feat: add generate sound effect #824
Conversation
📝 WalkthroughWalkthroughAdds a nullable IntegerChoices Estimated code review effort🎯 5 (Critical) | ⏱️ ~120 minutes Areas to pay extra attention:
Suggested reviewers
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
🧹 Nitpick comments (8)
daras_ai_v2/exceptions.py (1)
146-157: Add docstring/typing and consider unifying with ffprobe() to avoid duplication.Minor polish: annotate return type precisely and document the helper. Optionally, consolidate both wrappers into one that accepts a
showarg ("streams" vs "format") to reduce duplication.-def ffprobe_metadata(filename: str) -> dict: +def ffprobe_metadata(filename: str) -> dict[str, typing.Any]: + """Return ffprobe container-level metadata (format) as a parsed JSON dict.""" text = call_cmd( "ffprobe", "-v", "quiet", "-print_format", "json", "-show_format", filename, err_msg=FFMPEG_ERR_MSG, ) return json.loads(text)ai_models/models.py (1)
4-7: Set a default and index for category to match query patterns.You filter by
categoryin hot paths; adddb_index=True. To avoid future NULLs, considerdefault=Category.Video(once backfill is in place) andnull=False.- category = models.IntegerField( - choices=Category.choices, - help_text="Model category: generates Audio, Video, etc.", - null=True, - ) + category = models.IntegerField( + choices=Category.choices, + help_text="Model category: generates Audio, Video, etc.", + null=True, # switch to False after backfill + db_index=True, + # default=Category.Video, # enable after 0004 backfill + )Also applies to: 25-30
recipes/VideoGenPage.py (5)
159-166: Handle missingx-fal-order-properties.Fallback to schema properties if the ordering hint is absent.
- fields = model_output_schemas[0].get("x-fal-order-properties") + schema0 = model_output_schemas[0] + fields = schema0.get("x-fal-order-properties") or list( + schema0.get("properties", {}).keys() + )
149-153: Preserve traceback when re-raising to aid debugging.Use exception chaining and keep message concise.
- except Exception as e: - logger.error(f"Error getting output fields: {e}") - raise RuntimeError(f"Error getting output fields: {e}") + except Exception as e: + logger.error(f"Error getting output fields: {e}") + raise RuntimeError("Error getting output fields") from e
112-117: Type hint of generator return is inaccurate.This generator yields progress strings; update the annotation.
- ) -> typing.Iterator[dict[str, str]]: + ) -> typing.Iterator[str]:
485-487: Audio form: fallback when order hint is missing.Mirror the video form’s defensive logic.
- ordered_fields = schema.get("x-fal-order-properties") + ordered_fields = schema.get("x-fal-order-properties") or list( + schema.get("properties", {}).keys() + )
478-481: Wrap schema extraction in try/except like the video form.Prevents UI from breaking if OpenAPI is malformed.
- model_input_schemas = get_input_fields(models) - - if not model_input_schemas: - return + try: + model_input_schemas = get_input_fields(models) + except Exception as e: + logger.error(f"Error getting audio input fields: {e}") + return + if not model_input_schemas: + returnai_models/admin.py (1)
19-24: Use list_filter for category; drop it from search_fields.
search_fieldson an IntegerField can produce inefficient casts or DB errors. Filter by category instead.list_filter = [ - "pricing__category", + "category", + "pricing__category", "pricing__provider", "created_at", "updated_at", ] @@ - search_fields = ["name", "label", "model_id", "category"] + [ + search_fields = ["name", "label", "model_id"] + [ f"pricing__{field}" for field in ModelPricingAdmin.search_fields ]Also applies to: 26-28, 44-45
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (5)
ai_models/admin.py(2 hunks)ai_models/migrations/0003_videomodelspec_category.py(1 hunks)ai_models/models.py(2 hunks)daras_ai_v2/exceptions.py(1 hunks)recipes/VideoGenPage.py(10 hunks)
🧰 Additional context used
🧬 Code graph analysis (2)
ai_models/admin.py (1)
usage_costs/admin.py (1)
ModelPricingAdmin(62-80)
recipes/VideoGenPage.py (8)
ai_models/models.py (2)
VideoModelSpec(9-47)Category(4-6)daras_ai_v2/exceptions.py (4)
UserError(58-65)ffmpeg(128-129)ffprobe_metadata(146-157)raise_for_status(19-49)daras_ai_v2/fal_ai.py (1)
generate_on_fal(11-36)daras_ai_v2/functional.py (1)
get_initializer(137-145)daras_ai_v2/safety_checker.py (1)
safety_checker(13-19)daras_ai_v2/variables_widget.py (1)
render_prompt_vars(322-334)widgets/switch_with_section.py (1)
switch_with_section(5-36)daras_ai/image_input.py (1)
upload_file_from_bytes(59-66)
🪛 Ruff (0.14.1)
ai_models/migrations/0003_videomodelspec_category.py
8-10: Mutable class attributes should be annotated with typing.ClassVar
(RUF012)
12-22: Mutable class attributes should be annotated with typing.ClassVar
(RUF012)
recipes/VideoGenPage.py
4-4: Redefinition of unused json from line 3
Remove definition: json
(F811)
18-18: Redefinition of unused CaseInsensitiveDict from line 17
Remove definition: CaseInsensitiveDict
(F811)
28-28: Redefinition of unused SAFETY_CHECKER_MSG from line 27
Remove definition: SAFETY_CHECKER_MSG
(F811)
28-28: Redefinition of unused safety_checker from line 27
Remove definition: safety_checker
(F811)
32-32: Redefinition of unused CaseInsensitiveDict from line 18
Remove definition: CaseInsensitiveDict
(F811)
150-150: Do not catch blind exception: Exception
(BLE001)
152-152: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
152-152: Avoid specifying long messages outside the exception class
(TRY003)
155-157: Avoid specifying long messages outside the exception class
(TRY003)
185-187: Avoid specifying long messages outside the exception class
(TRY003)
216-216: Probable use of requests call without timeout
(S113)
221-221: Probable use of requests call without timeout
(S113)
330-330: Redefinition of unused DeforumSDPage from line 328
Remove definition: DeforumSDPage
(F811)
331-331: Redefinition of unused LipsyncPage from line 329
Remove definition: LipsyncPage
(F811)
417-417: Do not catch blind exception: Exception
(BLE001)
| operations = [ | ||
| migrations.AddField( | ||
| model_name="videomodelspec", | ||
| name="category", | ||
| field=models.IntegerField( | ||
| choices=[(1, "Video"), (2, "Audio")], | ||
| help_text="Model category: generates Audio, Video, etc.", | ||
| null=True, | ||
| ), | ||
| ), | ||
| ] |
There was a problem hiding this comment.
Backfill and index required to prevent empty model lists post-migration.
Adding a nullable category without a data migration will exclude all existing rows from UI queries that filter by category, breaking model selection until each row is edited. Add a RunPython migration to set existing rows to Video by default and consider indexing for frequent filtering.
Additional migration (new file example):
# ai_models/migrations/0004_backfill_videomodelspec_category.py
from django.db import migrations
def forwards(apps, schema_editor):
VideoModelSpec = apps.get_model("ai_models", "VideoModelSpec")
# Default all existing to Video (1). Adjust as needed.
VideoModelSpec.objects.filter(category__isnull=True).update(category=1)
def backwards(apps, schema_editor):
VideoModelSpec = apps.get_model("ai_models", "VideoModelSpec")
VideoModelSpec.objects.filter(category=1).update(category=None)
class Migration(migrations.Migration):
dependencies = [("ai_models", "0003_videomodelspec_category")]
operations = [migrations.RunPython(forwards, backwards)]If appropriate, follow up with:
migrations.AlterField(
model_name="videomodelspec",
name="category",
field=models.IntegerField(choices=[(1, "Video"), (2, "Audio")], help_text=..., null=False, default=1, db_index=True),
)🧰 Tools
🪛 Ruff (0.14.1)
12-22: Mutable class attributes should be annotated with typing.ClassVar
(RUF012)
🤖 Prompt for AI Agents
In ai_models/migrations/0003_videomodelspec_category.py lines 12-22, adding a
nullable category field without a backfill will make existing rows invisible to
category-filtering UI; add a follow-up RunPython migration that sets category=1
(Video) for all rows where category is NULL in the forwards function and
restores NULL in backwards, then optionally add a subsequent AlterField
migration to make category non-null with default=1 and db_index=True to enforce
the default and optimize queries.
| video_response = requests.get(video_url) | ||
| video_response.raise_for_status() | ||
| with open(video_path, "wb") as f: | ||
| f.write(video_response.content) | ||
|
|
||
| audio_response = requests.get(audio_url) | ||
| audio_response.raise_for_status() | ||
| with open(audio_path, "wb") as f: | ||
| f.write(audio_response.content) |
There was a problem hiding this comment.
🛠️ Refactor suggestion | 🟠 Major
Add HTTP timeouts to external downloads.
Avoid hanging request threads; optionally stream to reduce memory spikes.
- video_response = requests.get(video_url)
+ video_response = requests.get(video_url, timeout=(5, 30))
video_response.raise_for_status()
with open(video_path, "wb") as f:
f.write(video_response.content)
- audio_response = requests.get(audio_url)
+ audio_response = requests.get(audio_url, timeout=(5, 30))
audio_response.raise_for_status()
with open(audio_path, "wb") as f:
f.write(audio_response.content)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| video_response = requests.get(video_url) | |
| video_response.raise_for_status() | |
| with open(video_path, "wb") as f: | |
| f.write(video_response.content) | |
| audio_response = requests.get(audio_url) | |
| audio_response.raise_for_status() | |
| with open(audio_path, "wb") as f: | |
| f.write(audio_response.content) | |
| video_response = requests.get(video_url, timeout=(5, 30)) | |
| video_response.raise_for_status() | |
| with open(video_path, "wb") as f: | |
| f.write(video_response.content) | |
| audio_response = requests.get(audio_url, timeout=(5, 30)) | |
| audio_response.raise_for_status() | |
| with open(audio_path, "wb") as f: | |
| f.write(audio_response.content) |
🧰 Tools
🪛 Ruff (0.14.1)
216-216: Probable use of requests call without timeout
(S113)
221-221: Probable use of requests call without timeout
(S113)
🤖 Prompt for AI Agents
In recipes/VideoGenPage.py around lines 216 to 224, the HTTP GETs for video and
audio lack timeouts and streaming, which can hang and spike memory; update both
requests.get calls to include a sensible timeout (e.g., connect and read tuple)
and use stream=True, then write the response to file in chunked iterations
(e.g., response.iter_content with a small chunk size) while checking status via
raise_for_status before iterating; ensure the response is closed (use the with
context manager on the response object or call response.close()) to avoid leaks.
| { | ||
| model.name: model | ||
| for model in VideoModelSpec.objects.filter(category=Category.Video) | ||
| } |
There was a problem hiding this comment.
Don’t drop existing models with NULL category during rollout.
Until backfill is complete, include NULL for video models so the page doesn’t render empty.
- self.available_models = CaseInsensitiveDict(
- {
- model.name: model
- for model in VideoModelSpec.objects.filter(category=Category.Video)
- }
- )
+ self.available_models = CaseInsensitiveDict(
+ {
+ model.name: model
+ for model in VideoModelSpec.objects.filter(
+ Q(category=Category.Video) | Q(category__isnull=True)
+ )
+ }
+ )Also applies to: 262-267
🤖 Prompt for AI Agents
In recipes/VideoGenPage.py around lines 256-259 and 262-267, the queryset
currently only selects models with category=Category.Video which will drop
models that have NULL category during rollout; change the filter to include NULL
categories as well (e.g., use category__in=[Category.Video, None] or a Q filter
combining category=Category.Video OR category__isnull=True) so models with NULL
category are preserved and the page doesn’t render empty.
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (5)
recipes/VideoGenPage.py (5)
323-324: Remove duplicate imports.
DeforumSDPageandLipsyncPageare already imported on lines 321-322.Apply this diff:
from recipes.CompareText2Img import CompareText2ImgPage from recipes.DeforumSD import DeforumSDPage from recipes.Lipsync import LipsyncPage - from recipes.DeforumSD import DeforumSDPage - from recipes.Lipsync import LipsyncPage from recipes.VideoBots import VideoBotsPage
122-130: Guard against None before accessingaudio_inputs.
request.audio_inputscan beNone(per the type hint on line 48), so calling.get()directly will raise anAttributeError.Apply this diff:
+ audio_inputs = request.audio_inputs or {} for key in ["prompt", "negative_prompt"]: - if not request.audio_inputs.get(key): + val = audio_inputs.get(key) + if not val: continue - - # # Safety check if not disabled - if self.request.user.disable_safety_checker: + if not self.request.user.disable_safety_checker: - continue - yield "Running safety checker..." - safety_checker(text=request.audio_inputs[key]) + yield "Running safety checker..." + safety_checker(text=val)
134-138: Cast duration to float and guard against None in audio_inputs merge.
ffprobe_metadatareturns duration as a string. Also,request.audio_inputsmay beNone.Apply this diff:
model = VideoModelSpec.objects.get(name=request.selected_audio_model) raw_dur = ffprobe_metadata(video_url)["format"]["duration"] - payload = {"video_url": video_url, "duration": raw_dur} - payload = payload | request.audio_inputs + video_duration = float(raw_dur) if raw_dur not in (None, "N/A") else None + payload = {"video_url": video_url, "duration": video_duration} + payload |= (request.audio_inputs or {}) res = yield from generate_on_fal(model.model_id, payload)
209-217: Add HTTP timeouts to prevent hanging.Both
requests.getcalls lack timeouts, which can block request threads indefinitely. Optionally, use streaming to reduce memory usage for large files.Apply this diff:
- video_response = requests.get(video_url) + video_response = requests.get(video_url, timeout=(5, 30)) video_response.raise_for_status() with open(video_path, "wb") as f: f.write(video_response.content) - audio_response = requests.get(audio_url) + audio_response = requests.get(audio_url, timeout=(5, 30)) audio_response.raise_for_status() with open(audio_path, "wb") as f: f.write(audio_response.content)
248-260: Include models with NULL category during migration rollout.Until all existing models have their
categoryfield backfilled, filtering only bycategory=Category.Videowill exclude models wherecategoryisNULL, potentially rendering the page empty.Apply this diff:
self.available_models = CaseInsensitiveDict( { model.name: model - for model in VideoModelSpec.objects.filter(category=Category.Video) + for model in VideoModelSpec.objects.filter( + Q(category=Category.Video) | Q(category__isnull=True) + ) } ) self.available_audio_models = CaseInsensitiveDict( { model.name: model - for model in VideoModelSpec.objects.filter(category=Category.Audio) + for model in VideoModelSpec.objects.filter( + Q(category=Category.Audio) + ) } )
🧹 Nitpick comments (3)
recipes/VideoGenPage.py (3)
140-150: Refine exception handling to preserve context.Consider using
raise from eto maintain the exception chain, and narrow the exception type if possible.Apply this diff:
model_output_schemas = [] try: model_output_schemas = get_output_fields([model]) - except Exception as e: + except (KeyError, ValueError, TypeError) as e: logger.error(f"Error getting output fields: {e}") - raise RuntimeError(f"Error getting output fields: {e}") + raise RuntimeError(f"Error getting output fields: {e}") from e if not model_output_schemas: raise RuntimeError( f"No output fields returned from {request.selected_audio_model}" )
152-180: Consider more robust field type detection.The current string-matching approach (
"video" in field,"audio" in field) is fragile. Consider checking the schema's type information or using explicit field names from the model specification.Additionally, the warning log on line 175 is immediately followed by raising an exception, making the warning redundant:
else: - logger.warning( - f"Unknown field returned from {request.selected_audio_model}: {field.get('name')}" - ) raise RuntimeError( f"Unknown field returned from {request.selected_audio_model}" )
407-412: Consider showing error to user instead of silent failure.When
get_input_fieldsfails, the form silently fails to render (early return). Consider displaying an error message to the user usinggui.error()or narrowing the exception type.model_input_schemas = [] try: model_input_schemas = get_input_fields(models) - except Exception as e: + except (KeyError, ValueError, TypeError) as e: logger.error(f"Error getting input fields: {e}") + gui.error(f"Error loading model input fields: {e}") return
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
recipes/VideoGenPage.py(10 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
recipes/VideoGenPage.py (7)
ai_models/models.py (2)
VideoModelSpec(9-47)Category(4-6)daras_ai_v2/exceptions.py (4)
UserError(58-65)ffmpeg(128-129)ffprobe_metadata(146-157)raise_for_status(19-49)daras_ai_v2/fal_ai.py (1)
generate_on_fal(11-36)daras_ai_v2/functional.py (1)
get_initializer(137-145)daras_ai_v2/safety_checker.py (1)
safety_checker(13-19)widgets/switch_with_section.py (1)
switch_with_section(5-36)daras_ai/image_input.py (1)
upload_file_from_bytes(59-66)
🪛 Ruff (0.14.1)
recipes/VideoGenPage.py
143-143: Do not catch blind exception: Exception
(BLE001)
145-145: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
145-145: Avoid specifying long messages outside the exception class
(TRY003)
148-150: Avoid specifying long messages outside the exception class
(TRY003)
178-180: Avoid specifying long messages outside the exception class
(TRY003)
209-209: Probable use of requests call without timeout
(S113)
214-214: Probable use of requests call without timeout
(S113)
323-323: Redefinition of unused DeforumSDPage from line 321
Remove definition: DeforumSDPage
(F811)
324-324: Redefinition of unused LipsyncPage from line 322
Remove definition: LipsyncPage
(F811)
410-410: Do not catch blind exception: Exception
(BLE001)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: test (3.10.12, 1.8.3)
🔇 Additional comments (3)
recipes/VideoGenPage.py (3)
607-640: LGTM!The OpenAPI schema extraction logic is well-structured with safe dictionary traversal using
.get()with defaults. The function properly handles both input and output schema types and gracefully returnsNonewhen schemas are missing.
184-199: LGTM!The result normalization logic cleanly handles multiple result formats (None, list, dict, string) and provides a consistent URL output.
219-245: LGTM!The FFmpeg command is well-constructed with explicit stream mapping and the
-shortestflag to prevent duration mismatches. The merged video is properly uploaded to GCS with the correct content type.
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (7)
recipes/VideoGenPage.py (7)
323-324: Remove duplicate imports that are still present.Despite being marked as addressed, these duplicate imports of
DeforumSDPageandLipsyncPage(already imported at lines 321-322) remain in the code and will cause redefinition errors.Apply this diff:
from recipes.CompareText2Img import CompareText2ImgPage from recipes.DeforumSD import DeforumSDPage from recipes.Lipsync import LipsyncPage - from recipes.DeforumSD import DeforumSDPage - from recipes.Lipsync import LipsyncPage from recipes.VideoBots import VideoBotsPage
122-131: Fix NPE on audio_inputs - unresolved from previous review.
request.audio_inputscan beNone, causing anAttributeErrorwhen calling.get(key). Additionally, the safety checker is passed the wrong value on line 130.Apply this diff:
+ audio_inputs = request.audio_inputs or {} for key in ["prompt", "negative_prompt"]: - if not request.audio_inputs.get(key): + val = audio_inputs.get(key) + if not val: continue - - # # Safety check if not disabled - if self.request.user.disable_safety_checker: + if not self.request.user.disable_safety_checker: continue yield "Running safety checker..." - safety_checker(text=request.audio_inputs[key]) + safety_checker(text=val)
134-138: Cast duration to float and guard against None audio_inputs - unresolved from previous review.
ffprobe_metadatareturns duration as a string, andrequest.audio_inputsmay beNone, causing aTypeErrorwhen merging dictionaries.Apply this diff:
model = VideoModelSpec.objects.get(name=request.selected_audio_model) raw_dur = ffprobe_metadata(video_url)["format"]["duration"] - payload = {"video_url": video_url, "duration": raw_dur} - payload = payload | request.audio_inputs + video_duration = float(raw_dur) if raw_dur not in (None, "N/A") else None + payload = {"video_url": video_url, "duration": video_duration} + payload |= (request.audio_inputs or {})
209-217: Add HTTP timeouts to external downloads - unresolved from previous review.Missing timeouts can hang request threads. Add
timeoutparameter and optionally stream to reduce memory usage for large files.Apply this diff:
- video_response = requests.get(video_url) + video_response = requests.get(video_url, timeout=(5, 30)) video_response.raise_for_status() with open(video_path, "wb") as f: f.write(video_response.content) - audio_response = requests.get(audio_url) + audio_response = requests.get(audio_url, timeout=(5, 30)) audio_response.raise_for_status() with open(audio_path, "wb") as f: f.write(audio_response.content)
248-260: Include NULL category models during rollout - unresolved from previous review.Until all models have a category assigned, excluding
NULLvalues will drop existing video models and potentially render the page empty.Apply this diff:
self.available_models = CaseInsensitiveDict( { model.name: model - for model in VideoModelSpec.objects.filter(category=Category.Video) + for model in VideoModelSpec.objects.filter( + Q(category=Category.Video) | Q(category__isnull=True) + ) } ) self.available_audio_models = CaseInsensitiveDict( { model.name: model for model in VideoModelSpec.objects.filter(category=Category.Audio) } )
480-483: Guard against None before calling sort - unresolved from previous review.If
x-fal-order-propertiesis absent in the schema,ordered_fieldswill beNone, causing anAttributeErroron line 483.Apply this diff:
schema = model_input_schemas[0] required_fields = set(schema.get("required", [])) - ordered_fields = schema.get("x-fal-order-properties") + ordered_fields = schema.get("x-fal-order-properties") or [] ordered_fields.sort(key=lambda x: x not in required_fields)
152-153: Guard against None fields before iteration.If
x-fal-order-propertiesis missing from the schema,fieldswill beNone, causing aTypeErrorwhen attempting to iterate.Apply this diff:
- fields = model_output_schemas[0].get("x-fal-order-properties") + fields = model_output_schemas[0].get("x-fal-order-properties") or [] + if not fields: + raise RuntimeError( + f"No output field order found for {request.selected_audio_model}" + ) for field in fields:
🧹 Nitpick comments (1)
recipes/VideoGenPage.py (1)
140-150: Improve exception handling specificity.Catching bare
Exceptionand raising without chaining obscures the root cause. Consider catching specific exceptions (e.g.,KeyError,IndexError,requests.RequestException) and usingraise ... from eto preserve the traceback.Apply this diff:
model_output_schemas = [] try: model_output_schemas = get_output_fields([model]) - except Exception as e: + except (KeyError, ValueError, TypeError) as e: logger.error(f"Error getting output fields: {e}") - raise RuntimeError(f"Error getting output fields: {e}") + raise RuntimeError(f"Error getting output fields for {model.name}") from e if not model_output_schemas: raise RuntimeError( - f"No output fields returned from {request.selected_audio_model}" + f"No output fields found for {request.selected_audio_model}" )
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
recipes/VideoGenPage.py(10 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
recipes/VideoGenPage.py (6)
ai_models/models.py (2)
VideoModelSpec(9-47)Category(4-6)daras_ai_v2/exceptions.py (4)
UserError(58-65)ffmpeg(128-129)ffprobe_metadata(146-157)raise_for_status(19-49)daras_ai_v2/fal_ai.py (1)
generate_on_fal(11-36)daras_ai_v2/safety_checker.py (1)
safety_checker(13-19)widgets/switch_with_section.py (1)
switch_with_section(5-36)daras_ai/image_input.py (1)
upload_file_from_bytes(59-66)
🪛 Ruff (0.14.1)
recipes/VideoGenPage.py
143-143: Do not catch blind exception: Exception
(BLE001)
145-145: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
145-145: Avoid specifying long messages outside the exception class
(TRY003)
148-150: Avoid specifying long messages outside the exception class
(TRY003)
178-180: Avoid specifying long messages outside the exception class
(TRY003)
209-209: Probable use of requests call without timeout
(S113)
214-214: Probable use of requests call without timeout
(S113)
323-323: Redefinition of unused DeforumSDPage from line 321
Remove definition: DeforumSDPage
(F811)
324-324: Redefinition of unused LipsyncPage from line 322
Remove definition: LipsyncPage
(F811)
410-410: Do not catch blind exception: Exception
(BLE001)
475-475: Do not catch blind exception: Exception
(BLE001)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: test (3.10.12, 1.8.3)
| fields = model_output_schemas[0].get("x-fal-order-properties") | ||
| for field in fields: | ||
| if "video" in field: | ||
| video_result = res.get("video") | ||
| video_url = self.get_url_from_result(video_result) | ||
| if video_url: | ||
| merged_videos[video_model_name] = video_url | ||
| break | ||
| elif "audio" in field: | ||
| audio_result = res.get("audio") | ||
| audio_url = self.get_url_from_result(audio_result) | ||
| filename = ( | ||
| furl(video_url.strip("/")).path.segments[-1].rsplit(".", 1)[0] | ||
| + "_gooey_ai_" | ||
| + request.selected_audio_model | ||
| + ".mp4" | ||
| ) | ||
| if audio_url: | ||
| merged_videos[video_model_name] = self.merge_audio_and_video( | ||
| filename, audio_url, video_url | ||
| ) | ||
| break | ||
| else: | ||
| logger.warning( | ||
| f"Unknown field returned from {request.selected_audio_model}: {field.get('name')}" | ||
| ) | ||
| raise RuntimeError( | ||
| f"Unknown field returned from {request.selected_audio_model}" | ||
| ) |
There was a problem hiding this comment.
Fix type confusion in field iteration.
Line 154 checks if "video" in field assuming field is a string, but line 176 calls field.get('name') treating it as a dict. The x-fal-order-properties likely returns field names as strings, not dicts.
Apply this diff to fix the logic:
fields = model_output_schemas[0].get("x-fal-order-properties")
for field in fields:
- if "video" in field:
+ if "video" in field.lower():
video_result = res.get("video")
video_url = self.get_url_from_result(video_result)
if video_url:
merged_videos[video_model_name] = video_url
break
- elif "audio" in field:
+ elif "audio" in field.lower():
audio_result = res.get("audio")
audio_url = self.get_url_from_result(audio_result)
filename = (
furl(video_url.strip("/")).path.segments[-1].rsplit(".", 1)[0]
+ "_gooey_ai_"
+ request.selected_audio_model
+ ".mp4"
)
if audio_url:
merged_videos[video_model_name] = self.merge_audio_and_video(
filename, audio_url, video_url
)
break
else:
logger.warning(
- f"Unknown field returned from {request.selected_audio_model}: {field.get('name')}"
+ f"Unknown field returned from {request.selected_audio_model}: {field}"
)
raise RuntimeError(
f"Unknown field returned from {request.selected_audio_model}"
)🧰 Tools
🪛 Ruff (0.14.1)
178-180: Avoid specifying long messages outside the exception class
(TRY003)
Q/A checklist
How to check import time?
You can visualize this using tuna:
To measure import time for a specific library:
To reduce import times, import libraries that take a long time inside the functions that use them instead of at the top of the file:
Legal Boilerplate
Look, I get it. The entity doing business as “Gooey.AI” and/or “Dara.network” was incorporated in the State of Delaware in 2020 as Dara Network Inc. and is gonna need some rights from me in order to utilize my contributions in this PR. So here's the deal: I retain all rights, title and interest in and to my contributions, and by keeping this boilerplate intact I confirm that Dara Network Inc can use, modify, copy, and redistribute my contributions, under its choice of terms.