s5

base.py

@@ -4,6 +4,13 @@ import os
 
 RAW_DIR = os.path.join(os.getcwd(), 'raw', 'july')
 TARGET_DIR = os.path.join(os.getcwd(), 'build')
+MODEL_LIST = [
+    'PMSf', 'PMSD', 'PMHf', 'PMHD', 'PFSf',
+    'PFSD', 'PFHf', 'PFHD', 'IMSf', 'IMSD',
+    'IMHf', 'IMHD', 'IFSf', 'IFSD', 'IFHf',
+    'IFHD', 'CMSf', 'CMSD', 'CMHf', 'CMHD',
+    'CFSf', 'CFSD', 'CFHf', 'CFHD'
+]
 
 
 class Lo:

s5.py

@@ -0,0 +1,65 @@
+'''
+Step 5
+'''
+# -*- coding:utf-8 -*-
+#!/usr/bin/env python
+from __future__ import print_function
+from base import create_csv, TARGET_DIR, MODEL_LIST
+import os
+import csv
+import sys
+
+
+def get_sorted_list(model_name):
+    _m = model_name
+    result = []
+    fpath = os.path.join(TARGET_DIR, 'part4', '%s-step4.csv' % _m)
+    if _m not in MODEL_LIST:
+        return []
+    weight_index = MODEL_LIST.index(_m) + 5  # offset
+    with open(fpath, 'rb') as f:
+        rows = csv.reader(f)
+        for r in rows:
+            # result[r[0]] = [r[1], r[2], r[3], r[4], r[5], r[6]]
+            result.append([
+                r[0],  # no of tree
+                r[1],  # lo_id
+                r[2],  # ???
+                r[3],  # ??
+                r[4],  # something count
+                r[weight_index]
+            ])
+    sorted_result = sorted(result, key=lambda x: x[5], reverse=True)
+
+    return sorted_result
+
+
+def loop_thru_model():
+    result = {}
+    max_count = 0
+    for m in MODEL_LIST:
+        result[m] = get_sorted_list(m)
+        max_count = max_count if max_count > len(result[m]) else len(result[m])
+
+    return (result, max_count)
+
+
+sorted_by_weight, max_count = loop_thru_model()
+rows = []
+# fill up header since there is no way to know what model is
+row = ['|', ]
+for m in MODEL_LIST:
+    row += ['-', '-', m, '-', '-', '|']
+rows.append(row)
+
+for i in xrange(0, max_count):
+    row = [i, ]
+    for m in MODEL_LIST:
+        try:
+            row += sorted_by_weight[m][i]
+        except IndexError:
+            row += ['', '', '', '', '', '']
+
+    rows.append(row)
+
+create_csv('sorted_all.csv', rows, directory='part5')