mattdeitke
commited on
Commit
•
e3778f4
1
Parent(s):
495fbae
refactor github api
Browse files- objaverse_xl/github.py +52 -18
objaverse_xl/github.py
CHANGED
@@ -131,7 +131,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
131 |
handle_missing_object: Optional[Callable],
|
132 |
handle_new_object: Optional[Callable],
|
133 |
commit_hash: Optional[str],
|
134 |
-
) ->
|
135 |
"""Process a single repo.
|
136 |
|
137 |
Args:
|
@@ -144,12 +144,13 @@ class GitHubDownloader(ObjaverseSource):
|
|
144 |
{and the rest of the args are the same as download_objects}
|
145 |
|
146 |
Returns:
|
147 |
-
|
148 |
-
|
149 |
"""
|
150 |
# NOTE: assuming that the user has already checked that the repo doesn't exist,
|
151 |
org, repo = repo_id.split("/")
|
152 |
|
|
|
153 |
with tempfile.TemporaryDirectory() as temp_dir:
|
154 |
# clone the repo to a temp directory
|
155 |
target_directory = os.path.join(temp_dir, repo)
|
@@ -165,7 +166,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
165 |
sha256=sha256,
|
166 |
metadata=dict(github_organization=org, github_repo=repo),
|
167 |
)
|
168 |
-
return
|
169 |
|
170 |
# use the commit hash if specified
|
171 |
repo_commit_hash = self._get_commit_hash_from_local_git_dir(
|
@@ -225,6 +226,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
225 |
|
226 |
# handle the object under different conditions
|
227 |
if github_url in expected_objects:
|
|
|
228 |
if expected_objects[github_url] == file_hash:
|
229 |
if handle_found_object is not None:
|
230 |
handle_found_object(
|
@@ -265,8 +267,12 @@ class GitHubDownloader(ObjaverseSource):
|
|
265 |
# remove the .git directory
|
266 |
shutil.rmtree(os.path.join(target_directory, ".git"))
|
267 |
|
268 |
-
if save_repo_format is
|
269 |
-
|
|
|
|
|
|
|
|
|
270 |
# save the repo to a zip file
|
271 |
if save_repo_format == "zip":
|
272 |
shutil.make_archive(target_directory, "zip", target_directory)
|
@@ -295,10 +301,20 @@ class GitHubDownloader(ObjaverseSource):
|
|
295 |
os.path.join(temp_dir, f"{repo}.{save_repo_format}"),
|
296 |
os.path.join(dirname, f"{repo}.{save_repo_format}"),
|
297 |
)
|
|
|
|
|
|
|
|
|
|
|
298 |
else:
|
299 |
# move the repo to the correct location (with put)
|
300 |
fs.put(target_directory, dirname, recursive=True)
|
301 |
|
|
|
|
|
|
|
|
|
|
|
302 |
# get each object that was missing from the expected objects
|
303 |
if handle_missing_object is not None:
|
304 |
obtained_urls = {x["fileIdentifier"] for x in file_hashes}
|
@@ -310,7 +326,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
310 |
metadata=dict(github_organization=org, github_repo=repo),
|
311 |
)
|
312 |
|
313 |
-
return
|
314 |
|
315 |
def _list_files(self, root_dir: str) -> List[str]:
|
316 |
return [
|
@@ -344,7 +360,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
344 |
commit_hash = result.stdout.strip().decode("utf-8")
|
345 |
return commit_hash
|
346 |
|
347 |
-
def _parallel_process_repo(self, args) ->
|
348 |
"""Helper function to process a repo in parallel.
|
349 |
|
350 |
Note: This function is used to parallelize the processing of repos. It is not
|
@@ -354,8 +370,8 @@ class GitHubDownloader(ObjaverseSource):
|
|
354 |
args (Tuple): Tuple of arguments to pass to _process_repo.
|
355 |
|
356 |
Returns:
|
357 |
-
|
358 |
-
|
359 |
"""
|
360 |
|
361 |
(
|
@@ -391,7 +407,7 @@ class GitHubDownloader(ObjaverseSource):
|
|
391 |
def download_objects(
|
392 |
self,
|
393 |
objects: pd.DataFrame,
|
394 |
-
download_dir: str = "~/.objaverse",
|
395 |
processes: Optional[int] = None,
|
396 |
handle_found_object: Optional[Callable] = None,
|
397 |
handle_modified_object: Optional[Callable] = None,
|
@@ -400,15 +416,17 @@ class GitHubDownloader(ObjaverseSource):
|
|
400 |
save_repo_format: Optional[Literal["zip", "tar", "tar.gz", "files"]] = None,
|
401 |
handle_new_object: Optional[Callable] = None,
|
402 |
**kwargs,
|
403 |
-
) ->
|
404 |
"""Download the specified GitHub objects.
|
405 |
|
406 |
Args:
|
407 |
objects (pd.DataFrame): GitHub objects to download. Must have columns for
|
408 |
the object "fileIdentifier" and "sha256". Use the `get_annotations`
|
409 |
function to get the metadata.
|
410 |
-
download_dir (str, optional): Directory to download the GitHub
|
411 |
-
Supports all file systems supported by fsspec.
|
|
|
|
|
412 |
"~/.objaverse".
|
413 |
processes (Optional[int], optional): Number of processes to use for
|
414 |
downloading. If None, will use the number of CPUs on the machine.
|
@@ -467,12 +485,23 @@ class GitHubDownloader(ObjaverseSource):
|
|
467 |
GitHub organization and repo names.
|
468 |
Return is not used. Defaults to None.
|
469 |
|
|
|
|
|
|
|
|
|
470 |
Returns:
|
471 |
-
|
472 |
-
|
473 |
"""
|
474 |
if processes is None:
|
475 |
processes = multiprocessing.cpu_count()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
476 |
|
477 |
base_download_dir = os.path.join(download_dir, "github")
|
478 |
fs, path = fsspec.core.url_to_fs(base_download_dir)
|
@@ -558,7 +587,12 @@ class GitHubDownloader(ObjaverseSource):
|
|
558 |
tqdm(
|
559 |
pool.imap_unordered(self._parallel_process_repo, all_args),
|
560 |
total=len(all_args),
|
|
|
561 |
)
|
562 |
)
|
563 |
-
|
564 |
-
|
|
|
|
|
|
|
|
|
|
131 |
handle_missing_object: Optional[Callable],
|
132 |
handle_new_object: Optional[Callable],
|
133 |
commit_hash: Optional[str],
|
134 |
+
) -> Dict[str, Optional[str]]:
|
135 |
"""Process a single repo.
|
136 |
|
137 |
Args:
|
|
|
144 |
{and the rest of the args are the same as download_objects}
|
145 |
|
146 |
Returns:
|
147 |
+
Dict[str, Optional[str]]: A dictionary that maps from the "fileIdentifier" to the
|
148 |
+
path of the downloaded object.
|
149 |
"""
|
150 |
# NOTE: assuming that the user has already checked that the repo doesn't exist,
|
151 |
org, repo = repo_id.split("/")
|
152 |
|
153 |
+
out = {}
|
154 |
with tempfile.TemporaryDirectory() as temp_dir:
|
155 |
# clone the repo to a temp directory
|
156 |
target_directory = os.path.join(temp_dir, repo)
|
|
|
166 |
sha256=sha256,
|
167 |
metadata=dict(github_organization=org, github_repo=repo),
|
168 |
)
|
169 |
+
return {}
|
170 |
|
171 |
# use the commit hash if specified
|
172 |
repo_commit_hash = self._get_commit_hash_from_local_git_dir(
|
|
|
226 |
|
227 |
# handle the object under different conditions
|
228 |
if github_url in expected_objects:
|
229 |
+
out[github_url] = file[len(target_directory) + 1 :]
|
230 |
if expected_objects[github_url] == file_hash:
|
231 |
if handle_found_object is not None:
|
232 |
handle_found_object(
|
|
|
267 |
# remove the .git directory
|
268 |
shutil.rmtree(os.path.join(target_directory, ".git"))
|
269 |
|
270 |
+
if save_repo_format is None:
|
271 |
+
# remove the relative path, since it's not downloaded
|
272 |
+
for file_identifier in out:
|
273 |
+
out[file_identifier] = None
|
274 |
+
else:
|
275 |
+
logger.debug(f"Saving {org}/{repo} as {save_repo_format}")
|
276 |
# save the repo to a zip file
|
277 |
if save_repo_format == "zip":
|
278 |
shutil.make_archive(target_directory, "zip", target_directory)
|
|
|
301 |
os.path.join(temp_dir, f"{repo}.{save_repo_format}"),
|
302 |
os.path.join(dirname, f"{repo}.{save_repo_format}"),
|
303 |
)
|
304 |
+
|
305 |
+
for file_identifier in out:
|
306 |
+
out[file_identifier] = os.path.join(
|
307 |
+
dirname, f"{repo}.{save_repo_format}", out[file_identifier]
|
308 |
+
)
|
309 |
else:
|
310 |
# move the repo to the correct location (with put)
|
311 |
fs.put(target_directory, dirname, recursive=True)
|
312 |
|
313 |
+
for file_identifier in out:
|
314 |
+
out[file_identifier] = os.path.join(
|
315 |
+
dirname, repo, out[file_identifier]
|
316 |
+
)
|
317 |
+
|
318 |
# get each object that was missing from the expected objects
|
319 |
if handle_missing_object is not None:
|
320 |
obtained_urls = {x["fileIdentifier"] for x in file_hashes}
|
|
|
326 |
metadata=dict(github_organization=org, github_repo=repo),
|
327 |
)
|
328 |
|
329 |
+
return out
|
330 |
|
331 |
def _list_files(self, root_dir: str) -> List[str]:
|
332 |
return [
|
|
|
360 |
commit_hash = result.stdout.strip().decode("utf-8")
|
361 |
return commit_hash
|
362 |
|
363 |
+
def _parallel_process_repo(self, args) -> Dict[str, Optional[str]]:
|
364 |
"""Helper function to process a repo in parallel.
|
365 |
|
366 |
Note: This function is used to parallelize the processing of repos. It is not
|
|
|
370 |
args (Tuple): Tuple of arguments to pass to _process_repo.
|
371 |
|
372 |
Returns:
|
373 |
+
Dict[str, Optional[str]]: A dictionary that maps from the "fileIdentifier"
|
374 |
+
to the path of the downloaded object.
|
375 |
"""
|
376 |
|
377 |
(
|
|
|
407 |
def download_objects(
|
408 |
self,
|
409 |
objects: pd.DataFrame,
|
410 |
+
download_dir: Optional[str] = "~/.objaverse",
|
411 |
processes: Optional[int] = None,
|
412 |
handle_found_object: Optional[Callable] = None,
|
413 |
handle_modified_object: Optional[Callable] = None,
|
|
|
416 |
save_repo_format: Optional[Literal["zip", "tar", "tar.gz", "files"]] = None,
|
417 |
handle_new_object: Optional[Callable] = None,
|
418 |
**kwargs,
|
419 |
+
) -> Dict[str, Optional[str]]:
|
420 |
"""Download the specified GitHub objects.
|
421 |
|
422 |
Args:
|
423 |
objects (pd.DataFrame): GitHub objects to download. Must have columns for
|
424 |
the object "fileIdentifier" and "sha256". Use the `get_annotations`
|
425 |
function to get the metadata.
|
426 |
+
download_dir (Optional[str], optional): Directory to download the GitHub
|
427 |
+
objects to. Supports all file systems supported by fsspec. If None, the
|
428 |
+
repository will not be saved (note that save_repo_format must also be
|
429 |
+
None in this case, otherwise a ValueError is raised). Defaults to
|
430 |
"~/.objaverse".
|
431 |
processes (Optional[int], optional): Number of processes to use for
|
432 |
downloading. If None, will use the number of CPUs on the machine.
|
|
|
485 |
GitHub organization and repo names.
|
486 |
Return is not used. Defaults to None.
|
487 |
|
488 |
+
Raises:
|
489 |
+
ValueError: If download_dir is None and save_repo_format is not None.
|
490 |
+
Otherwise, we don't know where to save the repo!
|
491 |
+
|
492 |
Returns:
|
493 |
+
Dict[str, Optional[str]]: A dictionary that maps from the "fileIdentifier" to the
|
494 |
+
path of the downloaded object.
|
495 |
"""
|
496 |
if processes is None:
|
497 |
processes = multiprocessing.cpu_count()
|
498 |
+
if download_dir is None:
|
499 |
+
if save_repo_format is not None:
|
500 |
+
raise ValueError(
|
501 |
+
f"If {save_repo_format=} is not None, {download_dir=} must be specified."
|
502 |
+
)
|
503 |
+
# path doesn't matter if we're not saving the repo
|
504 |
+
download_dir = "~/.objaverse"
|
505 |
|
506 |
base_download_dir = os.path.join(download_dir, "github")
|
507 |
fs, path = fsspec.core.url_to_fs(base_download_dir)
|
|
|
587 |
tqdm(
|
588 |
pool.imap_unordered(self._parallel_process_repo, all_args),
|
589 |
total=len(all_args),
|
590 |
+
desc="Downloading repositories",
|
591 |
)
|
592 |
)
|
593 |
+
|
594 |
+
out_dict = {}
|
595 |
+
for x in out:
|
596 |
+
out_dict.update(x)
|
597 |
+
|
598 |
+
return out_dict
|