added dicomSort functionality to downloadSeries

.gitignore

@@ -137,4 +137,6 @@ old_src
 resources/*
 dicomsort.py
 NBIA-toolkit.code-workspace
-docs/data/*
+docs/data/*
+driver.py
+data/*

driver.py

@@ -14,43 +14,53 @@
     password=PASSWORD,
     log_level="DEBUG")
 
-series = client.getSeries(Collection="RADCURE")
+# series = client.getSeries(Collection="RADCURE")
 
-seriesList = [_["SeriesInstanceUID"] for _ in series]
-seriesList = seriesList[0:5]
+# seriesList = [_["SeriesInstanceUID"] for _ in series]
 
+series_one = "1.3.6.1.4.1.14519.5.2.1.245877810326980920641905130135164701440"
 
+client.downloadSeries(
+    SeriesInstanceUID=series_one, 
+    downloadDir="./data/rawdata",
+    filePattern = '%PatientName/%StudyDescription-%StudyDate/%SeriesNumber-%SeriesDescription-%SeriesInstanceUID/%InstanceNumber.dcm',
+    overwrite = True
+    )
 
 
-def download(series) -> bool:
-    response = client.downloadSeries(
-        SeriesInstanceUID = series,
-        downloadDir = "/home/bioinf/bhklab/jermiah/projects/rawdata")
-    return response
 
-if (True):
-    # iterate through each series and download
-    # print out the progressbar
-    with mp.Pool(processes=30) as pool:
-        for _ in tqdm(pool.imap_unordered(download, seriesList), total=len(seriesList)):
-            pass
-else:
-    _log = logging.getLogger(__name__)
-    for handler in logging.root.handlers[:]:
-        logging.root.removeHandler(handler)
+# def download(series) -> bool:
+#     response = client.downloadSeries(
+#         SeriesInstanceUID = series,
+#         downloadDir = "./rawdata2")
+#     return response
 
-    USERNAME = "sejinkim"
-    PASSWORD = "q6mJyLD8cPeGTwg!"
-    nbia.getToken(user=USERNAME, pw=PASSWORD)
+# download(seriesList[0])
 
-    collection_name = "RADCURE"
-    metadata = nbia.getSeries(collection = collection_name, format = "df", api_url = "restricted")
+
+# # if (True):
+# #     # iterate through each series and download
+# #     # print out the progressbar
+# #     with mp.Pool(processes=30) as pool:
+# #         for _ in tqdm(pool.imap_unordered(download, seriesList), total=len(seriesList)):
+# #             pass
+# # else:
+#     _log = logging.getLogger(__name__)
+#     for handler in logging.root.handlers[:]:
+#         logging.root.removeHandler(handler)
+
+#     USERNAME = "sejinkim"
+#     PASSWORD = "q6mJyLD8cPeGTwg!"
+#     nbia.getToken(user=USERNAME, pw=PASSWORD)
+
+#     collection_name = "RADCURE"
+#     metadata = nbia.getSeries(collection = collection_name, format = "df", api_url = "restricted")
 
-    nbia.downloadSeries(series_data = seriesList, path = "/home/bioinf/bhklab/jermiah/projects/rawdata", input_type = "list",  api_url = "restricted")
-# with mp.Pool(processes=30) as pool:
-#     for _ in tqdm(pool.imap_unordered(download, seriesList), total=len(seriesList)):
-#         pass
-# python dicomsort.py -u /home/bioinf/bhklab/jermiah/projects/rawdata /home/bioinf/bhklab/jermiah/projects/NBIA-toolkit/resources/rawdata/%PatientID/%StudyDate-%StudyID-%StudyDescription-%StudyInstanceUID/%SeriesNumber-%SeriesDescription-%SeriesIntanceUID/%InstanceNumber-%SOPInstanceUID.dcm
+#     nbia.downloadSeries(series_data = seriesList, path = "/home/bioinf/bhklab/jermiah/projects/rawdata", input_type = "list",  api_url = "restricted")
+# # with mp.Pool(processes=30) as pool:
+# #     for _ in tqdm(pool.imap_unordered(download, seriesList), total=len(seriesList)):
+# #         pass
+# # python dicomsort.py -u /home/bioinf/bhklab/jermiah/projects/rawdata /home/bioinf/bhklab/jermiah/projects/NBIA-toolkit/resources/rawdata/%PatientID/%StudyDate-%StudyID-%StudyDescription-%StudyInstanceUID/%SeriesNumber-%SeriesDescription-%SeriesIntanceUID/%InstanceNumber-%SOPInstanceUID.dcm
 
 
 

src/nbiatoolkit/dicomsort/__init__.py

@@ -0,0 +1,15 @@
+# from .logger import setup_logger
+# from .nbia_endpoints import NBIA_ENDPOINTS
+# from .md5 import validateMD5
+# __all__ = [
+#     "setup_logger",
+#     "NBIA_ENDPOINTS",
+#     "validateMD5"
+# ]
+
+from .dicomsort import DICOMSorter
+from .helper_functions import parseDICOMKeysFromFormat, sanitizeFileName, truncateUID
+
+__all__ = [
+    "DICOMSorter"
+]

src/nbiatoolkit/nbia.py

@@ -2,6 +2,8 @@
 from nbiatoolkit.utils.nbia_endpoints import NBIA_ENDPOINTS
 from nbiatoolkit.utils.logger import setup_logger
 from nbiatoolkit.utils.md5 import validateMD5
+from nbiatoolkit.dicomsort import DICOMSorter
+
 import requests
 from requests.exceptions import JSONDecodeError as JSONDecodeError
 import io, zipfile, os
@@ -141,27 +143,57 @@ def getSeries(self,
 
         return response
 
+
     def downloadSeries(
-        self, SeriesInstanceUID: str, downloadDir: str,
+        self, SeriesInstanceUID: str, downloadDir: str, 
+        filePattern: str = '%PatientName/%StudyDescription-%StudyDate/%SeriesNumber-%SeriesDescription-%SeriesInstanceUID/%InstanceNumber.dcm',
+        overwrite: bool = False
         ) -> bool:
 
+        # create temporary directory
+        from tempfile import TemporaryDirectory
+
         params = dict()
         params["SeriesInstanceUID"] = SeriesInstanceUID
 
         response = self.query_api(
             endpoint = NBIA_ENDPOINTS.DOWNLOAD_SERIES,
             params = params)
 
-        if isinstance(response, bytes):
-            file = zipfile.ZipFile(io.BytesIO(response))
-            seriesDir = os.path.join(downloadDir, SeriesInstanceUID)
-            file.extractall(path=seriesDir)
+        if not isinstance(response, bytes):
+            # Handle the case where the expected binary data is not received
+            # Log error or raise an exception
+            return False
+
+        file = zipfile.ZipFile(io.BytesIO(response))
+
+        with TemporaryDirectory() as tempDir:
+            file.extractall(path=tempDir)
+            assert validateMD5(seriesDir=tempDir) == True
+
+            # Create an instance of DICOMSorter with the desired target pattern
+            sorter = DICOMSorter(
+                sourceDir = tempDir,
+                destinationDir=downloadDir,
+                targetPattern=filePattern,
+                truncateUID=True,
+                sanitizeFilename=True
+                )
+
+            sorter.sortDICOMFiles(option="move", overwrite=overwrite)
+
+
+
+        # if isinstance(response, bytes):
+        #     file = zipfile.ZipFile(io.BytesIO(response))
+        #     seriesDir = os.path.join(downloadDir, SeriesInstanceUID)
+        #     file.extractall(path=seriesDir)
 
-            validateMD5(seriesDir=seriesDir)
-        else:
-        # Handle the case where the expected binary data is not received
-        # Log error or raise an exception
-            pass
+        #     validateMD5(seriesDir=seriesDir)
+        # else:
+        # # Handle the case where the expected binary data is not received
+        # # Log error or raise an exception
+        #     pass
 
         return True