Modify spark_print to be min based, not zero based

The existing implementation was difficult to use when the range relative
to the integers was small. This change moves the minimum of the spark
line to allow greater resolution in the steps.

spark.py

@@ -12,17 +12,28 @@
 ticks = u'▁▂▃▅▆▇'
 
 
-def spark_string(ints):
-    """Returns a spark string from given iterable of ints."""
-    step = ((max(ints)) / float(len(ticks) - 1)) or 1
-    return u''.join(ticks[int(round(i / step))] for i in ints)
-
-
-def spark_print(ints, stream=None):
+def spark_string(ints, fit_min=False):
+    """Returns a spark string from given iterable of ints.
+    
+    Keyword Arguments:
+    fit_min: Matches the range of the sparkline to the input integers
+             rather than the default of zero. Useful for large numbers with
+             relatively small differences between the positions
+    """
+    min_range = 0
+    if fit_min:
+        min_range = min(ints)
+
+    step_range = max(ints) - min_range
+    step = ((step_range) / float(len(ticks) - 1)) or 1
+    return u''.join(ticks[int(round((i - min_range) / step))] for i in ints)
+
+
+def spark_print(ints, stream=None, fit_min=False):
     """Prints spark to given stream."""
     if stream is None:
         stream = sys.stdout
-    stream.write(spark_string(ints).encode('utf-8'))
+    stream.write(spark_string(ints, fit_min=fit_min).encode('utf-8'))
 
 if __name__ == '__main__':
     if len(sys.argv) > 1: