Q/jZ#ddlZddlmZddlmZmZmZmZmZm Z m Z m Z m Z ddl mZddlmZddlmZddlmZddlmZmZd d gZe d d ZedGdd eeZdZedGdd eZy)N) namedtuple) AnyCallableDictIteratorListOptionalSizedTypeVarUnion)default_collate)functional_datapipe)dataframe_wrapper) IterDataPipe)_check_unpickable_fnvalidate_input_colCollatorIterDataPipeMapperIterDataPipe_T_coT) covariantmapcpeZdZUdZeed<eed< d dededdffd ZdZde e fdZ de fd Z xZS) ra Applies a function over each item from the source DataPipe (functional name: ``map``). The function can be any regular Python function or partial object. Lambda function is not recommended as it is not supported by pickle. Args: datapipe: Source Iterable DataPipe fn: Function being applied over each item input_col: Index or indices of data which ``fn`` is applied, such as: - ``None`` as default to apply ``fn`` to the data directly. - Integer(s) is used for list/tuple. - Key(s) is used for dict. output_col: Index of data where result of ``fn`` is placed. ``output_col`` can be specified only when ``input_col`` is not ``None`` - ``None`` as default to replace the index that ``input_col`` specified; For ``input_col`` with multiple indices, the left-most one is used, and other indices will be removed. - Integer is used for list/tuple. ``-1`` represents to append result at the end. - Key is used for dict. New key is acceptable. Example: >>> # xdoctest: +SKIP >>> from torchdata.datapipes.iter import IterableWrapper, Mapper >>> def add_one(x): ... return x + 1 >>> dp = IterableWrapper(range(10)) >>> map_dp_1 = dp.map(add_one) # Invocation via functional form is preferred >>> list(map_dp_1) [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] >>> # We discourage the usage of `lambda` functions as they are not serializable with `pickle` >>> # Use `functools.partial` or explicitly define the function instead >>> map_dp_2 = Mapper(dp, lambda x: x + 1) >>> list(map_dp_2) [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] datapipefnNreturnct|||_t|||_||_| | t dt|ttfrt|dkDr t d|d}||_ t||y)Nz3`output_col` must be None when `input_col` is None.z3`output_col` must be a single-element list or tupler) super__init__rrr input_col ValueError isinstancelisttuplelen output_colr)selfrrr r& __class__s k/root/repos/pocs/video-evals/.venv/lib/python3.12/site-packages/torch/utils/data/datapipes/iter/callable.pyrzMapperIterDataPipe.__init__Es   R "  !7RS S j4- 0:" !VWW#AJ$2y)c |j|j|jS|j|j}nlt|jtt fr.t fd|jD}|j|}n|j|j}tt rd}t nd}|jet|jtt fr5||jd<t |jdddD]}|=n@||j<n0|jdk(rj|n||j<|r t SS)Nc3(K|] }| ywN).0coldatas r) z/MapperIterDataPipe._apply_fn..cs=sc=sTFrr)reverse)r r&rr"r#r$sortedappend)r'r1resargst_flagidxs ` r) _apply_fnzMapperIterDataPipe._apply_fn\sF >> !doo&=774= >> !''$-C u 6=dnn==D$''4.C''$t~~./C dE "F:DF ?? "$..4-8*-T^^A&'!$.."4dC"CS "(+T^^$"$ C (+T__%%uT{.$.r*c#TK|jD]}|j|ywr-)rr;)r'r1s r)__iter__zMapperIterDataPipe.__iter__s(MM 'D..& & 's&(ct|jtrt|jSt t |j d)Nz# instance doesn't have valid length)r"rr r% TypeErrortype__name__)r's r)__len__zMapperIterDataPipe.__len__s= dmmU +t}}% %4:..//RSTTr*)NN)rA __module__ __qualname____doc__r__annotations__rrr;rrr=intrB __classcell__r(s@r)rrsc%NL  ** *  *.!/F'(5/'UUr*c.t|jdkDr td|d}tj|}g}g}|j D]}||vstd|D]|}||vrt ||s td||}n! ddlm}|jj}|jt||||} |j| ~td|} | |} | S#t$r} td| d} ~ wwxYw)Nrz%Only supports one DataFrame per batchrzConversion keys missmatchz5Collate (DF)DataPipe requires callable as dict valuesz?unable to import default collation function from the TorchArrow CollateResult)r%items RuntimeError df_wrapper get_columnskeyscallabletorcharrow.pytorchpytorchrecDefault Exceptionr6strr) conversionitemdf columns_name tuple_names tuple_valuesname collation_fntapevaluetpl_clsr$s r)_collate_helperrds7 4::BCC aB))"-LKL!< | #:; ;<# : Jt,-"K&d+L 0"ww0 3t9%RX&E")#0+6G \ "E L "U s C:: D DDcollatec teZdZdZedfdedeedefe ee efeeeffdfde eddffd Z xZ S) raf Collates samples from DataPipe to Tensor(s) by a custom collate function (functional name: ``collate``). By default, it uses :func:`torch.utils.data.default_collate`. .. note:: While writing a custom collate function, you can import :func:`torch.utils.data.default_collate` for the default behavior and `functools.partial` to specify any additional arguments. Args: datapipe: Iterable DataPipe being collated collate_fn: Customized collate function to collect and combine data or a batch of data. Default function collates to Tensor(s) based on data type. Example: >>> # xdoctest: +SKIP >>> # Convert integer data to float Tensor >>> class MyIterDataPipe(torch.utils.data.IterDataPipe): ... def __init__(self, start, end): ... super(MyIterDataPipe).__init__() ... assert end > start, "this example code only works with end >= start" ... self.start = start ... self.end = end ... ... def __iter__(self): ... return iter(range(self.start, self.end)) ... ... def __len__(self): ... return self.end - self.start ... >>> ds = MyIterDataPipe(start=3, end=7) >>> print(list(ds)) [3, 4, 5, 6] >>> def collate_fn(batch): ... return torch.tensor(batch, dtype=torch.float) ... >>> collated_ds = CollateIterDataPipe(ds, collate_fn=collate_fn) >>> print(list(collated_ds)) [tensor(3.), tensor(4.), tensor(5.), tensor(6.)] NrrX. collate_fnrc|t|||yt|rt|||ytjt |}t|||y)N)r)rrrQ functoolspartialrd)r'rrXrgr(s r)rzCollatorIterDataPipe.__init__s]  ! G X*  5 # j 9'.. K  j 9r*)rArCrDrEr rr rrrrWr rrHrIs@r)rrsy'\ )- :: S#X U38_eHcM6J%J KT Q : X& : ::r*)ri collectionsrtypingrrrrrr r r r torch.utils.data._utils.collater %torch.utils.data.datapipes._decoratorr$torch.utils.data.datapipes.dataframerrN#torch.utils.data.datapipes.datapiper'torch.utils.data.datapipes.utils.commonrr__all__rrrdrr.r*r)rss"WWW;EP<  4(UlUe,lUlU^(VY<:-<: <:r*