aggr.py: started working on script that combines cluster and code-export data...

aggr.py: started working on script that combines cluster and code-export data to produce final report

mossutils/aggr.py

+import logging
+import csv
+import pandas as pd
+from .utils import logging as logutils
+
+DEFAULT_CLUSTERS_SUMMARY_CSV_FILE="clusters.csv"
+DEFAULT_CX_COURSE_STUDENTS_CSV_FILE="cx_students.csv"
+
+def main(
+    clusters_summary_csv_file=DEFAULT_CLUSTERS_SUMMARY_CSV_FILE,
+    cx_course_students_csv_file=DEFAULT_CX_COURSE_STUDENTS_CSV_FILE):
+
+  clusters_csv = pd.read_csv(clusters_summary_csv_file)
+  students_csv = pd.read_csv(cx_course_students_csv_file)
+
+  clusters = clusters_csv.groupby("cluster_id")
+  for name, cluster in clusters:
+    ids = pd.concat([cluster["id1"], cluster["id2"]], ignore_index=False)
+    # print(cluster[["id1", "id2"]].unique())
+    print(ids.unique())
+
+  # for wtf in clusters[["id1", "id2"]]:
+  #   print(wtf)
+
+  # for cluster in clusters.groups:
+  #   print(cluster)
+
+  # for cluster in clusters.groupby("cluster_id"):
+  #   print(cluster["id1"])
+
+if __name__ == "__main__":
+  main()

setup.py

@@ -22,6 +22,7 @@ setup(
     'dataclass-csv',
     'pydot',
     'networkx',
+    'pandas',
     'Jinja2'
   ],
   # scripts=['bin/mossutils-moss'],
@@ -29,7 +30,8 @@ setup(
     "console_scripts": [
       'mu-moss = mossutils.moss:main',
       'mu-revise = mossutils.revise:main',
-      'mu-cluster = mossutils.cluster:main'
+      'mu-cluster = mossutils.cluster:main',
+      'mu-aggr = mossutils.aggr:main'
     ]
   },
   zip_safe=False)