Create action dict that's created when updating data

* gn3/db/sample_data.py (__extract_actions): An update on a vector of data can
contain: inserts, deletes and updates. This functions extracts these actions
during an update.
* tests/unit/db/test_sample_data.py (test_extract_actions): Add test-case for
the above.

2 files changed, 53 insertions, 0 deletions

diff --git a/gn3/db/sample_data.py b/gn3/db/sample_data.py
index ccb03f2..06c3cc5 100644
--- a/gn3/db/sample_data.py
+++ b/gn3/db/sample_data.py
@@ -11,6 +11,38 @@ _MAP = {
 }
 
 
+def __extract_actions(original_data: str,
+                      updated_data: str,
+                      csv_header: str) -> dict:
+    original_data = original_data.strip().split(",")
+    updated_data = updated_data.strip().split(",")
+    csv_header = csv_header.strip().split(",")
+    result = {
+        "delete": {"data": [], "csv_header": []},
+        "insert": {"data": [], "csv_header": []},
+        "update": {"data": [], "csv_header": []},
+    }
+    for _o, _u, _h in zip(original_data, updated_data, csv_header):
+        if _o == _u:  # No change
+            continue
+        elif _o and _u == "x":  # Deletion
+            result["delete"]["data"].append(_o)
+            result["delete"]["csv_header"].append(_h)
+        elif _o == "x" and _u:  # Insert
+            result["insert"]["data"].append(_u)
+            result["insert"]["csv_header"].append(_h)
+        elif _o and _u:  # Update
+            result["update"]["data"].append(_u)
+            result["update"]["csv_header"].append(_h)
+    for key, val in result.items():
+        if not val["data"]:
+            result[key] = None
+        else:
+            result[key]["data"] = ",".join(result[key]["data"])
+            result[key]["csv_header"] = ",".join(result[key]["csv_header"])
+    return result
+
+
 def get_trait_csv_sample_data(conn: Any,
                               trait_name: int, phenotype_id: int) -> str:
     """Fetch a trait and return it as a csv string"""
diff --git a/tests/unit/db/test_sample_data.py b/tests/unit/db/test_sample_data.py
index 4c03dbd..4015ba5 100644
--- a/tests/unit/db/test_sample_data.py
+++ b/tests/unit/db/test_sample_data.py
@@ -2,6 +2,7 @@ import pytest
 
 from gn3.db.sample_data import insert_sample_data
 from gn3.db.sample_data import delete_sample_data
+from gn3.db.sample_data import __extract_actions
 
 
 @pytest.mark.unit_test
@@ -73,3 +74,23 @@ def test_delete_sample_data(mocker):
                              (strain_id, "Sex", inbredset_id)),
                  ]
         cursor.execute.assert_has_calls(calls, any_order=False)
+
+
+@pytest.mark.unit_test
+def test_extract_actions():
+    assert (__extract_actions(original_data="BXD1,18,x,0,x",
+                              updated_data="BXD1,x,2,1,F",
+                              csv_header="Strain Name,Value,SE,Count,Sex") ==
+            {
+                "delete": {"data": "18", "csv_header": "Value"},
+                "insert": {"data": "2,F", "csv_header": "SE,Sex"},
+                "update": {"data": "1", "csv_header": "Count"},
+            })
+    assert(__extract_actions(original_data="BXD1,18,x,0,x",
+                             updated_data="BXD1,19,2,1,F",
+                             csv_header="Strain Name,Value,SE,Count,Sex") ==
+           {
+               "delete": None,
+               "insert": {"data": "2,F", "csv_header": "SE,Sex"},
+               "update": {"data": "19,1", "csv_header": "Value,Count"},
+           })