From 270ef94653db30ff0b02733816f755b0296f99da Mon Sep 17 00:00:00 2001
From: Dan Levitas <dlevitas@iu.edu>
Date: Thu, 25 Jul 2024 16:50:28 +0000
Subject: [PATCH] [ENH] Provide abridged ezBIDS information for telemetry

---
 handler/bids.sh                    |  4 ++
 handler/ezBIDS_core/ezBIDS_core.py |  2 +-
 handler/telemetry.py               | 84 ++++++++++++++++++++++++++++++
 3 files changed, 89 insertions(+), 1 deletion(-)
 create mode 100755 handler/telemetry.py

diff --git a/handler/bids.sh b/handler/bids.sh
index dd44de6..79bec9b 100755
--- a/handler/bids.sh
+++ b/handler/bids.sh
@@ -33,5 +33,9 @@ bids-validator --json "$rootDir" > $root/validator.json || true
 echo "Copying finalized.json file to ezBIDS_template.json"
 cp -r $root/finalized.json $root/ezBIDS_template.json
 
+# Create telemetry-specific files
+echo "Creating ezBIDS telemetry files"
+./telemetry.py $root
+
 # Telemetry (to hard-coded pet2bids server)
 curl -H 'Content-Type: application/json' -d @$root/validator.json -X POST http://52.87.154.236/telemetry/
\ No newline at end of file
diff --git a/handler/ezBIDS_core/ezBIDS_core.py b/handler/ezBIDS_core/ezBIDS_core.py
index cbd5c4b..ed0f813 100755
--- a/handler/ezBIDS_core/ezBIDS_core.py
+++ b/handler/ezBIDS_core/ezBIDS_core.py
@@ -788,7 +788,7 @@ def correct_pe(pe_direction, ornt, correction):
     -------
     proper_pe_direction : string
         pe_direction, in "ijk" format.
-    
+
     correction: boolean
     """
     # axes = (("R", "L"), ("A", "P"), ("S", "I"))
diff --git a/handler/telemetry.py b/handler/telemetry.py
new file mode 100755
index 0000000..5092250
--- /dev/null
+++ b/handler/telemetry.py
@@ -0,0 +1,84 @@
+#!/usr/bin/env python3
+
+"""
+The purpose of this code is to filter BIDS information (data type, suffix, and entity labels) discerned from the
+ezBIDS Core. This will be passed to the telemetry service, so that comparisons between what ezBIDS guessed vs the
+final product (with user modifications) can be made.
+
+@author: dlevitas
+"""
+
+import os
+import sys
+import json
+
+# Begin
+DATA_DIR = sys.argv[1]
+os.chdir(DATA_DIR)
+
+# Create telemetry information from the ezBIDS Core json (ezBIDS guesses) as the finalized json (including user edits)
+
+
+def gather_telemetry(dtype):
+    if dtype == 'core':
+        json_path = f'{DATA_DIR}/ezBIDS_core.json'
+        output_file = "ezBIDS_core_telemetry.json"
+        error_message = 'There is no ezBIDS_core.json file, indicating failure during the ezBIDS Core processing. \
+            Please contact support for assistance'
+    else:
+        json_path = f'{DATA_DIR}/finalized.json'
+        output_file = "ezBIDS_finalized_telemetry.json"
+        error_message = 'There is no ezBIDS finalized file, indicating either failure during ezBIDS or inability to \
+            finalize dataset. Please contact support for assistance'
+
+    if os.path.isfile(json_path):
+        json_data = open(json_path)
+        json_data = json.load(json_data, strict=False)
+
+        for obj in json_data['objects']:
+
+            subject_idx = obj['subject_idx']
+            session_idx = obj['session_idx']
+            series_idx = obj['series_idx']
+
+            idx = str(subject_idx) + str(session_idx) + str(series_idx)
+
+            seq_file_name = json_data['series'][series_idx]['nifti_path']
+
+            ezBIDS_type = json_data['series'][series_idx]['type']
+            if '/' in ezBIDS_type:
+                data_type = ezBIDS_type.split('/')[0]
+                suffix = ezBIDS_type.split('/')[-1]
+            else:
+                data_type = 'exclude'
+                suffix = 'exclude'
+
+            rationale = json_data['series'][series_idx]['message']
+
+            entities = json_data['series'][series_idx]['entities']
+            known_entities = {}
+            for entity_key in entities:
+                if entities[entity_key] != '':
+                    entity_val = entities[entity_key]
+                    known_entities[entity_key] = entity_val
+
+            ezBIDS_telemetry_info_list.append([idx, seq_file_name, data_type, suffix, rationale, known_entities])
+
+    else:
+        print(error_message)
+
+    if len(ezBIDS_telemetry_info_list) > 1:
+        header = ezBIDS_telemetry_info_list[0]
+        rows = ezBIDS_telemetry_info_list[1:]
+        ezBIDS_telemetry_info_json = []
+        for row in rows:
+            ezBIDS_telemetry_info_json.append(dict(zip(header, row)))
+
+        with open(output_file, "w") as fp:
+            json.dump(ezBIDS_telemetry_info_json, fp, indent=3)
+
+
+# Execute functionality
+for dtype in ['core', 'finalized']:
+    ezBIDS_telemetry_info_list = [['idx', 'seq_file_name', 'data_type', 'suffix', 'rationale', 'known_entities']]
+    gather_telemetry(dtype)