From c140aeaef17c2079a53aedfbccc471f22e9c6711 Mon Sep 17 00:00:00 2001
From: sebastianzwickl <64072650+sebastianzwickl@users.noreply.github.com>
Date: Wed, 25 Nov 2020 12:20:48 +0100
Subject: [PATCH] Add unit validation to `validate()` function (#84)

---
 nomenclature/__init__.py                      | 30 ++++++++++++++++++-
 .../definitions/variable/economy/economy.yaml |  2 +-
 nomenclature/tests/test_validate.py           |  4 +++
 3 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/nomenclature/__init__.py b/nomenclature/__init__.py
index 40c58e98..7e7aef11 100644
--- a/nomenclature/__init__.py
+++ b/nomenclature/__init__.py
@@ -179,7 +179,24 @@ def validate(df):
 
     # validate all (other) columns
     for col, codelist, ext in cols:
-        invalid = [c for c in df.data[col].unique() if c not in codelist]
+        invalid = []
+
+        # check variables for name and unit
+        if col == 'variable':
+            for c in df.data[col].unique():
+                # check if name is in codelist
+                # and unit in the .yaml file description
+                if (c not in codelist) or not(
+                        all(_s in variables[c][
+                            'unit'] for _s in df.data.loc[
+                            df.data['variable'] == c]['unit'].values)):
+                    invalid.append(c)
+                    success = False
+            # check if only unit is not valid
+            invalid = _validate_unit(invalid)
+        else:
+            invalid = [c for c in df.data[col].unique() if c not in codelist]
+
 
         # check if entries in the invalid list are related to directional data
         if col == 'region' and invalid:
@@ -265,3 +282,14 @@ def _validate_directional(x):
     """Utility function to check whether region-to-region code is valid"""
     x = x.split('>')
     return len(x) == 2 and all([i in regions for i in x])
+
+
+def _validate_unit(x):
+    # sub function to filter out variables with valid name
+    for i in reversed(x):  # iterate list reversely due to 'remove' method
+        if i in variables.keys():
+            _valid_units = variables[i]['unit']
+            logger.warning(
+                f'Unit for variable {i} is not in {_valid_units}!')
+            x.remove(i)
+    return x
diff --git a/nomenclature/definitions/variable/economy/economy.yaml b/nomenclature/definitions/variable/economy/economy.yaml
index 032cd481..9a11b94a 100644
--- a/nomenclature/definitions/variable/economy/economy.yaml
+++ b/nomenclature/definitions/variable/economy/economy.yaml
@@ -184,7 +184,7 @@ Price|Final Energy|Residential|Electricity:
                 Prices should include the effect of carbon prices
                 Mean price should reflect the variability of different prices that are accessible
                 to end-users (including regulated prices, prices proposed by different competiting retailers...)
-   unit: US$2010/GJ
+   unit: [euro/kWh, US$2010/GJ]
 
 Price|Final Energy|Residential|Gases|Natural Gas:
    description: Mean Natural gas price at the final level in the residential sector.
diff --git a/nomenclature/tests/test_validate.py b/nomenclature/tests/test_validate.py
index 97b5b0c5..85210e36 100644
--- a/nomenclature/tests/test_validate.py
+++ b/nomenclature/tests/test_validate.py
@@ -56,3 +56,7 @@ def test_validate_time_entry():
         replace([2005, 2010], value=['2005-06-17 00:00+01:00',
                                      '2010-07-21 12:00+01:00'])
     assert validate(IamDataFrame(df_sub))
+
+
+def test_validate_unit_entry():
+    assert not (validate(df.rename(unit={'EJ/yr': 'MWh'})))