def get_selection(available, selection, base='/scif/apps'):
'''we compare the basename (the exp_id) of the selection and available,
regardless of parent directories'''
if isinstance(selection, str):
selection = selection.split(',')
available = [os.path.basename(x) for x in available]
selection = [os.path.basename(x) for x in selection]
finalset = [x for x in selection if x in available]
if len(finalset) == 0:
bot.warning("No user experiments selected, providing all %s" %(len(available)))
finalset = available
return ["%s/%s" %(base,x) for x in finalset]
def get_selection(available, selection, base="/scif/apps"):
"""we compare the basename (the exp_id) of the selection and available,
regardless of parent directories"""
if isinstance(selection, str):
selection = selection.split(",")
available = [os.path.basename(x) for x in available]
selection = [os.path.basename(x) for x in selection]
finalset = [x for x in selection if x in available]
if len(finalset) == 0:
bot.warning("No user experiments selected, providing all %s" %
(len(available)))
finalset = available
return ["%s/%s" % (base, x) for x in finalset]
def main(args, parser, subparser=None):
"""this is the main entrypoint for a container based web server, with
most of the variables coming from the environment. See the Dockerfile
template for how this function is executed.
"""
# First priority to args.base
base = args.base
if base is None:
base = os.environ.get("EXPFACTORY_BASE")
# Does the base folder exist?
if base is None:
bot.error("You must set a base of experiments with --base" % base)
sys.exit(1)
if not os.path.exists(base):
bot.error("Base folder %s does not exist." % base)
sys.exit(1)
# Export environment variables for the client
experiments = args.experiments
if experiments is None:
experiments = " ".join(glob("%s/*" % base))
os.environ["EXPFACTORY_EXPERIMENTS"] = experiments
# If defined and file exists, set runtime variables
if args.vars is not None:
if os.path.exists(args.vars):
os.environ["EXPFACTORY_RUNTIME_VARS"] = args.vars
os.environ["EXPFACTORY_RUNTIME_DELIM"] = args.delim
else:
bot.warning("Variables file %s not found." % args.vars)
subid = os.environ.get("EXPFACTORY_STUDY_ID")
if args.subid is not None:
subid = args.subid
os.environ["EXPFACTORY_SUBID"] = subid
os.environ["EXPFACTORY_RANDOM"] = str(args.disable_randomize)
os.environ["EXPFACTORY_BASE"] = base
from expfactory.server import start
start(port=5000)
def main(args,parser,subparser=None):
'''this is the main entrypoint for a container based web server, with
most of the variables coming from the environment. See the Dockerfile
template for how this function is executed.
'''
# First priority to args.base
base = args.base
if base is None:
base = os.environ.get('EXPFACTORY_BASE')
# Does the base folder exist?
if base is None:
bot.error("You must set a base of experiments with --base" % base)
sys.exit(1)
if not os.path.exists(base):
bot.error("Base folder %s does not exist." % base)
sys.exit(1)
# Export environment variables for the client
experiments = args.experiments
if experiments is None:
experiments = " ".join(glob("%s/*" % base))
os.environ['EXPFACTORY_EXPERIMENTS'] = experiments
# If defined and file exists, set runtime variables
if args.vars is not None:
if os.path.exists(args.vars):
os.environ['EXPFACTORY_RUNTIME_VARS'] = args.vars
os.environ['EXPFACTORY_RUNTIME_DELIM'] = args.delim
else:
bot.warning('Variables file %s not found.' %args.vars)
subid = os.environ.get('EXPFACTORY_STUDY_ID')
if args.subid is not None:
subid = args.subid
os.environ['EXPFACTORY_SUBID'] = subid
os.environ['EXPFACTORY_RANDOM'] = str(args.disable_randomize)
os.environ['EXPFACTORY_BASE'] = base
from expfactory.server import start
start(port=5000)
def validate(self, folder):
'''validate is the first entrypoint function for running an experiment
or survey robot. It ensures that the content is valid,
and then calls _validate (should be defined in subclass)'''
validator = ExperimentValidator()
valid = validator.validate(folder)
if valid is True:
# IF missing favicon, add
self._check_favicon(folder)
valid = self._validate(folder)
bot.log("[done] stopping web server...")
self.httpd.server_close()
else:
bot.warning('%s is not valid, skipping robot testing.' %folder)
def initdb(self):
'''initdb will check for writability of the data folder, meaning
that it is bound to the local machine. If the folder isn't bound,
expfactory runs in demo mode (not saving data)
'''
self.database = EXPFACTORY_DATABASE
bot.info("DATABASE: %s" %self.database)
# Supported database options
valid = ('sqlite', 'postgres', 'mysql', 'filesystem')
if not self.database.startswith(valid):
bot.warning('%s is not yet a supported type, saving to filesystem.' % self.database)
self.database = 'filesystem'
# Add functions specific to database type
self.init_db() # uses url in self.database
bot.log("Data base: %s" % self.database)
def initdb(self):
'''initdb will check for writability of the data folder, meaning
that it is bound to the local machine. If the folder isn't bound,
expfactory runs in demo mode (not saving data)
'''
self.database = EXPFACTORY_DATABASE
bot.info("DATABASE: %s" %self.database)
# Supported database options
valid = ('sqlite', 'postgres', 'mysql', 'filesystem')
if not self.database.startswith(valid):
bot.warning('%s is not yet a supported type, saving to filesystem.' % self.database)
self.database = 'filesystem'
# Add functions specific to database type
self.init_db() # uses url in self.database
bot.log("Data base: %s" % self.database)
def _validate_preview(self, url):
bot.test("Experiment url: %s" % url)
org, repo = url.split("/")[-2:]
if repo.endswith(".git"):
repo = repo.replace(".git", "")
github_pages = "https://%s.github.io/%s" % (org, repo)
bot.test("Github Pages url: %s" % github_pages)
response = requests.get(github_pages)
if response.status_code == 404:
bot.error("""Preview not found at %s. You must publish a static
preview from the master branch of your repository to
add to the library.""" % github_pages)
return False
index = response.text
tmpdir = tempfile.mkdtemp()
repo_master = clone(url, tmpdir)
contenders = glob("%s/*" % repo_master)
license = False
found = False
for test in contenders:
if os.path.isdir(test):
continue
print("...%s" % test)
if "LICENSE" in os.path.basename(test):
license = True
if os.path.basename(test) == "index.html":
bot.test("Found index file in repository.")
found = True
break
if license is False:
bot.warning(
"LICENSE file not found. This will be required for future experiments!"
)
self._print_valid(found)
return found
def _validate_preview(self, url):
bot.test('Experiment url: %s' %url)
org,repo = url.split('/')[-2:]
if repo.endswith('.git'):
repo = repo.replace('.git','')
github_pages = "https://%s.github.io/%s" %(org,repo)
bot.test('Github Pages url: %s' %github_pages)
response = requests.get(github_pages)
if response.status_code == 404:
bot.error('''Preview not found at %s. You must publish a static
preview from the master branch of your repository to
add to the library.''' % github_pages)
return False
index = response.text
tmpdir = tempfile.mkdtemp()
repo_master = clone(url, tmpdir)
contenders = glob('%s/*' %repo_master)
license = False
found = False
for test in contenders:
if os.path.isdir(test):
continue
print('...%s' %test)
if "LICENSE" in os.path.basename(test):
license = True
if os.path.basename(test) == "index.html":
bot.test('Found index file in repository.')
found = True
break
if license is False:
bot.warning("LICENSE file not found. This will be required for future experiments!")
self._print_valid(found)
return found
def save_data(self, session, exp_id, content):
'''save data will obtain the current subid from the session, and save it
depending on the database type. Currently we just support flat files'''
subid = session.get('subid')
# We only attempt save if there is a subject id, set at start
data_file = None
if subid is not None:
if os.path.exists(self.data_base): # /scif/data
data_base = "%s/%s" % (self.data_base, subid)
# expfactory/00001
if not os.path.exists(data_base):
os.mkdir(data_base)
data_file = "%s/%s/%s-results.json" % (self.data_base, subid,
exp_id)
if os.path.exists(data_file):
bot.warning('%s exists, and is being overwritten.' % data_file)
write_json(content, data_file)
return data_file
def _validate_row(row, sep=',', required_length=None):
'''validate_row will ensure that a row has the proper length, and is
not empty and cleaned of extra spaces.
Parameters
==========
row: a single row, not yet parsed.
Returns a valid row, or None if not valid
'''
if not isinstance(row, list):
row = _parse_row(row, sep)
if required_length:
length = len(row)
if length != required_length:
bot.warning('Row should have length %s (not %s)' %(required_length,
length))
bot.warning(row)
row = None
return row
def _validate_row(row, sep=",", required_length=None):
"""validate_row will ensure that a row has the proper length, and is
not empty and cleaned of extra spaces.
Parameters
==========
row: a single row, not yet parsed.
Returns a valid row, or None if not valid
"""
if not isinstance(row, list):
row = _parse_row(row, sep)
if required_length:
length = len(row)
if length != required_length:
bot.warning("Row should have length %s (not %s)" %
(required_length, length))
bot.warning(row)
row = None
return row
def main(args, parser, subparser):
# List of experiments is required
template = get_template('build/docker/Dockerfile.template')
# For now, only one database provided
database = args.database
studyid = args.studyid
experiments = args.experiments
template = sub_template(template, "{{studyid}}", studyid)
template = sub_template(template, "{{database}}", database)
library = get_library(key='name')
apps = "\n"
# Add local experiments to library, first preference
local_installs = 0
for experiment in experiments:
if os.path.exists(experiment):
bot.info('local experiment %s found, validating...' % experiment)
# Is the experiment valid?
cli = ExperimentValidator()
valid = cli.validate(experiment, cleanup=False)
if valid is True:
local_installs += 1
config = load_experiment(experiment)
exp_id = config['exp_id']
# If we aren't building in the experiment directory, we need to copy there
output_dir = "%s/%s" % (os.path.abspath(
os.path.dirname(args.output)), exp_id)
experiment_dir = os.path.abspath(experiment)
if output_dir != experiment_dir:
copy_directory(experiment_dir, output_dir)
config['local'] = os.path.abspath(experiment)
library[exp_id] = config
# Warn the user that local installs are not reproducible (from recipe)
if local_installs > 0:
bot.warning(
"%s local installs detected: build is not reproducible without experiment folders"
% local_installs)
# Build Image with Experiments
for experiment in experiments:
exp_id = os.path.basename(experiment)
if exp_id in library:
config = library[exp_id]
app = "LABEL EXPERIMENT_%s /scif/apps/%s\n" % (exp_id, exp_id)
# Here add custom build routine, should be list of lines
if "install" in config:
commands = "\n".join(
["RUN %s " % s for x in config['install']]).strip('\n')
app = "%s%s\n" % (app, commands)
# The final installation step, either from Github (url) or local folder
if "local" in config:
app = "%sADD %s /scif/apps/%s\n" % (app, exp_id, exp_id)
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" % (
app, exp_id)
else:
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" % (
app, config['github'])
apps = "%s%s\n" % (apps, app)
else:
bot.warning('%s not in library, check spelling & punctuation.' %
exp_id)
if apps == "\n":
bot.error('No valid experiments found, cancelling recipe build.')
sys.exit(1)
template = sub_template(template, "{{experiments}}", apps)
outfile = write_file(args.output, template)
bot.log("Recipe written to %s" % outfile)
def generate_runtime_vars(variable_file=None, sep=","):
"""generate a lookup data structure from a
delimited file. We typically obtain the file name and delimiter from
the environment by way of EXPFACTORY_RUNTIME_VARS, and
EXPFACTORY_RUNTIME_DELIM, respectively, but the user can also parse
from a custom variable file by way of specifying it to the function
(preference is given here). The file should be csv, with the
only required first header field as "token" and second as "exp_id" to
distinguish the participant ID and experiment id. The subsequent
columns should correspond to experiment variable names. No special parsing
of either is done.
Parameters
==========
variable_file: full path to the tabular file with token, exp_id, etc.
sep: the default delimiter to use, if not set in enironment.
Returns
=======
varset: a dictionary lookup by exp_id and then participant ID.
{ 'test-parse-url': {
'123': {
'color': 'red',
'globalname': 'globalvalue',
'words': 'at the thing'
},
'456': {'color': 'blue',
'globalname': 'globalvalue',
'words': 'omg tacos'}
}
}
"""
# First preference goes to runtime, then environment, then unset
if variable_file is None:
if EXPFACTORY_RUNTIME_VARS is not None:
variable_file = EXPFACTORY_RUNTIME_VARS
if variable_file is not None:
if not os.path.exists(variable_file):
bot.warning("%s is set, but not found" % variable_file)
return variable_file
# If still None, no file
if variable_file is None:
return variable_file
# If we get here, we have a variable file that exists
delim = sep
if EXPFACTORY_RUNTIME_DELIM is not None:
delim = EXPFACTORY_RUNTIME_DELIM
bot.debug("Delim for variables file set to %s" % sep)
# Read in the file, generate config
varset = dict()
rows = _read_runtime_vars(variable_file)
if len(rows) > 0:
# When we get here, we are sure to have
# 'exp_id', 'var_name', 'var_value', 'token'
for row in rows:
exp_id = row[0].lower() # exp-id must be lowercase
var_name = row[1]
var_value = row[2]
token = row[3]
# Level 1: Experiment ID
if exp_id not in varset:
varset[exp_id] = {}
# Level 2: Participant ID
if token not in varset[exp_id]:
varset[exp_id][token] = {}
# If found global setting, courtesy debug message
if token == "*":
bot.debug("Found global variable %s" % var_name)
# Level 3: is the variable, issue warning if already defined
if var_name in varset[exp_id][token]:
bot.warning("%s defined twice %s:%s" %
(var_name, exp_id, token))
varset[exp_id][token][var_name] = var_value
return varset
def generate_runtime_vars(variable_file=None, sep=','):
'''generate a lookup data structure from a
delimited file. We typically obtain the file name and delimiter from
the environment by way of EXPFACTORY_RUNTIME_VARS, and
EXPFACTORY_RUNTIME_DELIM, respectively, but the user can also parse
from a custom variable file by way of specifying it to the function
(preference is given here). The file should be csv, with the
only required first header field as "token" and second as "exp_id" to
distinguish the participant ID and experiment id. The subsequent
columns should correspond to experiment variable names. No special parsing
of either is done.
Parameters
==========
variable_file: full path to the tabular file with token, exp_id, etc.
sep: the default delimiter to use, if not set in enironment.
Returns
=======
varset: a dictionary lookup by exp_id and then participant ID.
{ 'test-parse-url': {
'123': {
'color': 'red',
'globalname': 'globalvalue',
'words': 'at the thing'
},
'456': {'color': 'blue',
'globalname': 'globalvalue',
'words': 'omg tacos'}
}
}
'''
# First preference goes to runtime, then environment, then unset
if variable_file is None:
if EXPFACTORY_RUNTIME_VARS is not None:
variable_file = EXPFACTORY_RUNTIME_VARS
if variable_file is not None:
if not os.path.exists(variable_file):
bot.warning('%s is set, but not found' %variable_file)
return variable_file
# If still None, no file
if variable_file is None:
return variable_file
# If we get here, we have a variable file that exists
delim = sep
if EXPFACTORY_RUNTIME_DELIM is not None:
delim = EXPFACTORY_RUNTIME_DELIM
bot.debug('Delim for variables file set to %s' %sep)
# Read in the file, generate config
varset = dict()
rows = _read_runtime_vars(variable_file)
if len(rows) > 0:
# When we get here, we are sure to have
# 'exp_id', 'var_name', 'var_value', 'token'
for row in rows:
exp_id = row[0].lower() # exp-id must be lowercase
var_name = row[1]
var_value = row[2]
token = row[3]
# Level 1: Experiment ID
if exp_id not in varset:
varset[exp_id] = {}
# Level 2: Participant ID
if token not in varset[exp_id]:
varset[exp_id][token] = {}
# If found global setting, courtesy debug message
if token == "*":
bot.debug('Found global variable %s' %var_name)
# Level 3: is the variable, issue warning if already defined
if var_name in varset[exp_id][token]:
bot.warning('%s defined twice %s:%s' %(var_name, exp_id, token))
varset[exp_id][token][var_name] = var_value
return varset
def main(args, parser, subparser):
template = "build/docker/Dockerfile.template"
# Full path to template is required if provided via input
if args.input is not None:
template = args.input
template = get_template(template)
# For now, only one database provided
database = args.database
studyid = args.studyid
experiments = args.experiments
branch = "-b %s" % os.environ.get("EXPFACTORY_BRANCH", "master")
headless = "false"
if args.headless is True:
headless = "true"
template = sub_template(template, "{{studyid}}", studyid)
template = sub_template(template, "{{database}}", database)
template = sub_template(template, "{{headless}}", headless)
template = sub_template(template, "{{branch}}", branch)
if args.headless is True:
bot.info(
"Headless build detected, you will need to generate tokens for application entry with expfactory users --new"
)
library = get_library(key="name")
apps = "\n"
# Add local experiments to library, first preference
local_installs = 0
for experiment in experiments:
if os.path.exists(experiment):
bot.info("local experiment %s found, validating..." % experiment)
# Is the experiment valid?
cli = ExperimentValidator()
valid = cli.validate(experiment, cleanup=False)
if valid is True:
local_installs += 1
config = load_experiment(experiment)
exp_id = config["exp_id"]
# If we aren't building in the experiment directory, we need to copy there
output_dir = "%s/%s" % (
os.path.abspath(os.path.dirname(args.output)),
exp_id,
)
experiment_dir = os.path.abspath(experiment)
if output_dir != experiment_dir:
copy_directory(experiment_dir, output_dir)
config["local"] = os.path.abspath(experiment)
library[exp_id] = config
# Warn the user that local installs are not reproducible (from recipe)
if local_installs > 0:
bot.warning(
"%s local installs detected: build is not reproducible without experiment folders"
% local_installs
)
# Build Image with Experiments
for experiment in experiments:
exp_id = os.path.basename(experiment)
if exp_id in library:
config = library[exp_id]
app = "LABEL EXPERIMENT_%s /scif/apps/%s\n" % (exp_id, exp_id)
# Here add custom build routine, should be list of lines
if "install" in config:
commands = "\n".join(["RUN %s " % s for x in config["install"]]).strip(
"\n"
)
app = "%s%s\n" % (app, commands)
# The final installation step, either from Github (url) or local folder
if "local" in config:
app = "%sADD %s /scif/apps/%s\n" % (app, exp_id, exp_id)
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" % (
app,
exp_id,
)
else:
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" % (
app,
config["github"],
)
apps = "%s%s\n" % (apps, app)
else:
bot.warning("%s not in library, check spelling & punctuation." % exp_id)
if apps == "\n":
bot.error("No valid experiments found, cancelling recipe build.")
sys.exit(1)
template = sub_template(template, "{{experiments}}", apps)
outfile = write_file(args.output, template)
bot.log("Recipe written to %s" % outfile)
def main(args,parser,subparser):
template = 'build/docker/Dockerfile.template'
# Full path to template is required if provided via input
if args.input is not None:
template = args.input
template = get_template(template)
# For now, only one database provided
database = args.database
studyid = args.studyid
experiments = args.experiments
branch = "-b %s" %os.environ.get('EXPFACTORY_BRANCH','master')
headless = "false"
if args.headless is True:
headless = "true"
template = sub_template(template,"{{studyid}}",studyid)
template = sub_template(template,"{{database}}",database)
template = sub_template(template,"{{headless}}",headless)
template = sub_template(template,"{{branch}}",branch)
if args.headless is True:
bot.info("Headless build detected, you will need to generate tokens for application entry with expfactory users --new")
library = get_library(key='name')
apps = "\n"
# Add local experiments to library, first preference
local_installs = 0
for experiment in experiments:
if os.path.exists(experiment):
bot.info('local experiment %s found, validating...' %experiment)
# Is the experiment valid?
cli = ExperimentValidator()
valid = cli.validate(experiment, cleanup=False)
if valid is True:
local_installs +=1
config = load_experiment(experiment)
exp_id = config['exp_id']
# If we aren't building in the experiment directory, we need to copy there
output_dir = "%s/%s" %(os.path.abspath(os.path.dirname(args.output)), exp_id)
experiment_dir = os.path.abspath(experiment)
if output_dir != experiment_dir:
copy_directory(experiment_dir, output_dir)
config['local'] = os.path.abspath(experiment)
library[exp_id] = config
# Warn the user that local installs are not reproducible (from recipe)
if local_installs > 0:
bot.warning("%s local installs detected: build is not reproducible without experiment folders" %local_installs)
# Build Image with Experiments
for experiment in experiments:
exp_id = os.path.basename(experiment)
if exp_id in library:
config = library[exp_id]
app = "LABEL EXPERIMENT_%s /scif/apps/%s\n" %(exp_id, exp_id)
# Here add custom build routine, should be list of lines
if "install" in config:
commands = "\n".join(["RUN %s "%s for x in config['install']]).strip('\n')
app = "%s%s\n" %(app, commands)
# The final installation step, either from Github (url) or local folder
if "local" in config:
app = "%sADD %s /scif/apps/%s\n" %(app, exp_id, exp_id)
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" %(app, exp_id)
else:
app = "%sWORKDIR /scif/apps\nRUN expfactory install %s\n" %(app, config['github'])
apps = "%s%s\n" %(apps,app)
else:
bot.warning('%s not in library, check spelling & punctuation.' %exp_id)
if apps == "\n":
bot.error('No valid experiments found, cancelling recipe build.')
sys.exit(1)
template = sub_template(template,"{{experiments}}",apps)
outfile = write_file(args.output,template)
bot.log("Recipe written to %s" %outfile)