modelbox.Buffer

构造方法

构造Buffer对象。

modelbox.Buffer(device, data)

args:

• device (modelbox.Device) —— 构造当前Buffer所在的modelbox.Device对象
• data (numpy.array) —— 当前Buffer包含的numpy数据

modelbox.Buffer(device, string)

args:

• device (modelbox.Device) —— 构造当前Buffer所在的modelbox.Device对象
• string (str) —— 当前Buffer包含的string数据

modelbox.Buffer(device, list_item)

args:

• device (modelbox.Device) —— 构造当前Buffer所在的modelbox.Device对象
• list_item (str) —— 当前Buffer包含的list数据，其中每一个元素必须同一类型

return:

modelbox.Buffer

example:

   import numpy as np
   ...
   def process(self, data_ctx):
       infer_data = np.ones((5,5))
       numpy_buffer = modelbox.Buffer(self.get_bind_device(), infer_data)
       str_buffer = modelbox.Buffer(self.get_bind_device(), "test")
       list_buffer = modelbox.Buffer(self.get_bind_device(), [3.1, 3.2, 3.3])
       ...

   return modelbox.Status()

modelbox.Buffer.as_object

将ModelBox Buffer对象自动转换成Python原始对象，如Buffer是由numpy类型转为而来，则调用as_object后返回numpy类型对象。目前支持基础类型、numpy.array、str类型。

args:

无

return:

基础类型、str 或者 numpy.array对象

example:

    ...
    def process(self, data_ctx):
        buf_list = data_ctx.input("input")
        for buf in buf_list:
            data = buf.as_object()
            print(data, type(data))
            ...

        return modelbox.Status()

modelbox.Buffer.has_error

判断当前Buffer是否存在处理异常。

args:

无

return:

bool, 是否存在处理异常

example:

    ...
    def process(self, data_ctx):
        buf_list = data_ctx.input("input")
        for buf in buf_list:
            res = buf.has_error()
            ...

        return modelbox.Status()

modelbox.Buffer.get_error

获取当前Buffer的第一个异常信息对象。

args:

无

return:

modelbox.FlowUnitError

example:

    ...
    def process(self, data_ctx):
        buf_list = data_ctx.input("input")
        for buf in buf_list:
            error = buf.get_error()
            ...

        return modelbox.Status()

modelbox.Buffer.get_bytes

获取当前Buffer的字节数

args:

无

return:

int64, Buffer的字节数

example:

    import numpy as np
    ...
    def process(self, data_ctx):
        infer_data = np.ones((5,5))
        new_buffer = modelbox.Buffer(self.get_bind_device(), infer_data)
        bytes = new_buffer.get_bytes()
        print(bytes)
        ...

        return modelbox.Status()

result:

字节数为 5*5*8 = 200

modelbox.Buffer.set

设置当前Buffer的某个Meta值

args:

• key (str) —— Meta的key值

• obj (int, str, double, bool, modelbox.ModelBoxDataType, list[str], list[int], list[double], list[bool], list[list], numpy) —— Meta的value值

return:

无

modelbox.Buffer.get

获取当前Buffer的某个Meta值

args:

• key (str) ——Meta的key值

return:

Python object 获取key对应的value值

type: int, double, str, bool, list[int], list[str], list[double], list[bool], list[list], numpy

example:

   ...
   def process(self, data_ctx):
        infer_data = np.ones((5,5))
        new_buffer = modelbox.Buffer(self.get_bind_device(), infer_data)
        res = new_buffer.set("key", "test")
        print(res)
        print(new_buffer.get("key"))
        ...

        return modelbox.Status()

result:

true test

modelbox.Buffer.copy_meta

把参数的Buffer的所有Meta信息拷贝给当前Buffer

args:

• buffer (modelbox.Buffer) —— 源Buffer

return:

modelbox.Status

example:

    import numpy as np
    ...
    def process(self, data_ctx):
        buf_list = data_ctx.input("input")
        for buf in buf_list:
            infer_data = np.ones((5,5))
            new_buffer = modelbox.Buffer(self.get_bind_device(), infer_data)
            #  new_buffer具有和buf相同的Meta信息
            status = new_buffer.copy_meta(buf)
            ...

        return modelbox.Status()