bi>#HddlZddlZddlmZmZmZmZmZddlZddl m Z ddl m Z m Z e rJddlZddlmZddlmZe rddlZej*ej,ej.ej0d Zdd lmZd ed efd Z d+dedddeeddedee dee deddeddeddededdeddee!d ee!fd!Z" d,d"ejFdeeddeddeddef d#Z$ d-dddeeddedee dee deddeddedee!d ee!fd$Z% d+dedddeeddedee dee deddeddeddededdeddee!d ee!d eej6eej6edffd%Z& d.ded&edej6fdee dee fd'Z'd"ejFfd(Z( d.d&edej6fdee dee fd)Z) d.ded&edej6fdee dee d df d*Z*y)/N)ListLiteralOptionalUnioncast) deprecate)is_safetensors_availableis_torch_available)VaeImageProcessor)VideoProcessor)float16float32bfloat16uint8)Imagedatareturnc|jdry|jdry|jds|jdry|jdry y ) NsjpegsPNG  pngsGIF87asGIF89agifsBMbmpunknown) startswith)rs W/home/cdr/jupyterlab/.venv/lib/python3.12/site-packages/diffusers/utils/remote_utils.pydetect_image_typer.sM {# - .  #ty'A   binaryendpointtensorz torch.Tensor processor)r r do_scalingscaling_factor shift_factor output_type)mp4pilpt return_type image_format)rjpgpartial_postprocessinput_tensor_typeoutput_tensor_typeheightwidthc|jdk(r| | td|dk(r(|dk(r#| s!t|ttfs td|r|t dddd yyy) Nz1`height` and `width` required for packed latents.r*r)z`processor` is required.r$1.0.0zQ`do_scaling` is deprecated, pass `scaling_factor` and `shift_factor` if required.F standard_warn)ndim ValueError isinstancer rr )r!r"r#r$r%r&r'r+r,r.r/r0r1r2s rcheck_inputs_decoder;:sv {{aFNu}LMMt 5 #9'8.&IJ344n,   _  -zrresponsec|dk(s|dk(ro|m|j}|j}tj|d}|d}t|} t j t|| j|}|dk(r|rS|dk(rED cgc]%} tj| j'} } t| dk(r| d} | S|dk(r} | S||dk(r} | St|tr7tt tj|j#dd} | Sttj|j%dd}  S|dk(rf|dk(ra|_tj&t)j*|jj-d } t/|j} | | _| S|dk(r||dk(rrj3dd d dj5jd zj7j9d D cgc]} tj| } } | S|dk(r}  S|dk(r|dk(r |j}  Scc} wcc} w)Nr*r)shapedtyper?rr)r'RGBr r4rr()contentheadersjsonloads DTYPE_MAPtorch frombuffer bytearrayreshaper fromarraynumpylenr:rrrpostprocess_video postprocessopenioBytesIOconvertrformatpermutefloatroundastype) r<r#r'r+r. output_tensor parametersr>r? torch_dtypeimageoutputdetected_formats rpostprocess_decoder`\sd{e3 8M (( %%  :g./7#& ((=)AU]]^cd d e#FSTU%//%++-8TTv;!##AYF> M=$&: M7 K4$7&4 M1i8!U[[)!33Mu3UVWXF. M%" !--m-OPQRF$ M  +"69;LBJJx'7'789AA%H+H,<,<='  M  )"7 % ,33Aq!Q?EEGMMORUU\\^eefmn&F M D "F M  +"6!! MCU2s *I I%c i} |||t|jt|jj ddd} |r||| d<|r||| d<|r||| d<n |r |||| d<| | || d<| | d<d | d <d | d <|d k(r |d k(r|d| d <n|d k(r |dk(r|d| d <n |dk(rd| d <t j j|d} | | | dS)N.)r,r'r.r>r?r%r&r$r1r2z tensor/binaryz Content-TypeAcceptr)r-z image/jpegrz image/pngr(z text/plainr"rparamsrD)listr>strr?split safetensorsrH_tobytes) r"r#r$r%r&r'r,r.r1r2rDr[ tensor_datas rprepare_decoderms7G$"2fll#V\\"((-b1 Jn0'5 #$l.%1 >"n,#- < .<3G#- <  e/% 8# 7-GN'GHe  5):K(  ,%"7IK:' JJrcN| dk(rtddddd} | dk(rtddd dd} t|||||||||| | | | | t|||||||| | | }tj|fi|}|j st |jt||||| }|S) aM Hugging Face Hybrid Inference that allow running VAE decode remotely. Args: endpoint (`str`): Endpoint for Remote Decode. tensor (`torch.Tensor`): Tensor to be decoded. processor (`VaeImageProcessor` or `VideoProcessor`, *optional*): Used with `return_type="pt"`, and `return_type="pil"` for Video models. do_scaling (`bool`, default `True`, *optional*): **DEPRECATED**. **pass `scaling_factor`/`shift_factor` instead.** **still set do_scaling=None/do_scaling=False for no scaling until option is removed** When `True` scaling e.g. `latents / self.vae.config.scaling_factor` is applied remotely. If `False`, input must be passed with scaling applied. scaling_factor (`float`, *optional*): Scaling is applied when passed e.g. [`latents / self.vae.config.scaling_factor`](https://github.com/huggingface/diffusers/blob/7007febae5cff000d4df9059d9cf35133e8b2ca9/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py#L1083C37-L1083C77). - SD v1: 0.18215 - SD XL: 0.13025 - Flux: 0.3611 If `None`, input must be passed with scaling applied. shift_factor (`float`, *optional*): Shift is applied when passed e.g. `latents + self.vae.config.shift_factor`. - Flux: 0.1159 If `None`, input must be passed with scaling applied. output_type (`"mp4"` or `"pil"` or `"pt", default `"pil"): **Endpoint** output type. Subject to change. Report feedback on preferred type. `"mp4": Supported by video models. Endpoint returns `bytes` of video. `"pil"`: Supported by image and video models. Image models: Endpoint returns `bytes` of an image in `image_format`. Video models: Endpoint returns `torch.Tensor` with partial `postprocessing` applied. Requires `processor` as a flag (any `None` value will work). `"pt"`: Support by image and video models. Endpoint returns `torch.Tensor`. With `partial_postprocess=True` the tensor is postprocessed `uint8` image tensor. Recommendations: `"pt"` with `partial_postprocess=True` is the smallest transfer for full quality. `"pt"` with `partial_postprocess=False` is the most compatible with third party code. `"pil"` with `image_format="jpg"` is the smallest transfer overall. return_type (`"mp4"` or `"pil"` or `"pt", default `"pil"): **Function** return type. `"mp4": Function returns `bytes` of video. `"pil"`: Function returns `PIL.Image.Image`. With `output_type="pil" no further processing is applied. With `output_type="pt" a `PIL.Image.Image` is created. `partial_postprocess=False` `processor` is required. `partial_postprocess=True` `processor` is **not** required. `"pt"`: Function returns `torch.Tensor`. `processor` is **not** required. `partial_postprocess=False` tensor is `float16` or `bfloat16`, without denormalization. `partial_postprocess=True` tensor is `uint8`, denormalized. image_format (`"png"` or `"jpg"`, default `jpg`): Used with `output_type="pil"`. Endpoint returns `jpg` or `png`. partial_postprocess (`bool`, default `False`): Used with `output_type="pt"`. `partial_postprocess=False` tensor is `float16` or `bfloat16`, without denormalization. `partial_postprocess=True` tensor is `uint8`, denormalized. input_tensor_type (`"binary"`, default `"binary"`): Tensor transfer type. output_tensor_type (`"binary"`, default `"binary"`): Tensor transfer type. height (`int`, **optional**): Required for `"packed"` latents. width (`int`, **optional**): Required for `"packed"` latents. Returns: output (`Image.Image` or `List[Image.Image]` or `bytes` or `torch.Tensor`). base64zinput_tensor_type='base64'r5z9input_tensor_type='base64' is deprecated. Using `binary`.Fr6r zoutput_tensor_type='base64'z:output_tensor_type='base64' is deprecated. Using `binary`.) r"r#r$r%r&r'r,r.r1r2)r<r#r'r+r.) r r;rmrequestspostok RuntimeErrorrEr`)r!r"r#r$r%r&r'r+r,r.r/r0r1r2kwargsr<r^s r remote_decoderusxH$ (  G  %X% )  H  &  %!!/ F}}X00H ;;8==?++ / F Mrr]cy)N)r!r]r%r&s rcheck_inputs_encoderxQs  rc|j}|j}tj|d}|d}t|}t j t||j|}|S)Nr>r?r@) rCrDrErFrGrHrIrJrK)r<rZr[r>r?r\s rpostprocess_encoderzZsj$$M!!J JJz'* +E w EE"K$$Y}%=[QYYZ_`M rci}i}|||d<|||d<t|tjrqtjj |j d}t |j|d<t|jjdd|d<n7tj}|j|d |j}|||d S) Nr%r&r"r>rbrcr?PNG)rUre)r:rHTensorrjrk contiguousrgr>rhr?rirRrSsavegetvalue)r]r%r&rDr[rbuffers rprepare_encoderfs GJ!'5 #$%1 >"%&  ))%*:*:*rs  77(F30! ====NN I E c IM&*$(/4/4*/ %+3,4    CDE   UO  5/  +, +, ,'  x(  )  SM  C= HIM/4/4 % 22CDE2+,2+, 2  2nIM&*$(/4*/ % (K (KCDE(K(KUO (K 5/ (K +, (K,'(K(K SM(K C=(K\IM&*$(/4/4*/ %+3,4 RR RCDER R UO R 5/ R+,R+,R,'RRx(R )R SMR C=R 5;;U[[)5. @ARp'+$(   , - UO 5/    '+$(D , -DUOD5/D2'+$( -- , --UO-5/ -  -r