GEP 662 Final project - Python on a Bike

gep662project.py

"""
-------------------------------------------------------------------
LEHMAN COLLEGE 
Geographic Information Science Program
Course: GEP 662 Introduction to Programming for GISc
Semester:	Spring 2014
Instructor:	Prof. Jennifer Brisbane

Student: Ragnar Heidar Thrastarson

Final project - Python on a Bike

Description:
This script will perform the following tasks
-Ask the user for CSV table that is a Citi Bike System data table
-Make changes to timestamps in the file
-Export the changes to a new CSV file
-Ask the user to supply a name for a new File GDB
-Import the converted CSV table into the new File GDB
-Perform multiple tasks and create both tables and feature classes
-Add a feature class with Citi Bike stations to an existing MXD file
-Export a JPG map with the Citi Bike stations from the MXD file
-------------------------------------------------------------------
"""

from tempfile import NamedTemporaryFile # Import tempfile module to create temporary files
import shutil # Import shutil module for high level operations (copy/paste)
import csv # Import the CSV module
import arcpy
from arcpy import env
import os

current_path = os.getcwd() + "\\" # check the current path and add a single backslash
writefile = current_path + "trips_imported.csv" # name for the exported CSV table

def askforinput(): # prompt user and get name of input table
    print "Please specify input filename including CSV extension"
    print "(assuming file is located in " + current_path + ")"
    print ">",
    user_file = raw_input() # if nothing is entered and enter is pressed, the code will throw an error! need to fix
    converttrips(user_file) # pass the input to the converttrips function

def converttrips (user_file):
    check_file = os.path.exists(current_path + user_file) # check if the input table exists
    check_file_whitespace = len(user_file) == 0 # check for white-space
    if check_file == True and check_file_whitespace != True : # if true, run what follows
        print '-' * 10
        print "Converting timestamps..."
    
        readfile = user_file # input CSV file
        writefile = current_path + "trips_imported.csv" # output file, created if not there, overwritten if it's already there.
        tempfile = NamedTemporaryFile(delete=False) # a temporary file

        with open(readfile, 'rb') as csvfilein, tempfile: # open CSV files
            reader = csv.reader(csvfilein, delimiter=',', quotechar='"') # Create a CSV reader object
            writer = csv.writer(tempfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_ALL) # Create a CSV writer object

            # add the header row into the writefile
            writer.writerow(["tripduration","starttime","stoptime","start station id","start station name","start station latitude","start station longitude","end station id","end station name","end station latitude","end station longitude","bikeid","usertype","birth year","gender"])

            next(reader) # skip the header row for the readfile
            for row in reader: # for each row, split row into variables
                tripduration = row[0] # trip time in seconds
                starttime = row[1] # time stamp for trip start
                stoptime = row[2] # time stamp from trip end
                ssid = row[3] # id number start station
                ssname = row [4] # name of start station
                sslat = row [5] # latitude of start station
                sslon= row [6] # longitude of start station
                esid = row [7] # id number of end station
                esname = row [8] # name of end station
                eslat = row[9] # latitude of end station
                eslon = row[10] # longitude of end station
                bikeid= row[11] # id number for bike
                usertype = row[12] # subscriber or non-subscriber
                birth = row[13] # birth year of rider, only for subscribers
                gender = row[14] # gender of rider, only for subscribers
             
                # split the start timestamp into day, month and year+time
                s_month, s_day, s_yeartime = starttime.split('/')
                # split the year+time into year and time
                s_year, s_time = s_yeartime.split(' ')
                # merge start timestamp into new format
                s_timestamp = s_year + "-" + s_month + "-" + s_day + " " + s_time
            
                # split the end timestamp into day, month and year+time
                e_month, e_day, e_yeartime = stoptime.split('/')
                # split the year+time into year and time
                e_year, e_time = e_yeartime.split(' ')
                # merge start timestamp into new format
                e_timestamp = e_year + "-" + e_month + "-" + e_day + " " + e_time
            
                # write each row into a tempfile with new timestamp formats
                writer.writerow([tripduration, s_timestamp, e_timestamp, ssid, ssname, sslat, sslon, esid, esname, eslat, eslon, bikeid, usertype, birth, gender])
            
        # write contents of tempfile into a new file
        shutil.move(tempfile.name, writefile)
        print "CSV output file saved here: " + writefile
    else: #  # if false, run what fallows
        print '-' * 28
        print "| This file was not found! |" # notify the user that input file in not found
        print '-' * 28
        askforinput() # go back to the askforinput function and get a new input table

def askforfgdb():
    print "Please specify name for File Geodatabase"
    print ">",
    user_filegdb = raw_input() # ask for user input
    user_filegdb.replace(" ", "") # if there are spaces in the name, remove
    user_filegdb = user_filegdb + ".gdb" # add the extension for the file geodatabase
    gisanalysis(user_filegdb) # pass the name for the new file geodatabase to the gisanalysis function
        
def gisanalysis (user_filegdb):
    check_for_gdb = os.path.isdir(user_filegdb) # check if the file geodatabase exists
    if check_for_gdb == False: # if file geodatabase does NOT exist, run the following
        env.overwriteOutput = True #allow overwrites
        
        arcpy.CreateFileGDB_management(current_path, user_filegdb, "CURRENT") # create file geodatabase

        print "File Geodatabase created"
        env.workspace = current_path + user_filegdb #define workspace

        print "Importing table..."
        # importin CSV table into file geodatabase
        arcpy.TableToGeodatabase_conversion(writefile, env.workspace)

        # count the number of records imported
        count_resault = arcpy.GetCount_management("trips_imported")
        count_trips = int(count_resault.getOutput(0))
        print "Total trips imported: " + str(count_trips)
        print "Summarizing..."
        # create summary tables for the fields gender and stationsIDs
        arcpy.Statistics_analysis("trips_imported", "summary_gender", "gender COUNT", "gender")
        arcpy.Statistics_analysis("trips_imported", "summary_stations", "start_station_latitude FIRST;start_station_longitude FIRST;start_station_id COUNT;end_station_id COUNT", "start_station_id")
        print "Summary tables created"
        print "Creating feature class..."
        # create an event layer from station summary table
        arcpy.MakeXYEventLayer_management("summary_stations", "FIRST_start_station_longitude", "FIRST_start_station_latitude", "summary_stations_Layer", "GEOGCS['GCS_WGS_1984',DATUM['D_WGS_1984',SPHEROID['WGS_1984',6378137.0,298.257223563]],PRIMEM['Greenwich',0.0],UNIT['Degree',0.0174532925199433]];-400 -400 1000000000;-100000 10000;-100000 10000;8.98315284119522E-09;0.001;0.001;IsHighPrecision", "")
        print "XY Event layer created"
        # Copy event layer and create a new feature class - Citi Bike Stations
        arcpy.CopyFeatures_management("summary_stations_Layer", "bike_stations_wgs84", "", "0", "0", "0")

        print "XY Event Layer saved to new feature class"
        # project the Citi Bike Stations feature class from WGS84 to State Plane-NY-LI-feet
        arcpy.Project_management("bike_stations_wgs84", "bike_stations", "PROJCS['NAD_1983_StatePlane_New_York_Long_Island_FIPS_3104_Feet',GEOGCS['GCS_North_American_1983',DATUM['D_North_American_1983',SPHEROID['GRS_1980',6378137.0,298.257222101]],PRIMEM['Greenwich',0.0],UNIT['Degree',0.0174532925199433]],PROJECTION['Lambert_Conformal_Conic'],PARAMETER['False_Easting',984250.0],PARAMETER['False_Northing',0.0],PARAMETER['Central_Meridian',-74.0],PARAMETER['Standard_Parallel_1',40.66666666666666],PARAMETER['Standard_Parallel_2',41.03333333333333],PARAMETER['Latitude_Of_Origin',40.16666666666666],UNIT['Foot_US',0.3048006096012192]]", "WGS_1984_(ITRF00)_To_NAD_1983", "GEOGCS['GCS_WGS_1984',DATUM['D_WGS_1984',SPHEROID['WGS_1984',6378137.0,298.257223563]],PRIMEM['Greenwich',0.0],UNIT['Degree',0.0174532925199433]]")

        print "Bike stations projected from WGS-84 to New York Long Island State Plane"
        print "Adding fields..."
        # add a field for the total number of trips
        arcpy.AddField_management("bike_stations", "total_trips", "LONG", "", "", "", "", "NULLABLE", "NON_REQUIRED", "")
        # populate the total number of trips field
        arcpy.CalculateField_management("bike_stations", "total_trips", "[COUNT_start_station_id] + [COUNT_end_station_id]", "VB", "")

        # add Citi Bike Stations feature class to an MXD file
        bike_stations_lyr = arcpy.MakeFeatureLayer_management("bike_stations", "bike_stations_lyr")
        mxd = arcpy.mapping.MapDocument(current_path + "Citibikemap.mxd") # this MXD needs to exists and be in the same folder as script
        df = arcpy.mapping.ListDataFrames(mxd, "Layers")[0] # add to data frame that is called layers
        addLayer = arcpy.mapping.Layer("bike_stations_lyr") # add the bike station layer to the data frame
        arcpy.mapping.AddLayer(df, addLayer, "AUTO_ARRANGE") # zoom to extent

        arcpy.mapping.ExportToJPEG(mxd, current_path + "NewMap.jpg") # export JPG map
        print "Map exported"

        del mxd, addLayer # clean up and delete these two variables
        print "Done!"
    else:  # if file geodatabase does exist, run the following
        print '-' * 39
        print "|This File Geodatabase already exists!|"  # notify the user that the file geodatabase does exist
        print '-' * 39
        askforfgdb() # call the askforfgdb again to get a new name for a file geodatabase
    
askforinput() # run function askforinput

askforfgdb() # run function askforfgdb