url = 'https://samples.ffmpeg.org/MPEG-4/video.mp4'
+
+    with pytest.raises(errors.DownloadTooLargeError):
+        net.download(url)
+
+
+def test_video_download():
     url = 'https://www.youtube.com/watch?v=C0DPdy98e4c'
     expected_sha1 = '508f89ee85bc6186e18cfaa4f4d0279bcf2418ab'
 
@@ -64,12 +74,15 @@ def test_video_download(tmpdir, config_injector):
     assert get_sha1(actual_content) == expected_sha1
 
 
-def test_failed_video_download(tmpdir, config_injector):
-    config_injector({
-        'user_agent': None,
-        'data_dir': str(tmpdir.mkdir('data'))
-    })
-    url = 'http://info.cern.ch/hypertext/WWW/TheProject.html'
+def test_failed_video_download():
+    url = 'https://samples.ffmpeg.org/flac/short.flac'
 
-    with pytest.raises(ThirdPartyError):
+    with pytest.raises(errors.ThirdPartyError):
+        net.download(url, use_video_downloader=True)
+
+
+def test_too_large_video_download():
+    url = 'https://www.youtube.com/watch?v=dQw4w9WgXcQ'
+
+    with pytest.raises(errors.DownloadTooLargeError):
         net.download(url, use_video_downloader=True)