From 911a43bfad38bb40513ba0f2bafce85fb4f1d159 Mon Sep 17 00:00:00 2001
From: Inga Ulusoy <inga.ulusoy@uni-heidelberg.de>
Date: Tue, 15 Aug 2023 13:07:41 +0200
Subject: [PATCH] check for missing keys (#144)

* check for missing keys

* fix code smells
---
 ammico/test/test_utils.py | 32 +++++++++++++++++++++++++
 ammico/utils.py           | 49 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 81 insertions(+)

diff --git a/ammico/test/test_utils.py b/ammico/test/test_utils.py
index 28dae4a..0ab8382 100644
--- a/ammico/test/test_utils.py
+++ b/ammico/test/test_utils.py
@@ -84,6 +84,38 @@ def test_initialize_dict(get_path):
     assert mydict == out_dict
 
 
+def test_check_for_missing_keys():
+    mydict = {
+        "file1": {"faces": "Yes", "text_english": "Something"},
+        "file2": {"faces": "No", "text_english": "Otherthing"},
+    }
+    # check that dict is not changed
+    mydict2 = ut.check_for_missing_keys(mydict)
+    assert mydict2 == mydict
+    # check that dict is updated if key is missing
+    mydict = {
+        "file1": {"faces": "Yes", "text_english": "Something"},
+        "file2": {"faces": "No"},
+    }
+    mydict2 = ut.check_for_missing_keys(mydict)
+    assert mydict2["file2"] == {"faces": "No", "text_english": None}
+    # check that dict is updated if more than one key is missing
+    mydict = {"file1": {"faces": "Yes", "text_english": "Something"}, "file2": {}}
+    mydict2 = ut.check_for_missing_keys(mydict)
+    assert mydict2["file2"] == {"faces": None, "text_english": None}
+    # now test the exceptions
+    with pytest.raises(ValueError):
+        ut.check_for_missing_keys({"File": "path"})
+    with pytest.raises(ValueError):
+        ut.check_for_missing_keys({"File": {}})
+    mydict = {
+        "file1": {"faces": "Yes"},
+        "file2": {"faces": "No", "text_english": "Something"},
+    }
+    with pytest.raises(ValueError):
+        ut.check_for_missing_keys(mydict)
+
+
 def test_append_data_to_dict(get_path):
     with open(get_path + "example_append_data_to_dict_in.json", "r") as file:
         mydict = json.load(file)
diff --git a/ammico/utils.py b/ammico/utils.py
index 6d0c014..0a9a380 100644
--- a/ammico/utils.py
+++ b/ammico/utils.py
@@ -154,6 +154,55 @@ def initialize_dict(filelist: list) -> dict:
     return mydict
 
 
+def check_for_missing_keys(mydict: dict) -> dict:
+    """Check the nested dictionary for any missing keys in the subdicts.
+
+    Args:
+        mydict(dict): The nested dictionary with keys to check.
+    Returns:
+        dict: The dictionary with keys appended."""
+    # check that we actually got a nested dict
+    if not isinstance(mydict[next(iter(mydict))], dict):
+        raise ValueError(
+            "Please provide a nested dictionary - you provided {}".format(
+                next(iter(mydict))
+            )
+        )
+    # gather all existing keys of first item in a list
+    subdict = mydict[next(iter(mydict))]
+    if len(list(subdict.keys())) < 1:
+        raise ValueError(
+            "Could not get any keys to compare to - please check if your nested dict is empty!"
+        )
+    for key in mydict.keys():
+        # compare keys of next item with first item
+        if subdict.keys() != mydict[key].keys():
+            # print a warning if key is not found and set to None
+            keys_a = set(subdict.keys())
+            keys_b = set(mydict[key].keys())
+            missing_keys_in_b = keys_a - keys_b
+            if missing_keys_in_b:
+                print(
+                    "Found missing key(s) {} in subdict {} - setting to None.".format(
+                        missing_keys_in_b, key
+                    )
+                )
+                for missing_key in missing_keys_in_b:
+                    mydict[key][missing_key] = None
+            # check that there are no other keys in the subdicts -
+            # this would only happen if there is a key missing in the first subdict
+            # then we would need to start over so best to
+            # abort if this happens - this is a very unlikely case
+            missing_keys_in_a = keys_b - keys_a
+            if missing_keys_in_a:
+                raise ValueError(
+                    "Could not update missing keys - first item already missing {}".format(
+                        missing_keys_in_a
+                    )
+                )
+    return mydict
+
+
 def append_data_to_dict(mydict: dict) -> dict:
     """Append entries from nested dictionaries to keys in a global dict."""