def main():
import spiper
from path import Path
from spiper.runner import get_all_files, get_changed_files, cache_run
from pprint import pprint
spiper.rcParams['dir_layout'] = 'flat'
prefix = Path('/tmp/test_import/root')
prefix.dirname().rmtree_p()
print('#### [Note] currently the package change is not recorded')
fs = get_changed_files(test_import, prefix)
pprint(fs)
fs = get_all_files(test_import,prefix)
pprint(fs)
cache_run(test_import,prefix)
pprint(fs)
print('#### [Note] Remote workflow should detect file changes')
fs = get_changed_files(simple_flow, prefix)
pprint(fs)
assert fs != []
print('### run actual workflow')
cache_run(simple_flow,prefix)
fs = get_changed_files(simple_flow, prefix)
pprint(fs)
assert fs == []
def main():
import spiper
from path import Path
from spiper.runner import get_all_files, get_changed_files, cache_run
from pprint import pprint
spiper.rcParams['dir_layout'] = 'flat'
prefix = Path('/tmp/test_import/root')
prefix.dirname().rmtree_p()
print('#### [Note] currently the package change is not recorded')
fs = get_changed_files(test_import, prefix)
pprint(fs)
fs = get_all_files(test_import, prefix)
pprint(fs)
cache_run(test_import, prefix)
pprint(fs)
print('#### [Note] Remote workflow should detect file changes')
fs = get_changed_files(simple_flow, prefix)
pprint(fs)
assert fs != []
print('### run actual workflow')
cache_run(simple_flow, prefix)
fs = get_changed_files(simple_flow, prefix)
pprint(fs)
#### The Flow() execution will never be skipped
#### hence The self.output.log will always be changed
assert fs == [
File('/tmp/test_import/root.workflow.log'),
File('/tmp/test_import/root_backup.output..log')
]
def get_tree1(self, http=0):
if http:
tups = (
http_job2,
self.DIR / 'root',
)
res0 = cache_run(*tups, verbose=0)
tups = (dimple_job, self.DIR / 'root', 'ATG', res0.output.cache,
'/tmp/letter.txt')
else:
tups = (dimple_job, self.DIR / 'root', 'ATG', '/tmp/digit.txt',
'/tmp/letter.txt')
res1 = force_run(*tups, verbose=0)
tups = (dimple_job, self.DIR / 'job2', 'ATG', res1.output.out_txt,
'/tmp/digit.txt')
res2 = force_run(*tups, verbose=0)
tups = (dimple_job, self.DIR / 'job3', 'ATG', res1.output.out_txt,
res2.output.out_txt)
res3 = force_run(*tups, verbose=0)
fn = File('/tmp/digit.txt')
f2 = res3.output.out_txt
tree = None
# tree = spiper.runner.get_downstream_tree(fn, flat=0,strict=0)
return fn, f2, tree
def test_cacherun_use_cache(self):
_ = '''
if the input files / params to a simple node changed,
then trigger a recalc
'''
tups = (simple_job, self.DIR / 'root', 'ATG', '/tmp/digit.txt')
force_run(*tups, verbose=0)
result = cache_run(*tups, verbose=0)
from path import Path
Path('/tmp/some_node/root').dirname().rmtree_p()
fs = get_all_files(some_node, '/tmp/some_node/root', '/tmp/input_file.txt')
print('##### all files governed by this node #######')
pprint(fs)
fs = get_changed_files(some_node, '/tmp/some_node/root',
'/tmp/input_file.txt')
print('#### files changed in the next execution of this node #######')
pprint(fs)
print('#### write some input file ###')
with open('/tmp/input_file.txt', 'w') as f:
f.write('barbarfoo\n')
print('#### actual execution #####')
res = cache_run(some_node, '/tmp/some_node/root', '/tmp/input_file.txt')
print('#### the second execution is skipped ###')
res = cache_run(some_node, '/tmp/some_node/root', '/tmp/input_file.txt')
print()
#### the fact that the execution is skipped implies
#### that executing this node would not change any file
fs = get_changed_files(some_node, '/tmp/some_node/root',
'/tmp/input_file.txt')
print('#### files changed in the next execution of this node #######')
pprint(fs)
assert fs == []
from path import Path
from spiper.runner import cache_run, force_run
def simplest_node(self, prefix, _output=[]):
print('Running node:%r' % self)
return self ### return Runtime Caller as output is beneficial
if __name__ == '__main__':
print('\n### running')
prefix = Path('/tmp/bulid_spiper/root')
prefix.dirname().rmtree_p()
cache_run(simplest_node, prefix)
cache_run(simplest_node, prefix)
print()
s = '''
### comment
expect:
Running node:spiper.runner.Caller(dotname='__main__.simplest_node',prefix_named=File('/tmp/bulid_spiper/root.simplest_node'))
Running node:spiper.runner.Caller(dotname='__main__.simplest_node',prefix_named=File('/tmp/bulid_spiper/root.simplest_node'))
got:
Running node:spiper.runner.Caller(dotname='__main__.simplest_node',prefix_named=File('/tmp/bulid_spiper/root.simplest_node'))
The second run accessed cache
'''
print(s)
pass
def main(self=None,
prefix = None):
from spiper.runner import cache_run, mock_run, get_changed_files, get_all_files
from spiper.shell import LoggedShellCommand
from spiper.types import File,CacheFile
from pprint import pprint
spiper.rcParams['dir_layout']='clean'
# if prefix is None:
prefix = Path('/tmp/spiper.symbolic/root')
# backup_prefix = File('/home/user/.temp/backup_03_mock_flow/root')
backup_prefix = File('~/.temp/backup_03_mock_flow/root').expand()
prefix.dirname().rmtree_p()
backup_prefix.dirname().rmtree_p()
print('\n...[start]%r'%prefix)
#### once a workflow is defined, we can view the proposed file changes
fs = get_changed_files(workflow, prefix, 1, 100, verbose=0)
pprint(fs)
assert fs ==[
File('/tmp/spiper.symbolic/root.workflow.log'),
File('/tmp/spiper.symbolic/root.random_seq.seq'),
File('/tmp/spiper.symbolic/root.random_seq_const.seq'),
File('/tmp/spiper.symbolic/root.transcribe.fasta'),
File('/tmp/spiper.symbolic/root.mutate.fasta'),
File('/tmp/spiper.symbolic/root.source.py'),
# File('/home/user/.temp/backup_03_mock_flow/root.source.py')
]
### backup is conveniently defined as a workflow taking an executed workflow as an input.
### To check the proposed backup, mock_run() the workflow first.
workflow_out = mock_run(workflow, prefix, 1, 100)
fs = get_changed_files(backup, backup_prefix, workflow_out)
pprint(fs)
assert fs == [
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq_const.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.transcribe.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.mutate.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.output.log'),
# File('/tmp/spiper.symbolic/root.source.py')
File('/home/user/.temp/backup_03_mock_flow/root.source.py')
]
### a convenient Flow may be defined to execute the two in chain
### If there is certain change to the workflow,
### the backup can also be runned
fs = get_changed_files (run_and_backup, prefix, 1, 100, backup_prefix, verbose=0)
pprint(fs)
assert fs == [
File('/tmp/spiper.symbolic/root.workflow.log'),
File('/tmp/spiper.symbolic/root.random_seq.seq'),
File('/tmp/spiper.symbolic/root.random_seq_const.seq'),
File('/tmp/spiper.symbolic/root.transcribe.fasta'),
File('/tmp/spiper.symbolic/root.mutate.fasta'),
File('/tmp/spiper.symbolic/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq_const.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.transcribe.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.mutate.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.output.log'),
File('/home/user/.temp/backup_03_mock_flow/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_json'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_dot_txt'),
]
###### constants that are preserved between runs should be detected unchanged
_ = cache_run (run_and_backup, prefix, 1, 100, backup_prefix, verbose=0)
fs = get_changed_files (run_and_backup, prefix, 2, 200, backup_prefix, verbose=0)
pprint(fs)
assert fs == [File('/tmp/spiper.symbolic/root.workflow.log'),
File('/tmp/spiper.symbolic/root.random_seq.seq'),
# File('/tmp/spiper.symbolic/root.random_seq_const.seq'),
File('/tmp/spiper.symbolic/root.transcribe.fasta'),
File('/tmp/spiper.symbolic/root.mutate.fasta'),
# File('/tmp/spiper.symbolic/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq.output.seq'),
# File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq_const.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.transcribe.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.mutate.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.output.log'),
# File('/home/user/.temp/backup_03_mock_flow/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_json'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_dot_txt'),
]
##### get_all_files() return a leaf file regardless of whether is is changed
fs = get_all_files (run_and_backup, prefix, 2, 200, backup_prefix, verbose=0)
pprint(fs)
assert fs == [
File('/tmp/spiper.symbolic/root.workflow.log'),
File('/tmp/spiper.symbolic/root.random_seq.seq'),
File('/tmp/spiper.symbolic/root.random_seq_const.seq'),
File('/tmp/spiper.symbolic/root.transcribe.fasta'),
File('/tmp/spiper.symbolic/root.mutate.fasta'),
File('/tmp/spiper.symbolic/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.random_seq_const.output.seq'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.transcribe.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.subflow.mutate.output.fasta'),
File('/home/user/.temp/backup_03_mock_flow/root.output.log'),
File('/home/user/.temp/backup_03_mock_flow/root.source.py'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_json'),
File('/home/user/.temp/backup_03_mock_flow/root.plot_graph.deptree_dot_txt'),
]
_ = cache_run (run_and_backup, prefix, 2, 200, backup_prefix, verbose=0)
def runner(*a, **kw):
return cache_run(*a, **kw, check_only=2)
from spiper.types import Node,Flow
from spiper.types import RemotePythonObject as RPO
package_path = 'spiper_mock_flow@https://github.com/shouldsee/spiper_mock_flow/tarball/8cd0f6b'
@Flow
def simple_flow(
self,prefix,
_main= RPO(package_path, None, 'run_and_backup'),
_output=[]
):
func = _main.loaded()
self.runner( func, prefix, 1, 20, prefix+'_backup')
return self
if __name__ == '__main__':
from path import Path
from spiper.runner import get_changed_files,get_all_files,cache_run
from pprint import pprint
prefix = Path('/tmp/test_import/root')
prefix.dirname().rmtree_p()
fs = get_changed_files( simple_flow, prefix)
pprint(fs)
cache_run( simple_flow, prefix)
fs = get_changed_files( simple_flow, prefix)
pprint(fs)
'gzip -d | cut -f2- >',
self.output.genepred,
]
LoggedShellCommand(CMD, self.output.cmd, mode='w')
CMD = ['genePredToGtf', 'file', self.output.genepred, self.output.gtf]
LoggedSingularityCommand(CMD, _IMAGE, self.output.cmd, mode='a')
# return
if __name__ == '__main__':
from spiper.runner import force_run, cache_run
data = {}
data['fasta'] = cache_run(
get_fasta,
'~/.temp/0305',
)
data['gtf'] = cache_run(
get_genepred,
'~/.temp/0305',
)
data['flow1'] = cache_run(
workflow,
'~/.temp/0305.sample1',
'~/.temp/hisat2/',
data['fasta'].output.fasta,
'wuhan-ncov19',
data['gtf'].output.gtf,
'./tests/data/test_R1_.fastq',
'./tests/data/test_R2_.fastq',
2,
_output=[]):
curr = self.runner(get_fasta, prefix)
curr = self.runner(get_genepred, prefix)
curr = self.runner(workflow, prefix + '.sample1',
prefix + '.hisat2/wuhan-ncov19',
self.subflow['get_fasta'].output.fasta,
self.subflow['get_genepred'].output.gtf,
'./test_data/test_R1_.fastq',
'./test_data/test_R2_.fastq', _THREADS)
return self
if __name__ == '__main__':
from pprint import pprint
from spiper.runner import force_run, cache_run
cache_run(test_job, '_temp_build/root')
'python3 -m spiper run $PACKAGE TOPLEVEL:test_job _temp_build/root'
######
if 0:
################################### TBC afterwards ############################
def get_htseq():
_ = '''
#### htseq-count is too slow and not used
htseq-count
-s reverse
-f bam 809_S1.bam
/home/feng/ref/Arabidopsis_thaliana_TAIR10/annotation/genes.gtf
-r pos -o 809_S1.htseq.sam >809_S1.htseq.count 2>809_S1.htseq.log
'''