使用Google Cloud DataFlow python sdk读取一组xml文件

Question

我正在尝试从GCS存储桶中读取XML文件的集合并处理它们，其中集合中的每个元素都是表示整个文件的字符串，但我找不到关于如何实现此目的的正确示例，我也无法理解它来自Apache Beam文档，主要是关于Java版本。

我当前的管道如下所示：

p = beam.Pipeline(options=PipelineOptions(pipeline_args))

(p
 | 'Read from a File' >> beam.io.Read(training_files_folder)
 | 'String To BigQuery Row' >> beam.Map(lambda s:
                                        data_ingestion.parse_method(s))
 | 'Write to BigQuery' >> beam.io.Write(
            beam.io.BigQuerySink(
                known_args.output,
                schema='title:STRING,text:STRING,id:STRING',
                create_disposition=beam.io.BigQueryDisposition.CREATE_IF_NEEDED,
                write_disposition=beam.io.BigQueryDisposition.WRITE_TRUNCATE)))
p.run().wait_until_finish()

我收到的错误信息是：

File "C:\Program Files\JetBrains\PyCharm Community Edition 2018.2.1\helpers\pydev\pydevd.py", line 1664, in <module>
main()

File "C:\Program Files\JetBrains\PyCharm Community Edition 2018.2.1\helpers\pydev\pydevd.py", line 1658, in main
globals = debugger.run(setup['file'], None, None, is_module)

File "C:\Program Files\JetBrains\PyCharm Community Edition 2018.2.1\helpers\pydev\pydevd.py", line 1068, in run
pydev_imports.execfile(file, globals, locals)  # execute the script

File "C:/Users/Tomer/PycharmProjects/hyperpartisan/cloud-version/data_ingestion.py", line 135, in <module>
run()

File "C:/Users/Tomer/PycharmProjects/hyperpartisan/cloud-version/data_ingestion.py", line 130, in run
p.run().wait_until_finish()
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\direct_runner.py", line 421, in wait_until_finish
self._executor.await_completion()
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\executor.py", line 398, in await_completion
self._executor.await_completion()
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\executor.py", line 444, in await_completion
six.reraise(t, v, tb)
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\executor.py", line 341, in call
finish_state)
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\executor.py", line 366, in attempt_call
side_input_values)
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\transform_evaluator.py", line 109, in get_evaluator
input_committed_bundle, side_inputs)
File "C:\Users\Tomer\anaconda\envs\hyperpartisan\lib\site-packages\apache_beam\runners\direct\transform_evaluator.py", line 283, in __init__
self._source.pipeline_options = evaluation_context.pipeline_options
AttributeError: 'str' object has no attribute 'pipeline_options'

非常感谢任何帮助。 谢谢Tomer

解决了第一个问题：事实证明这不适用于DirectRunner，将运行器更改为DataFlowRunner并用ReadFromText替换Read解决了异常：

p = beam.Pipeline（options = PipelineOptions（pipeline_args））

(p
 | 'Read from a File' >> beam.io.ReadFromText(training_files_folder)
 | 'String To BigQuery Row' >> beam.Map(lambda s:
                                        data_ingestion.parse_method(s))
 | 'Write to BigQuery' >> beam.io.Write(
            beam.io.BigQuerySink(
                known_args.output,
                schema='title:STRING,text:STRING,id:STRING',
                create_disposition=beam.io.BigQueryDisposition.CREATE_IF_NEEDED,
                write_disposition=beam.io.BigQueryDisposition.WRITE_TRUNCATE)))
p.run().wait_until_finish() 

但是现在我看到这种方法给了我一行来自每个文件作为管道元素，而我希望将整个文件作为字符串作为每个元素。 不知道该怎么做。 我发现这篇文章，但它是在java中，并不确定它如何与python和gcs版本具体。

所以看起来ReadFromText对我的用例不起作用，我不知道如何创建文件管道。

解决方案：感谢Ankur的帮助，我修改了代码以包含从MatchResult对象列表转换所需的步骤，这些对象是GCSFileSystem返回到字符串的pCollection，每个字符串代表一个文件。

p = beam.Pipeline(options=PipelineOptions(pipeline_args))
gcs = GCSFileSystem(PipelineOptions(pipeline_args))
gcs_reader = GCSFileReader(gcs)

(p
 | 'Read Files' >> beam.Create([m.metadata_list for m in gcs.match([training_files_folder])])
 | 'metadata_list to filepath' >> beam.FlatMap(lambda metadata_list: [metadata.path for metadata in metadata_list])
 | 'string To BigQuery Row' >> beam.Map(lambda filepath:
                                        data_ingestion.parse_method(gcs_reader.get_string_from_filepath(filepath)))
 | 'Write to BigQuery' >> beam.io.Write(
            beam.io.BigQuerySink(
                known_args.output,
                schema='title:STRING,text:STRING,id:STRING',
                # Creates the table in BigQuery if it does not yet exist.
                create_disposition=beam.io.BigQueryDisposition.CREATE_IF_NEEDED,
                # Appends data to the BigQuery table
                write_disposition=beam.io.BigQueryDisposition.WRITE_APPEND)))
p.run().wait_until_finish()

代码使用此帮助程序类来读取gcs文件：

class GCSFileReader:
  """Helper class to read gcs files"""
  def __init__(self, gcs):
      self.gcs = gcs

  def get_string_from_filepath(self,filepath):
      with self.gcs.open(filepath) as reader:
          res = reader.read()

      return res

Answer 1

ReadFromText在给定路径中逐行读取文件。 你想要的是一个文件列表，然后使用GcsFileSystem在ParDo中一次读取一个文件https://github.com/apache/beam/blob/master/sdks/python/apache_beam/io/gcp/gcsfilesystem.py然后将内容写入BigQuery。

您也可以参考类似主题的邮件主题https://lists.apache.org/thread.html/85da22a845cef8edd942fcc4906a7b47040a4ae8e10aef4ef00be233@%3Cuser.beam.apache.org%3E