Using python regex to create a JSON array of temperature data

Question

I have a log file that stores temperatures in the format:

2013/09/30 11:23:01 Temperature 41.34F 5.19C
2013/09/30 11:23:01 Temperature 99.84F 37.69C
2013/09/30 11:23:01 Temperature 65.86F 18.81C
2013/09/30 11:25:02 Temperature 41.67F 5.38C
2013/09/30 11:25:02 Temperature 65.64F 18.69C
2013/09/30 11:25:02 Temperature 98.83F 37.12C

There are a variable number of values corresponding to a given minute, from 1-3. How would I use Python regular expressions to convert the data to JSON format, such that a series of values is given for each time and Fahrenheit value?

{"c":[{"v":"Date(2013, 8, 30, 11, 23)"},{"v":41.34},{"v":99.84},{"v":65.86}]},

So the script would open "temperatures.log", read through the file, take the time value and put it in the format:

{"c":[{"v":"Date(2013, 8, 30, 11, 23)"}, 

(with the month offset by -1)

and then loop through all the temperature values at that time and include each like:

{"v":41.34},

Until it found a date/time expression that was different from the previous line, and then close the expression with

]}, 

write the output file, and start the next series, until the end of the log file.

Answer 1

You don't need regular expressions for this, since your data is pretty straightforward. First, note that you can organize the data without even parsing the date, because you can use simple string comparison:

def proc_lines(lines):
    cur_date = None
    cur_temps = []

    results = []

    for line in lines:
        parts = line.split()
        date = "%s %s" % (parts[0], parts[1])
        if date != cur_date:
            if cur_temps:
                #save current data
                results.append((cur_date, cur_temps))
            #reset state
            cur_date = date
            cur_temps = []
        #add the line's temperature in fahrenheit, stripping out the 'F'
        cur_temps.append(float(parts[3][:-1]))

    #process the last line
    if cur_temps:
        results.append((cur_date, cur_temps))

    return results

Now results will be a list of (date, temperature) tuples with an unparsed date:

>>> lines = """2013/09/30 11:23:01 Temperature 41.34F 5.19C
2013/09/30 11:23:01 Temperature 99.84F 37.69C
2013/09/30 11:23:01 Temperature 65.86F 18.81C
2013/09/30 11:25:02 Temperature 41.67F 5.38C
2013/09/30 11:25:02 Temperature 65.64F 18.69C
2013/09/30 11:25:02 Temperature 98.83F 37.12C""".split("\n")
>>> results = proc_lines(lines)
>>> results
[('2013/09/30 11:23:01', [41.340000000000003, 99.840000000000003, 
                          65.859999999999999]), 
 ('2013/09/30 11:25:02', [41.670000000000002, 65.640000000000001, 
                          98.829999999999998])]

You can use datetime.datetime.strptime to actually parse the date and process the date (subtracting the month as you asked):

>>> import datetime
>>> def proc_datestr(date):
        dt = datetime.datetime.strptime(date, "%Y/%m/%d %H:%M:%S")
    return "Date(%d, %d, %d, %d, %d, %d)" % (
        dt.year, dt.month - 1, dt.day, dt.hour, dt.minute, dt.second)

>>> proc_datestr(results[0][0])
'Date(2013, 8, 30, 11, 23, 1)'

Note the format string "%Y/%m/%d %H:%M:%S" which parses dates as detailed here . This lovely built-in function obviates the need for you to write your own regexp to deal with the date.

Then you just process the results & dump to json as follows:

>>> import json
>>> def proc_result(result):
    date, temps = result
    res = {'c': [{'v': proc_datestr(date)}]}
    for temp in temps:
        res['c'].append({'v': temp})
    return json.dumps(res)

>>> proc_result(results[0])
'{"c": [{"v": "Date(2013, 8, 30, 11, 23, 1)"}, {"v": 41.340000000000003}, {"v": 99.840000000000003}, {"v": 65.859999999999999}]}'
>>> proc_result(results[1])
'{"c": [{"v": "Date(2013, 8, 30, 11, 25, 2)"}, {"v": 41.670000000000002}, {"v": 65.640000000000001}, {"v": 98.829999999999998}]}'