I've been thinking for a while now about how nice it would be to automatically create a Markdown post of the Instagram photos I take. I looked around at Instagram's API's and, while I could probably set up a way to pull directly from the API, I wanted to throw something together quickly. I'm on my way to Maui for vacation, so I spent an evening making this handy little Python script that'll enable me to easily post my Instagram photos to jayhickey.com.
import os
import sys
import re
from glob import glob
import urllib
from time import localtime, strftime
def read_file(file):
# Check for filename for Instagram file from IFTTT
f = open("%s" % (files), mode="r")
fileLines = f.readlines()
fileDict = {}
# Create a dictionary with the Instagram info
for line in fileLines:
x = re.search(r'([\w\@\.]+)\s*:\s*(.*)', line)
if x != None:
fileDict[x.group(1)] = x.group(2)
f.close()
return fileDict
def create_draft(fileDict, draftLoc, imgLink):
# Replace non alphanumerics with dashes for filename
c = re.sub(r'[\t !"#$:;%&\'()*\-/<=>?@\[\\\]^_`{|},.]+',"-",fileDict['Caption'])
while c.endswith('-'):
c = c[:-1]
c = c.lower()
# Embed the photo with Markdown
draft = open(draftLoc + "/%s.md" % (c), mode="w")
draft.write(fileDict['Caption'] + '\n')
draft.write("=====================\n")
draft.write("Link: %s" % (fileDict['URL']) + "\n")
draft.write("publish-not-yet\n\n")
print imgLink
draft.write("s)\n\n" % {"1" : fileDict['Caption'], "2" : imgLink})
draft.write("(Via [Instagram](http://instagram.com))")
draft.close()
if __name__ == '__main__':
# These might not be used, so make them empty
Local_Image_URL_Path = ''
Website = ''
# Read input arguments
IFTTT_Read_Path = sys.argv[1]
Draft_Write_Path = sys.argv[2]
# These parameters are optional
if len(sys.argv) >= 4:
Local_Image_URL_Path = sys.argv[3]
if len(sys.argv) == 5:
Website = sys.argv[4]
# Make sure the file is a text file from Instagram
fileList = glob(IFTTT_Read_Path + '*instagr.am*.txt')
for files in fileList:
# Read the Instagram data
fileDict = read_file(files)
if Local_Image_URL_Path != '' or Website != '':
# Make a local copy of the image and date it
image = urllib.URLopener()
eventTime = strftime("%Y-%m-%d_%H%M%S", localtime())
fileName, fileExtension = os.path.splitext(fileDict['Source'])
imgLinkPath = IFTTT_Read_Path + eventTime + fileExtension
image.retrieve(fileDict['Source'], imgLinkPath)
imgURL = Website + Local_Image_URL_Path + eventTime + fileExtension
else:
# Use the image hosted by Instagram
imgURL = fileDict['Source']
# Create a Markdown draft
create_draft(fileDict, Draft_Write_Path, imgURL)
# Delete the Instagram text file from IFTTT
os.remove(files)
Installation
Here's the input to run it (the last two parameters are optional):
python MarkdownifyInstagram.py {{IFTTT_Read_Path}}, {{Draft_Write_Path}}, {{Local_Image_URL_Path}}, {{Website}}
Example:
python MarkdownifyInstagram.py /home/blog/secondcrack/www/media/instagram/ /home/blog/Dropbox/Blog/drafts/ /media/instagram/ https://jayhickey.com
The downside to using IFTTT, according to their hilariously titled /wtf page, is that it only polls for new data every 15 minutes. So this won't happen instantly.
GitHub is where you can find all the instructions for setting up the IFTTT recipe, running Markdownify Instagram, and even how to use a shell script and iNotify to automate the process. I won't go into any of that here. However, I do think it's interesting to dive a little deeper and see how the script works.
Implementation
Here's a look at a few of the design decisions I made while writing Markdownify Instagram.
Reading the text file from IFTTT
When triggered, the IFTTT recipe will create a plain text file with a name like http-instagr.ampoefb-ihvv0.txt and these contents:
URL: http://instagram.com/p/JF1n/
Source: http://distillery.s3.amazonaws.com/media/2010/11/03/217c074328864f76b5d730837403f371_7.jpg
Caption: Deadly.
Python's built in glob module is first used to create a list of all the Instagram text files in your {{IFTTT_Read_Path}}:
fileList = glob.glob(IFTTT_Read_Path + '*instagr.am*.txt')
The * will match zero or more characters, so a file that starts with anything, contains instagr.am, ends with .txt, and has anything in between will be appended to fileList. This is the first time I've actually used the glob module. Although less powerful than regular expressions, it's perfect for finding specific files in directories.
Creating a dictionary from the data
After getting a list of files, read_file uses readlines() to split the text into individual lines. The regex pattern r'([\w\@\.]+)\s*:\s*(.*)' creates matchobject's out of those lines, then lastly the dictionary fileDict is created from the matches:
for line in fileLines:
x = re.search(r'([\w\@\.]+)\s*:\s*(.*)', line)
if x != None:
fileDict[x.group(1)] = x.group(2)
Regex is hard to look at, but it's not too bad once you grasp the syntax. Google's regular expressions tutorial is a good place to start, and you can learn a whole lot more just by searching the web. Each set of parenthesis corresponds to a group match that will be stored in an additional element of x. Because there are two sets of parenthesis, x will have a group size of 3. x.group(0) will contain all of line, x.group(1) will contain the text before the : (the first set of parenthesis in the regex), and x.group(2) will have everything after (the second set). Like this:
x.group(0) = 'URL: http://instagram.com/p/JF1n/'x.group(1) = 'URL'x.group(2) = 'http://instagram.com/p/JF1n/'
After the for loop completes, fileDict will look something like this
fileDict = {'URL':'http://instagram.com/p/JF1n/', 'Source':'http://distillery.s3.amazonaws.com/media/2010/11/03/217c074328864f76b5d730837403f371_7.jpg', 'Caption':'Deadly.'}
So printing the URL is as simple as typing print fileDict['URL'].
Saving the local image file
If the last two parameters are entered, your photo will be saved locally—so you aren't relying on Instagram's S3 hosting. If Facebook ever decides to shut down Instagram, you won't have a broken embedded image. To accomplish this, I used the Python urllib module to save an image from a URL by setting image = urllib.URLopener(), then
image.retrieve(fileDict['Source'], imgLinkPath)
where fileDict['Source'] is the link to the Instagram image and imgLinkPath is where the image saved will be saved, e.g., /home/blog/secondcrack/www/media/instagram/2012-11-30_062759.jpg
The local image is named using strftime("%Y-%m-%d_%H%M%S", localtime()). This creates a string formatted something like 2012-11-30_062759, where:
- %Y is the year with century as a decimal (use %y for without century, resulting in
12 instead of 2012)
- %m is the month from
[01-12]
- %d is the day of the month from
[01-31]
- %H is the hour in 24-hour time (use
%I for 12-hour time)
- %M is the minute from
[00-59]
- %S is the second from
[00-61]
You can see the see the other strftime directives here
I've been using this convention ever since Gabe over at MacDrifter recommended it. It's very easy to both sort and quickly look at. I love it so much I even set this as a TextExpander shortcut for naming files and photos on my Mac:
Creating the Markdown draft
The final function, create_draft, is pretty straightforward and will create a Markdown file that looks like this:
Deadly.
=====================
type: Link
publish-not-yet
(Via [Instagram](http://instagram.com))
Conclusion
Overall, I'm pretty happy with the way Markdownify Instagram turned out. It produces a nice, simple post that's easy to modify. There's also tons of possiblities for new features and expansions. I find it extremely useful, so get prepared for an influx of pictures posted here. Aloha!
Update: I've made a small addition to the script's create_draft function. Non-alphanumerics are now replaced with dashes in the filename. So any spaces or special characters will be turned into dashes for the .md slug.
I used a little more regex magic to make the substitution. Here's what it looks like:
# Repace non-alphanumerics with dashes for filename
c = re.sub(r'[\t !"#$:;%&\'()*\-/<=>?@\[\\\]^_`{|},.]+',"-",fileDict['Caption'])
while c.endswith('-'):
c = c[:-1]
c = c.lower()
Make sure to grab the latest commit off GitHub so you have this feature.
I remember the commotion around ultra efficient Speedo LZR Racer during the Beijing Olympics, but I didn't realize they were banned in '09:
At the 2009 World Championships in Rome, swimmers set 43 world records in the last major meet for the super suits, which were banned at the end of that year. Since then, only five swimming world records have been broken — the three thus far in London and two at the 2011 World Championships in Shanghai.
How they work
The suits are a mix of woven elastane-nylon and polyurethane panels that were wind tunnel tested by NASA—making it smoother than shaven skin. And instead of sewed together, the the seams are ultrasonically welded (yes, using sound).
These factors contribute to the hydrodynamics by reducing drag, and the LZR Racer then compresses the body and traps air for added buoyancy. On the downside, they can only be worn once or twice. Because of the thinness of the materials, a rip can happen at any moment. But when you have 19 Olympic Medals, that's just a business expense.
A small tidbit for you fashionistas—the cosmetics of the LZR were designed by Comme De Garçons.
Banning
Swimmers started to wear multiple LZR's overtop one another for an added buoyancy effect. Therese Alshammar of Sweden had her world record in the 50m butterfly taken from her because she used two suits, claiming the second was backup for a possible "wardrobe malfunction" if the top suit tore.
But FINA stepped in, and started limiting the suits' buoyancy and thickness. Buoyancy must now be below 0.5 Newton and have a thickness no greater than 0.8mm. Also no more than 50% of the suit can be covered in non-permeable materials (such as the NASA polyurethane panels).
I understand the reasoning for the ban, but it's a shame that such a fascinating use of technology will be stunted by Olympic rules.
The technology is accurate to between 10 centimeters and two meters, uses your phone's compass, and no upfront infrastructure is needed:
Indoor Atlas's technology works by analyzing the magnetic field inside a building. The structure of a building causes disturbances to the Earth's magnetic field. Once these disturbances are mapped, people can be pinpointed within them through their phone's magnetometer. Indoor Atlas's product arose from research findings that showed the signature magnetic field within buildings was sufficiently varied and stable to be used for navigation, says company founder and computer science professor Janne Haverinen.
If this works as advertised, the possibilities are endless. An app could route you around a store based on your shopping list, automatically checking off the items as you go. Or let you find exactly where your friend is in a busy shopping mall. And businesses don't have to drop the dough to fill the place with WiFi or NFC antennas. I often use the excellent aisle411 app for in-store navigation at my local Kroger. But the lack of a "blue dot" showing my present location within in the store results in me staring too long at my phone, trying not to block the isle.
This would dramatically improve geofencing too. I constantly take advantage of the nearby contexts features of Omnifocus for iPhone, such as setting reminders for leaving or arriving at my work and home. It's also great for setting specific searches as locations, such as "Grocery Store" or "Walgreens" (I look forward to giving the just released, "iOS Reminders on streoids" app, Checkmark, a spin too). There's a constant vexation with this geofencing though. To save battery, it doesn't use full GPS. So it isn't accurate enough, especially indoors. I find myself only setting "arriving" reminders, because the "leaving" reminder to grab the package I need to take to FedEx doesn't end up going off until I'm already in my car and heading down the road. But add integration with this Indoor Atlas technology, and the alert could go off the second I step out the front door. We're at the tip of the iceberg when it comes to the potential of geofencing, and as a impressively forgetful person, I can't wait to see this take off.
