Python question relating to dated values in a csv file

Question

I am loading a data file with dated data into a csv.DictReader

The data looks like this:

date,   weight, blood_pressure, sugar_level
1/1/01, 120.1,  100.1,          25.2
2/1/01, 130.1,  102.1,          26.2
3/1/01, 110.1,  120.1,          24.2
4/1/01, 130.1,  130.1,          28.2
5/1/01, 160.1,  104.1,          27.0
6/1/01, 150.5,  100.1,          22.5
7/1/01, 120.2,  129.1,          25.2
... etc...   

I want to write a function that allows me to pass in the dictionary (read from the csv file), and return the previous value of a column an epoch ago. Since there are no enumerations in Python, I will use enums to 'constrain' the permissible epochs and column names:

The code snippet I have so far looks something like this:

import csv;

headers = ['date', 'weight', 'blood_pressure', 'sugar_level']
data = csv.DictReader(csvfile, headers);

epoch_type   = (week_ago, fortnight_ago, month_ago);
column_names = headers[1:];

def function get_previous_value(data, column_name, epoch_name):
  """ Note: data is assumed to be sorted in date ascending order  """
  """ Returns the previous value in the data, using the specified """
  """ column name and epoch type                                  """

I would be grateful if someone could show how to implement this function

[Edit] Changed argument name from 'lookback' to 'epoch_name' to aid clarity

Answer 1

>>> epochdict = {'week_ago': 7, 'fortnight_ago': 14, 'month_ago': 30}
>>> datetime.datetime.strptime('1/1/01', '%m/%d/%y').date() + datetime.timedelta(epochdict['week_ago']) < datetime.date.today()
True

Answer 2

First of all, please don't use ; after commands in python.

import datetime
import csv

WEEK = datetime.timedelta(weeks=1)
DAY = datetime.timedelta(days=1)
MONTH = datetime.timedelta(days=30)

# read the entire file to memory in a dict keyed by date
data = {}
with open('file.csv') as csvfile:
    for row in csv.DictReader(csvfile):
        data[datetime.datetime.strptime(row['date'], '%d/%m/%y').date()] = row

now just query the data:

# blood pressure one week ago:
print data[(datetime.datetime.now() - WEEK).date()]['blood_pressure']

# the entire data for two months ago:
print data[(datetime.datetime.now() - 2 * MONTH).date()]

As an added bonus, you could use dateutil's relativedeltas to make stuff like "last friday in last month".

---

EDIT: If the file is too big to fit memory, the best solution would be to use a database.

Read the entire csv file to a sqlite database. You could use python to do this, but sqlite can already import csv files to table format on its own, and using it is faster than using python to read and parse the file.

Then you can just query the database:

import sqlite3
import datetime

# connect to db
con = sqlite3.connect('myfile.db', detect_types=sqlite3.PARSE_DECLTYPES)

date_i_want = datetime.date(2001, 1, 6)
cur.execute('SELECT * FROM data WHERE date = ?', (date_i_want,))
row = cur.fetchone()