%%file birds.txt
robin,kentucky,may 23
robin,west virginia,may 24
seagull,maine,may 24
chickadee,kentucky,june 1
seagull,maine,june 1

# Let's start by reading in each line of the file, one by one.
fp = file('birds.txt', 'rb')
for line in fp:
    print line

import csv
fp = file('birds.txt', 'rb')
reader = csv.reader(fp)
for bird, state, day in reader:
    print bird, '/', state, '/', day
    

import csv
fp = file('birds.txt', 'rb')
reader = csv.reader(fp)

n = 0
for bird, state, day in reader:
    if bird == 'chickadee':
        n += 1

print 'we saw', n, 'chickadees'

# define an empty list
birdlist = []

fp = file('birds.txt', 'rb')
reader = csv.reader(fp)

n = 0
for bird, state, day in reader:
    birdlist.append(bird)

print 'we saw', birdlist.count('chickadee'), 'chickadees'

print 'we saw', birdlist.count('robin'), 'robins'

# to do this, define functions
def read_birdlist(filename):
    birdlist = []

    fp = file('birds.txt', 'rb')
    reader = csv.reader(fp)

    n = 0
    for bird, state, day in reader:
        birdlist.append(bird)
        
    return birdlist

def print_birdcount(birdlist, bird):
    print 'we saw', birdlist.count(bird), 'of bird type', bird
    

birdlist = read_birdlist('birds.txt')
print_birdcount(birdlist, 'robin')
print_birdcount(birdlist, 'chickadee')

def read_birdlist2(filename):
    birdlist = []

    fp = file(filename, 'rb')
    reader = csv.reader(fp)

    n = 0
    for bird, state, day in reader:
        birdlist.append((bird, state, day))
        
    return birdlist

birdlist2 = read_birdlist2('birds.txt')
print birdlist2

for bird, state, day in birdlist2:
    print bird, state, day

def print_birdcount2(birdlist2, bird_to_count):
    n = 0
    for bird, state, day in birdlist2:
        if bird == bird_to_count:
            n += 1
            
    print 'we saw', n, 'of bird type', bird_to_count


print_birdcount2(birdlist2, 'robin')

def get_entries_by_state(birdlist2, query_state):
    newlist = []
    for bird, state, day in birdlist2:
        if state == query_state:
            newlist.append((bird, state, day))
    return newlist

get_entries_by_state(birdlist2, 'kentucky')

kentucky_birds = get_entries_by_state(birdlist2, 'kentucky')
print_birdcount2(kentucky_birds, 'chickadee')

bigbirdlist = read_birdlist2('long-birds.csv')
print len(bigbirdlist)
print bigbirdlist[:10]

birdlist2 = read_birdlist2('birds.txt')
for bird, state, day in birdlist2:
    print_birdcount2(birdlist2, bird)

bird_types = set()
for bird, state, day in birdlist2:
    bird_types.add(bird)
    
print bird_types

# now we can just look at the unique set of birds
for bird_type in bird_types:
    print_birdcount2(birdlist2, bird_type)

d = {}
d['robin'] = 1
d['seagull'] = 2

print d['robin']

def make_birddict(birdlist2):
    # first, get the unique set of birds
    bird_types = set()
    for bird, state, day in birdlist2:
        bird_types.add(bird)
    
    # then, copy the code from print_birdcount2:
    d = {}
    
    for bird_type in bird_types:
        n = 0
        for bird, state, day in birdlist2:
            if bird == bird_type:
                n += 1 
                
        d[bird_type] = n
        
    return d

# tada!
print make_birddict(birdlist2)

def make_birddict2(birdlist2):
    # first, get the unique set of birds
    bird_types = set()
    for bird, state, day in birdlist2:
        bird_types.add(bird)
    
    # then, set the count for each bird to 0
    d = {}
    for bird in bird_types:
        d[bird] = 0
        
    # now, go through all of the entries in birdlist2 and, for each one, increment the count in the dictionary
    for bird, state, day in birdlist2:
        d[bird] = d[bird] + 1

    return d

make_birddict2(birdlist2)

from datetime import datetime

# strptime will convert strings into dates given the appropriate formatting string (see docs above)
print datetime.strptime('may 21', '%B %d')



# we should fix that year...
print datetime.strptime('may 21' + ' 2013', '%B %d %Y')


# ok -- and now we need to convert back to a straight up number.
# If all the dates are 2013, we can just ask for day of year...
date = datetime.strptime('may 21' + ' 2013', '%B %d %Y')

day_of_year = date.strftime('%j')
print day_of_year

# hmm, does that actually work!?
for bird, state, day in birdlist2:
    date = datetime.strptime(day + ' 2013', '%B %d %Y')
    day_of_year = date.strftime('%j')
    print day, date, day_of_year 

birdcount_by_day = {}

for bird, state, day in birdlist2:
    birdcount_by_day[day] = 0
    
for bird, state, day in birdlist2:
    birdcount_by_day[day] = birdcount_by_day[day] + 1

print birdcount_by_day


plotme_x = []
plotme_y = []

for day in birdcount_by_day:  # note, iterating over dictionaries gives you keys
    date = datetime.strptime(day + ' 2013', '%B %d %Y')
    day_of_year = date.strftime('%j')
    
    # trick: we need to convert day_of_year into an integer
    day_of_year = int(day_of_year)
    
    # retrieve birdcount
    count = birdcount_by_day[day]
    
    # now add day_of_year and birdcount
    plotme_x.append(day_of_year)
    plotme_y.append(count)

print plotme_x
print plotme_y

plot(plotme_x, plotme_y, 'ro')

plot(plotme_x, plotme_y, 'ro')
axis(ymin=0, xmin=140, xmax=160)

# if we plot with lines, we see that because the points aren't sorted, we don't get what we expect.
plot(plotme_x, plotme_y, 'r-')
axis(ymin=0, xmin=140, xmax=160)

# how do we fix this?  one way is to zip the two lists together, sort them, and pull them back out...
plotme_all = zip(plotme_x, plotme_y)
plotme_all.sort()

plotme_x = []
plotme_y = []
for (x, y) in plotme_all:
    plotme_x.append(x)
    plotme_y.append(y)

print plotme_x

# tada...
plot(plotme_x, plotme_y, 'r-')
axis(ymin=0, xmin=140, xmax=160)

bigbirdcount_by_day = {}

for bird, state, day in bigbirdlist:
    bigbirdcount_by_day[day] = 0
    
for bird, state, day in bigbirdlist:
    bigbirdcount_by_day[day] = bigbirdcount_by_day[day] + 1


plotme = []

for day in bigbirdcount_by_day:  # note, iterating over dictionaries gives you keys
    date = datetime.strptime(day + ' 2013', '%B %d %Y')
    day_of_year = date.strftime('%j')
    
    # trick: we need to convert day_of_year into an integer
    day_of_year = int(day_of_year)
    
    # retrieve birdcount
    count = bigbirdcount_by_day[day]
    
    # now add day_of_year and birdcount
    plotme.append((day_of_year, count))

    
plotme.sort()
print plotme[:5]

plotme_x = []
plotme_y = []
for (x, y) in plotme:
    plotme_x.append(x)
    plotme_y.append(y)



plot(plotme_x, plotme_y, 'r-')

birdlist = read_birdlist2('long-birds.csv')
bird_dict = make_birddict2(birdlist)

labels=[]
counts=[]
nbirds = len(bird_dict)
for bird,count in bird_dict.iteritems():
    labels.append(bird)
    counts.append(count)

fig = plt.figure()
ax = fig.add_subplot(1,1,1)

xlocations = numpy.arange(nbirds)
print xlocations
width = 0.6 
bars = ax.bar(xlocations, counts, width=width)
xtickNames = ax.set_xticklabels(labels)
ax.set_xticks(xlocations-0.1)
plt.setp(xtickNames, rotation=45)
plt.show()
ax.set_title("Bird counts")

# start by picking out the birds we want to save:
birdlist2 = read_birdlist2('birds.txt')
print birdlist2

kentucky_birds = []
for bird, state, date in birdlist2:
    if state == 'kentucky':
        kentucky_birds.append((bird, date))
        
print kentucky_birds

import csv

# now, open a file for writing, and use 'csv.writer' to write rows to it.
# just for grins, use tab for a delimiter, instead of a comma.
fp = file('kentucky-birds.tsv', 'wb')
w = csv.writer(fp, delimiter='\t') # <- use tab instead of the default, which is a comma

for bird, date in kentucky_birds:
    row = (bird, date)
    w.writerow(row)
fp.close()

# check it out -- did it work?
print open('kentucky-birds.tsv').read()

%%file bird-weights.txt
robin,5
seagull,12
chickadee,4

# first, load in the bird weights
fp = file('bird-weights.txt', 'rb')
r = csv.reader(fp)

weight_dict = {}
for bird, weight in r:
    weight_dict[bird] = weight
    


# now, go through all of the birdlist2 entries
birdlist2 = read_birdlist2('birds.txt')

new_birdlist = []
for bird, state, day in birdlist2:
    weight = weight_dict[bird]
    new_birdlist.append((bird,state,day,weight))
    
print new_birdlist

# what if one of the birds doesn't have a weight in the file, though?
del weight_dict['robin']

new_birdlist = []
for bird, state, day in birdlist2:
    weight = weight_dict[bird]
    new_birdlist.append((bird,state,day,weight))
    
print new_birdlist

# change the code to be aware of potentially missing data:
new_birdlist = []
for bird, state, day in birdlist2:
    weight = weight_dict.get(bird, -1)  # use '-1' or something patently absurd if no weight in table
    new_birdlist.append((bird,state,day,weight))
    
print new_birdlist

# and now you can output to a file, or whatever.