o p i]>@sddlmZddlZddlmZmZddlZddlmZ ddl m Z ddl Z ddl mZmZddlmZddlmZmZmZddlmZdd lmZdd lmZmZd d lmZerndd l mZddl m!Z!m"Z"m#Z#m$Z$m%Z%gZ&d(ddZ'e  d)d*ddZ(GdddZ)d+d&d'Z*dS),) annotationsN) TYPE_CHECKINGAny)Self)Variablecore) check_type)convert_np_dtype_to_dtype_ in_pir_mode static_only) LayerHelper)DataType)get_current_insertion_pointset_insertion_point_setitem_static)Tensor) DTypeLike ShapeLikeSize1 TensorIndex TensorLikereturnboolcCst|dvS)N)falseoff0none)strlower)valr"Z/home/app/PaddleOCR-VL/.venv_paddleocr/lib/python3.10/site-packages/paddle/static/input.py evaluate_flag6sr$namershaperdtypeDTypeLike | None lod_levelint paddle.Tensorc Csdd}tdit}t|dttfdt|dttfdt|}tt|D](}||dur4d||<t ||t rP||dkrP||dkrPt d ||q(|durYt }tr|}t |tsjt jj|}t}|t j|||t} t|| S|j|||tjjjd |d d d } tjd d} t| rtdit}t |tjjst|}|j did | i||d|dd| S)a This function creates a variable on the global block. The global variable can be accessed by all the following operators in the graph. The variable is a placeholder that could be fed with input, such as Executor can feed input into the variable. When `dtype` is None, the dtype will get from the global dtype by `paddle.get_default_dtype()`. Args: name (str): The name/alias of the variable, see :ref:`api_guide_Name` for more details. shape (list|tuple): List|Tuple of integers declaring the shape. You can set None or -1 at a dimension to indicate the dimension can be of any size. For example, it is useful to set changeable batch size as None or -1. dtype (np.dtype|str, optional): The type of the data. Supported dtype: bool, float16, float32, float64, int8, int16, int32, int64, uint8. Default: None. When `dtype` is not set, the dtype will get from the global dtype by `paddle.get_default_dtype()`. lod_level (int, optional): The LoD level of the DenseTensor. Usually users don't have to set this value. Default: 0. Returns: Variable: The global variable that gives access to the data. Examples: .. code-block:: python >>> # doctest: +SKIP("This has diff in xdoctest env") >>> import numpy as np >>> import paddle >>> paddle.enable_static() # Creates a variable with fixed size [3, 2, 1] # User can only feed data of the same shape to x # the dtype is not set, so it will set "float32" by # paddle.get_default_dtype(). You can use paddle.get_default_dtype() to # change the global dtype >>> x = paddle.static.data(name='x', shape=[3, 2, 1]) # Creates a variable with changeable batch size -1. # Users can feed data of any batch size into y, # but size of each data sample has to be [2, 1] >>> y = paddle.static.data(name='y', shape=[-1, 2, 1], dtype='float32') >>> z = x + y # In this example, we will feed x and y with np-ndarray "1" # and fetch z, like implementing "1 + 1 = 2" in PaddlePaddle >>> feed_data = np.ones(shape=[3, 2, 1], dtype=np.float32) >>> exe = paddle.static.Executor(paddle.framework.CPUPlace()) >>> out = exe.run(paddle.static.default_main_program(), ... feed={ ... 'x': feed_data, ... 'y': feed_data ... }, ... fetch_list=[z.name]) # np-ndarray of shape=[3, 2, 1], dtype=float32, whose elements are 2 >>> print(out) [array([[[2.], [2.]], [[2.], [2.]], [[2.], [2.]]], dtype=float32)] cSsRtjj}|j}t|dkrdS|D]}|dkr&tj|dSqdS)Nrz pd_op.data) paddlepirrdefault_main_programZ global_blockopslenr%r)r.r/opr"r"r#_reset_data_op_insertion_points     z,data.._reset_data_op_insertion_pointdatar%r&NrzIOnly -1 can be used in shape to indicate unknown dimension, but received T)r%r&r'type stop_gradientr)Zis_dataZneed_check_feedZFLAGS_enable_pir_in_executorout)r&r'Zplacer%)r5ZinputsZoutputsattrs)r3)!r localsrbytesrlisttupleranger0 isinstancer* ValueErrorr,Zget_default_dtyper r r-rr rZ_pir_opsr3ZPlacerZcreate_global_variableZVarDescZVarTypeZ DENSE_TENSORosenvirongetr$Z append_op) r%r&r'r)r2helperiZir_dtypeZprev_insertion_pointr7Z is_pir_moder"r"r#r3:sdL  &    r3c@seZdZdZ   d/d0ddZddZd1ddZed2d3ddZe d2d4ddZ d5d!d"Z d6d#d$Z d%d&Z d7d(d)Z d8d+d,Zd9d-d.ZdS): InputSpeca@ InputSpec describes the signature information of the model input, such as ``shape`` , ``dtype`` , ``name`` . This interface is often used to specify input tensor information of models in high-level API. It's also used to specify the tensor information for each input parameter of the forward function decorated by `@paddle.jit.to_static`. Args: shape (tuple(integers)|list[integers]): List|Tuple of integers declaring the shape. You can set "None" or -1 at a dimension to indicate the dimension can be of any size. For example, it is useful to set changeable batch size as "None" or -1. dtype (np.dtype|str, optional): The type of the data. Supported dtype: bool, float16, float32, float64, int8, int16, int32, int64, uint8. Default: float32. name (str): The name/alias of the variable, see :ref:`api_guide_Name` for more details. stop_gradient (bool, optional): A boolean that mentions whether gradient should flow. Default is False, means don't stop calculate gradients. Examples: .. code-block:: python >>> import paddle >>> from paddle.static import InputSpec >>> input = InputSpec([None, 784], 'float32', 'x') >>> label = InputSpec([None, 1], 'int64', 'label') >>> print(input) InputSpec(shape=(-1, 784), dtype=paddle.float32, name=x, stop_gradient=False) >>> print(label) InputSpec(shape=(-1, 1), dtype=paddle.int64, name=label, stop_gradient=False) float32NFr&rr'rr% str | Noner6rrNonecCsB|||_|durt|tjtfrt|}||_||_||_dSN) _verifyr&r>npr'rr r%r6)selfr&r'r%r6r"r"r#__init__s  zInputSpec.__init__cCst|j|j|jdS)Nr&r')r3r%r&r'rLr"r"r#_create_feed_layerszInputSpec._create_feed_layerrc Cs0t|jd|jd|jd|jd|jd S)Nz(shape=z, dtype=z, name=z, stop_gradient=))r5__name__r&r'r%r6rOr"r"r#__repr__s0zInputSpec.__repr__tensorrrcCsDt|ttjjtjjfr||j|j |p|j St dt |j d)a Generates a InputSpec based on the description of input tensor. Args: tensor(Tensor): the source tensor to generate a InputSpec instance Returns: A InputSpec instance generated from Tensor. Examples: .. code-block:: python >>> import paddle >>> from paddle.static import InputSpec >>> paddle.disable_static() >>> x = paddle.ones([2, 2], dtype="float32") >>> x_spec = InputSpec.from_tensor(x, name='x') >>> print(x_spec) InputSpec(shape=(2, 2), dtype=paddle.float32, name=x, stop_gradient=False) z0Input `tensor` should be a Tensor, but received .)r>rreagerrr,r-Valuer&r'r%r?r5rR)clsrTr%r"r"r# from_tensors zInputSpec.from_tensorndarraynpt.NDArray[Any]cCs||j|j|S)a Generates a InputSpec based on the description of input np.ndarray. Args: tensor(Tensor): the source numpy ndarray to generate a InputSpec instance Returns: A InputSpec instance generated from Tensor. Examples: .. code-block:: python >>> import numpy as np >>> from paddle.static import InputSpec >>> x = np.ones([2, 2], np.float32) >>> x_spec = InputSpec.from_numpy(x, name='x') >>> print(x_spec) InputSpec(shape=(2, 2), dtype=paddle.float32, name=x, stop_gradient=False) rN)rXrZr%r"r"r# from_numpy$szInputSpec.from_numpy batch_size int | Size1cCs|t|ttfrt|dkrtd|dt|d|d}nt|ts/tdt|jd|gt|j }t||_ |S)ac Inserts `batch_size` in front of the `shape`. Args: batch_size(int): the inserted integer value of batch size. Returns: The original InputSpec instance by inserting `batch_size` in front of `shape`. Examples: .. code-block:: python >>> from paddle.static import InputSpec >>> x_spec = InputSpec(shape=[64], dtype='float32', name='x') >>> x_spec.batch(4) >>> print(x_spec) InputSpec(shape=(4, 64), dtype=paddle.float32, name=x, stop_gradient=False) zLength of batch_size: z shall be 1, but received rUrz.type(batch_size) shall be `int`, but received ) r>r;r<r0r?r* TypeErrorr5rRr&)rLr]Z new_shaper"r"r#batch?s    zInputSpec.batchcCs0t|jdkr td||jdd|_|S)a: Removes the first element of `shape`. Returns: The original InputSpec instance by removing the first element of `shape` . Examples: .. code-block:: python >>> from paddle.static import InputSpec >>> x_spec = InputSpec(shape=[4, 64], dtype='float32', name='x') >>> x_spec.unbatch() >>> print(x_spec) # InputSpec(shape=(64,), dtype=paddle.float32, name=x) InputSpec(shape=(64,), dtype=paddle.float32, name=x, stop_gradient=False) rz8Not support to unbatch a InputSpec when len(shape) == 0.r_N)r0r&r?rJrOr"r"r#unbatchds zInputSpec.unbatchc Cst|ttfstdt|jdt|D]*\}}|dur4t|ts4td|dt|jd|d|dus<|dkr@d||<qt|S)zI Verifies the input shape and modifies `None` into `-1`. zJType of `shape` in InputSpec should be one of (tuple, list), but received rUNzshape[z$] should be an `int`, but received `z`:r4) r>r;r<r`r5rR enumerater*r?)rLr&rDZeler"r"r#rJ~s zInputSpec._verifyr*cCstt|j|j|jfSrI)hashr<r&r'r6rOr"r"r#__hash__szInputSpec.__hash__othercs0gd}ttuotfdd|DS)N)r&r'r%r6c3s$|] }t|t|kVqdSrI)getattr).0attrrfrLr"r# s z#InputSpec.__eq__..)r5all)rLrfslotsr"rjr#__eq__szInputSpec.__eq__cCs ||k SrIr")rLrfr"r"r#__ne__s zInputSpec.__ne__)rFNF) r&rr'rr%rGr6rrrH)rrrI)rTrr%rGrr)rZr[r%rGrr)r]r^rr)rr)rr*)rfrrrrr)rR __module__ __qualname____doc__rMrPrS classmethodrYr\rarbrJrernror"r"r"r#rEs&&    %  rExrindexrvaluercCs t|||S)a= x(Tensor): input Tensor. index(Scalar|Tuple|List|Tensor): Where should be set value. value(Scalar|Tensor): The value which is going to be set. [How to write index?] 1. ':' -> slice(), (1) a[:]=v -> setitem(a, slice(None,None,None), v) (2) a[1::2] -> setitem(a, slice(1,None,2), v) 2. if there are multiple indexes for axes, use TUPLE (Not LIST) to pack them. (1) a[1, 2]=v -> setitem(a, (1, 2), v) (2) a[[1,2],[2,3]]=v -> setitem(a, ([1,2],[2,3]), v) (3) a[1,:, 3] = v -> setitem(a, (1, slice(None,None,None),3), v) (4) a[1, ..., 2]=v -> setitem(a, (1, ..., 2), v) 3. You can always use TUPLE as index input, even there is only one index. (1) a[Tensor([10,10])]=v -> setitem(a, (Tensor([10,10]),), v) (2) a[1] = v -> setitem(a, (1,), v) r)rurvrwr"r"r#setitems rxrp)Nr) r%rr&rr'r(r)r*rr+)rurrvrrwrrr)+ __future__rr@typingrrnumpyrKZ numpy.typingZnptZtyping_extensionsrr,Z paddle.baserrZpaddle.base.data_feederrZpaddle.base.frameworkr r r Zpaddle.base.layer_helperr Zpaddle.base.libpaddler Zpaddle.base.libpaddle.pirrrZbase.variable_indexrrZpaddle._typingrrrrr__all__r$r3rErxr"r"r"r#s6         f