month end automation

etl/month_end_automation.py

@@ -71,6 +71,7 @@ filtered_dfs = []
 ra = df[
     df["ra"].str.lower().isin(["completed rdsap 10", "completed rdsap 9.9"])
 ].copy()
+ra["job_type"] = "RA"
 filtered_dfs.append(ra)
 
 
@@ -78,68 +79,94 @@ filtered_dfs.append(ra)
 att = df[
     df["att"].str.lower().isin(["completed"])
 ].copy()
+att["job_type"] = "ATT"
 filtered_dfs.append(att)
 
 # V1 Coordination
 v1 = df[
     df["v1 coordination status"].str.lower().isin(["rc complete"])
 ].copy()
+v1["job_type"] = "V1 Coordination"
 filtered_dfs.append(v1)
 
 # V2 Coordination
 _ = df[df["v2 invoiced"].fillna('').str.lower().isin(['to be invoiced'])]
-v2 = _[_["v2 dc/ima/pas"] > 0]
+v2 = _[_["v2 dc/ima/pas"] > 0].copy()
+v2["job_type"] = "V2 Coordination"
 filtered_dfs.append(v2)
 
 # V3 Coordination
 v3 = df[
     df["v3  invoiced"].str.lower().isin(["to be invoiced"])
 ].copy()
+v3["job_type"] = "V3 Coordination"
 filtered_dfs.append(v3)
 
 # Design stage 1
 design1 = df[
     df["design invoice status"].str.lower().isin(["to invoice"])
-]
+].copy()
+design1["job_type"] = "Design"
 filtered_dfs.append(design1)
 
 # Design stage 2
 design2 = df[
     df["revision 2 design invoice"].str.lower().isin(["to invoice"])
-]
+].copy()
+design2["job_type"] = "Design Revision"
 filtered_dfs.append(design2)
 
 # Lodgement Phase 1
 lodg1 = df[
     df["lodg. phase 1 invoice status"].str.lower().isin(["to invoice"])
-]
+].copy()
+lodg1["job_type"] = "Lodgement Phase 1"
 filtered_dfs.append(lodg1)
 
 # Full Lodgement Phase
 lodg2 = df[
     df["full lodgement invoice status"].str.lower().isin(["to invoice"])
-]
+].copy()
+lodg2["job_type"] = "Full Lodgement"
 filtered_dfs.append(lodg2)
 
 # POST EPC
-# !!! Need Kevins input ( how does he determine invoice in this one)
+post_epc = df[
+    df["post-epc status"].str.lower().isin(["epc pics uploaded"])
+].copy()
+post_epc["job_type"] = "POST EPC"
+filtered_dfs.append(post_epc)
+
+
+# POST EPR
+post_epr = df[
+    df["post-epc status"].str.lower().isin(["post epr completed"])
+].copy()
+post_epr["job_type"] = "POST ATT"
+filtered_dfs.append(post_epr)
+
+
 
 # Post ATT
-# !!! Check with kevin, if he uses epc pic uploaded ( to invoice ) or uses anything else
-filtered_df = df[
+post_att = df[
     df["post-att"].str.lower().isin(["post-att uploaded"])
-]
+].copy()
+post_att["job_type"] = "POST ATT"
+filtered_dfs.append(post_att)
+
 
 # Retrofit Evaluation
 retro = df[
     df["retrofit evaluation"].str.lower().isin(["complete"])
-]
+].copy()
+retro["job_type"] = "Retrofit Evaluation"
 filtered_dfs.append(retro)
 
 # RA NO Show
 ra_ns  = df[
     df["ra no show evidence"].fillna(-9999) != df["ra no show invoice"].fillna(-9999)
-]
+].copy()
+ra_ns["job_type"] = "RA NO SHOW"
 filtered_dfs.append(ra_ns)
 
 
@@ -147,15 +174,18 @@ filtered_dfs.append(ra_ns)
 # ATT NO Show
 att_ns = df[
     df["att no show evidence"].fillna(-9999) != df["att no show invoice"].fillna(-9999)
-]
+].copy()
+att_ns["job_type"] = "ATT NO SHOW"
 filtered_dfs.append(att_ns)
 
 
 # Post visit no show
 epc_ns = df[
     df["epc no show evidence"].fillna(-9999) != df["epc no show invoice"].fillna(-9999)
-]
+].copy()
+epc_ns["job_type"] = "EPC NO SHOW"
 filtered_dfs.append(epc_ns)
 
 final_df = pd.concat(filtered_dfs).drop_duplicates().reset_index(drop=True)
 
+final_df[['address', 'client', 'job_type']]