I guess step 3 is done

.gitignore

@@ -2,3 +2,4 @@
 
 .DS_Store
 *.pyc
+build/*

s3.py

@@ -1,3 +1,6 @@
+'''
+WTH
+'''
 # -*- coding:utf-8 -*-
 #!/usr/bin/env python
 from __future__ import print_function
@@ -18,6 +21,9 @@ fs = ('Case1_LS.csv', 'Case1_Gender.csv', 'Case1_Level.csv', 'Case1_SciF.csv')
 
 
 def process_s2_data():
+    '''
+    docstring huhu?
+    '''
     xm_data = {}
     for ii in xmodels.keys():
         fname = '%s-2.csv' % ii
@@ -30,29 +36,77 @@ def process_s2_data():
                     rank = r[2]
                     continue
                 if ii not in xm_data:
-                    xm_data[ii] = {}
+                    xm_data[ii] = {'all': set()}
                 if rank not in xm_data[ii]:
                     xm_data[ii][rank] = {}
-                xm_data[ii][rank] = r[1].split(',')
+                if r[0] not in xm_data[ii][rank]:
+                    xm_data[ii][rank][r[0]] = []
+                xm_data[ii][rank][r[0]] = r[1].split(',')
+                xm_data[ii]['all'] = xm_data[ii]['all'].union(set(r[1].split(',')))
     return xm_data
 
 
 def produce_match_all(xdata):
     rows = []
     for ii in xmodels.keys():
-        los = set()
+        _los = set()
         for _rank in xdata[ii]:
-            los = los.union(set(xdata[ii][_rank]))
+            for i in ii:
-        rows.append([ii, ','.join(list(los))])
+                if i in xdata[ii][_rank]:
+                    _los = _los.union(set(xdata[ii][_rank][i]))
+        rows.append([ii, ','.join(sorted(list(_los)))])
     _f = 'step3_match_all.csv'
     create_csv(_f, rows, directory='part3')
 
 
+def produce_match_all_mra(xdata):
+    for ii in xmodels.keys():
+        lo_count = {}
+        for j in xdata[ii]['all']:
+            if j not in lo_count:
+                lo_count[j] = 0
+                for i in ii:
+                    for rank in '1234':
+                        if i in xdata[ii][rank] and j in xdata[ii][rank][i]:
+                            lo_count[j] += 1
+        rows = [
+            ['4/4', ','.join([k for (k, v) in lo_count.items() if v == 4])],
+            ['3/4', ','.join([k for (k, v) in lo_count.items() if v == 3])],
+            ['2/4', ','.join([k for (k, v) in lo_count.items() if v == 2])],
+            ['1/4', ','.join([k for (k, v) in lo_count.items() if v == 1])],
+        ]
+        _f = '%s-all.csv' % (ii,)
+        create_csv(_f, rows, directory='part3')
+
+
+
+def produce_match_rank_mra(xdata, rank):
+    for ii in xmodels.keys():
+        lo_count = {}
+        for j in xdata[ii]['all']:
+            if j not in lo_count:
+                lo_count[j] = 0
+                for i in ii:
+                    if i in xdata[ii][rank] and j in xdata[ii][rank][i]:
+                        lo_count[j] += 1
+        rows = [
+            ['4/4', ','.join([k for (k, v) in lo_count.items() if v == 4])],
+            ['3/4', ','.join([k for (k, v) in lo_count.items() if v == 3])],
+            ['2/4', ','.join([k for (k, v) in lo_count.items() if v == 2])],
+            ['1/4', ','.join([k for (k, v) in lo_count.items() if v == 1])],
+        ]
+        _f = '%s-rank-%s.csv' % (ii, rank)
+        create_csv(_f, rows, directory='part3')
+
+
 def main(*argv):
-    xm = process_s2_data()
+    xmm = process_s2_data()
-    print(xm['CMHf'].keys())
+    print(xmm['CMHf']['1'].keys())
-    produce_match_all(xm)
+    produce_match_all(xmm)
-    pass
+    produce_match_all_mra(xmm)
+    for i in '1234':
+        produce_match_rank_mra(xmm, i)
+
 
 if __name__ == '__main__':
     if len(sys.argv) > 1: