From 30884c2c819ff509b983fad8b3284a369449dd52 Mon Sep 17 00:00:00 2001
From: Abhishek Dasgupta <abhishek.dasgupta@dtc.ox.ac.uk>
Date: Mon, 7 Aug 2023 12:59:26 +0100
Subject: [PATCH] transformations: add wordSubstituteSet (#80)

---
 adtl/transformations.py       | 24 ++++++++++++++++++++++++
 tests/test_transformations.py | 33 +++++++++++++++++++++++++++++++++
 2 files changed, 57 insertions(+)

diff --git a/adtl/transformations.py b/adtl/transformations.py
index 6ee1807..3ed8258 100644
--- a/adtl/transformations.py
+++ b/adtl/transformations.py
@@ -20,6 +20,30 @@ def textIfNotNull(field, return_val):
     return return_val if field not in [None, ""] else None
 
 
+def wordSubstituteSet(value, params):
+    """
+    For a value that can have multiple words, use substitutions from params.
+    params is a list of 2-tuples, in the form
+
+        [(w1, s1), (w2, s2), ... (w_n, s_n)]
+
+    where w1 is replaced by s1, w2 is replaced by s2.
+
+    Word matches are regular expressions, delimited by the `\b` word boundary
+    delimiter so can have arbitrary regular expressions to match. Any match of
+    regex w_n will use substitute s_n. Case is ignored when matching.
+    """
+    out = []
+    for i in params:
+        if not isinstance(i, (tuple, list)):
+            raise ValueError("wordSubstituteSet: params item not a tuple or list")
+        sub_map = dict(params)
+        for match, subst in sub_map.items():
+            if re.search(r"\b" + match + r"\b", value, re.IGNORECASE):
+                out.append(subst)
+    return sorted(set(out)) if out else None
+
+
 def getFloat(value, set_decimal=None, separator=None):
     """
     In cases where the decimal seperators is not a . you can
diff --git a/tests/test_transformations.py b/tests/test_transformations.py
index f674bed..fe9049a 100644
--- a/tests/test_transformations.py
+++ b/tests/test_transformations.py
@@ -10,6 +10,39 @@ def test_isNotNull(test_input, expected):
     assert transform.isNotNull(test_input) == expected
 
 
+@pytest.mark.parametrize(
+    "test_input,expected",
+    [
+        (
+            (
+                "Metilprednisolona - Dexametasona - Fluticasona",
+                [
+                    ("Metilprednisolona", "Methylprednisolone"),
+                    ("Fluticasona", "Fluticasone"),
+                    ("Dexametasona", "Dexamethasone"),
+                ],
+            ),
+            ["Dexamethasone", "Fluticasone", "Methylprednisolone"],
+        ),
+        (
+            (
+                "Hidrocortisona - Fluticasona",
+                [("Hidrocortisona", "Hydrocortisone"), ("Fluticasona", "Fluticasone")],
+            ),
+            ["Fluticasone", "Hydrocortisone"],
+        ),
+        ((("Hidrocortisona - Fluticasona"), [("Cortisona", "Cortisone")]), None),
+    ],
+)
+def test_wordSubstituteSet(test_input, expected):
+    assert transform.wordSubstituteSet(*test_input) == expected
+
+
+def test_wordSubstituteSet_error():
+    with pytest.raises(ValueError):
+        transform.wordSubstituteSet("value", [20, 30])
+
+
 @pytest.mark.parametrize(
     "test_date_birth, test_date_current, expected",
     [