input_data_parser.py « presentation « tools « resources - csit - Integration tests

# Copyright (c) 2021 Cisco and/or its affiliates.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at:
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""Data pre-processing

- extract data from output.xml files generated by Jenkins jobs and store in
  pandas' Series,
- provide access to the data.
- filter the data using tags,
"""

import re
import copy
import resource
import logging

from collections import OrderedDict
from os import remove, walk, listdir
from os.path import isfile, isdir, join
from datetime import datetime as dt
from datetime import timedelta
from json import loads
from json.decoder import JSONDecodeError

import hdrh.histogram
import hdrh.codec
import prettytable
import pandas as pd

from robot.api import ExecutionResult, ResultVisitor
from robot import errors

from resources.libraries.python import jumpavg
from input_data_files import download_and_unzip_data_file
from pal_errors import PresentationError


# Separator used in file names
SEPARATOR = u"__"


class ExecutionChecker(ResultVisitor):
    """Class to traverse through the test suite structure.

    The functionality implemented in this class generates a json structure:

    Performance tests:

    {
        "metadata": {
            "generated": "Timestamp",
            "version": "SUT version",
            "job": "Jenkins job name",
            "build": "Information about the build"
        },
        "suites": {
            "Suite long name 1": {
                "name": Suite name,
                "doc": "Suite 1 documentation",
                "parent": "Suite 1 parent",
                "level": "Level of the suite in the suite hierarchy"
            }
            "Suite long name N": {
                "name": Suite name,
                "doc": "Suite N documentation",
                "parent": "Suite 2 parent",
                "level": "Level of the suite in the suite hierarchy"
            }
        }
        "tests": {
            # NDRPDR tests:
            "ID": {
                "name": "Test name",
                "parent": "Name of the parent of the test",
                "doc": "Test documentation",
                "msg": "Test message",
                "conf-history": "DUT1 and DUT2 VAT History",
                "show-run": "Show Run",
                "tags": ["tag 1", "tag 2", "tag n"],
                "type": "NDRPDR",
                "status": "PASS" | "FAIL",
                "throughput": {
                    "NDR": {
                        "LOWER": float,
                        "UPPER": float
                    },
                    "PDR": {
                        "LOWER": float,
                        "UPPER": float
                    }
                },
                "latency": {
                    "NDR": {
                        "direction1": {
                            "min": float,
                            "avg": float,
                            "max": float,
                            "hdrh": str
                        },
                        "direction2": {
                            "min": float,
                            "avg": float,
                            "max": float,
                            "hdrh": str
                        }
                    },
                    "PDR": {
                        "direction1": {
                            "min": float,
                            "avg": float,
                            "max": float,
                            "hdrh": str
                        },
                        "direction2": {
                            "min": float,
                            "avg": float,
                            "max": float,
                            "hdrh": str
                        }
                    }
                }
            }

            # TCP tests:
            "ID": {
                "name": "Test name",
                "parent": "Name of the parent of the test",
                "doc": "Test documentation",
                "msg": "Test message",
                "tags": ["tag 1", "tag 2", "tag n"],
                "type": "TCP",
                "status": "PASS" | "FAIL",
                "result": int
            }

            # MRR, BMRR tests:
            "ID": {
                "name": "Test name",
                "parent": "Name of the parent of the test",
                "doc": "Test documentation",
                "msg": "Test message",
                "tags": ["tag 1", "tag 2", "tag n"],
                "type": "MRR" | "BMRR",
                "status": "PASS" | "FAIL",
                "result": {
                    "receive-rate": float,
                    # Average of a list, computed using AvgStdevStats.
                    # In CSIT-1180, replace with List[float].
                }
            }

            "ID" {
                # next test
            }
        }
    }


    Functional tests:

    {
        "metadata": {  # Optional
            "version": "VPP version",
            "job": "Jenkins job name",
            "build": "Information about the build"
        },
        "suites": {
            "Suite name 1": {
                "doc": "Suite 1 documentation",
                "parent": "Suite 1 parent",
                "level": "Level of the suite in the suite hierarchy"
            }
            "Suite name N": {
                "doc": "Suite N documentation",
                "parent": "Suite 2 parent",
                "level": "Level of the suite in the suite hierarchy"
            }
        }
        "tests": {
            "ID": {
                "name": "Test name",
                "parent": "Name of the parent of the test",
                "doc": "Test documentation"
                "msg": "Test message"
                "tags": ["tag 1", "tag 2", "tag n"],
                "conf-history": "DUT1 and DUT2 VAT History"
                "show-run": "Show Run"
                "status": "PASS" | "FAIL"
            },
            "ID" {
                # next test
            }
        }
    }

    .. note:: ID is the lowercase full path to the test.
    """

    REGEX_PLR_RATE = re.compile(
        r'PLRsearch lower bound::?\s(\d+.\d+).*\n'
        r'PLRsearch upper bound::?\s(\d+.\d+)'
    )
    REGEX_NDRPDR_RATE = re.compile(
        r'NDR_LOWER:\s(\d+.\d+).*\n.*\n'
        r'NDR_UPPER:\s(\d+.\d+).*\n'
        r'PDR_LOWER:\s(\d+.\d+).*\n.*\n'
        r'PDR_UPPER:\s(\d+.\d+)'
    )
    REGEX_NDRPDR_GBPS = re.compile(
        r'NDR_LOWER:.*,\s(\d+.\d+).*\n.*\n'
        r'NDR_UPPER:.*,\s(\d+.\d+).*\n'
        r'PDR_LOWER:.*,\s(\d+.\d+).*\n.*\n'
        r'PDR_UPPER:.*,\s(\d+.\d+)'
    )
    REGEX_PERF_MSG_INFO = re.compile(
        r'NDR_LOWER:\s(\d+.\d+)\s.*\s(\d+.\d+)\s.*\n.*\n.*\n'
        r'PDR_LOWER:\s(\d+.\d+)\s.*\s(\d+.\d+)\s.*\n.*\n.*\n'
        r'Latency at 90% PDR:.*\[\'(.*)\', \'(.*)\'\].*\n'
        r'Latency at 50% PDR:.*\[\'(.*)\', \'(.*)\'\].*\n'
        r'Latency at 10% PDR:.*\[\'(.*)\', \'(.*)\'\].*\n'
    )
    REGEX_CPS_MSG_INFO = re.compile(
        r'NDR_LOWER:\s(\d+.\d+)\s.*\s.*\n.*\n.*\n'
        r'PDR_LOWER:\s(\d+.\d+)\s.*\s.*\n.*\n.*'
    )
    REGEX_PPS_MSG_INFO = re.compile(
        r'NDR_LOWER:\s(\d+.\d+)\s.*\s(\d+.\d+)\s.*\n.*\n.*\n'
        r'PDR_LOWER:\s(\d+.\d+)\s.*\s(\d+.\d+)\s.*\n.*\n.*'
    )
    REGEX_MRR_MSG_INFO = re.compile(r'.*\[(.*)\]')

    REGEX_VSAP_MSG_INFO = re.compile(
        r'Transfer Rate: (\d*.\d*).*\n'
        r'Latency: (\d*.\d*).*\n'
        r'Completed requests: (\d*).*\n'
        r'Failed requests: (\d*).*\n'
        r'Total data transferred: (\d*).*\n'
        r'Connection [cr]ps rate:\s*(\d*.\d*)'
    )

    # Needed for CPS and PPS tests
    REGEX_NDRPDR_LAT_BASE = re.compile(
        r'LATENCY.*\[\'(.*)\', \'(.*)\'\]\s\n.*\n.*\n'
        r'LATENCY.*\[\'(.*)\', \'(.*)\'\]'
    )
    REGEX_NDRPDR_LAT = re.compile(
        r'LATENCY.*\[\'(.*)\', \'(.*)\'\]\s\n.*\n.*\n'
        r'LATENCY.*\[\'(.*)\', \'(.*)\'\]\s\n.*\n'
        r'Latency.*\[\'(.*)\', \'(.*)\'\]\s\n'
        r'Latency.*\[\'(.*)\', \'(.*)\'\]\s\n'
        r'Latency.*\[\'(.*)\', \'(.*)\'\]\s\n'
        r'Latency.*\[\'(.*)\', \'(.*)\'\]'
    )

    REGEX_VERSION_VPP = re.compile(
        r"(return STDOUT Version:\s*|"
        r"VPP Version:\s*|VPP version:\s*)(.*)"
    )
    REGEX_VERSION_DPDK = re.compile(
        r"(DPDK version:\s*|DPDK Version:\s*)(.*)"
    )
    REGEX_TCP = re.compile(
        r'Total\s(rps|cps|throughput):\s(\d*).*$'
    )
    REGEX_MRR = re.compile(
        r'MaxReceivedRate_Results\s\[pkts/(\d*)sec\]:\s'
        r'tx\s(\d*),\srx\s(\d*)'
    )
    REGEX_BMRR = re.compile(
        r'.*trial results.*: \[(.*)\]'
    )
    REGEX_RECONF_LOSS = re.compile(
        r'Packets lost due to reconfig: (\d*)'
    )
    REGEX_RECONF_TIME = re.compile(
        r'Implied time lost: (\d*.[\de-]*)'
    )
    REGEX_TC_TAG = re.compile(r'\d+[tT]\d+[cC]')

    REGEX_TC_NAME_NEW = re.compile(r'-\d+[cC]-')

    REGEX_TC_NUMBER = re.compile(r'tc\d{2}-')

    REGEX_TC_PAPI_CLI = re.compile(r'.*\((\d+.\d+.\d+.\d+.) - (.*)\)')

    REGEX_SH_RUN_HOST = re.compile(
        r'hostname=\"(\d{1,3}.\d{1,3}.\d{1,3}.\d{1,3})\",hook=\"(.*)\"'
    )

    def __init__(self, metadata, mapping, ignore, for_output):
        """Initialisation.

        :param metadata: Key-value pairs to be included in "metadata" part of
            JSON structure.
        :param mapping: Mapping of the old names of test cases to the new
            (actual) one.
        :param ignore: List of TCs to be ignored.
        :param for_output: Output to be generated from downloaded data.
        :type metadata: dict
        :type mapping: dict
        :type ignore: list
        :type for_output: str
        """

        # Type of message to parse out from the test messages
        self._msg_type = None

        # VPP version
        self._version = None

        # Timestamp
        self._timestamp = None

        # Testbed. The testbed is identified by TG node IP address.
        self._testbed = None

        # Mapping of TCs long names
        self._mapping = mapping

        # Ignore list
        self._ignore = ignore

        self._for_output = for_output

        # Number of PAPI History messages found:
        # 0 - no message
        # 1 - PAPI History of DUT1
        # 2 - PAPI History of DUT2
        self._conf_history_lookup_nr = 0

        self._sh_run_counter = 0
        self._telemetry_kw_counter = 0
        self._telemetry_msg_counter = 0

        # Test ID of currently processed test- the lowercase full path to the
        # test
        self._test_id = None

        # The main data structure
        self._data = {
            u"metadata": OrderedDict(),
            u"suites": OrderedDict(),
            u"tests": OrderedDict()
        }

        # Save the provided metadata
        for key, val in metadata.items():
            self._data[u"metadata"][key] = val

        # Dictionary defining the methods used to parse different types of
        # messages
        self.parse_msg = {
            u"vpp-version": self._get_vpp_version,
            u"dpdk-version": self._get_dpdk_version,
            u"teardown-papi-history": self._get_papi_history,
            u"test-show-runtime": self._get_show_run,
            u"testbed": self._get_testbed,
            u"test-telemetry": self._get_telemetry
        }

    @property
    def data(self):
        """Getter - Data parsed from the XML file.

        :returns: Data parsed from the XML file.
        :rtype: dict
        """
        return self._data

    def _get_data_from_mrr_test_msg(self, msg):
        """Get info from message of MRR performance tests.

        :param msg: Message to be processed.
        :type msg: str
        :returns: Processed message or original message if a problem occurs.
        :rtype: str
        """

        groups = re.search(self.REGEX_MRR_MSG_INFO, msg)
        if not groups or groups.lastindex != 1:
            return u"Test Failed."

        try:
            data = groups.group(1).split(u", ")
        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

        out_str = u"["
        try:
            for item in data:
                out_str += f"{(float(item) / 1e6):.2f}, "
            return out_str[:-2] + u"]"
        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

    def _get_data_from_cps_test_msg(self, msg):
        """Get info from message of NDRPDR CPS tests.

        :param msg: Message to be processed.
        :type msg: str
        :returns: Processed message or "Test Failed." if a problem occurs.
        :rtype: str
        """

        groups = re.search(self.REGEX_CPS_MSG_INFO, msg)
        if not groups or groups.lastindex != 2:
            return u"Test Failed."

        try:
            return (
                f"1. {(float(groups.group(1)) / 1e6):5.2f}\n"
                f"2. {(float(groups.group(2)) / 1e6):5.2f}"
            )
        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

    def _get_data_from_pps_test_msg(self, msg):
        """Get info from message of NDRPDR PPS tests.

        :param msg: Message to be processed.
        :type msg: str
        :returns: Processed message or "Test Failed." if a problem occurs.
        :rtype: str
        """

        groups = re.search(self.REGEX_PPS_MSG_INFO, msg)
        if not groups or groups.lastindex != 4:
            return u"Test Failed."

        try:
            return (
                f"1. {(float(groups.group(1)) / 1e6):5.2f}      "
                f"{float(groups.group(2)):5.2f}\n"
                f"2. {(float(groups.group(3)) / 1e6):5.2f}      "
                f"{float(groups.group(4)):5.2f}"
            )
        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

    def _get_data_from_perf_test_msg(self, msg):
        """Get info from message of NDRPDR performance tests.

        :param msg: Message to be processed.
        :type msg: str
        :returns: Processed message or "Test Failed." if a problem occurs.
        :rtype: str
        """

        groups = re.search(self.REGEX_PERF_MSG_INFO, msg)
        if not groups or groups.lastindex != 10:
            return u"Test Failed."

        try:
            data = {
                u"ndr_low": float(groups.group(1)),
                u"ndr_low_b": float(groups.group(2)),
                u"pdr_low": float(groups.group(3)),
                u"pdr_low_b": float(groups.group(4)),
                u"pdr_lat_90_1": groups.group(5),
                u"pdr_lat_90_2": groups.group(6),
                u"pdr_lat_50_1": groups.group(7),
                u"pdr_lat_50_2": groups.group(8),
                u"pdr_lat_10_1": groups.group(9),
                u"pdr_lat_10_2": groups.group(10),
            }
        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

        def _process_lat(in_str_1, in_str_2):
            """Extract min, avg, max values from latency string.

            :param in_str_1: Latency string for one direction produced by robot
                framework.
            :param in_str_2: Latency string for second direction produced by
                robot framework.
            :type in_str_1: str
            :type in_str_2: str
            :returns: Processed latency string or None if a problem occurs.
            :rtype: tuple
            """
            in_list_1 = in_str_1.split('/', 3)
            in_list_2 = in_str_2.split('/', 3)

            if len(in_list_1) != 4 and len(in_list_2) != 4:
                return None

            in_list_1[3] += u"=" * (len(in_list_1[3]) % 4)
            try:
                hdr_lat_1 = hdrh.histogram.HdrHistogram.decode(in_list_1[3])
            except hdrh.codec.HdrLengthException:
                return None

            in_list_2[3] += u"=" * (len(in_list_2[3]) % 4)
            try:
                hdr_lat_2 = hdrh.histogram.HdrHistogram.decode(in_list_2[3])
            except hdrh.codec.HdrLengthException:
                return None

            if hdr_lat_1 and hdr_lat_2:
                hdr_lat = (
                    hdr_lat_1.get_value_at_percentile(50.0),
                    hdr_lat_1.get_value_at_percentile(90.0),
                    hdr_lat_1.get_value_at_percentile(99.0),
                    hdr_lat_2.get_value_at_percentile(50.0),
                    hdr_lat_2.get_value_at_percentile(90.0),
                    hdr_lat_2.get_value_at_percentile(99.0)
                )

                if all(hdr_lat):
                    return hdr_lat

            return None

        try:
            out_msg = (
                f"1. {(data[u'ndr_low'] / 1e6):5.2f}      "
                f"{data[u'ndr_low_b']:5.2f}"
                f"\n2. {(data[u'pdr_low'] / 1e6):5.2f}      "
                f"{data[u'pdr_low_b']:5.2f}"
            )
            latency = (
                _process_lat(data[u'pdr_lat_10_1'], data[u'pdr_lat_10_2']),
                _process_lat(data[u'pdr_lat_50_1'], data[u'pdr_lat_50_2']),
                _process_lat(data[u'pdr_lat_90_1'], data[u'pdr_lat_90_2'])
            )
            if all(latency):
                max_len = len(str(max((max(item) for item in latency))))
                max_len = 4 if max_len < 4 else max_len

                for idx, lat in enumerate(latency):
                    if not idx:
                        out_msg += u"\n"
                    out_msg += (
                        f"\n{idx + 3}. "
                        f"{lat[0]:{max_len}d} "
                        f"{lat[1]:{max_len}d} "
                        f"{lat[2]:{max_len}d}      "
                        f"{lat[3]:{max_len}d} "
                        f"{lat[4]:{max_len}d} "
                        f"{lat[5]:{max_len}d} "
                    )

            return out_msg

        except (AttributeError, IndexError, ValueError, KeyError):
            return u"Test Failed."

    def _get_testbed(self, msg):
        """Called when extraction of testbed IP is required.
        The testbed is identified by TG node IP address.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """

        if msg.message.count(u"Setup of TG node") or \
                msg.message.count(u"Setup of node TG host"):
            reg_tg_ip = re.compile(
                r'.*TG .* (\d{1,3}.\d{1,3}.\d{1,3}.\d{1,3}).*')
            try:
                self._testbed = str(re.search(reg_tg_ip, msg.message).group(1))
            except (KeyError, ValueError, IndexError, AttributeError):
                pass
            finally:
                self._data[u"metadata"][u"testbed"] = self._testbed
                self._msg_type = None

    def _get_vpp_version(self, msg):
        """Called when extraction of VPP version is required.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """

        if msg.message.count(u"return STDOUT Version:") or \
                msg.message.count(u"VPP Version:") or \
                msg.message.count(u"VPP version:"):
            self._version = str(
                re.search(self.REGEX_VERSION_VPP, msg.message).group(2)
            )
            self._data[u"metadata"][u"version"] = self._version
            self._msg_type = None

    def _get_dpdk_version(self, msg):
        """Called when extraction of DPDK version is required.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """

        if msg.message.count(u"DPDK Version:"):
            try:
                self._version = str(re.search(
                    self.REGEX_VERSION_DPDK, msg.message).group(2))
                self._data[u"metadata"][u"version"] = self._version
            except IndexError:
                pass
            finally:
                self._msg_type = None

    def _get_papi_history(self, msg):
        """Called when extraction of PAPI command history is required.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """
        if msg.message.count(u"PAPI command history:"):
            self._conf_history_lookup_nr += 1
            if self._conf_history_lookup_nr == 1:
                self._data[u"tests"][self._test_id][u"conf-history"] = str()
            else:
                self._msg_type = None
            text = re.sub(
                r"\d{1,3}.\d{1,3}.\d{1,3}.\d{1,3} PAPI command history:",
                u"",
                msg.message,
                count=1
            ).replace(u'"', u"'")
            self._data[u"tests"][self._test_id][u"conf-history"] += (
                f"**DUT{str(self._conf_history_lookup_nr)}:** {text}"
            )

    def _get_show_run(self, msg):
        """Called when extraction of VPP operational data (output of CLI command
        Show Runtime) is required.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """

        if not msg.message.count(u"stats runtime"):
            return

        # Temporary solution
        if self._sh_run_counter > 1:
            return

        if u"show-run" not in self._data[u"tests"][self._test_id].keys():
            self._data[u"tests"][self._test_id][u"show-run"] = dict()

        groups = re.search(self.REGEX_TC_PAPI_CLI, msg.message)
        if not groups:
            return
        try:
            host = groups.group(1)
        except (AttributeError, IndexError):
            host = u""
        try:
            sock = groups.group(2)
        except (AttributeError, IndexError):
            sock = u""

        dut = u"dut{nr}".format(
            nr=len(self._data[u'tests'][self._test_id][u'show-run'].keys()) + 1)

        self._data[u'tests'][self._test_id][u'show-run'][dut] = \
            copy.copy(
                {
                    u"host": host,
                    u"socket": sock,
                    u"runtime": str(msg.message).replace(u' ', u'').
                                replace(u'\n', u'').replace(u"'", u'"').
                                replace(u'b"', u'"').replace(u'u"', u'"').
                                split(u":", 1)[1]
                }
            )

    def _get_telemetry(self, msg):
        """Called when extraction of VPP telemetry data is required.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """

        if self._telemetry_kw_counter > 1:
            return
        if not msg.message.count(u"# TYPE vpp_runtime_calls"):
            return

        if u"telemetry-show-run" not in \
                self._data[u"tests"][self._test_id].keys():
            self._data[u"tests"][self._test_id][u"telemetry-show-run"] = dict()

        self._telemetry_msg_counter += 1
        groups = re.search(self.REGEX_SH_RUN_HOST, msg.message)
        if not groups:
            return
        try:
            host = groups.group(1)
        except (AttributeError, IndexError):
            host = u""
        try:
            sock = groups.group(2)
        except (AttributeError, IndexError):
            sock = u""
        runtime = {
            u"source_type": u"node",
            u"source_id": host,
            u"msg_type": u"metric",
            u"log_level": u"INFO",
            u"timestamp": msg.timestamp,
            u"msg": u"show_runtime",
            u"host": host,
            u"socket": sock,
            u"data": list()
        }
        for line in msg.message.splitlines():
            if not line.startswith(u"vpp_runtime_"):
                continue
            try:
                params, value, timestamp = line.rsplit(u" ", maxsplit=2)
                cut = params.index(u"{")
                name = params[:cut].split(u"_", maxsplit=2)[-1]
                labels = eval(
                    u"dict" + params[cut:].replace('{', '(').replace('}', ')')
                )
                labels[u"graph_node"] = labels.pop(u"name")
                runtime[u"data"].append(
                    {
                        u"name": name,
                        u"value": value,
                        u"timestamp": timestamp,
                        u"labels": labels
                    }
                )
            except (TypeError, ValueError, IndexError):
                continue
        self._data[u'tests'][self._test_id][u'telemetry-show-run']\
            [f"dut{self._telemetry_msg_counter}"] = copy.copy(
                {
                    u"host": host,
                    u"socket": sock,
                    u"runtime": runtime
                }
            )

    def _get_ndrpdr_throughput(self, msg):
        """Get NDR_LOWER, NDR_UPPER, PDR_LOWER and PDR_UPPER from the test
        message.

        :param msg: The test message to be parsed.
        :type msg: str
        :returns: Parsed data as a dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """

        throughput = {
            u"NDR": {u"LOWER": -1.0, u"UPPER": -1.0},
            u"PDR": {u"LOWER": -1.0, u"UPPER": -1.0}
        }
        status = u"FAIL"
        groups = re.search(self.REGEX_NDRPDR_RATE, msg)

        if groups is not None:
            try:
                throughput[u"NDR"][u"LOWER"] = float(groups.group(1))
                throughput[u"NDR"][u"UPPER"] = float(groups.group(2))
                throughput[u"PDR"][u"LOWER"] = float(groups.group(3))
                throughput[u"PDR"][u"UPPER"] = float(groups.group(4))
                status = u"PASS"
            except (IndexError, ValueError):
                pass

        return throughput, status

    def _get_ndrpdr_throughput_gbps(self, msg):
        """Get NDR_LOWER, NDR_UPPER, PDR_LOWER and PDR_UPPER in Gbps from the
        test message.

        :param msg: The test message to be parsed.
        :type msg: str
        :returns: Parsed data as a dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """

        gbps = {
            u"NDR": {u"LOWER": -1.0, u"UPPER": -1.0},
            u"PDR": {u"LOWER": -1.0, u"UPPER": -1.0}
        }
        status = u"FAIL"
        groups = re.search(self.REGEX_NDRPDR_GBPS, msg)

        if groups is not None:
            try:
                gbps[u"NDR"][u"LOWER"] = float(groups.group(1))
                gbps[u"NDR"][u"UPPER"] = float(groups.group(2))
                gbps[u"PDR"][u"LOWER"] = float(groups.group(3))
                gbps[u"PDR"][u"UPPER"] = float(groups.group(4))
                status = u"PASS"
            except (IndexError, ValueError):
                pass

        return gbps, status

    def _get_plr_throughput(self, msg):
        """Get PLRsearch lower bound and PLRsearch upper bound from the test
        message.

        :param msg: The test message to be parsed.
        :type msg: str
        :returns: Parsed data as a dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """

        throughput = {
            u"LOWER": -1.0,
            u"UPPER": -1.0
        }
        status = u"FAIL"
        groups = re.search(self.REGEX_PLR_RATE, msg)

        if groups is not None:
            try:
                throughput[u"LOWER"] = float(groups.group(1))
                throughput[u"UPPER"] = float(groups.group(2))
                status = u"PASS"
            except (IndexError, ValueError):
                pass

        return throughput, status

    def _get_ndrpdr_latency(self, msg):
        """Get LATENCY from the test message.

        :param msg: The test message to be parsed.
        :type msg: str
        :returns: Parsed data as a dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """
        latency_default = {
            u"min": -1.0,
            u"avg": -1.0,
            u"max": -1.0,
            u"hdrh": u""
        }
        latency = {
            u"NDR": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
            u"PDR": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
            u"LAT0": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
            u"PDR10": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
            u"PDR50": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
            u"PDR90": {
                u"direction1": copy.copy(latency_default),
                u"direction2": copy.copy(latency_default)
            },
        }

        groups = re.search(self.REGEX_NDRPDR_LAT, msg)
        if groups is None:
            groups = re.search(self.REGEX_NDRPDR_LAT_BASE, msg)
        if groups is None:
            return latency, u"FAIL"

        def process_latency(in_str):
            """Return object with parsed latency values.

            TODO: Define class for the return type.

            :param in_str: Input string, min/avg/max/hdrh format.
            :type in_str: str
            :returns: Dict with corresponding keys, except hdrh float values.
            :rtype dict:
            :throws IndexError: If in_str does not have enough substrings.
            :throws ValueError: If a substring does not convert to float.
            """
            in_list = in_str.split('/', 3)

            rval = {
                u"min": float(in_list[0]),
                u"avg": float(in_list[1]),
                u"max": float(in_list[2]),
                u"hdrh": u""
            }

            if len(in_list) == 4:
                rval[u"hdrh"] = str(in_list[3])

            return rval

        try:
            latency[u"NDR"][u"direction1"] = process_latency(groups.group(1))
            latency[u"NDR"][u"direction2"] = process_latency(groups.group(2))
            latency[u"PDR"][u"direction1"] = process_latency(groups.group(3))
            latency[u"PDR"][u"direction2"] = process_latency(groups.group(4))
            if groups.lastindex == 4:
                return latency, u"PASS"
        except (IndexError, ValueError):
            pass

        try:
            latency[u"PDR90"][u"direction1"] = process_latency(groups.group(5))
            latency[u"PDR90"][u"direction2"] = process_latency(groups.group(6))
            latency[u"PDR50"][u"direction1"] = process_latency(groups.group(7))
            latency[u"PDR50"][u"direction2"] = process_latency(groups.group(8))
            latency[u"PDR10"][u"direction1"] = process_latency(groups.group(9))
            latency[u"PDR10"][u"direction2"] = process_latency(groups.group(10))
            latency[u"LAT0"][u"direction1"] = process_latency(groups.group(11))
            latency[u"LAT0"][u"direction2"] = process_latency(groups.group(12))
            if groups.lastindex == 12:
                return latency, u"PASS"
        except (IndexError, ValueError):
            pass

        return latency, u"FAIL"

    @staticmethod
    def _get_hoststack_data(msg, tags):
        """Get data from the hoststack test message.

        :param msg: The test message to be parsed.
        :param tags: Test tags.
        :type msg: str
        :type tags: list
        :returns: Parsed data as a JSON dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """
        result = dict()
        status = u"FAIL"

        msg = msg.replace(u"'", u'"').replace(u" ", u"")
        if u"LDPRELOAD" in tags:
            try:
                result = loads(msg)
                status = u"PASS"
            except JSONDecodeError:
                pass
        elif u"VPPECHO" in tags:
            try:
                msg_lst = msg.replace(u"}{", u"} {").split(u" ")
                result = dict(
                    client=loads(msg_lst[0]),
                    server=loads(msg_lst[1])
                )
                status = u"PASS"
            except (JSONDecodeError, IndexError):
                pass

        return result, status

    def _get_vsap_data(self, msg, tags):
        """Get data from the vsap test message.

        :param msg: The test message to be parsed.
        :param tags: Test tags.
        :type msg: str
        :type tags: list
        :returns: Parsed data as a JSON dict and the status (PASS/FAIL).
        :rtype: tuple(dict, str)
        """
        result = dict()
        status = u"FAIL"

        groups = re.search(self.REGEX_VSAP_MSG_INFO, msg)
        if groups is not None:
            try:
                result[u"transfer-rate"] = float(groups.group(1)) * 1e3
                result[u"latency"] = float(groups.group(2))
                result[u"completed-requests"] = int(groups.group(3))
                result[u"failed-requests"] = int(groups.group(4))
                result[u"bytes-transferred"] = int(groups.group(5))
                if u"TCP_CPS"in tags:
                    result[u"cps"] = float(groups.group(6))
                elif u"TCP_RPS" in tags:
                    result[u"rps"] = float(groups.group(6))
                else:
                    return result, status
                status = u"PASS"
            except (IndexError, ValueError):
                pass

        return result, status

    def visit_suite(self, suite):
        """Implements traversing through the suite and its direct children.

        :param suite: Suite to process.
        :type suite: Suite
        :returns: Nothing.
        """
        if self.start_suite(suite) is not False:
            suite.suites.visit(self)
            suite.tests.visit(self)
            self.end_suite(suite)

    def start_suite(self, suite):
        """Called when suite starts.

        :param suite: Suite to process.
        :type suite: Suite
        :returns: Nothing.
        """

        try:
            parent_name = suite.parent.name
        except AttributeError:
            return

        self._data[u"suites"][suite.longname.lower().
                              replace(u'"', u"'").
                              replace(u" ", u"_")] = {
                                  u"name": suite.name.lower(),
                                  u"doc": suite.doc,
                                  u"parent": parent_name,
                                  u"level": len(suite.longname.split(u"."))
                              }

        suite.keywords.visit(self)

    def end_suite(self, suite):
        """Called when suite ends.

        :param suite: Suite to process.
        :type suite: Suite
        :returns: Nothing.
        """

    def visit_test(self, test):
        """Implements traversing through the test.

        :param test: Test to process.
        :type test: Test
        :returns: Nothing.
        """
        if self.start_test(test) is not False:
            test.keywords.visit(self)
            self.end_test(test)

    def start_test(self, test):
        """Called when test starts.

        :param test: Test to process.
        :type test: Test
        :returns: Nothing.
        """

        self._sh_run_counter = 0
        self._telemetry_kw_counter = 0
        self._telemetry_msg_counter = 0

        longname_orig = test.longname.lower()

        # Check the ignore list
        if longname_orig in self._ignore:
            return

        tags = [str(tag) for tag in test.tags]
        test_result = dict()

        # Change the TC long name and name if defined in the mapping table
        longname = self._mapping.get(longname_orig, None)
        if longname is not None:
            name = longname.split(u'.')[-1]
            logging.debug(
                f"{self._data[u'metadata']}\n{longname_orig}\n{longname}\n"
                f"{name}"
            )
        else:
            longname = longname_orig
            name = test.name.lower()

        # Remove TC number from the TC long name (backward compatibility):
        self._test_id = re.sub(self.REGEX_TC_NUMBER, u"", longname)
        # Remove TC number from the TC name (not needed):
        test_result[u"name"] = re.sub(self.REGEX_TC_NUMBER, "", name)

        test_result[u"parent"] = test.parent.name.lower()
        test_result[u"tags"] = tags
        test_result["doc"] = test.doc
        test_result[u"type"] = u""
        test_result[u"status"] = test.status
        test_result[u"starttime"] = test.starttime
        test_result[u"endtime"] = test.endtime

        if test.status == u"PASS":
            if u"NDRPDR" in tags:
                if u"TCP_PPS" in tags or u"UDP_PPS" in tags:
                    test_result[u"msg"] = self._get_data_from_pps_test_msg(
                        test.message)
                elif u"TCP_CPS" in tags or u"UDP_CPS" in tags:
                    test_result[u"msg"] = self._get_data_from_cps_test_msg(
                        test.message)
                else:
                    test_result[u"msg"] = self._get_data_from_perf_test_msg(
                        test.message)
            elif u"MRR" in tags or u"FRMOBL" in tags or u"BMRR" in tags:
                test_result[u"msg"] = self._get_data_from_mrr_test_msg(
                    test.message)
            else:
                test_result[u"msg"] = test.message
        else:
            test_result[u"msg"] = test.message

        if u"PERFTEST" in tags and u"TREX" not in tags:
            # Replace info about cores (e.g. -1c-) with the info about threads
            # and cores (e.g. -1t1c-) in the long test case names and in the
            # test case names if necessary.
            tag_count = 0
            tag_tc = str()
            for tag in test_result[u"tags"]:
                groups = re.search(self.REGEX_TC_TAG, tag)
                if groups:
                    tag_count += 1
                    tag_tc = tag

            if tag_count == 1:
                self._test_id = re.sub(
                    self.REGEX_TC_NAME_NEW, f"-{tag_tc.lower()}-",
                    self._test_id, count=1
                )
                test_result[u"name"] = re.sub(
                    self.REGEX_TC_NAME_NEW, f"-{tag_tc.lower()}-",
                    test_result["name"], count=1
                )
            else:
                test_result[u"status"] = u"FAIL"
                self._data[u"tests"][self._test_id] = test_result
                logging.debug(
                    f"The test {self._test_id} has no or more than one "
                    f"multi-threading tags.\n"
                    f"Tags: {test_result[u'tags']}"
                )
                return

        if u"DEVICETEST" in tags:
            test_result[u"type"] = u"DEVICETEST"
        elif u"NDRPDR" in tags:
            if u"TCP_CPS" in tags or u"UDP_CPS" in tags:
                test_result[u"type"] = u"CPS"
            else:
                test_result[u"type"] = u"NDRPDR"
            if test.status == u"PASS":
                test_result[u"throughput"], test_result[u"status"] = \
                    self._get_ndrpdr_throughput(test.message)
                test_result[u"gbps"], test_result[u"status"] = \
                    self._get_ndrpdr_throughput_gbps(test.message)
                test_result[u"latency"], test_result[u"status"] = \
                    self._get_ndrpdr_latency(test.message)
        elif u"MRR" in tags or u"FRMOBL" in tags or u"BMRR" in tags:
            if u"MRR" in tags:
                test_result[u"type"] = u"MRR"
            else:
                test_result[u"type"] = u"BMRR"
            if test.status == u"PASS":
                test_result[u"result"] = dict()
                groups = re.search(self.REGEX_BMRR, test.message)
                if groups is not None:
                    items_str = groups.group(1)
                    items_float = [
                        float(item.strip().replace(u"'", u""))
                        for item in items_str.split(",")
                    ]
                    # Use whole list in CSIT-1180.
                    stats = jumpavg.AvgStdevStats.for_runs(items_float)
                    test_result[u"result"][u"samples"] = items_float
                    test_result[u"result"][u"receive-rate"] = stats.avg
                    test_result[u"result"][u"receive-stdev"] = stats.stdev
                else:
                    groups = re.search(self.REGEX_MRR, test.message)
                    test_result[u"result"][u"receive-rate"] = \
                        float(groups.group(3)) / float(groups.group(1))
        elif u"SOAK" in tags:
            test_result[u"type"] = u"SOAK"
            if test.status == u"PASS":
                test_result[u"throughput"], test_result[u"status"] = \
                    self._get_plr_throughput(test.message)
        elif u"HOSTSTACK" in tags:
            test_result[u"type"] = u"HOSTSTACK"
            if test.status == u"PASS":
                test_result[u"result"], test_result[u"status"] = \
                    self._get_hoststack_data(test.message, tags)
        elif u"LDP_NGINX" in tags:
            test_result[u"type"] = u"LDP_NGINX"
            test_result[u"result"], test_result[u"status"] = \
                self._get_vsap_data(test.message, tags)
        # elif u"TCP" in tags:  # This might be not used
        #     test_result[u"type"] = u"TCP"
        #     if test.status == u"PASS":
        #         groups = re.search(self.REGEX_TCP, test.message)
        #         test_result[u"result"] = int(groups.group(2))
        elif u"RECONF" in tags:
            test_result[u"type"] = u"RECONF"
            if test.status == u"PASS":
                test_result[u"result"] = None
                try:
                    grps_loss = re.search(self.REGEX_RECONF_LOSS, test.message)
                    grps_time = re.search(self.REGEX_RECONF_TIME, test.message)
                    test_result[u"result"] = {
                        u"loss": int(grps_loss.group(1)),
                        u"time": float(grps_time.group(1))
                    }
                except (AttributeError, IndexError, ValueError, TypeError):
                    test_result[u"status"] = u"FAIL"
        else:
            test_result[u"status"] = u"FAIL"

        self._data[u"tests"][self._test_id] = test_result

    def end_test(self, test):
        """Called when test ends.

        :param test: Test to process.
        :type test: Test
        :returns: Nothing.
        """

    def visit_keyword(self, keyword):
        """Implements traversing through the keyword and its child keywords.

        :param keyword: Keyword to process.
        :type keyword: Keyword
        :returns: Nothing.
        """
        if self.start_keyword(keyword) is not False:
            self.end_keyword(keyword)

    def start_keyword(self, keyword):
        """Called when keyword starts. Default implementation does nothing.

        :param keyword: Keyword to process.
        :type keyword: Keyword
        :returns: Nothing.
        """
        try:
            if keyword.type == u"setup":
                self.visit_setup_kw(keyword)
            elif keyword.type == u"teardown":
                self.visit_teardown_kw(keyword)
            else:
                self.visit_test_kw(keyword)
        except AttributeError:
            pass

    def end_keyword(self, keyword):
        """Called when keyword ends. Default implementation does nothing.

        :param keyword: Keyword to process.
        :type keyword: Keyword
        :returns: Nothing.
        """

    def visit_test_kw(self, test_kw):
        """Implements traversing through the test keyword and its child
        keywords.

        :param test_kw: Keyword to process.
        :type test_kw: Keyword
        :returns: Nothing.
        """
        for keyword in test_kw.keywords:
            if self.start_test_kw(keyword) is not False:
                self.visit_test_kw(keyword)
                self.end_test_kw(keyword)

    def start_test_kw(self, test_kw):
        """Called when test keyword starts. Default implementation does
        nothing.

        :param test_kw: Keyword to process.
        :type test_kw: Keyword
        :returns: Nothing.
        """
        if self._for_output == u"trending":
            return

        if test_kw.name.count(u"Run Telemetry On All Duts"):
            self._msg_type = u"test-telemetry"
            self._telemetry_kw_counter += 1
        elif test_kw.name.count(u"Show Runtime On All Duts"):
            self._msg_type = u"test-show-runtime"
            self._sh_run_counter += 1
        else:
            return
        test_kw.messages.visit(self)

    def end_test_kw(self, test_kw):
        """Called when keyword ends. Default implementation does nothing.

        :param test_kw: Keyword to process.
        :type test_kw: Keyword
        :returns: Nothing.
        """

    def visit_setup_kw(self, setup_kw):
        """Implements traversing through the teardown keyword and its child
        keywords.

        :param setup_kw: Keyword to process.
        :type setup_kw: Keyword
        :returns: Nothing.
        """
        for keyword in setup_kw.keywords:
            if self.start_setup_kw(keyword) is not False:
                self.visit_setup_kw(keyword)
                self.end_setup_kw(keyword)

    def start_setup_kw(self, setup_kw):
        """Called when teardown keyword starts. Default implementation does
        nothing.

        :param setup_kw: Keyword to process.
        :type setup_kw: Keyword
        :returns: Nothing.
        """
        if setup_kw.name.count(u"Show Vpp Version On All Duts") \
                and not self._version:
            self._msg_type = u"vpp-version"
        elif setup_kw.name.count(u"Install Dpdk Framework On All Duts") and \
                not self._version:
            self._msg_type = u"dpdk-version"
        elif setup_kw.name.count(u"Setup Framework") and not self._testbed:
            self._msg_type = u"testbed"
        else:
            return
        setup_kw.messages.visit(self)

    def end_setup_kw(self, setup_kw):
        """Called when keyword ends. Default implementation does nothing.

        :param setup_kw: Keyword to process.
        :type setup_kw: Keyword
        :returns: Nothing.
        """

    def visit_teardown_kw(self, teardown_kw):
        """Implements traversing through the teardown keyword and its child
        keywords.

        :param teardown_kw: Keyword to process.
        :type teardown_kw: Keyword
        :returns: Nothing.
        """
        for keyword in teardown_kw.keywords:
            if self.start_teardown_kw(keyword) is not False:
                self.visit_teardown_kw(keyword)
                self.end_teardown_kw(keyword)

    def start_teardown_kw(self, teardown_kw):
        """Called when teardown keyword starts

        :param teardown_kw: Keyword to process.
        :type teardown_kw: Keyword
        :returns: Nothing.
        """
        if teardown_kw.name.count(u"Show Papi History On All Duts"):
            self._conf_history_lookup_nr = 0
            self._msg_type = u"teardown-papi-history"
            teardown_kw.messages.visit(self)

    def end_teardown_kw(self, teardown_kw):
        """Called when keyword ends. Default implementation does nothing.

        :param teardown_kw: Keyword to process.
        :type teardown_kw: Keyword
        :returns: Nothing.
        """

    def visit_message(self, msg):
        """Implements visiting the message.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """
        if self.start_message(msg) is not False:
            self.end_message(msg)

    def start_message(self, msg):
        """Called when message starts. Get required information from messages:
        - VPP version.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """
        if self._msg_type:
            self.parse_msg[self._msg_type](msg)

    def end_message(self, msg):
        """Called when message ends. Default implementation does nothing.

        :param msg: Message to process.
        :type msg: Message
        :returns: Nothing.
        """


class InputData:
    """Input data

    The data is extracted from output.xml files generated by Jenkins jobs and
    stored in pandas' DataFrames.

    The data structure:
    - job name
      - build number
        - metadata
          (as described in ExecutionChecker documentation)
        - suites
          (as described in ExecutionChecker documentation)
        - tests
          (as described in ExecutionChecker documentation)
    """

    def __init__(self, spec, for_output):
        """Initialization.

        :param spec: Specification.
        :param for_output: Output to be generated from downloaded data.
        :type spec: Specification
        :type for_output: str
        """

        # Specification:
        self._cfg = spec

        self._for_output = for_output

        # Data store:
        self._input_data = pd.Series()

    @property
    def data(self):
        """Getter - Input data.

        :returns: Input data
        :rtype: pandas.Series
        """
        return self._input_data

    def metadata(self, job, build):
        """Getter - metadata

        :param job: Job which metadata we want.
        :param build: Build which metadata we want.
        :type job: str
        :type build: str
        :returns: Metadata
        :rtype: pandas.Series
        """
        return self.data[job][build][u"metadata"]

    def suites(self, job, build):
        """Getter - suites

        :param job: Job which suites we want.
        :param build: Build which suites we want.
        :type job: str
        :type build: str
        :returns: Suites.
        :rtype: pandas.Series
        """
        return self.data[job][str(build)][u"suites"]

    def tests(self, job, build):
        """Getter - tests

        :param job: Job which tests we want.
        :param build: Build which tests we want.
        :type job: str
        :type build: str
        :returns: Tests.
        :rtype: pandas.Series
        """
        return self.data[job][build][u"tests"]

    def _parse_tests(self, job, build):
        """Process data from robot output.xml file and return JSON structured
        data.

        :param job: The name of job which build output data will be processed.
        :param build: The build which output data will be processed.
        :type job: str
        :type build: dict
        :returns: JSON data structure.
        :rtype: dict
        """

        metadata = {
            u"job": job,
            u"build": build
        }

        with open(build[u"file-name"], u'r') as data_file:
            try:
                result = ExecutionResult(data_file)
            except errors.DataError as err:
                logging.error(
                    f"Error occurred while parsing output.xml: {repr(err)}"
                )
                return None
        checker = ExecutionChecker(
            metadata, self._cfg.mapping, self._cfg.ignore, self._for_output
        )
        result.visit(checker)

        checker.data[u"metadata"][u"tests_total"] = \
            result.statistics.total.all.total
        checker.data[u"metadata"][u"tests_passed"] = \
            result.statistics.total.all.passed
        checker.data[u"metadata"][u"tests_failed"] = \
            result.statistics.total.all.failed
        checker.data[u"metadata"][u"elapsedtime"] = result.suite.elapsedtime
        checker.data[u"metadata"][u"generated"] = result.suite.endtime[:14]

        return checker.data

    def _download_and_parse_build(self, job, build, repeat, pid=10000):
        """Download and parse the input data file.

        :param pid: PID of the process executing this method.
        :param job: Name of the Jenkins job which generated the processed input
            file.
        :param build: Information about the Jenkins build which generated the
            processed input file.
        :param repeat: Repeat the download specified number of times if not
            successful.
        :type pid: int
        :type job: str
        :type build: dict
        :type repeat: int
        """

        logging.info(f"Processing the job/build: {job}: {build[u'build']}")

        state = u"failed"
        success = False
        data = None
        do_repeat = repeat
        while do_repeat:
            success = download_and_unzip_data_file(self._cfg, job, build, pid)
            if success:
                break
            do_repeat -= 1
        if not success:
            logging.error(
                f"It is not possible to download the input data file from the "
                f"job {job}, build {build[u'build']}, or it is damaged. "
                f"Skipped."
            )
        if success:
            logging.info(f"  Processing data from build {build[u'build']}")
            data = self._parse_tests(job, build)
            if data is None:
                logging.error(
                    f"Input data file from the job {job}, build "
                    f"{build[u'build']} is damaged. Skipped."
                )
            else:
                state = u"processed"

            try:
                remove(build[u"file-name"])
            except OSError as err:
                logging.error(
                    f"Cannot remove the file {build[u'file-name']}: {repr(err)}"
                )

        # If the time-period is defined in the specification file, remove all
        # files which are outside the time period.
        is_last = False
        timeperiod = self._cfg.environment.get(u"time-period", None)
        if timeperiod and data:
            now = dt.utcnow()
            timeperiod = timedelta(int(timeperiod))
            metadata = data.get(u"metadata", None)
            if metadata:
                generated = metadata.get(u"generated", None)
                if generated:
                    generated = dt.strptime(generated, u"%Y%m%d %H:%M")
                    if (now - generated) > timeperiod:
                        # Remove the data and the file:
                        state = u"removed"
                        data = None
                        is_last = True
                        logging.info(
                            f"  The build {job}/{build[u'build']} is "
                            f"outdated, will be removed."
                        )
        return {
            u"data": data,
            u"state": state,
            u"job": job,
            u"build": build,
            u"last": is_last
        }

    def download_and_parse_data(self, repeat=1):
        """Download the input data files, parse input data from input files and
        store in pandas' Series.

        :param repeat: Repeat the download specified number of times if not
            successful.
        :type repeat: int
        """

        logging.info(u"Downloading and parsing input files ...")

        for job, builds in self._cfg.input.items():
            for build in builds:

                result = self._download_and_parse_build(job, build, repeat)
                if result[u"last"]:
                    break
                build_nr = result[u"build"][u"build"]

                if result[u"data"]:
                    data = result[u"data"]
                    build_data = pd.Series({
                        u"metadata": pd.Series(
                            list(data[u"metadata"].values()),
                            index=list(data[u"metadata"].keys())
                        ),
                        u"suites": pd.Series(
                            list(data[u"suites"].values()),
                            index=list(data[u"suites"].keys())
                        ),
                        u"tests": pd.Series(
                            list(data[u"tests"].values()),
                            index=list(data[u"tests"].keys())
                        )
                    })

                    if self._input_data.get(job, None) is None:
                        self._input_data[job] = pd.Series()
                    self._input_data[job][str(build_nr)] = build_data
                    self._cfg.set_input_file_name(
                        job, build_nr, result[u"build"][u"file-name"]
                    )
                self._cfg.set_input_state(job, build_nr, result[u"state"])

                mem_alloc = \
                    resource.getrusage(resource.RUSAGE_SELF).ru_maxrss / 1000
                logging.info(f"Memory allocation: {mem_alloc:.0f}MB")

        logging.info(u"Done.")

        msg = f"Successful downloads from the sources:\n"
        for source in self._cfg.environment[u"data-sources"]:
            if source[u"successful-downloads"]:
                msg += (
                    f"{source[u'url']}/{source[u'path']}/"
                    f"{source[u'file-name']}: "
                    f"{source[u'successful-downloads']}\n"
                )
        logging.info(msg)

    def process_local_file(self, local_file, job=u"local", build_nr=1,
                           replace=True):
        """Process local XML file given as a command-line parameter.

        :param local_file: The file to process.
        :param job: Job name.
        :param build_nr: Build number.
        :param replace: If True, the information about jobs and builds is
            replaced by the new one, otherwise the new jobs and builds are
            added.
        :type local_file: str
        :type job: str
        :type build_nr: int
        :type replace: bool
        :raises: PresentationError if an error occurs.
        """
        if not isfile(local_file):
            raise PresentationError(f"The file {local_file} does not exist.")

        try:
            build_nr = int(local_file.split(u"/")[-1].split(u".")[0])
        except (IndexError, ValueError):
            pass

        build = {
            u"build": build_nr,
            u"status": u"failed",
            u"file-name": local_file
        }
        if replace:
            self._cfg.input = dict()
        self._cfg.add_build(job, build)

        logging.info(f"Processing {job}: {build_nr:2d}: {local_file}")
        data = self._parse_tests(job, build)
        if data is None:
            raise PresentationError(
                f"Error occurred while parsing the file {local_file}"
            )

        build_data = pd.Series({
            u"metadata": pd.Series(
                list(data[u"metadata"].values()),
                index=list(data[u"metadata"].keys())
            ),
            u"suites": pd.Series(
                list(data[u"suites"].values()),
                index=list(data[u"suites"].keys())
            ),
            u"tests": pd.Series(
                list(data[u"tests"].values()),
                index=list(data[u"tests"].keys())
            )
        })

        if self._input_data.get(job, None) is None:
            self._input_data[job] = pd.Series()
        self._input_data[job][str(build_nr)] = build_data

        self._cfg.set_input_state(job, build_nr, u"processed")

    def process_local_directory(self, local_dir, replace=True):
        """Process local directory with XML file(s). The directory is processed
        as a 'job' and the XML files in it as builds.
        If the given directory contains only sub-directories, these
        sub-directories processed as jobs and corresponding XML files as builds
        of their job.

        :param local_dir: Local directory to process.
        :param replace: If True, the information about jobs and builds is
            replaced by the new one, otherwise the new jobs and builds are
            added.
        :type local_dir: str
        :type replace: bool
        """
        if not isdir(local_dir):
            raise PresentationError(
                f"The directory {local_dir} does not exist."
            )

        # Check if the given directory includes only files, or only directories
        _, dirnames, filenames = next(walk(local_dir))

        if filenames and not dirnames:
            filenames.sort()
            # local_builds:
            # key: dir (job) name, value: list of file names (builds)
            local_builds = {
                local_dir: [join(local_dir, name) for name in filenames]
            }

        elif dirnames and not filenames:
            dirnames.sort()
            # local_builds:
            # key: dir (job) name, value: list of file names (builds)
            local_builds = dict()
            for dirname in dirnames:
                builds = [
                    join(local_dir, dirname, name)
                    for name in listdir(join(local_dir, dirname))
                    if isfile(join(local_dir, dirname, name))
                ]
                if builds:
                    local_builds[dirname] = sorted(builds)

        elif not filenames and not dirnames:
            raise PresentationError(f"The directory {local_dir} is empty.")
        else:
            raise PresentationError(
                f"The directory {local_dir} can include only files or only "
                f"directories, not both.\nThe directory {local_dir} includes "
                f"file(s):\n{filenames}\nand directories:\n{dirnames}"
            )

        if replace:
            self._cfg.input = dict()

        for job, files in local_builds.items():
            for idx, local_file in enumerate(files):
                self.process_local_file(local_file, job, idx + 1, replace=False)

    @staticmethod
    def _end_of_tag(tag_filter, start=0, closer=u"'"):
        """Return the index of character in the string which is the end of tag.

        :param tag_filter: The string where the end of tag is being searched.
        :param start: The index where the searching is stated.
        :param closer: The character which is the tag closer.
        :type tag_filter: str
        :type start: int
        :type closer: str
        :returns: The index of the tag closer.
        :rtype: int
        """
        try:
            idx_opener = tag_filter.index(closer, start)
            return tag_filter.index(closer, idx_opener + 1)
        except ValueError:
            return None

    @staticmethod
    def _condition(tag_filter):
        """Create a conditional statement from the given tag filter.

        :param tag_filter: Filter based on tags from the element specification.
        :type tag_filter: str
        :returns: Conditional statement which can be evaluated.
        :rtype: str
        """
        index = 0
        while True:
            index = InputData._end_of_tag(tag_filter, index)
            if index is None:
                return tag_filter
            index += 1
            tag_filter = tag_filter[:index] + u" in tags" + tag_filter[index:]

    def filter_data(self, element, params=None, data=None, data_set=u"tests",
                    continue_on_error=False):
        """Filter required data from the given jobs and builds.

        The output data structure is:
        - job 1
          - build 1
            - test (or suite) 1 ID:
              - param 1
              - param 2
              ...
              - param n
            ...
            - test (or suite) n ID:
            ...
          ...
          - build n
        ...
        - job n

        :param element: Element which will use the filtered data.
        :param params: Parameters which will be included in the output. If None,
            all parameters are included.
        :param data: If not None, this data is used instead of data specified
            in the element.
        :param data_set: The set of data to be filtered: tests, suites,
            metadata.
        :param continue_on_error: Continue if there is error while reading the
            data. The Item will be empty then
        :type element: pandas.Series
        :type params: list
        :type data: dict
        :type data_set: str
        :type continue_on_error: bool
        :returns: Filtered data.
        :rtype pandas.Series
        """

        try:
            if data_set == "suites":
                cond = u"True"
            elif element[u"filter"] in (u"all", u"template"):
                cond = u"True"
            else:
                cond = InputData._condition(element[u"filter"])
            logging.debug(f"   Filter: {cond}")
        except KeyError:
            logging.error(u"  No filter defined.")
            return None

        if params is None:
            params = element.get(u"parameters", None)
            if params:
                params.extend((u"type", u"status"))

        data_to_filter = data if data else element[u"data"]
        data = pd.Series()
        try:
            for job, builds in data_to_filter.items():
                data[job] = pd.Series()
                for build in builds:
                    data[job][str(build)] = pd.Series()
                    try:
                        data_dict = dict(
                            self.data[job][str(build)][data_set].items())
                    except KeyError:
                        if continue_on_error:
                            continue
                        return None

                    for test_id, test_data in data_dict.items():
                        if eval(cond, {u"tags": test_data.get(u"tags", u"")}):
                            data[job][str(build)][test_id] = pd.Series()
                            if params is None:
                                for param, val in test_data.items():
                                    data[job][str(build)][test_id][param] = val
                            else:
                                for param in params:
                                    try:
                                        data[job][str(build)][test_id][param] =\
                                            test_data[param]
                                    except KeyError:
                                        data[job][str(build)][test_id][param] =\
                                            u"No Data"
            return data

        except (KeyError, IndexError, ValueError) as err:
            logging.error(
                f"Missing mandatory parameter in the element specification: "
                f"{repr(err)}"
            )
            return None
        except AttributeError as err:
            logging.error(repr(err))
            return None
        except SyntaxError as err:
            logging.error(
                f"The filter {cond} is not correct. Check if all tags are "
                f"enclosed by apostrophes.\n{repr(err)}"
            )
            return None

    def filter_tests_by_name(self, element, params=None, data_set=u"tests",
                             continue_on_error=False):
        """Filter required data from the given jobs and builds.

        The output data structure is:
        - job 1
          - build 1
            - test (or suite) 1 ID:
              - param 1
              - param 2
              ...
              - param n
            ...
            - test (or suite) n ID:
            ...
          ...
          - build n
        ...
        - job n

        :param element: Element which will use the filtered data.
        :param params: Parameters which will be included in the output. If None,
        all parameters are included.
        :param data_set: The set of data to be filtered: tests, suites,
        metadata.
        :param continue_on_error: Continue if there is error while reading the
        data. The Item will be empty then
        :type element: pandas.Series
        :type params: list
        :type data_set: str
        :type continue_on_error: bool
        :returns: Filtered data.
        :rtype pandas.Series
        """

        include = element.get(u"include", None)
        if not include:
            logging.warning(u"No tests to include, skipping the element.")
            return None

        if params is None:
            params = element.get(u"parameters", None)
            if params and u"type" not in params:
                params.append(u"type")

        cores = element.get(u"core", None)
        if cores:
            tests = list()
            for core in cores:
                for test in include:
                    tests.append(test.format(core=core))
        else:
            tests = include

        data = pd.Series()
        try:
            for job, builds in element[u"data"].items():
                data[job] = pd.Series()
                for build in builds:
                    data[job][str(build)] = pd.Series()
                    for test in tests:
                        try:
                            reg_ex = re.compile(str(test).lower())
                            for test_id in self.data[job][
                                    str(build)][data_set].keys():
                                if re.match(reg_ex, str(test_id).lower()):
                                    test_data = self.data[job][
                                        str(build)][data_set][test_id]
                                    data[job][str(build)][test_id] = pd.Series()
                                    if params is None:
                                        for param, val in test_data.items():
                                            data[job][str(build)][test_id]\
                                                [param] = val
                                    else:
                                        for param in params:
                                            try:
                                                data[job][str(build)][
                                                    test_id][param] = \
                                                    test_data[param]
                                            except KeyError:
                                                data[job][str(build)][
                                                    test_id][param] = u"No Data"
                        except KeyError as err:
                            if continue_on_error:
                                logging.debug(repr(err))
                                continue
                            logging.error(repr(err))
                            return None
            return data

        except (KeyError, IndexError, ValueError) as err:
            logging.error(
                f"Missing mandatory parameter in the element "
                f"specification: {repr(err)}"
            )
            return None
        except AttributeError as err:
            logging.error(repr(err))
            return None

    @staticmethod
    def merge_data(data):
        """Merge data from more jobs and builds to a simple data structure.

        The output data structure is:

        - test (suite) 1 ID:
          - param 1
          - param 2
          ...
          - param n
        ...
        - test (suite) n ID:
        ...

        :param data: Data to merge.
        :type data: pandas.Series
        :returns: Merged data.
        :rtype: pandas.Series
        """

        logging.info(u"    Merging data ...")

        merged_data = pd.Series()
        for builds in data.values:
            for item in builds.values:
                for item_id, item_data in item.items():
                    merged_data[item_id] = item_data
        return merged_data

    def print_all_oper_data(self):
        """Print all operational data to console.
        """

        for job in self._input_data.values:
            for build in job.values:
                for test_id, test_data in build[u"tests"].items():
                    print(f"{test_id}")
                    if test_data.get(u"show-run", None) is None:
                        continue
                    for dut_name, data in test_data[u"show-run"].items():
                        if data.get(u"runtime", None) is None:
                            continue
                        runtime = loads(data[u"runtime"])
                        try:
                            threads_nr = len(runtime[0][u"clocks"])
                        except (IndexError, KeyError):
                            continue
                        threads = OrderedDict(
                            {idx: list() for idx in range(threads_nr)})
                        for item in runtime:
                            for idx in range(threads_nr):
                                if item[u"vectors"][idx] > 0:
                                    clocks = item[u"clocks"][idx] / \
                                             item[u"vectors"][idx]
                                elif item[u"calls"][idx] > 0:
                                    clocks = item[u"clocks"][idx] / \
                                             item[u"calls"][idx]
                                elif item[u"suspends"][idx] > 0:
                                    clocks = item[u"clocks"][idx] / \
                                             item[u"suspends"][idx]
                                else:
                                    clocks = 0.0

                                if item[u"calls"][idx] > 0:
                                    vectors_call = item[u"vectors"][idx] / \
                                                   item[u"calls"][idx]
                                else:
                                    vectors_call = 0.0

                                if int(item[u"calls"][idx]) + int(
                                        item[u"vectors"][idx]) + \
                                        int(item[u"suspends"][idx]):
                                    threads[idx].append([
                                        item[u"name"],
                                        item[u"calls"][idx],
                                        item[u"vectors"][idx],
                                        item[u"suspends"][idx],
                                        clocks,
                                        vectors_call
                                    ])

                        print(f"Host IP: {data.get(u'host', '')}, "
                              f"Socket: {data.get(u'socket', '')}")
                        for thread_nr, thread in threads.items():
                            txt_table = prettytable.PrettyTable(
                                (
                                    u"Name",
                                    u"Nr of Vectors",
                                    u"Nr of Packets",
                                    u"Suspends",
                                    u"Cycles per Packet",
                                    u"Average Vector Size"
                                )
                            )
                            avg = 0.0
                            for row in thread:
                                txt_table.add_row(row)
                                avg += row[-1]
                            if len(thread) == 0:
                                avg = u""
                            else:
                                avg = f", Average Vector Size per Node: " \
                                      f"{(avg / len(thread)):.2f}"
                            th_name = u"main" if thread_nr == 0 \
                                else f"worker_{thread_nr}"
                            print(f"{dut_name}, {th_name}{avg}")
                            txt_table.float_format = u".2"
                            txt_table.align = u"r"
                            txt_table.align[u"Name"] = u"l"
                            print(f"{txt_table.get_string()}\n")