erjemin/instaloader
erjemin/instaloader
1
0
Fork 0
Code Issues Wiki Activity

Fix efficiency bug regarding re-downloading targets

Browse Source 
Remove unnecessary queries when a file in a target does not have to be
re-downloaded.

This significantly speeds up the download in cases where the majority of media
is already present locally.

Co-Authored-By: André Koch-Kramer <koch-kramer@web.de>
This commit is contained in:
Alexander Graf 2021-01-30 16:54:35 +01:00
parent 43c521afa9
commit 672b0b1840
2 changed files with 78 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

90
instaloader/instaloader.py
View File

@ -551,6 +551,27 @@ class Instaloader:
:return: True if something was downloaded, False otherwise, i.e. file was already there :return: True if something was downloaded, False otherwise, i.e. file was already there
""" """
def _already_downloaded(path: str) -> bool:
if not os.path.isfile(path):
return False
else:
self.context.log(path + ' exists', end=' ', flush=True)
return True
def _all_already_downloaded(path_base, is_videos_enumerated) -> bool:
if '{filename}' in self.filename_pattern:
# full URL needed to evaluate actual filename, cannot determine at
# this point if all sidecar nodes were already downloaded.
return False
for idx, is_video in is_videos_enumerated:
if self.download_pictures and (not is_video or self.download_video_thumbnails):
if not _already_downloaded("{0}_{1}.jpg".format(path_base, idx)):
return False
if is_video and self.download_videos:
if not _already_downloaded("{0}_{1}.mp4".format(path_base, idx)):
return False
return True
dirname = _PostPathFormatter(post).format(self.dirname_pattern, target=target) dirname = _PostPathFormatter(post).format(self.dirname_pattern, target=target)
filename_template = os.path.join(dirname, self.format_filename(post, target=target)) filename_template = os.path.join(dirname, self.format_filename(post, target=target))
filename = self.__prepare_filename(filename_template, lambda: post.url) filename = self.__prepare_filename(filename_template, lambda: post.url)
@ -559,37 +580,45 @@ class Instaloader:
downloaded = True downloaded = True
if post.typename == 'GraphSidecar': if post.typename == 'GraphSidecar':
if self.download_pictures or self.download_videos: if self.download_pictures or self.download_videos:
for edge_number, sidecar_node in enumerate( if not _all_already_downloaded(
post.get_sidecar_nodes(self.slide_start, self.slide_end), filename_template, enumerate(
start=post.mediacount if self.slide_start < 0 else self.slide_start + 1 (post.get_is_videos()[i]
for i in range(self.slide_start % post.mediacount, self.slide_end % post.mediacount + 1)),
start=self.slide_start % post.mediacount + 1
)
): ):
if self.download_pictures and (not sidecar_node.is_video or self.download_video_thumbnails): for edge_number, sidecar_node in enumerate(
suffix = str(edge_number) post.get_sidecar_nodes(self.slide_start, self.slide_end),
start=self.slide_start % post.mediacount + 1
):
suffix = str(edge_number) # type: Optional[str]
if '{filename}' in self.filename_pattern: if '{filename}' in self.filename_pattern:
suffix = '' suffix = None
# pylint:disable=cell-var-from-loop if self.download_pictures and (not sidecar_node.is_video or self.download_video_thumbnails):
filename = self.__prepare_filename(filename_template, lambda: sidecar_node.display_url) # pylint:disable=cell-var-from-loop
# Download sidecar picture or video thumbnail (--no-pictures implies --no-video-thumbnails) filename = self.__prepare_filename(filename_template, lambda: sidecar_node.display_url)
downloaded &= self.download_pic(filename=filename, url=sidecar_node.display_url, # Download sidecar picture or video thumbnail (--no-pictures implies --no-video-thumbnails)
mtime=post.date_local, filename_suffix=suffix) downloaded &= self.download_pic(filename=filename, url=sidecar_node.display_url,
if sidecar_node.is_video and self.download_videos: mtime=post.date_local, filename_suffix=suffix)
suffix = str(edge_number) if sidecar_node.is_video and self.download_videos:
if '{filename}' in self.filename_pattern: # pylint:disable=cell-var-from-loop
suffix = '' filename = self.__prepare_filename(filename_template, lambda: sidecar_node.video_url)
# pylint:disable=cell-var-from-loop # Download sidecar video if desired
filename = self.__prepare_filename(filename_template, lambda: sidecar_node.video_url) downloaded &= self.download_pic(filename=filename, url=sidecar_node.video_url,
# Download sidecar video if desired mtime=post.date_local, filename_suffix=suffix)
downloaded &= self.download_pic(filename=filename, url=sidecar_node.video_url, else:
mtime=post.date_local, filename_suffix=suffix) downloaded = False
elif post.typename == 'GraphImage': elif post.typename == 'GraphImage':
# Download picture # Download picture
if self.download_pictures: if self.download_pictures:
downloaded = self.download_pic(filename=filename, url=post.url, mtime=post.date_local) downloaded = (not _already_downloaded(filename + ".jpg") and
self.download_pic(filename=filename, url=post.url, mtime=post.date_local))
elif post.typename == 'GraphVideo': elif post.typename == 'GraphVideo':
# Download video thumbnail (--no-pictures implies --no-video-thumbnails) # Download video thumbnail (--no-pictures implies --no-video-thumbnails)
if self.download_pictures and self.download_video_thumbnails: if self.download_pictures and self.download_video_thumbnails:
with self.context.error_catcher("Video thumbnail of {}".format(post)): with self.context.error_catcher("Video thumbnail of {}".format(post)):
downloaded = self.download_pic(filename=filename, url=post.url, mtime=post.date_local) downloaded = (not _already_downloaded(filename + ".jpg") and
self.download_pic(filename=filename, url=post.url, mtime=post.date_local))
else: else:
self.context.error("Warning: {0} has unknown typename: {1}".format(post, post.typename)) self.context.error("Warning: {0} has unknown typename: {1}".format(post, post.typename))
@ -600,7 +629,8 @@ class Instaloader:
# Download video if desired # Download video if desired
if post.is_video and self.download_videos: if post.is_video and self.download_videos:
downloaded &= self.download_pic(filename=filename, url=post.video_url, mtime=post.date_local) downloaded &= (not _already_downloaded(filename + ".mp4") and
self.download_pic(filename=filename, url=post.video_url, mtime=post.date_local))
# Download geotags if desired # Download geotags if desired
if self.download_geotags and post.location: if self.download_geotags and post.location:
@ -693,17 +723,25 @@ class Instaloader:
:return: True if something was downloaded, False otherwise, i.e. file was already there :return: True if something was downloaded, False otherwise, i.e. file was already there
""" """
def _already_downloaded(path: str) -> bool:
if not os.path.isfile(path):
return False
else:
self.context.log(path + ' exists', end=' ', flush=True)
return True
date_local = item.date_local date_local = item.date_local
dirname = _PostPathFormatter(item).format(self.dirname_pattern, target=target) dirname = _PostPathFormatter(item).format(self.dirname_pattern, target=target)
filename_template = os.path.join(dirname, self.format_filename(item, target=target)) filename_template = os.path.join(dirname, self.format_filename(item, target=target))
filename = self.__prepare_filename(filename_template, lambda: item.url) filename = self.__prepare_filename(filename_template, lambda: item.url)
downloaded = False downloaded = False
if not item.is_video or self.download_video_thumbnails is True: if not item.is_video or self.download_video_thumbnails is True:
url = item.url downloaded = (not _already_downloaded(filename + ".jpg") and
downloaded = self.download_pic(filename=filename, url=url, mtime=date_local) self.download_pic(filename=filename, url=item.url, mtime=date_local))
if item.is_video and self.download_videos is True: if item.is_video and self.download_videos is True:
filename = self.__prepare_filename(filename_template, lambda: str(item.video_url)) filename = self.__prepare_filename(filename_template, lambda: str(item.video_url))
downloaded |= self.download_pic(filename=filename, url=item.video_url, mtime=date_local) downloaded |= (not _already_downloaded(filename + ".mp4") and
self.download_pic(filename=filename, url=item.video_url, mtime=date_local))
# Save caption if desired # Save caption if desired
metadata_string = _ArbitraryItemFormatter(item).format(self.storyitem_metadata_txt_pattern).strip() metadata_string = _ArbitraryItemFormatter(item).format(self.storyitem_metadata_txt_pattern).strip()
if metadata_string: if metadata_string:

17
instaloader/structures.py
View File

@ -272,6 +272,17 @@ class Post:
return len(edges) return len(edges)
return 1 return 1
def get_is_videos(self) -> List[bool]:
"""
Return a list containing the ``is_video`` property for each media in the post.
.. versionadded:: 4.7
"""
if self.typename == 'GraphSidecar':
edges = self._field('edge_sidecar_to_children', 'edges')
return [edge['node']['is_video'] for edge in edges]
return [self.is_video]
def get_sidecar_nodes(self, start=0, end=-1) -> Iterator[PostSidecarNode]: def get_sidecar_nodes(self, start=0, end=-1) -> Iterator[PostSidecarNode]:
""" """
Sidecar nodes of a Post with typename==GraphSidecar. Sidecar nodes of a Post with typename==GraphSidecar.
@ -281,13 +292,13 @@ class Post:
""" """
if self.typename == 'GraphSidecar': if self.typename == 'GraphSidecar':
edges = self._field('edge_sidecar_to_children', 'edges') edges = self._field('edge_sidecar_to_children', 'edges')
if any(edge['node']['is_video'] for edge in edges):
# video_url is only present in full metadata, issue #558.
edges = self._full_metadata['edge_sidecar_to_children']['edges']
if end < 0: if end < 0:
end = len(edges)-1 end = len(edges)-1
if start < 0: if start < 0:
start = len(edges)-1 start = len(edges)-1
if any(self.get_is_videos()[start:(end+1)]):
# video_url is only present in full metadata, issue #558.
edges = self._full_metadata['edge_sidecar_to_children']['edges']
for idx, edge in enumerate(edges): for idx, edge in enumerate(edges):
if start <= idx <= end: if start <= idx <= end:
node = edge['node'] node = edge['node']