From 6899c67fa4f55e9d7a82f6b2cd780bc43185f3fc Mon Sep 17 00:00:00 2001
From: Dominic Ricottone <me@dominic-ricottone.com>
Date: Wed, 11 May 2022 22:27:35 -0500
Subject: [PATCH] With data scraping complete, moving on to analysis

Basic summation of projects, as proof of concept

Basic SAS program for importing CSV data and storing as time series data
---
 analysis/timesheets.sas | 26 ++++++++++++++++++++++++++
 analysis/totals.py      | 32 ++++++++++++++++++++++++++++++++
 main.py                 |  4 ++++
 3 files changed, 62 insertions(+)
 create mode 100644 analysis/timesheets.sas
 create mode 100644 analysis/totals.py

diff --git a/analysis/timesheets.sas b/analysis/timesheets.sas
new file mode 100644
index 0000000..a6047ca
--- /dev/null
+++ b/analysis/timesheets.sas
@@ -0,0 +1,26 @@
+proc import datafile="/home/u44593168/timesheets_sas.csv"
+		out=work.raw
+		dbms=csv
+		replace;
+	getnames=no;
+run;
+
+data work.clean;
+	set work.raw;
+	project = VAR1;
+	date = VAR2;
+	hours = VAR3;
+	drop VAR1 VAR2 VAR3;
+run;
+
+proc sort data=work.clean;
+	by project date;
+run;
+
+proc timeseries data=work.clean
+		out=work.final;
+	by project;
+	id date interval=day accumulate=total;
+	var hours;
+run;
+
diff --git a/analysis/totals.py b/analysis/totals.py
new file mode 100644
index 0000000..8df1a8d
--- /dev/null
+++ b/analysis/totals.py
@@ -0,0 +1,32 @@
+#!/usr/bin/env python3
+
+def totals(timesheets):
+    projects = {}
+
+    for timesheet in timesheets:
+        for entry in timesheet:
+            key = entry.project
+            if entry.time_code in ("HOL", "OTU", "VAC", "OPL", ):
+                key = entry.time_code
+
+            if key not in projects.keys():
+                projects[key] = {}
+                projects[key]["name"] = entry.label
+                projects[key]["hours"] = 0
+
+            for hours in entry.data.values():
+                projects[key]["hours"] += hours
+
+    for project, data in projects.items():
+        print(f"{project:20} {data['name']:100} {data['hours']}")
+
+def total_ocps2020(timesheets):
+    total = 0
+    for timesheet in timesheets:
+        for entry in timesheet:
+            if entry.project == "20032.001.20.005":
+                for date, hours in entry.data.items():
+                    total += hours
+                break
+    print(f"{total} hours spent on OCPS 2020")
+
diff --git a/main.py b/main.py
index b2cea10..b9a9f4c 100644
--- a/main.py
+++ b/main.py
@@ -10,6 +10,8 @@ from parser.timesheet import parse as parse_timesheet
 
 from exporter.long_csv import export
 
+from analysis.totals import totals, total_ocps2020
+
 def main(filelist):
     timesheets = []
 
@@ -26,6 +28,8 @@ def main(filelist):
     dest_filename = pathlib.Path("analysis/timesheets_sas.csv")
     export(dest_filename, timesheets)
 
+    total_ocps2020(timesheets)
+
 if __name__ == "__main__":
     filelist = []
     for filename in sys.argv[1:]:
-- 
2.45.2