// -*- html -*-
// $Id$
require_once("docutil.php");
page_head("Python Framework");
?>
See the section on Python in the Software
Prerequisites.
Structure
The directory boinc/py/Boinc
contains the Boinc
module. This means if you have boinc/py/
in your python path you can
write for example:
from Boinc.setup_project import *
To ensure boinc/py/
is in your python path:
import boinc_path_config
This is a special module that configure
places in relevant
directories which then modifies sys.path
appropriately.
Project-specific settings
The module boinc_project_path
is imported to get the paths
for config.xml
and run_state.xml
. The default
paths for these are the parent directory of the invocation script. You can
override these defaults
- modify this file directly (if you have only one project on your server
or have separate copies for each)
- create a new boinc_project_path.py and place it earlier in PYTHONPATH
than the default one
- define environment variables
Example boinc_project_path.py
config_xml_filename = '/etc/boinc/yetiathome/config.xml'
run_state_xml_filename = '/var/lib/boinc/yetiathome/run_state.xml'
See the source of file boinc/py/Boinc/boinc_project_path.py
for
details.
Directories containing python scripts
boinc/py/ | Boinc/*.py |
Main BOINC python modules
|
boinc/sched/ | start |
BOINC start / Super Cron program
|
boinc/tools/ | add |
Adds objects to the database.
TODO: document
|
boinc/tools/ | make_project |
Creates a project
|
boinc/tools/ | update_versions |
Adds all relevant core client and application executables to download
directory and database
|
boinc/test/ | test*.py cgiserver.py |
Test scripts: see the testing framework.
|
Python modules in boinc/py/Boinc/
boinc_path_config.py.in |
Configure puts boinc_path_config.py in all
directories that need it; see above
|
boinc_project_path.py |
sets where config.xml et al can be found; see above.
|
configxml.py |
reads and writes config.xml and run_state.xml
- see its pydoc for more information
|
boinc_db.py |
auto-generated file that contains database constant definitions,
e.g. RESULT_OUTCOME_SUCCESS = 1
|
setup_project.py |
internal module for creating a project. See make_project and test scripts.
|
database.py |
defines database backend functions and database operations; see below.
|
db_mid.py |
"middle-end": optional mix-in to ease debugging by allowing printing of
database objects directly
|
util.py |
miscellaneous functions
|
version.py.in |
version and platform-specific definitions snarfed by configure
|
Python database access
Database.py
defines database backend library and database table
and object relationships to allow easy data manipulation.
All database tables have a corresponding class and
its rows have classes, where each column is a member of that class. Ids are
automatically translated to and from objects. To begin, import
the database
module:
from Boinc import database
Connect to the database:
database.connect_default_config()
Table classes can be indexed using the [ ] operator to retrieve an object by
id; e.g.
# executes 'select * from project where id=1'.
# exception is raised if project is not found
project_with_id_1 = database.Projects[1]
Table classes have a find
function that builds and executes a
MySQL query based on its arguments:
# this could return any number (0, 1, 2, ...) of platforms
# executes 'select * from platform where user_friendly_name="commodore 64"'
list_of_platforms_called_c64 = database.Platforms.find(
user_friendly_name = 'Commodore 64')
Find can take any number of arguments; they are ANDed. For more advanced
usage such as custom SQL queries (anything is possible :) see the pydoc.
all_apps = database.Apps.find()
finished_yeti_wus = database.Workunits.find(
app = database.Apps.find(name='YETI@home')[0],
assimilate_state = ASSIMILATE_DONE)
Objects (table rows) have their column data as members so you can access and
modify them directly.
user_quarl = database.users.find(email_addr='quarl@quarl.org')[0]
print "name =", user_quarl.name
user_quarl.postal_code = 97404
To create a new database object, create a Python object and give all values
as parameters to the initializer:
new_app = database.App(name='SPAGHETTI@home',
min_version=1,
create_time=time.time())
To commit any changes (including a new object), call commit()
(the tool boinc/tools/add.py
is a command-line interface to
this):
user_quarl.commit() # executes an UPDATE
new_app.commit() # executes an INSERT
To remove an object, call remove()
:
team_eric_test = database.Teams(name="Eric's Test Team")[0]
team_eric_test.remove()
# OR
for team in database.Teams(name="Eric's Test Team"):
team.remove()
# OR
map(database.Team.remove,database.Teams(name="Eric's Test Team"))
To access objects related by id, access the field name without "id" suffix:
(the result
table has columns 'workunitid
' and
'hostid
'; the host
table has
column userid
)
wu_1234 = database.Workunits.find(name='1234.wu')[0]
results_of_wu_1234 = database.Results.find(workunit=wu_1234)
for result in results_of_wu_1234:
os.system("echo 'you are crunching %s' | mail '%s'" %(
result.name, result.host.user.email_addr))
Table | Python table object | Python row object
class |
project | Projects | Project |
platform | Platforms | Platform |
core_version | CoreVersions | CoreVersion |
app | Apps | App |
app_version | AppVersions | AppVersion |
user | Users | User |
team | Teams | Team |
host | Hosts | Host |
workunit | Workunits | Workunit |
result | Results | Result |
workseq | Workseqs | Workseq |
page_tail();
?>