From 8b0ccfaff2c4f93bd383d6b73e5c299deee4c51f Mon Sep 17 00:00:00 2001
From: Electricity Maps <electricitymaps@Electricitys-MacBook-Pro.local>
Date: Mon, 19 Aug 2024 11:48:06 +0200
Subject: [PATCH 1/2] fix(parsers):India parser updated for new data format

---
 parsers/IN.py | 40 ++++++++++++++++++++++++++++------------
 1 file changed, 28 insertions(+), 12 deletions(-)

diff --git a/parsers/IN.py b/parsers/IN.py
index 33db9728a3..7fbf8ca817 100644
--- a/parsers/IN.py
+++ b/parsers/IN.py
@@ -347,31 +347,46 @@ def fetch_consumption(
     ).to_list()
 
 
-def format_ren_production_data(url: str, zone_key: str) -> dict[str, Any]:
+def format_ren_production_data(url: str, zone_key: str, target_datetime: datetime) -> dict[str, Any]:
     """Formats daily renewable production data for each zone"""
     df_ren = pd.read_excel(url, engine="openpyxl", header=5, skipfooter=2)
     df_ren = df_ren.dropna(axis=0, how="all")
-    df_ren = df_ren.rename(
-        columns={
-            df_ren.columns[1]: "region",
-            df_ren.columns[2]: "wind",
-            df_ren.columns[3]: "solar",
-            df_ren.columns[4]: "unknown",
-        }
-    )
+    
+    #They changed format of the data from 2024/07/01
+    if target_datetime < datetime(2024, 7, 1, 0, 0, tzinfo=IN_TZ):
+        df_ren = df_ren.rename(
+            columns={
+                df_ren.columns[1]: "region",
+                df_ren.columns[2]: "wind",
+                df_ren.columns[3]: "solar",
+                df_ren.columns[4]: "unknown",
+            }
+        )
+    else:
+        df_ren = df_ren.rename(
+            columns={
+                df_ren.columns[0]: "region",
+                df_ren.columns[1]: "wind",
+                df_ren.columns[2]: "solar",
+                df_ren.columns[3]: "unknown",
+            }
+        )
+    
     df_ren.loc[:, "zone_key"] = (
         df_ren["region"].apply(lambda x: x if "Region" in x else np.nan).backfill()
     )
+    
     df_ren["zone_key"] = df_ren["zone_key"].str.strip()
     df_ren["zone_key"] = df_ren["zone_key"].map(CEA_REGION_MAPPING)
-
+    
     zone_data = df_ren.loc[
         (df_ren.zone_key == zone_key) & (~df_ren.region.str.contains("Region"))
     ][["wind", "solar", "unknown"]].sum()
-
+    
     renewable_production = {
         key: round(zone_data.get(key) / CONVERSION_GWH_MW, 3) for key in zone_data.index
     }
+    
     return renewable_production
 
 
@@ -395,6 +410,7 @@ def fetch_cea_production(
             for elem in all_data
             if target_datetime.strftime("%Y-%m-%d") in elem["date"]
         ]
+        
         if len(target_elem) > 0:
             if target_elem[0]["link"] == "file_not_found":
                 raise ParserException(
@@ -406,7 +422,7 @@ def fetch_cea_production(
                 formatted_url = target_url.split("^")[0]
                 r: Response = session.get(formatted_url)
                 renewable_production = format_ren_production_data(
-                    url=r.url, zone_key=zone_key
+                    url=r.url, zone_key=zone_key, target_datetime=target_datetime
                 )
                 return renewable_production
     else:

From 8eeff54a806580fd3dfd2d341e6a91a05a50cbf0 Mon Sep 17 00:00:00 2001
From: Electricity Maps <electricitymaps@Electricitys-MacBook-Pro.local>
Date: Mon, 19 Aug 2024 11:57:51 +0200
Subject: [PATCH 2/2] fix(parsers):Reformatted

---
 parsers/IN.py | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/parsers/IN.py b/parsers/IN.py
index 7fbf8ca817..28e2272ab1 100644
--- a/parsers/IN.py
+++ b/parsers/IN.py
@@ -347,12 +347,14 @@ def fetch_consumption(
     ).to_list()
 
 
-def format_ren_production_data(url: str, zone_key: str, target_datetime: datetime) -> dict[str, Any]:
+def format_ren_production_data(
+    url: str, zone_key: str, target_datetime: datetime
+) -> dict[str, Any]:
     """Formats daily renewable production data for each zone"""
     df_ren = pd.read_excel(url, engine="openpyxl", header=5, skipfooter=2)
     df_ren = df_ren.dropna(axis=0, how="all")
-    
-    #They changed format of the data from 2024/07/01
+
+    # They changed format of the data from 2024/07/01
     if target_datetime < datetime(2024, 7, 1, 0, 0, tzinfo=IN_TZ):
         df_ren = df_ren.rename(
             columns={
@@ -371,22 +373,22 @@ def format_ren_production_data(url: str, zone_key: str, target_datetime: datetim
                 df_ren.columns[3]: "unknown",
             }
         )
-    
+
     df_ren.loc[:, "zone_key"] = (
         df_ren["region"].apply(lambda x: x if "Region" in x else np.nan).backfill()
     )
-    
+
     df_ren["zone_key"] = df_ren["zone_key"].str.strip()
     df_ren["zone_key"] = df_ren["zone_key"].map(CEA_REGION_MAPPING)
-    
+
     zone_data = df_ren.loc[
         (df_ren.zone_key == zone_key) & (~df_ren.region.str.contains("Region"))
     ][["wind", "solar", "unknown"]].sum()
-    
+
     renewable_production = {
         key: round(zone_data.get(key) / CONVERSION_GWH_MW, 3) for key in zone_data.index
     }
-    
+
     return renewable_production
 
 
@@ -410,7 +412,7 @@ def fetch_cea_production(
             for elem in all_data
             if target_datetime.strftime("%Y-%m-%d") in elem["date"]
         ]
-        
+
         if len(target_elem) > 0:
             if target_elem[0]["link"] == "file_not_found":
                 raise ParserException(