This is an automated email from the ASF dual-hosted git repository.
yhu pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/beam.git
The following commit(s) were added to refs/heads/master by this push:
new b4c23b32f2b Correct Examples of utilizing `UploadToDicomStore` (#30008)
b4c23b32f2b is described below
commit b4c23b32f2b80ce052c8a235e5064c69f37df992
Author: Svetak Sundhar <[email protected]>
AuthorDate: Thu Jan 18 14:54:31 2024 +0000
Correct Examples of utilizing `UploadToDicomStore` (#30008)
* Create HealthcareUtils file with shared resources
* revert
* This PR introduces the correct transforms to utilize when calling
UploadToDicomStore()
* This PR introduces the correct transforms to utilize when calling
UploadToDicomStore()
* indent
* indent
---
.../apache_beam/io/gcp/healthcare/dicomio.py | 31 ++++++++++++----------
1 file changed, 17 insertions(+), 14 deletions(-)
diff --git a/sdks/python/apache_beam/io/gcp/healthcare/dicomio.py
b/sdks/python/apache_beam/io/gcp/healthcare/dicomio.py
index b31e0d3a78d..a73de19d5a3 100644
--- a/sdks/python/apache_beam/io/gcp/healthcare/dicomio.py
+++ b/sdks/python/apache_beam/io/gcp/healthcare/dicomio.py
@@ -77,7 +77,7 @@ transform. Here is a sample usage:
In the example above, the pipeline is listening to a pubsub topic and waiting
for messages from DICOM API. When a new DICOM file comes into the storage, the
-pipeline will receive a pubsub message, convert it to a Qido request dict and
+pipeline will receive a pubsub message, convert it to a Qido request dict, and
feed it to DicomSearch() PTransform. As a result, users can get the metadata
for
every new DICOM file. Note that not every pubsub message received is from DICOM
API, so we to filter the results first.
@@ -85,23 +85,26 @@ API, so we to filter the results first.
Store a DICOM file in a DICOM storage
===================================================
UploadToDicomStore() wraps store request API and users can use it to send a
-DICOM file to a DICOM store. It supports two types of input: 1.file data in
-byte[] 2.fileio object. Users should set the 'input_type' when initialzing
+DICOM file to a DICOM store. It supports two types of input: 1. fileio object
+2. file data in byte[]. Users should set the 'input_type' when initialzing
this PTransform. Here are the examples:
+ input_dict = {'project_id': 'abc123', 'type': 'instances',...}
+ str_input = json.dumps(dict_input)
+ temp_dir = '%s%s' % (self._new_tempdir(), os.sep)
+ self._create_temp_file(dir=temp_dir, content=str_input)
with Pipeline() as p:
- input_dict = {'project_id': 'abc123', 'type': 'instances',...}
- path = "gcs://bucketname/something/a.dcm"
- match = p | fileio.MatchFiles(path)
- fileio_obj = match | fileio.ReadAll()
- results = fileio_obj | UploadToDicomStore(input_dict, 'fileio')
-
+ results = (
+ p | beam.Create([FileSystems.join(temp_dir, '*')])
+ | fileio.MatchAll() | fileio.ReadMatches()
+ | UploadToDicomStore(input_dict, 'fileio'))
+
+ input_dict = {'project_id': 'abc123', 'type': 'instances',...}
+ str_input = json.dumps(dict_input)
+ bytes_input = bytes(str_input.encode("utf-8"))
with Pipeline() as p:
- input_dict = {'project_id': 'abc123', 'type': 'instances',...}
- f = open("abc.dcm", "rb")
- dcm_file = f.read()
- byte_file = p | 'create byte file' >> beam.Create([dcm_file])
- results = byte_file | UploadToDicomStore(input_dict, 'bytes')
+ results = (
+ p | beam.Create([bytes_input]) | UploadToDicomStore(input_dict, 'bytes'))
The first example uses a PCollection of fileio objects as input.
UploadToDicomStore will read DICOM files from the objects and send them
