working with rasterio and Apache Spark

gistfile1.py

>>> '''launch python repl with ./bin/pyspark'''
'launch python repl with ./bin/pyspark'
>>> import rasterio as rio
>>>
>>> def img_size(path):
...     with rio.drivers():
...         with rio.open(path) as img:
...             return [img.width, img.height]
... 
>>> 
>>> f = sc.textFile("tifs.txt")
>>> wc = f.map(img_size)
>>> wc.collect()
[[4800, 4800], [4800, 4800], [4800, 4800]]

tifs.txt

/tmp/Hansen_GFC2013_lossyear_50N_080W.tif
/tmp/Hansen_GFC2013_lossyear_60N_040E.tif
/tmp/Hansen_GFC2013_lossyear_60N_050E.tif