pyiron_base.database.filetable module

File based database interface

class pyiron_base.database.filetable.FileTable(index_from)

Bases: IsDatabase

File table should behave to the user like a database, but it infers project hierarchy directly from the file system hierarchy.

Because indexing the file system can be expensive, and projects sometimes get re-initialized, it is important to keep the (re)instantiation cost for this class as minimal as possible.

Parameters:

• index_from (str) – The file path to start indexing at, i.e. the project path.

• fileindex (PyFileIndex) – In case the file path in index_from is already indexed, then the index can be provided as additional input parameter.

add_item_dict(par_dict)

Create a new database item

Parameters:

par_dict (dict) –

Dictionary with the item values and column names as keys, like:

{‘chemicalformula’: ‘BO’,

’computer’: ‘localhost’, ‘hamilton’: ‘VAMPS’, ‘hamversion’: ‘1.1’, ‘job’: ‘testing’, ‘subjob’ : ‘SubJob’, ‘parentid’: 0L, ‘myCol’: ‘Blubbablub’, ‘project’: ‘database.testing’, ‘projectpath’: ‘/root/directory/tmp’, ‘status’: ‘KAAAA’, ‘timestart’: datetime(2016, 5, 2, 11, 31, 4, 253377), ‘timestop’: datetime(2016, 5, 2, 11, 31, 4, 371165), ‘totalcputime’: 0.117788, ‘username’: ‘Test’}

Returns:

Database ID of the item created as an int, like: 3

Return type:

int

delete_item(item_id)

Delete Item from database

Parameters:

item_id (int) – Databse Item ID (Integer), like: 38

force_reset(fileindex=None)

Reset cache of the FileTable object

Parameters:

fileindex (PyFileIndex) – File index for the current directory

get_child_ids(job_specifier, project=None, status=None)

Get the childs for a specific job

Parameters:

• job_specifier (str) – name of the master job or the master jobs job ID

• project (str) – project_path - this is in contrast to the project_path in GenericPath

• status (str) – filter childs which match a specific status - None by default

Returns:

list of child IDs

Return type:

list

static get_extract(path, mtime)

get_item_by_id(item_id)

Get item from database by searching for a specific item Id.

Parameters:

item_id (int) – Databse Item ID (Integer), like: 38

Returns:

Dictionary where the key is the column name, like:

{‘chemicalformula’: u’BO’,

’computer’: u’localhost’, ‘hamilton’: u’VAMPS’, ‘hamversion’: u’1.1’, ‘id’: 1, ‘job’: u’testing’, ‘masterid’: None, ‘parentid’: 0, ‘project’: u’database.testing’, ‘projectpath’: u’/root/directory/tmp’, ‘status’: u’KAAAA’, ‘subjob’: u’SubJob’, ‘timestart’: datetime.datetime(2016, 5, 2, 11, 31, 4, 253377), ‘timestop’: datetime.datetime(2016, 5, 2, 11, 31, 4, 371165), ‘totalcputime’: 0.117788, ‘username’: u’Test’}

Return type:

dict

get_items_dict(item_dict, return_all_columns=True)

Get list of jobs which fulfills the query in the dictionary

Parameters:

• item_dict (dict) –

  a dict type, which has a certain syntax for this function: a normal dict like {‘hamilton’: ‘VAMPE’, ‘hamversion’: ‘1.1’} has similarities with a simple query like

  select * from table_name where hamilton = ‘VAMPE AND hamversion = ‘1.1’

  as seen it puts an AND for every key, value combination in the dict and searches for it.

  another syntax is for an OR statement, simply: {‘hamilton’: [‘VAMPE’, ‘LAMMPS’]}, the query would be:

  select * from table_name where hamilton = ‘VAMPE’ OR hamilton = ‘LAMMPS’

  and lastly for a LIKE statement, simply: {‘project’: ‘database.%’}, the query would be

  select * from table_name where project LIKE ‘database.%’

  that means you can simply add the syntax for a like statement like ‘%’ and it will automatically operate a like-search

  of course you can also use a more complex select method, with everything in use:

  {‘hamilton’: [‘VAMPE’, ‘LAMMPS’],

  ’project’: ‘databse%’, ‘hamversion’: ‘1.1’}

  select * from table_name where (hamilton = ‘VAMPE’ Or hamilton = ‘LAMMPS’) AND

  (project LIKE ‘database%’) AND hamversion = ‘1.1’

• return_all_columns (bool) – return all columns or only the ‘id’ - still the format stays the same.

Returns:

the function returns a list of dicts like get_items_sql, but it does not format datetime:

[{‘chemicalformula’: u’Ni108’,

’computer’: u’mapc157’, ‘hamilton’: u’LAMMPS’, ‘hamversion’: u’1.1’, ‘id’: 24, ‘job’: u’DOF_1_0’, ‘parentid’: 21L, ‘project’: u’lammps.phonons.Ni_fcc’, ‘projectpath’: u’D:/PyIron/PyIron_data/projects’, ‘status’: u’finished’, ‘timestart’: datetime.datetime(2016, 6, 24, 10, 17, 3, 140000), ‘timestop’: datetime.datetime(2016, 6, 24, 10, 17, 3, 173000), ‘totalcputime’: 0.033, ‘username’: u’test’},

{‘chemicalformula’: u’Ni108’,

’computer’: u’mapc157’, ‘hamilton’: u’LAMMPS’, ‘hamversion’: u’1.1’, ‘id’: 21, ‘job’: u’ref’, ‘parentid’: 20L, ‘project’: u’lammps.phonons.Ni_fcc’, ‘projectpath’: u’D:/PyIron/PyIron_data/projects’, ‘status’: u’finished’, ‘timestart’: datetime.datetime(2016, 6, 24, 10, 17, 2, 429000), ‘timestop’: datetime.datetime(2016, 6, 24, 10, 17, 2, 463000), ‘totalcputime’: 0.034, ‘username’: u’test’},…….]

Return type:

list

get_job_id(job_specifier, project=None)

Get job ID from filetable

Parameters:

• job_specifier (str) – Job ID or job name

• project (str/ None) – project_path as string

Returns:

job ID

Return type:

int/ None

get_job_status(job_id)

Get status of a given job selected by its job ID

Parameters:

job_id (int) – job ID as integer

Returns:

status of the job

Return type:

str

get_job_working_directory(job_id)

Get the working directory of a particular job

Parameters:

job_id (int) – job ID as integer

Returns:

working directory as absolute path

Return type:

str

init_table(fileindex, working_dir_lst=None)

Initialize the filetable class

Parameters:

• fileindex (pandas.DataFrame) – file system index for the current project path

• working_dir_lst (list/ None) – list of working directories

Returns:

list of dictionaries

Return type:

list

set_job_status(job_id, status)

Set job status

Parameters:

• job_id (int) – job ID as integer

• status (str) – job status

update()

Update the filetable cache

class pyiron_base.database.filetable.FileTableSingleton(name, bases, namespace, /, **kwargs)

Bases: ABCMeta

Indexing the file system for each FileTable can be expensive, so we use a singleton system that does this once for each path instead.

pyiron_base.database.filetable.filter_function(file_name)

pyiron_base.database.filetable.get_hamilton_from_file(hdf5_file, job_name)

pyiron_base.database.filetable.get_hamilton_version_from_file(hdf5_file, job_name)

pyiron_base.database.filetable.get_job_status_from_file(hdf5_file, job_name)