def webauthorprofile_daemon():
""" Constructs the webauthorprofile bibtask. """
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="WebAuthorProfile Task Submission",
description="""
Purpose:
Precompute WebAuthorProfile caches.
Examples:
$webauthorprofile -u admin --all
""",
help_specific_usage="""
webauthorprofile [OPTIONS]
OPTIONS
Options for update personid
(default) Computes all caches for all persons with at least one expired cache
--all Computes all caches for all persons
--mp Enables multiprocessing computation
""",
version="Invenio WebAuthorProfile v 1.0",
specific_params=("i:", ["all", "mp"]),
task_submit_elaborate_specific_parameter_fnc=_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibclassify_daemon():
"""Constructs the BibClassify bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="BibClassify Task Submission",
description="Extract keywords and create a BibUpload "
"task.\nExamples:\n"
" $ bibclassify\n"
" $ bibclassify -i 79 -k HEP\n"
" $ bibclassify -c 'Articles' -k HEP\n",
help_specific_usage=" -i, --recid\t\tkeywords are extracted from "
"this record\n"
" -c, --collection\t\tkeywords are extracted from this collection\n"
" -k, --taxonomy\t\tkeywords are based on that reference",
version="Invenio BibClassify v%s" % bconfig.VERSION,
specific_params=("i:c:k:f",
[
"recid=",
"collection=",
"taxonomy=",
"force"
]),
task_submit_elaborate_specific_parameter_fnc=
_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def webauthorprofile_daemon():
"""Constructs the webauthorprofile bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="WebAuthorProfile Task Submission",
description="""
Purpose:
Precompute WebAuthorProfile caches.
Examples:
$webauthorprofile -u admin --all
""",
help_specific_usage="""
webauthorprofile [OPTIONS]
OPTIONS
Options for update personid
(default) Computes all caches for all persons with at least one expired cache
--all Computes all caches for all persons
--mp Enables multiprocessing computation
""",
version="Invenio WebAuthorProfile v 1.0",
specific_params=("i:",
[
"all",
"mp"
]),
task_submit_elaborate_specific_parameter_fnc=_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibauthorid_daemon():
"""Constructs the Bibauthorid bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="Bibauthorid Task Submission",
description="""
Purpose:
Disambiguate Authors and find their identities.
Examples:
- Process all records that hold an author with last name 'Ellis':
$ bibauthorid -u admin --lastname 'Ellis'
- Process all records and regard all authors:
$ bibauthorid -u admin --process-all
- Prepare job packages in folder 'gridfiles' with the sub directories
prefixed with 'task' and a maximum number of 2000 records per package:
$ bibauthorid -u admin --prepare-grid -d gridfiles -p task -m 2000
""",
help_specific_usage="""
NOTE: Options -n, -a, -U, -G and -R are mutually exclusive (XOR)!
-n, --lastname=STRING Process only authors with this last name.
-a, --process-all The option for cleaning all authors.
-U, --update-universe Update bibauthorid universe. Find modified and
newly entered records and process all the authors
on these records.
-G, --prepare-grid Prepares a set of files that supply the
pre-clustered data needed for stand alone job to
run (e.g. needed on the grid). The behavior of
this export can be controlled with the
options -d (required), -p and -m (both optional).
-R, --load-grid-results Loads the results from the grid jobs
and writes them to the database. The behavior of
this import can be controlled with the
options -d (required).
-d, --data-dir=DIRNAME Specifies the data directory, in which the data for
the grid preparation will be stored to or loaded
from. It requires the -G or -R switch.
-p, --prefix=STRING Specifies the prefix of the directories created
under the --data-dir directory. Optional.
Defaults to 'job'. It requires the -G switch.
-m, --max-records Specifies the number of records that
shall be stored per job package. Optional.
Defaults to 4000 and requires -G switch.
--update-cache Updates caches to the newly introduced changes
(new and modified documents).
This should be called daily or better more then
once per day, to ensure the correct operation of
the frontend (and the backend).
--clean-cache Clean the cache from out of date contents
(deleted documents).
""",
version="Invenio Bibauthorid v%s" % bconfig.VERSION,
specific_params=("d:n:p:m:GRa", [
"data-dir=", "lastname=", "prefix=", "max-records=",
"process-all", "prepare-grid", "load-grid-results",
"update-universe", "update-cache", "clean-cache"
]),
task_submit_elaborate_specific_parameter_fnc=
_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibauthorid_daemon():
"""Constructs the Bibauthorid bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="Bibauthorid Task Submission",
description="""
Purpose:
Disambiguate Authors and find their identities.
Examples:
- Process all records that hold an author with last name 'Ellis':
$ bibauthorid -u admin --lastname 'Ellis'
- Process all records and regard all authors:
$ bibauthorid -u admin --process-all
- Prepare job packages in folder 'gridfiles' with the sub directories
prefixed with 'task' and a maximum number of 2000 records per package:
$ bibauthorid -u admin --prepare-grid -d gridfiles -p task -m 2000
""",
help_specific_usage="""
--repair-personid Deletes untouched person entities to then
re-create and updated these entities.
--fast-update-personid Updates personid adding not yet assigned papers to the system,
in a fast, best effort basis. Use -r to limit to a comma separated
set of records.
--personid-gc Garbage collects personid for stale records. Use -r to limit to a comma
separated set of records.
-r, --record-ids Specifies a list of record ids. To use as on option
for --update-universe to limit the update to the
selected records
--all-records To use as on option for --update-universe to
perform the update an all existing record ids. Be
WARNED that this will empty and re-fill all aid*
tables in the process!
""",
version="Invenio Bibauthorid v%s" % bconfig.VERSION,
specific_params=("r:",
[
"record-ids=",
"all-records",
"repair-personid",
"fast-update-personid",
"personid-gc"
]),
task_submit_elaborate_specific_parameter_fnc
=_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc
=_task_submit_check_options,
task_run_fnc
=_task_run_core)
def bibclassify_daemon():
"""Constructs the BibClassify bibtask."""
bibtask.task_init(
authorization_action='runbibclassify',
authorization_msg="BibClassify Task Submission",
description="Extract keywords and create a BibUpload "
"task.\nExamples:\n"
" $ bibclassify\n"
" $ bibclassify -i 79 -k HEP\n"
" $ bibclassify -c 'Articles' -k HEP\n",
help_specific_usage=" -i, --recid\t\tkeywords are extracted from "
"this record\n"
" -c, --collection\t\tkeywords are extracted from this collection\n"
" -k, --taxonomy\t\tkeywords are based on that reference",
version="Invenio BibClassify v%s" % bconfig.VERSION,
specific_params=("i:c:k:f",
["recid=", "collection=", "taxonomy=", "force"]),
task_submit_elaborate_specific_parameter_fnc=
_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibauthorid_daemon():
"""Constructs the Bibauthorid bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="Bibauthorid Task Submission",
description="""
Purpose:
Disambiguate Authors and find their identities.
Examples:
- Process all records that hold an author with last name 'Ellis':
$ bibauthorid -u admin --lastname 'Ellis'
- Process all records and regard all authors:
$ bibauthorid -u admin --process-all
- Prepare job packages in folder 'gridfiles' with the sub directories
prefixed with 'task' and a maximum number of 2000 records per package:
$ bibauthorid -u admin --prepare-grid -d gridfiles -p task -m 2000
""",
help_specific_usage="""
NOTE: Options -n, -a, -U, -G and -R are mutually exclusive (XOR)!
-n, --lastname=STRING Process only authors with this last name.
-a, --process-all The option for cleaning all authors.
-U, --update-universe Update bibauthorid universe. Find modified and
newly entered records and process all the authors
on these records.
-G, --prepare-grid Prepares a set of files that supply the
pre-clustered data needed for stand alone job to
run (e.g. needed on the grid). The behavior of
this export can be controlled with the
options -d (required), -p and -m (both optional).
-R, --load-grid-results Loads the results from the grid jobs
and writes them to the database. The behavior of
this import can be controlled with the
options -d (required).
-d, --data-dir=DIRNAME Specifies the data directory, in which the data for
the grid preparation will be stored to or loaded
from. It requires the -G or -R switch.
-p, --prefix=STRING Specifies the prefix of the directories created
under the --data-dir directory. Optional.
Defaults to 'job'. It requires the -G switch.
-m, --max-records Specifies the number of records that
shall be stored per job package. Optional.
Defaults to 4000 and requires -G switch.
--update-cache Updates caches to the newly introduced changes
(new and modified documents).
This should be called daily or better more then
once per day, to ensure the correct operation of
the frontend (and the backend).
--clean-cache Clean the cache from out of date contents
(deleted documents).
""",
version="Invenio Bibauthorid v%s" % bconfig.VERSION,
specific_params=("d:n:p:m:GRa",
[
"data-dir=",
"lastname=",
"prefix=",
"max-records=",
"process-all",
"prepare-grid",
"load-grid-results",
"update-universe",
"update-cache",
"clean-cache"
]),
task_submit_elaborate_specific_parameter_fnc=
_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibauthorid_daemon():
"""Constructs the Bibauthorid bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="Bibauthorid Task Submission",
description="""
Purpose:
Disambiguate Authors and find their identities.
Examples:
- Process all records that hold an author with last name 'Ellis':
$ bibauthorid -u admin --update-personid --all-records
- Disambiguate all records on a fresh installation
$ bibauthorid -u admin --disambiguate --from-scratch
""",
help_specific_usage="""
bibauthorid [COMMAND] [OPTIONS]
COMMAND
You can choose only one from the following:
--update-personid Updates personid adding not yet assigned papers
to the system, in a fast, best effort basis.
Cleans the table from stale records.
--disambiguate Disambiguates all signatures in the database
using the tortoise/wedge algorithm. This usually
takes a LOT of time so the results are stored in
a special table. Use --merge to use the results.
--merge Updates the personid tables with the results from
the --disambiguate algorithm.
OPTIONS
Options for update personid
(default) Will update only the modified records since last
run.
-i, --record-ids Force the procedure to work only on the specified
records. This option is exclusive with --all-records.
--all-records Force the procedure to work on all records. This
option is exclusive with --record-ids.
Options for disambiguate
(default) Performs full disambiguation of all records in the
current personid tables with respect to the user
decisions.
--from-scratch Ignores the current information in the personid
tables and disambiguates everything from scratch.
There are no options for the merger.
""",
version="Invenio Bibauthorid v%s" % bconfig.VERSION,
specific_params=("i:",
[
"record-ids=",
"disambiguate",
"merge",
"all-records",
"update-personid",
"from-scratch"
]),
task_submit_elaborate_specific_parameter_fnc=_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
def bibauthorid_daemon():
"""Constructs the Bibauthorid bibtask."""
bibtask.task_init(authorization_action='runbibclassify',
authorization_msg="Bibauthorid Task Submission",
description="""
Purpose:
Disambiguate Authors and find their identities.
Examples:
- Process all records that hold an author with last name 'Ellis':
$ bibauthorid -u admin --update-personid --all-records
- Disambiguate all records on a fresh installation
$ bibauthorid -u admin --disambiguate --from-scratch
""",
help_specific_usage="""
bibauthorid [COMMAND] [OPTIONS]
COMMAND
You can choose only one from the following:
--update-personid Updates personid adding not yet assigned papers
to the system, in a fast, best effort basis.
Cleans the table from stale records.
--disambiguate Disambiguates all signatures in the database
using the tortoise/wedge algorithm. This usually
takes a LOT of time so the results are stored in
a special table. Use --merge to use the results.
--merge Updates the personid tables with the results from
the --disambiguate algorithm.
OPTIONS
Options for update personid
(default) Will update only the modified records since last
run.
-i, --record-ids Force the procedure to work only on the specified
records. This option is exclusive with --all-records.
--all-records Force the procedure to work on all records. This
option is exclusive with --record-ids.
Options for disambiguate
(default) Performs full disambiguation of all records in the
current personid tables with respect to the user
decisions.
--from-scratch Ignores the current information in the personid
tables and disambiguates everything from scratch.
There are no options for the merger.
""",
version="Invenio Bibauthorid v%s" % bconfig.VERSION,
specific_params=("i:",
[
"record-ids=",
"disambiguate",
"merge",
"all-records",
"update-personid",
"from-scratch"
]),
task_submit_elaborate_specific_parameter_fnc=_task_submit_elaborate_specific_parameter,
task_submit_check_options_fnc=_task_submit_check_options,
task_run_fnc=_task_run_core)
