WIP field definitions

main.py

@@ -2,16 +2,15 @@
 import json
 from utils import csv_parser as cps, data_functions as df
 from datetime import datetime
-from collections import defaultdict
 
 
 def conversion(csv_file: str):
-    """[summary]
+    """Converts and transforms a CSV file from the Sleep as Android app into a JSON file.
 
     Parameters
     ----------
     csv_file : str
-        [description]
+        Name / location of the CSV file to be processed.
     """
     first_pass = []
 
@@ -34,6 +33,7 @@ def conversion(csv_file: str):
     # let's identify each part and convert it into
     # something much more useable
     i = 0
+    id = 0
     json_array = []
     for record in first_pass:
         headers = []
@@ -43,10 +43,10 @@ def conversion(csv_file: str):
 
         for key in record:
             val = record[key]
-            id = 0
 
             header = df.process_header(key)
 
+            # these headers contain datetimes in various forms, so we want to standardise them
             if header in ('id', 'tracking_start', 'tracking_end', 'alarm_scheduled'):
                 datetime_value = df.process_dates(header, val)
 
@@ -56,7 +56,7 @@ def conversion(csv_file: str):
                     val = datetime.strftime(datetime_value, '%Y-%m-%d %H:%M')
 
             elif header in ('tracking_hours'):
-                val = df.process_numbers(header, val)
+                val = df.process_numbers(val)
 
             elif header.startswith('event'):
                 event = df.process_event(val)

utils/csv_parser.py

@@ -1,4 +1,28 @@
 
+saa_fields = {
+    'Id': {
+        'name': 'id',
+        'type': 'unix timestamp'
+    },
+    'Tz': {
+        'name': 'timezone',
+        'type': 'string'
+    },
+    'From': {
+        'name': 'tracking_start',
+        'type': 'datetime'
+    },
+    'To': {
+        'name': 'tracking_end',
+        'type': 'datetime'
+    },
+    'Sched': {
+        'name': 'alarm_scheduled',
+        'type': 'datetime'
+    },
+}
+
+
 def csv_headers(headers):
     """[summary]
 
@@ -43,3 +67,13 @@ def combine_record(headers, row):
     record = dict(zip_it)
 
     return record
+
+
+def get_instructions(header):
+    instruction = saa_fields[header]
+
+    return instruction
+
+
+def saa_field_parser(header, value):
+    data_type = get_instructions(header)

utils/data_functions.py

@@ -2,7 +2,7 @@
 
 
 def process_header(header: str) -> str:
-    """[summary]
+    """Converts all headers to lowercase and makes them a bit more descriptive.
 
     Parameters
     ----------
@@ -19,10 +19,9 @@ def process_header(header: str) -> str:
     Exception
         [description]
     """
-    # convert all headers to lowercase
+
     header = header.lower()
 
-    # rename some of the headings to be a bit more descriptive
     try:
         if header == 'tz':
             header = 'timezone'
@@ -33,29 +32,29 @@ def process_header(header: str) -> str:
         elif header == 'sched':
             header = 'alarm_scheduled'
         elif header == 'hours':
-            header = 'tracking_hours'
+            header = 'hours_tracked'
     except Exception as e:
         raise Exception(
             "An error occurred processing header '{}': {}".format(header, e))
 
     return header
 
 
-def process_dates(header, detail):
+def process_dates(header: str, detail: str) -> datetime:
     """[summary]
 
     Parameters
     ----------
-    header : [type]
+    header : str
         [description]
-    detail : [type]
+    detail : str
         [description]
 
     Returns
     -------
-    [type]
+    datetime
         [description]
-    """
+    """    
     if header == 'id':
         datetime_value = datetime.fromtimestamp(int(detail)/1000)
     else:
@@ -64,7 +63,7 @@ def process_dates(header, detail):
     return datetime_value
 
 
-def process_numbers(header, detail):
+def process_numbers(detail: str) -> float:
     """[summary]
 
     Parameters
@@ -79,9 +78,9 @@ def process_numbers(header, detail):
     [type]
         [description]
     """
-    detail = float(detail)
+    value = float(detail)
 
-    return detail
+    return value
 
 
 def process_event(event):