a dJ @s@ddlZddlZddlZddlmZddlmZddlmZddlm Z m Z m Z m Z m Z mZmZmZmZddlmZddlmZmZmZmZmZddlmZdd lmZdd lmZdd l m!Z!Gd d d eZ"ee#dddZ$eee efe#dddZ%eeee fedddZ&d1eeee fe#eee e e'e ffdddZ(d2eeee fe#e e'e fdddZ)ee*ddddZ+dde e eee e d d!d"Z,d3e ee'e d#d$d%Z-e e"e d&d'd(Z.ed4eee'e d)d*d+d,Z/e'e ee d-fe e'e fe e'e fee'd-fee#ee d-fe e'e ffd.d/d0Z0dS)5N) OrderedDict)contextmanager)partial) AnyCallableDict GeneratorIterableOptionalTupleTypeUnion)get_all_subclasses) BatchSampler DataLoaderDatasetIterableDatasetSampler) LightningEnum)MisconfigurationException)rank_zero_warn)pl_worker_init_functionc@s$eZdZdZdZeddddZdS) _WrapAttrTagsetdelN)argsreturncGs||jkrt}nt}||SN)SETsetattrdelattr)selfrfnr#h/var/www/html/stable-diffusion-webui/venv/lib/python3.9/site-packages/lightning_fabric/utilities/data.py__call__$s z_WrapAttrTag.__call__)__name__ __module__ __qualname__rDELrr%r#r#r#r$r sr) dataloaderrcCst|dot|jtS)Ndataset)hasattr isinstancer+r)r*r#r#r$has_iterable_dataset-sr.c Csfz(t|dkr"td|jjdd}WnttfyBd}Yn0|rbt|trbt|rbtd|S)z}Checks if a given Dataloader has ``__len__`` method implemented i.e. if it is a finite dataloader or infinite dataloader.r`z>` returned 0 length. Please make sure this was your intention.TFzYour `IterableDataset` has `__len__` defined. In combination with multi-process data loading (when num_workers > 1), `__len__` could be inaccurate if each worker is not configured independently to avoid having duplicate data.) lenr __class__r& TypeErrorNotImplementedErrorr-rr.)r*has_lenr#r#r$r41s  r4)r*samplerrcCs(t||\}}t|g|Ri|}|Sr)$_get_dataloader_init_args_and_kwargs_reinstantiate_wrapped_cls)r*r5dl_args dl_kwargsr#r#r$_update_dataloaderHsr:F)r*r5disallow_batch_samplerrc st|tstd|dt|d}|rB|j}|j|j|j}n(ddt| Dd}|j d<dt t |jj}tdd |D}|r|r|d dt tjj Dn |t tjj|d d|sfd d | Ddfdd Dd}d|}t|trHdd<dd<nt|||fdd |D} | rt| } |jj} ddd | D} td| d| d| d| d |sttB|} | rt| }|jj} td| d|d| d|fS)NzThe dataloader z0 needs to subclass `torch.utils.data.DataLoader`__pl_saved_argscSs i|]\}}|ds||qS)_) startswith.0kvr#r#r$ ^z8_get_dataloader_init_args_and_kwargs..multiprocessing_contextr#css|]}|j|juVqdSr)kind VAR_KEYWORDr@pr#r#r$ hrDz7_get_dataloader_init_args_and_kwargs..cSs"i|]\}}|j|jur||qSr#)defaultemptyr?r#r#r$rCprDr!cs*h|]"\}}|vr|j|ur|qSr#)rK)r@namerI)attrsr#r$ xrDz7_get_dataloader_init_args_and_kwargs..r+csi|]\}}|vr||qSr#r#r?) non_defaultsr#r$rC}rD batch_samplerr5csDh|]<}|j|j|jfvr|j|jur|jvr|jvr|jqSr#)rFPOSITIONAL_ONLYPOSITIONAL_OR_KEYWORDrKrLrMrH) arg_namesr9r#r$rOs    z, css|]}d|dVqdS)z`self.r/Nr#)r@Zarg_namer#r#r$rJrDz,Trying to inject custom `Sampler` into the `z` instance. This would fail as some of the `__init__` arguments are not available as instance attributes. The missing attributes are z. If you instantiate your `zZ` inside a `*_dataloader` hook of your module, we will do this for you. Otherwise, define z inside your `__init__`.z&Trying to inject parameters into the `z{` instance. This would fail as it doesn't expose all its attributes in the `__init__` signature. The missing arguments are z. HINT: If you wrote the `zA` class, add the `__init__` arguments or allow passing `**kwargs`) r-r ValueErrorr,r<__pl_saved_kwargs__pl_saved_arg_namesZ __datasetvarsitemsrEdictinspect signature__init__ parametersanyvaluesupdatepopaddgetr'_dataloader_init_kwargs_resolve_samplersortedr1r&joinrrkeysr2)r*r5r;Z was_wrappedr8Zoriginal_datasetparamsZhas_variadic_kwargsr+Z required_argsZsorted_required_argsZdataloader_cls_nameZmissing_args_messageZmissing_kwargsZsorted_missing_kwargsr#)rTrNr9rPr$r6Nsx          r6c CsFt|d}|dur:|rDt|tur:|j|kr:|j|jksBtdnt|tur:t|}t|dr|j}|j}|j }|j }t d|||||\} }}| st d|j dt|g|Ri|}nhz|||j|jd}WnPt y(} z6d dl} | d t| } | s t d | WYd} ~ n d} ~ 00dd |d d dS|d ddS)zThis function is used to handle the sampler, batch_sampler arguments associated within a DataLoader for its re-instantiation.rQNzdIt is not possible to have a batch sampler in your dataloader, when running on multiple IPU devices.r<r5zYTrying to inject a modified sampler into the batch sampler; however, it seems the class `z` does not have an argument called `sampler.` To mitigate this, expose an argument `sampler` in the `__init__` method of your custom class.) batch_size drop_lastrz:.*__init__\(\) (got multiple values)|(missing \d required)zWe tried to re-instantiate your custom batch sampler and failed. To mitigate this, either follow the API of `BatchSampler` or instantiate your custom batch sampler inside `*_dataloader` hooks of your module.F)r5shufflerQrjrk)r5rmrQ)getattrtyperr5rjrr,r<rV__pl_saved_default_kwargsrW_replace_value_in_saved_argsr2r(r7rkrematchstr) r*r5r;rQZbatch_sampler_clsrkwargsdefault_kwargsrTsuccesserrrsr#r#r$resh       re)r*rankrcCs.ttjddr*|jdur*tt|d|_dS)NZPL_SEED_WORKERSr)ry)intosenvironrdZworker_init_fnrr)r*ryr#r#r$_auto_add_worker_init_fnsr}) explicit_cls) orig_objectrr~rurc Os|durt|n|}z||i|}Wnxty}z`ddl}|dt|}|sT|d} d|jd| d| d| d } t| |WYd}~n d}~00t|dt } | D]\}} | |g|Rq|S) Nrz-.*__init__\(\) got multiple values .* '(\w+)'zThe zd implementation has an error where more than one `__init__` argument can be passed to its parent's `zr=...` `__init__` argument. This is likely caused by allowing passing both a custom argument that will map to the `zc` argument as well as `**kwargs`. `kwargs` should be filtered to make sure they don't contain the `zR` key. This argument was automatically passed to your object by PyTorch Lightning.__pl_attrs_record) ror2rrrsrtgroupsr&rrnlist) rr~rru constructorresultrxrrrsargumentmessage attrs_recordr"r#r#r$r7s,    r7)initstore_explicit_argrcs(ttttddfdd }|S)zWraps the ``__init__`` method of classes (currently :class:`~torch.utils.data.DataLoader` and :class:`~torch.utils.data.BatchSampler`) in order to enable re-instantiation of custom subclasses.N)objrrurcs"t|dd}t|ddtj}tdd|D}t|dt |fdd| D}t |dst|d|t|d t|d t|d |durvrt|d | n vrt|d |g|Rit|d|dS) N__pl_inside_initFTcss6|].}|jdkr|j|j|jfvr|j|jfVqdS)r!N)rMrFVAR_POSITIONALrGrK)r@paramr#r#r$rJ&sz5_wrap_init_method..wrapper..cs2i|]*\}}|vr|vr|tjjkr||qSr#)r[ ParameterrL)r@rMvalueruZ param_namesr#r$rC.sz6_wrap_init_method..wrapper..r<rVrWrp__) rnobject __setattr__r[r\r^rr`tupler0rYr,index)rrruZold_inside_initriZparameters_defaultsrvrrrr$wrappers,     z"_wrap_init_method..wrapper functoolswrapsr)rrrr#rr$_wrap_init_methods'r)methodtagrcs&tttddfdd }|S)zWraps the ``__setattr__`` or ``__delattr__`` method of classes (currently :class:`~torch.utils.data.DataLoader` and :class:`~torch.utils.data.BatchSampler`) in order to enable re- instantiation of custom subclasses.N)rrrcs|^}}t|dd\}}||ko&|k }t|d|f|g|R|rt|ddst|dt}||ft|d|t|d||fdS)NZ__pl_current_call)NrrTr)rnrrrappend)rrrMr=Zprev_call_nameZprev_call_methodZ first_callrrrr#r$rNsz"_wrap_attr_method..wrapperr)rrrr#rr$_wrap_attr_methodIsr)NNN)base_clsrrc cst||hB}|D]}d|jvr6|j|_t|j||_dtjfdtjffD]N\}}||jvsd||urJd|}t||t ||t||t t |||qJqdV|D]F}dD]<}d||jvrt||t |d|t |d|qqdS)zThis context manager is used to add support for re-instantiation of custom (subclasses) of `base_cls`. It patches the ``__init__``, ``__setattr__`` and ``__delattr__`` methods. r]r __delattr__Z__oldN)rrr]) r__dict__r]Z __old__init__rrrr)rrnrr )rrclassesclsZ patch_fn_namerZ saved_nameZ patched_namer#r#r$_replace_dunder_methodses   r.) replace_key replace_valuerrurvrTrcCsj||vr>||}|d||f||dd}d||fS||vsN||vr`|||<d||fSd||fS)zTries to replace an argument value in a saved list of args and kwargs. Returns a tuple indicating success of the operation and modified saved args and kwargs NrlTF)r)rrrrurvrTZ replace_indexr#r#r$rqs  "  rq)F)F)N)N)1rr[r{ collectionsr contextlibrrtypingrrrrr r r r r Z$lightning_utilities.core.inheritancerZtorch.utils.datarrrrrZ lightning_fabric.utilities.enumsrZ%lightning_fabric.utilities.exceptionsrZ$lightning_fabric.utilities.rank_zerorZlightning_fabric.utilities.seedrrboolr.r4r:rtr6rerzr}r7rrrrqr#r#r#r$sX   ,         a  J /