data-manipulation-utilities 0.2.8.dev714__py3-none-any.whl → 0.2.8.dev720__py3-none-any.whl

dmu_data/stats/parameters/data.yaml

@@ -1,178 +0,0 @@
-exp:
-  c:
-    val : -0.010
-    low : -0.020
-    high: -0.0001
-hypexp:
-  mu:
-    val : 5000
-    low : 4000
-    high: 6000
-  ap:
-    val : 0.020
-    low : 0
-    high: 0.10
-  bt:
-    val : 0.002
-    low : 0.0001
-    high: 0.003
-modexp:
-  mu:
-    val : 4250
-    low : 4250
-    high: 4500
-  ap:
-    val : 0.002
-    low : 0.002
-    high: 0.026
-  bt:
-    val : 0.002
-    low : 0.002
-    high: 0.020
-pol1:
-  a:
-    val : -0.005
-    low : -0.95
-    high: 0.00
-pol2:
-  a:
-    val : -0.005
-    low : -0.95
-    high: 0.00
-  b:
-    val : 0.000
-    low : -0.95
-    high: 0.95
-pol3:
-  a:
-    val : -0.005
-    low : -0.95
-    high: 0.00
-  b:
-    val : 0.000
-    low : -0.95
-    high: 0.95
-  c:
-    val : 0.000
-    low : -0.95
-    high: 0.95
-cbr:
-  mu:
-    val : 5300
-    low : 5100
-    high: 5500
-  sg:
-    val : 50
-    low : 2
-    high: 300
-  ac:
-    val : -2
-    low : -14.
-    high: -0.1
-  nc:
-    val : 1
-    low : 0.5
-    high: 150
-suj:
-  mu:
-    val : 4000
-    low : 3500
-    high: 5000
-  sg:
-    val : 10
-    low : 2
-    high: 5000
-  gm:
-    val : 1
-    low : -10
-    high: 10
-  dl:
-    val : 1
-    low : 0.1
-    high: 40
-cbl:
-  mu:
-    val : 5300
-    low : 5100
-    high: 5500
-  sg:
-    val : 50
-    low : 2
-    high: 300
-  ac:
-    val : 2
-    low : 0.0
-    high: 14.
-  nc:
-    val : 1
-    low : 0.5
-    high: 150
-gauss:
-  mu:
-    val : 5300
-    low : 5100
-    high: 5500
-  sg:
-    val : 50
-    low : 2
-    high: 300
-dscb:
-  mu:
-    val : 5300
-    low : 5000
-    high: 5400
-  sg:
-    val : 50
-    low :  5 
-    high: 500
-  ar:
-    val : 1
-    low : 0
-    high: 5
-  al:
-    val : 1
-    low : 0
-    high: 5
-  nr:
-    val : 2
-    low : 1
-    high: 150
-  nl:
-    val : 2
-    low : 0
-    high: 150
-voigt:
-  mu:
-    val : 5280
-    low : 5040
-    high: 5500
-  sg:
-    val : 20
-    low : 10
-    high: 400
-  gm:
-    val : 4
-    low : 0.1
-    high: 100
-qgauss:
-  mu:
-    val : 5280
-    low : 5040
-    high: 5500
-  sg:
-    val : 20
-    low : 10
-    high: 400
-  q:
-    val : 1
-    low : 1
-    high: 3
-cauchy:
-  mu:
-    val : 5280
-    low : 5040
-    high: 5500
-  gm:
-    val : 150
-    low : 50
-    high: 500

dmu_data/tests/config.json

@@ -1,6 +0,0 @@
-{
-    "key" : [
-    "value1",
-    "value2",
-    "value3"]
-}

dmu_data/tests/config.yaml

@@ -1,4 +0,0 @@
-key:
-  - value1
-  - value2
-  - value3

dmu_data/tests/pdf_to_tex.txt

@@ -1,34 +0,0 @@
---------------------
-PDF: SumPDF
-OBS: <zfit Space obs=('B_M_brem_track_2',), axes=(0,), limits=(array([[4500.]]), array([[7000.]])), binned=False>
-Name                                                        Value            Low           HighFloating               Constraint
---------------------
-ar_dscb_Signal_002_1_reso_flt                           1.000e+00      2.000e-01      5.000e+00    1      9.062e-01___1.910e-02
-c_exp_cmb_1                                            -1.000e-02     -2.000e-02     -1.000e-04    1                       none
-frac_brem_001                                           3.000e-01      0.000e+00      1.000e+00    1      3.273e-01___1.157e-05
-mu_Signal_001_scale_flt                                 0.000e+00     -1.000e+02      1.000e+02    1     -1.876e+01___7.881e-01
-mu_Signal_002_scale_flt                                 0.000e+00     -1.000e+02      1.000e+02    1     -1.228e+01___4.835e-01
-nBu_JpsiK_ee_eq_DPC                                     0.000e+00      0.000e+00      1.000e+04    1                       none
-ncmb                                                    0.000e+00      0.000e+00      2.000e+04    1                       none
-nl_dscb_Signal_001_1_reso_flt                           1.000e+00      2.000e-01      5.000e+00    1      8.314e+00___1.204e+00
-nr_dscb_Signal_002_1_reso_flt                           1.000e+00      2.000e-01      5.000e+00    1      1.770e+00___5.338e-02
-nsig                                                    0.000e+00      0.000e+00      1.000e+04    1                       none
-sBd_Kstee_eq_btosllball05_DPC                           0.000e+00      0.000e+00      1.000e+01    1      1.547e-01___1.552e-02
-sBu_Kstee_Kpi0_eq_btosllball05_DPC                      0.000e+00      0.000e+00      1.000e+01    1      1.698e-01___2.266e-02
-sg_Signal_001_reso_flt                                  1.000e+00      2.000e-01      5.000e+00    1      1.399e+00___2.266e-02
-sg_Signal_002_reso_flt                                  1.000e+00      2.000e-01      5.000e+00    1      1.327e+00___1.401e-02
-
-al_dscb_Signal_001_1                                    3.297e-01      0.000e+00      5.000e+00    0                     none
-al_dscb_Signal_002_1                                    3.684e-01      0.000e+00      5.000e+00    0                     none
-ar_dscb_Signal_001_1                                    1.274e+00      0.000e+00      5.000e+00    0                     none
-ar_dscb_Signal_002_1                                    8.425e-01      0.000e+00      5.000e+00    0                     none
-frc_Signal_001_1                                        2.911e-02      0.000e+00      1.000e+00    0                     none
-frc_Signal_002_1                                        1.074e-01      0.000e+00      1.000e+00    0                     none
-mu_Signal_001                                           5.231e+03      5.100e+03      5.500e+03    0                     none
-mu_Signal_002                                           5.230e+03      5.100e+03      5.500e+03    0                     none
-nl_dscb_Signal_001_1                                    9.130e+00      0.000e+00      1.500e+02    0                     none
-nl_dscb_Signal_002_1                                    5.152e+01      0.000e+00      1.500e+02    0                     none
-nr_dscb_Signal_001_1                                    2.483e+00      1.000e+00      1.500e+02    0                     none
-nr_dscb_Signal_002_1                                    3.801e+00      1.000e+00      1.500e+02    0                     none
-sg_Signal_001                                           5.155e+01      2.000e+00      3.000e+02    0                     none
-sg_Signal_002                                           7.496e+01      2.000e+00      3.000e+02    0                     none

dmu_data/text/transform.toml

@@ -1,4 +0,0 @@
-[trf]
-[trf.append]
-'primes are'=['2', '3', '5']
-'days are'=['Monday', 'Tuesday', 'Wednesday']

dmu_data/text/transform.txt

@@ -1,6 +0,0 @@
-the
-first
-primes are
-and
-the first
-days are

dmu_data/text/transform_set.toml

@@ -1,8 +0,0 @@
-[settings]
-as_substring=true
-format      ='--> {} <--'
-
-[trf]
-[trf.append]
-'primes are'=['2', '3', '5']
-'days are'=['Monday', 'Tuesday', 'Wednesday']

dmu_data/text/transform_set.txt

@@ -1,6 +0,0 @@
-the
-first
-primes are:
-and
-the first
-days are:

dmu_data/text/transform_trf.txt

@@ -1,12 +0,0 @@
-the
-first
-primes are
-2
-3
-5
-and
-the first
-days are
-Monday
-Tuesday
-Wednesday

dmu_scripts/git/publish

@@ -1,89 +0,0 @@
-#!/usr/bin/env bash
-
-# --------------------------
-display_help()
-{
-    echo "Script meant to:"
-    echo ""
-    echo "1. Check if version in pyproject.toml has been modified"
-    echo "2. If it has create new tag following version name"
-    echo "3. Push to remote "
-}
-# --------------------------
-get_opts()
-{
-    while getopts :hf: option; do 
-        case "${option}" in
-            h)  
-                display_help
-                exit 0
-                ;;  
-           \?)  echo "Invalid option: -${OPTARG}"
-                display_help
-                exit 1
-                ;;  
-            :)  echo "$0: Arguments needed"
-                display_help
-                exit 1
-                ;;  
-        esac
-    done
-}
-# --------------------------
-# Picks up version from pyproject.toml
-get_version()
-{
-    if [[ ! -f pyproject.toml ]];then
-        echo "Cannot find pyproject.toml"
-        exit 1
-    fi
-
-    VERSION_LINE=$(grep version pyproject.toml)
-
-    if [[ $? -ne 0 ]];then
-        ehco "Could not extract version from pyproject.toml"
-        exit 1
-    fi
-
-    if [[ "$VERSION_LINE" =~ .*([0-9]\.[0-9]\.[0-9]).* ]];then
-        VERSION=${BASH_REMATCH[1]}
-        echo "Using version: $VERSION"
-        return
-    fi
-
-    echo "Could not extract version from: $VERSION_LINE"
-    exit 1
-}
-# --------------------------
-create_tag()
-{
-    git tag -n | grep $VERSION
-
-    if [[ $? -eq 0 ]];then
-        echo "Version found among tags, not tagging"
-        return
-    fi
-
-    echo "Version $VERSION not found among tags, creating new tag"
-
-    git tag -a $VERSION
-}
-# --------------------------
-push_all()
-{
-    for REMOTE in $(git remote);do
-        echo "Pushing tags and commits to remote: $REMOTE"
-        git add pyproject.toml
-        git commit -m "Publication commit"
-
-        git pull    $REMOTE HEAD
-        git push -u $REMOTE HEAD
-        git push    $REMOTE --tags
-    done
-}
-# --------------------------
-get_opts "$@"
-
-get_version
-create_tag
-push_all

dmu_scripts/kerberos/check_expiration

@@ -1,21 +0,0 @@
-#!/usr/bin/env bash
-
-: '
-This script is meant to check the expiration date of a grid certificate
-
-Usage:
-
-./check_expiration
-'
-
-check()
-{
-    PEMFILE=$1
-    if [[ ! -f $PEMFILE ]];then
-        echo "Cannot find PEM file: $PEMFILE"
-    fi
-
-    openssl x509 -enddate -noout -in $PEMFILE 
-}
-
-check usercert.pem

dmu_scripts/kerberos/convert_certificate

@@ -1,22 +0,0 @@
-#!/usr/bin/env bash
-
-: '
-This script is used to convert p12 grid certificate files into PEM files
-
-Usage:
-
-./convert_certificate cert.p12
-'
-
-CERTIFICATE=$1
-
-if [[ ! -f $CERTIFICATE  ]];then
-    echo "ERROR::Certificate \"$CERTIFICATE\" does not exist."
-    kill -INT $$
-fi
-
-openssl pkcs12 -in $CERTIFICATE -clcerts -nokeys -out usercert.pem
-openssl pkcs12 -in $CERTIFICATE -nocerts -out userkey.pem
-chmod 400 userkey.pem
-chmod 444 usercert.pem
-

dmu_scripts/ml/compare_classifiers.py

@@ -1,85 +0,0 @@
-'''
-Script used to compare performance of classifiers
-'''
-import os
-import argparse
-import yaml
-import mplhep
-import matplotlib.pyplot as plt
-import pandas            as pnd
-
-from sklearn.metrics       import auc
-from dmu.logging.log_store import LogStore
-
-log=LogStore.add_logger('dmu:ml:compare_classifiers')
-# ------------------------------
-class Data:
-    '''
-    Data class
-    '''
-    out_path : str
-    cfg_path : str
-    logl     : int
-    cfg      : dict
-
-    plt.style.use(mplhep.style.LHCb2)
-# ------------------------------
-def _initialize() -> None:
-    log.info(f'Loading settings from: {Data.cfg_path}')
-    with open(Data.cfg_path, encoding='utf-8') as ifile:
-        Data.cfg = yaml.safe_load(ifile)
-
-    Data.out_path = Data.cfg['out_dir']
-    os.makedirs(Data.out_path, exist_ok=True)
-# ------------------------------
-def _parse_args():
-    parser = argparse.ArgumentParser(description='Used to perform comparisons of classifier performances')
-    parser.add_argument('-c', '--conf' , help='Path to configuration path', required=True)
-    parser.add_argument('-l', '--logl' , help='Logging level', choices=[10, 20, 30], default=20)
-    args = parser.parse_args()
-
-    Data.cfg_path = args.conf
-    Data.logl     = args.logl
-# ------------------------------
-def _plot_roc(name : str, path : str) -> None:
-    roc_path = f'{path}/fold_all/roc.json'
-    df = pnd.read_json(roc_path)
-
-    plt.figure(num='ROC')
-    xval = df['x'].to_numpy()
-    yval = df['y'].to_numpy()
-    area = auc(xval, yval)
-
-    plt.plot(xval, yval, label=f'{name}: {area:.3f}')
-# ------------------------------
-def _compare():
-    for name, cls_path in Data.cfg['classifiers'].items():
-        _plot_roc(name=name, path=cls_path)
-
-    _save_roc()
-# ------------------------------
-def _save_roc():
-    d_set = Data.cfg['roc']
-    if 'xrange' in d_set:
-        plt.xlim(d_set['xrange'])
-
-    if 'yrange' in d_set:
-        plt.ylim(d_set['yrange'])
-
-    plt.figure(num='ROC')
-    plt.legend()
-    plt.grid()
-    plt.xlabel('Signal Efficiency')
-    plt.ylabel('Background Rejection')
-    plt.savefig(f'{Data.out_path}/roc.png')
-# ------------------------------
-def main():
-    '''
-    Start here
-    '''
-    _parse_args()
-    _initialize()
-    _compare()
-# ------------------------------
-if __name__ == '__main__':
-    main()

dmu_scripts/physics/check_truth.py

@@ -1,121 +0,0 @@
-'''
-Script meant to do truth matching checks
-'''
-import os
-import copy
-import argparse
-
-import yaml
-import mplhep
-import matplotlib.pyplot as plt
-
-from ROOT import RDataFrame
-
-from dmu.logging.log_store   import LogStore
-from dmu.plotting.plotter_1d import Plotter1D as Plotter
-
-log=LogStore.add_logger('dmu:physics:check_truth')
-# ----------------------------------
-def _set_logs() -> None:
-    LogStore.set_level('dmu:plotting:Plotter'  , 30)
-    LogStore.set_level('dmu:plotting:Plotter1D', 30)
-# ----------------------------------
-def _get_args() -> argparse.Namespace:
-    '''
-    Parse args
-    '''
-    parser = argparse.ArgumentParser(description='Script used to carry out checks on truth matching mechanisms for MC')
-    parser.add_argument('-c', '--conf' , type=str, help='Path to config file', required=True)
-    args = parser.parse_args()
-
-    return args
-# ----------------------------------
-def _get_config(args : argparse.Namespace) -> dict:
-    path = args.conf
-    if not os.path.isfile(path):
-        raise FileNotFoundError(f'Cannot find {path}')
-
-    with open(path, encoding='utf-8') as ifile:
-        cfg = yaml.safe_load(ifile)
-
-    return cfg
-# ----------------------------------
-def _get_rdf(file_path : str, tree_path : str) -> RDataFrame:
-    log.debug(f'Picking inputs from: {file_path}/{tree_path}')
-    rdf = RDataFrame(tree_path, file_path)
-
-    nentries = rdf.Count().GetValue()
-    log.debug(f'Found {nentries} entries')
-
-    return rdf
-# ----------------------------------
-def _preprocess_rdf(rdf : RDataFrame, cfg : dict) -> RDataFrame:
-    if 'max_entries' in cfg:
-        max_entries = cfg['max_entries']
-        rdf = rdf.Range(max_entries)
-
-    return rdf
-# ----------------------------------
-def _check(cfg : dict) -> None:
-    log.info(110 * '-')
-    log.info(f'{"Sample":<20}{"Method":<20}{"Initial":<15}{"":<15}{"Final":<15}{"":15}{"Efficiency":<10}')
-    log.info(110 * '-')
-
-    for sample_name in cfg['samples']:
-        file_path = cfg['samples'][sample_name]['file_path']
-        tree_path = cfg['samples'][sample_name]['tree_path']
-        rdf = _get_rdf(file_path, tree_path)
-        rdf = _preprocess_rdf(rdf, cfg)
-
-        d_cut_true = {}
-        d_cut_fake = {}
-        for method, cut in cfg['samples'][sample_name]['methods'].items():
-            _check_kind(rdf, sample_name, method, cut)
-
-            d_cut_true[method] = cut
-            d_cut_fake[method] = f'({cut}) == 0'
-        log.info('')
-
-        _plot_distributions(cfg, sample_name, rdf, d_cut_true, kind='matched')
-        _plot_distributions(cfg, sample_name, rdf, d_cut_fake, kind='anti_matched')
-# ----------------------------------
-def _plot_distributions(cfg : dict, sample_name : str, rdf : RDataFrame, d_cut : dict[str,str], kind : str) -> None:
-    cfg     = copy.deepcopy(cfg)
-    cfg_plt = cfg['samples'][sample_name]['plot']
-    cfg_plt = _add_suffix(cfg_plt, sample_name, kind)
-    d_rdf   = { method : rdf.Filter(cut) for method, cut in d_cut.items() }
-
-    ptr=Plotter(d_rdf=d_rdf, cfg=cfg_plt)
-    ptr.run()
-# ----------------------------------
-def _add_suffix(cfg : dict, sample_name : str, kind : str) -> dict:
-    d_var = cfg['plots']
-    for var in d_var:
-        d_var[var]['name']  = f'{var}_{kind}'
-        d_var[var]['title'] = f'{sample_name}; {kind}'
-
-    cfg['plots'] = d_var
-
-    return cfg
-# ----------------------------------
-def _check_kind(rdf : RDataFrame, sample : str, name : str, cut : str) -> RDataFrame:
-    nini = rdf.Count().GetValue()
-    rdf  = rdf.Filter(cut, name)
-    nfnl = rdf.Count().GetValue()
-    eff  = nfnl / nini * 100
-
-    log.info(f'{sample:<20}{name:<20}{nini:<15}{"":<15}{nfnl:<15}{"-->":15}{eff:10.2f}')
-# ----------------------------------
-def main():
-    '''
-    Script starts here
-    '''
-    _set_logs()
-    args = _get_args()
-    cfg  = _get_config(args)
-    plt.style.use(mplhep.style.LHCb2)
-
-    _check(cfg)
-# ----------------------------------
-if __name__ == '__main__':
-    main()