gistfile1.py

from TwitterAPI import TwitterAPI
api = TwitterAPI("QHCFXKcJeYvFMr8XBTes3eID8", 
                    "Zh3JhEBL8PXnfAEAAOivlB4xrX1PAfDEIdg6DsC78qs6O2Aruk", 
                    "2401578782-7P2QUySnDc8gFHQ1vCwTrckbLLSty8tOFjrizC6", 
                    "WNcilnQcbd0o7xnTYf1XBylsriyBb1KkNlxyTx4u4uB1s"
                    )

# instantiate empty lists that will hold Event data.
#  these would be populated by a periodic (e.g. every 1 minute) query of "active" Events.
event_ids = []
event_upper_longitudes = []
event_lower_longitudes = []
event_upper_latitudes = []
event_lower_latitudes = []

# helper function for testing (adds an Event to the list of target Events)
def addEvent(boxNeLat, boxNeLon, boxSwLat, boxSwLon, eventID):
    
    global event_ids
    global event_upper_longitudes
    global event_lower_longitudes
    global event_upper_latitudes
    global event_lower_latitudes
    
    event_ids.append(eventID)
    event_upper_longitudes.append(boxNeLon)
    event_lower_longitudes.append(boxSwLon)
    event_upper_latitudes.append(boxNeLat)
    event_lower_latitudes.append(boxSwLat)
    
    return


# add box around where I live
addEvent(32.819655, -96.802374, 32.817328, -96.807116, 2006)

# add box around Love Field
addEvent(32.861336, -96.830076, 32.829175, -96.869387, 737)

# add box around SMU
addEvent(32.847997, -96.776475, 32.834909, -96.789864, 342)

# add box around White Rock Lake
addEvent(32.852973, -96.706780, 32.814101, -96.742743, 1000)

# add box around Fort Worth Water Gardens
addEvent(32.749163, -97.324360, 32.746618, -97.328266, 420)

# add box around JPS Hospital in Fort Worth
addEvent(32.730951, -97.323065, 32.725734, -97.330017, 83)

# add box around NAS JRB Fort Worth
addEvent(32.788938, -97.413410, 32.746645, -97.463535, 1776)



# helper function to remove t.co link from tweets
def removeLink(tweetText):
    
    # clean up general tweet text
    tweetText = tweetText.replace('\n', ' ')
    tweetText = tweetText.replace('  ', ' ')
    
    link = False
    
    if "http://t.co/" in tweetText:
        target = "http://t.co/"
        link = True
        offset = 23
        
    elif "https://t.co/" in tweetText:
        target = "https://t.co/"
        link = True
        offset = 24
        
    if link:  
        i = tweetText.index(target)
        before = tweetText[:i]
        after = tweetText[(i + offset):]
        tweetText = before + after
    
    return tweetText


# helper function to extract video or photo URL (if media is attached to the tweet)
def extractMedia(tweet):
    tweetMediaURL = None
    
    # attempt to get video
    if 'extended_entities' in tweet:
        if 'media' in tweet['extended_entities']:
            if 'type' in tweet['extended_entities']['media'][0]:
                if tweet['extended_entities']['media'][0]['type'] == 'video':
                    video = tweet['extended_entities']['media'][0]
                    
                    # loop through variants (will get highest bitrate version of video)
                    for vid in video['video_info']['variants']:
                        if vid['content_type'] == 'video/mp4':
                            tweetMediaURL = vid['url']
    
    # if no video (more valuable), then attempt to get photo
    if tweetMediaURL is None:
        if 'entities' in tweet:
            if 'media' in tweet['entities']:
                tweetMediaURL = tweet['entities']['media'][0]['media_url_https']

    return tweetMediaURL


neLat = 0.0
neLon = 0.0
swLat = 0.0
swLon = 0.0

# a helper function to make it easier to work with lat/lon values copied from Google Maps 
def quickBox(boxNeLat, boxNeLon, boxSwLat, boxSwLon):
    global neLat
    global neLon
    global swLat
    global swLon
    
    neLat = boxNeLat
    neLon = boxNeLon
    swLat = boxSwLat
    swLon = boxSwLon
    
    return



# This function determines if a Tweet falls within a list of Events.
# Compares the coordinates of the [point-type] Tweet against pre-instantiated lists 
#   that contain the minimum and maximum latitudes & longitudes of all the target Events.
def eventsTweetWithin(tweetLat, tweetLon):
    
    # there's likely a better way to do this (bringing in the target lists)
    global event_ids
    global event_upper_longitudes
    global event_lower_longitudes
    global event_upper_latitudes
    global event_lower_latitudes
    
    # instantiate the "pointer" lists
    upperLonIndexes = []
    lowerLonIndexes = []
    upperLatIndexes = []
    lowerLatIndexes = []
    
    # Scan through the minimum & maximum lists of Event latitudes & longitudes
    #   We record each qualifying index into the respective "pointer" list
    for (index, event_upper_lon) in enumerate(event_upper_longitudes):
        if tweetLon < event_upper_lon:
            upperLonIndexes.append(index)
    
    for (index, event_lower_lon) in enumerate(event_lower_longitudes):
        if tweetLon > event_lower_lon:
            lowerLonIndexes.append(index)
            
    for (index, event_upper_lat) in enumerate(event_upper_latitudes):
        if tweetLat < event_upper_lat:
            upperLatIndexes.append(index)
            
    for (index, event_lower_lat) in enumerate(event_lower_latitudes):
        if tweetLat > event_lower_lat:
            lowerLatIndexes.append(index)
    
    # finally, we intersect the four "pointer" lists, thus producing (if any) the index (or indexes) that fit *all* four criteria 
    s = set(upperLonIndexes).intersection(lowerLonIndexes).intersection(upperLatIndexes).intersection(lowerLatIndexes)
    
    # if the resultant set > 0, then the Tweet must "fall within" at least one Event
    # - all that's left to do is, using the index (or indexes) retrieve the Event ID(s)
    if len(s) > 0:    
        events = []
        for index in list(s):
            events.append(event_ids[index])
        
        return events
    else:
        return None

# this is where we easily paste in our NE and SW coordinates from Google Maps
quickBox(32.953848, -96.611259, 32.647951, -96.998527)

# notice the odd ordering of lat/lon values Twitter uses -- lower left *and then* upper right using lon, lat (versus lat, lon)
stream = api.request('statuses/filter', {'locations': str(swLon) +','+ str(swLat) +','+ str(neLon) +','+ str(neLat) })
print "monitoring the box bounded by "+ str(neLat) +", "+ str(neLon) +" and "+ str(swLat) +", "+ str(swLon) +" ..."
for tweet in stream:
    
    # verify the tweet has geo information (and a usable point-type coordinate pair)
    if 'geo' in tweet:
        if tweet['geo'] is not None:
            if tweet['geo']['type'] == 'Point':
                
                # get the tweet's coordinates
                tweetLat = tweet['geo']['coordinates'][0]
                tweetLon = tweet['geo']['coordinates'][1]
                
                # sanity check to ensure the tweet is actually within our market bounding box
                if tweetLat < neLat and tweetLat > swLat and tweetLon < neLon and tweetLon > swLon:
                    # analyze the tweet to see if it falls within any of our mock events
                    tweet_events = eventsTweetWithin(tweetLat, tweetLon)
                    
                    if tweet_events is not None:
                        tweetText = removeLink(tweet['text'])
                        tweetUser = tweet['user']['screen_name']
                        tweetMediaURL = extractMedia(tweet)
                        
                        tweetProfilePicULR = tweet['user']['profile_image_url']
                        # tweetUTC = parser.parse(tweet['created_at'])
                        
                        print tweetUser +" is on-scene Event "+ str(tweet_events[0]) +" at "+ str(tweetLat) +", "+ str(tweetLon) + " "+ tweetText
                        
                        if tweetMediaURL is not None:
                            print ' * '+ tweetMediaURL