waveform80

# Working (overlays image on videotestsrc)
gst-launch-1.0 videomixer name=mix ! videoconvert ! xvimagesink   multifilesrc location="test.png" caps="image/png,framerate=0/1" ! pngdec ! imagefreeze ! alpha method=0 alpha=0.5 ! mix.  videotestsrc ! "video/x-raw,format=AYUV,framerate=25/1,width=320,height=240" ! mix.

# Fails (attempts to overlay image on webcam source)
gst-launch-1.0 videomixer name=mix ! videoconvert ! xvimagesink   multifilesrc location="test.png" caps="image/png,framerate=0/1" ! pngdec ! imagefreeze ! alpha method=0 alpha=0.5 ! mix.  v4l2src device=/dev/video0 ! "video/x-raw,format=YUY2,width=320,height=240" ! videoconvert ! mix.

# Output of failing command:
cat << EOF

Setting pipeline to PAUSED ...

waveform80

Demeter Installation

These instructions cover a non-root manual build of Bruce Ravel's Demeter package under Ubuntu 12.04. The instructions are deliberately written with a non-technical audience in mind and hence might read as somewhat patronizing for the technical reader. The main

waveform80

CASE WHEN ContractPrice < DiscountPrice
THEN
    CASE WHEN ContractPrice < ListPrice THEN ContractPrice ELSE ListPrice END
ELSE
    CASE WHEN DiscountPrice < ListPrice THEN DiscountPrice ELSE ListPrice END
END

waveform80

CREATE VIEW some_dates AS
WITH RECURSIVE dates(i, d) AS (
    VALUES (1, DATE('1970-01-01'))
    UNION ALL
    SELECT i + 1, d + 1 DAY FROM dates WHERE d < DATE('2070-12-31')
)
SELECT
    d.d        AS a_date,
    YEAR(d.d)  AS a_year,
    MONTH(d.d) AS a_month,

waveform80

WITH qdps_with_rows AS (
  SELECT
    ROW_NUMBER() OVER (PARTITION BY q.product_code) AS row_num,
    q.*
  FROM qdps q
)
SELECT
  p.product_code,
  SUM(CASE WHEN q.row_num = 1 THEN q.value END) AS qdp1,
  SUM(CASE WHEN q.row_num = 2 THEN q.value END) AS qdp2,

waveform80

#!/usr/bin/env python

import json
from itertools import izip, cycle, groupby
from operator import itemgetter, attrgetter

# Imagine this is the result of some DB query like:
# SELECT region, year, value FROM some_table ORDER BY region, year
data = [
    ('Southwark',  2009, 71.89),

waveform80

def process_iis_directory(d):
    print('Processing IIS directory %s' % d)
    for entry in os.listdir(d):
        entry = os.path.join(d, entry)
        if entry.endswith('.gz'):
            print('Processing %s into CSV' % entry)
            with gzip.open(entry, 'rb') as uncompressed, \
                    io.BufferedReader(uncompressed) as buffered, \
                    io.TextIOWrapper(buffered, encoding='latin1') as infile, \
                    io.open(os.path.splitext(entry)[0] + '.csv', 'wb') as outfile, \

waveform80

def factorial(i):                                                               
    if i > 1:                                                                   
        return factorial(i - 1) * i                                             
    elif i == 1:                                                                
        return 1                                                                
    else:                                                                       
        raise ValueError('Cannot take factorial of %d' % i)                     
                                                                                

waveform80

Capturing video both before and after motion detection

Assuming you want to record 10 seconds of video before motion is detected, and 10 seconds of video after motion is detected, you can simply create a circular buffer with enough space for 20 seconds of video, wait until motion is detected, then wait 10 seconds more, and only then write the contents of the circular buffer to disk. basic_circular.py demonstrates this.

The second example is a little more complicated and results in a couple of files per motion event: the 10 seconds before and the all seconds that motion is still occurring after first detection. Once motion is detected, we start recording subsequent frames to a file, and while that's going on, dump the circular buffer to another file. Once we stop detecting motion, we close the "after" file, and split recording back to the circular stream. This is demonstrated in advanced_circular.py.

waveform80

import io
import picamera

class MyOutput(object):
    def __init__(self):
        self.file_num = 0
        self.output = None

    def write(self, buf):
        if buf.startswith(b'\xff\xd8'):