aboutsummaryrefslogtreecommitdiffstats
path: root/resources/tools/presentation
diff options
context:
space:
mode:
Diffstat (limited to 'resources/tools/presentation')
-rw-r--r--resources/tools/presentation/input_data_files.py136
-rw-r--r--resources/tools/presentation/input_data_parser.py219
-rw-r--r--resources/tools/presentation/pal.py2
3 files changed, 255 insertions, 102 deletions
diff --git a/resources/tools/presentation/input_data_files.py b/resources/tools/presentation/input_data_files.py
index 9e0cfa6931..cde6d1acc4 100644
--- a/resources/tools/presentation/input_data_files.py
+++ b/resources/tools/presentation/input_data_files.py
@@ -16,11 +16,9 @@ Download all data.
"""
import re
-import logging
-from os import rename
+from os import rename, mkdir
from os.path import join
-from shutil import move
from zipfile import ZipFile, is_zipfile, BadZipfile
from httplib import responses
from requests import get, codes, RequestException, Timeout, TooManyRedirects, \
@@ -38,31 +36,33 @@ SEPARATOR = "__"
REGEX_RELEASE = re.compile(r'(\D*)(\d{4}|master)(\D*)')
-def _download_file(url, file_name):
+def _download_file(url, file_name, log):
"""Download a file with input data.
:param url: URL to the file to download.
:param file_name: Name of file to download.
+ :param log: List of log messages.
:type url: str
:type file_name: str
+ :type log: list of tuples (severity, msg)
:returns: True if the download was successful, otherwise False.
:rtype: bool
"""
success = False
try:
- logging.info(" Connecting to '{0}' ...".format(url))
+ log.append(("INFO", " Connecting to '{0}' ...".format(url)))
response = get(url, stream=True)
code = response.status_code
- logging.info(" {0}: {1}".format(code, responses[code]))
+ log.append(("INFO", " {0}: {1}".format(code, responses[code])))
if code != codes["OK"]:
return False
- logging.info(" Downloading the file '{0}' to '{1}' ...".
- format(url, file_name))
+ log.append(("INFO", " Downloading the file '{0}' to '{1}' ...".
+ format(url, file_name)))
file_handle = open(file_name, "wb")
for chunk in response.iter_content(chunk_size=CHUNK_SIZE):
@@ -71,38 +71,38 @@ def _download_file(url, file_name):
file_handle.close()
success = True
except ConnectionError as err:
- logging.error("Not possible to connect to '{0}'.".format(url))
- logging.debug(str(err))
+ log.append(("ERROR", "Not possible to connect to '{0}'.".format(url)))
+ log.append(("DEBUG", str(err)))
except HTTPError as err:
- logging.error("Invalid HTTP response from '{0}'.".format(url))
- logging.debug(str(err))
+ log.append(("ERROR", "Invalid HTTP response from '{0}'.".format(url)))
+ log.append(("DEBUG", str(err)))
except TooManyRedirects as err:
- logging.error("Request exceeded the configured number "
- "of maximum re-directions.")
- logging.debug(str(err))
+ log.append(("ERROR", "Request exceeded the configured number "
+ "of maximum re-directions."))
+ log.append(("DEBUG", str(err)))
except Timeout as err:
- logging.error("Request timed out.")
- logging.debug(str(err))
+ log.append(("ERROR", "Request timed out."))
+ log.append(("DEBUG", str(err)))
except RequestException as err:
- logging.error("Unexpected HTTP request exception.")
- logging.debug(str(err))
+ log.append(("ERROR", "Unexpected HTTP request exception."))
+ log.append(("DEBUG", str(err)))
except (IOError, ValueError, KeyError) as err:
- logging.error("Download failed.")
- logging.debug(str(err))
+ log.append(("ERROR", "Download failed."))
+ log.append(("DEBUG", str(err)))
- logging.info(" Download finished.")
+ log.append(("INFO", " Download finished."))
return success
-def _unzip_file(spec, job, build):
+def _unzip_file(spec, build, pid, log):
"""Unzip downloaded source file.
:param spec: Specification read form the specification file.
- :param job: Name of the Jenkins job.
:param build: Information about the build.
+ :param log: List of log messages.
:type spec: Specification
- :type job: str
:type build: dict
+ :type log: list of tuples (severity, msg)
:returns: True if the download was successful, otherwise False.
:rtype: bool
"""
@@ -110,44 +110,48 @@ def _unzip_file(spec, job, build):
data_file = spec.input["extract"]
file_name = build["file-name"]
directory = spec.environment["paths"]["DIR[WORKING,DATA]"]
+ tmp_dir = join(directory, str(pid))
+ try:
+ mkdir(tmp_dir)
+ except OSError:
+ pass
new_name = "{0}{1}{2}".format(file_name.rsplit('.')[-2],
SEPARATOR,
data_file.split("/")[-1])
- logging.info(" Unzipping: '{0}' from '{1}'.".
- format(data_file, file_name))
+
+ log.append(("INFO", " Unzipping: '{0}' from '{1}'.".
+ format(data_file, file_name)))
try:
with ZipFile(file_name, 'r') as zip_file:
- zip_file.extract(data_file, directory)
- logging.info(" Moving {0} to {1} ...".
- format(join(directory, data_file), directory))
- move(join(directory, data_file), directory)
- logging.info(" Renaming the file '{0}' to '{1}'".
- format(join(directory, data_file.split("/")[-1]),
- new_name))
- rename(join(directory, data_file.split("/")[-1]),
- new_name)
- spec.set_input_file_name(job, build["build"],
- new_name)
+ zip_file.extract(data_file, tmp_dir)
+ log.append(("INFO", " Renaming the file '{0}' to '{1}'".
+ format(join(tmp_dir, data_file), new_name)))
+ rename(join(tmp_dir, data_file), new_name)
+ build["file-name"] = new_name
return True
except (BadZipfile, RuntimeError) as err:
- logging.error("Failed to unzip the file '{0}': {1}.".
- format(file_name, str(err)))
+ log.append(("ERROR", "Failed to unzip the file '{0}': {1}.".
+ format(file_name, str(err))))
return False
except OSError as err:
- logging.error("Failed to rename the file '{0}': {1}.".
- format(data_file, str(err)))
+ log.append(("ERROR", "Failed to rename the file '{0}': {1}.".
+ format(data_file, str(err))))
return False
-def download_and_unzip_data_file(spec, job, build):
+def download_and_unzip_data_file(spec, job, build, pid, log):
"""Download and unzip a source file.
:param spec: Specification read form the specification file.
:param job: Name of the Jenkins job.
:param build: Information about the build.
+ :param pid: PID of the process executing this method.
+ :param log: List of log messages.
:type spec: Specification
:type job: str
:type build: dict
+ :type pid: int
+ :type log: list of tuples (severity, msg)
:returns: True if the download was successful, otherwise False.
:rtype: bool
"""
@@ -158,7 +162,7 @@ def download_and_unzip_data_file(spec, job, build):
elif spec.input["file-name"].endswith(".gz"):
url = spec.environment["urls"]["URL[NEXUS,LOG]"]
else:
- logging.error("Not supported file format.")
+ log.append(("ERROR", "Not supported file format."))
return False
elif job.startswith("hc2vpp-"):
url = spec.environment["urls"]["URL[JENKINS,HC]"]
@@ -174,27 +178,35 @@ def download_and_unzip_data_file(spec, job, build):
format(job=job, sep=SEPARATOR, build=build["build"],
name=file_name))
# Download the file from the defined source (Jenkins, logs.fd.io):
- success = _download_file(url, new_name)
+ success = _download_file(url, new_name, log)
+
+ if success and new_name.endswith(".zip"):
+ if not is_zipfile(new_name):
+ success = False
# If not successful, download from docs.fd.io:
if not success:
- logging.info(" Trying to download from https://docs.fd.io:")
+ log.append(("INFO", " Trying to download from https://docs.fd.io:"))
release = re.search(REGEX_RELEASE, job).group(2)
- nexus_file_name = "{job}{sep}{build}{sep}{name}". \
- format(job=job, sep=SEPARATOR, build=build["build"], name=file_name)
- try:
- release = "rls{0}".format(int(release))
- except ValueError:
- pass
- url = "{url}/{release}/{dir}/{file}". \
- format(url=spec.environment["urls"]["URL[NEXUS]"],
- release=release,
- dir=spec.environment["urls"]["DIR[NEXUS]"],
- file=nexus_file_name)
- success = _download_file(url, new_name)
+ for rls in (release, "master"):
+ nexus_file_name = "{job}{sep}{build}{sep}{name}". \
+ format(job=job, sep=SEPARATOR, build=build["build"],
+ name=file_name)
+ try:
+ rls = "rls{0}".format(int(rls))
+ except ValueError:
+ pass
+ url = "{url}/{release}/{dir}/{file}". \
+ format(url=spec.environment["urls"]["URL[NEXUS]"],
+ release=rls,
+ dir=spec.environment["urls"]["DIR[NEXUS]"],
+ file=nexus_file_name)
+ success = _download_file(url, new_name, log)
+ if success:
+ break
if success:
- spec.set_input_file_name(job, build["build"], new_name)
+ build["file-name"] = new_name
else:
return False
@@ -205,12 +217,14 @@ def download_and_unzip_data_file(spec, job, build):
else:
rename(new_name, new_name[:-3])
execute_command("gzip --keep {0}".format(new_name[:-3]))
- spec.set_input_file_name(job, build["build"], new_name[:-3])
+ build["file-name"] = new_name[:-3]
if new_name.endswith(".zip"):
if is_zipfile(new_name):
- return _unzip_file(spec, job, build)
+ return _unzip_file(spec, build, pid, log)
else:
+ log.append(("ERROR",
+ "Zip file '{0}' is corrupted.".format(new_name)))
return False
else:
return True
diff --git a/resources/tools/presentation/input_data_parser.py b/resources/tools/presentation/input_data_parser.py
index d0f9eed9a4..9428b2cdc9 100644
--- a/resources/tools/presentation/input_data_parser.py
+++ b/resources/tools/presentation/input_data_parser.py
@@ -18,10 +18,11 @@
- provide access to the data.
"""
+import multiprocessing
+import os
import re
import pandas as pd
import logging
-import xml.etree.ElementTree as ET
from robot.api import ExecutionResult, ResultVisitor
from robot import errors
@@ -698,7 +699,7 @@ class InputData(object):
self._cfg = spec
# Data store:
- self._input_data = None
+ self._input_data = pd.Series()
@property
def data(self):
@@ -749,14 +750,16 @@ class InputData(object):
return self.data[job][build]["tests"]
@staticmethod
- def _parse_tests(job, build):
+ def _parse_tests(job, build, log):
"""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.
+ :param log: List of log messages.
:type job: str
:type build: dict
+ :type log: list of tuples (severity, msg)
:returns: JSON data structure.
:rtype: dict
"""
@@ -770,54 +773,129 @@ class InputData(object):
try:
result = ExecutionResult(data_file)
except errors.DataError as err:
- logging.error("Error occurred while parsing output.xml: {0}".
- format(err))
+ log.append(("ERROR", "Error occurred while parsing output.xml: "
+ "{0}".format(err)))
return None
checker = ExecutionChecker(metadata)
result.visit(checker)
return checker.data
- def download_and_parse_data(self):
+ def _download_and_parse_build(self, pid, data_queue, job, build, repeat):
+ """Download and parse the input data file.
+
+ :param pid: PID of the process executing this method.
+ :param data_queue: Shared memory between processes. Queue which keeps
+ the result data. This data is then read by the main process and used
+ in further processing.
+ :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 data_queue: multiprocessing.Manager().Queue()
+ :type job: str
+ :type build: dict
+ :type repeat: int
+ """
+
+ logs = list()
+
+ logging.info(" Processing the job/build: {0}: {1}".
+ format(job, build["build"]))
+
+ logs.append(("INFO", " Processing the job/build: {0}: {1}".
+ format(job, build["build"])))
+
+ state = "failed"
+ success = False
+ data = None
+ do_repeat = repeat
+ while do_repeat:
+ success = download_and_unzip_data_file(self._cfg, job, build, pid,
+ logs)
+ if success:
+ break
+ do_repeat -= 1
+ if not success:
+ logs.append(("ERROR", "It is not possible to download the input "
+ "data file from the job '{job}', build "
+ "'{build}', or it is damaged. Skipped.".
+ format(job=job, build=build["build"])))
+ if success:
+ logs.append(("INFO", " Processing data from the build '{0}' ...".
+ format(build["build"])))
+ data = InputData._parse_tests(job, build, logs)
+ if data is None:
+ logs.append(("ERROR", "Input data file from the job '{job}', "
+ "build '{build}' is damaged. Skipped.".
+ format(job=job, build=build["build"])))
+ else:
+ state = "processed"
+
+ try:
+ remove(build["file-name"])
+ except OSError as err:
+ logs.append(("ERROR", "Cannot remove the file '{0}': {1}".
+ format(build["file-name"], err)))
+ logs.append(("INFO", " Done."))
+
+ result = {
+ "data": data,
+ "state": state,
+ "job": job,
+ "build": build,
+ "logs": logs
+ }
+ data_queue.put(result)
+
+ 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("Downloading and parsing input files ...")
- job_data = dict()
+ work_queue = multiprocessing.JoinableQueue()
+
+ manager = multiprocessing.Manager()
+
+ data_queue = manager.Queue()
+
+ cpus = multiprocessing.cpu_count()
+ workers = list()
+ for cpu in range(cpus):
+ worker = Worker(work_queue,
+ data_queue,
+ self._download_and_parse_build)
+ worker.daemon = True
+ worker.start()
+ workers.append(worker)
+ os.system("taskset -p -c {0} {1} > /dev/null 2>&1".
+ format(cpu, worker.pid))
+
for job, builds in self._cfg.builds.items():
- logging.info(" Processing data from the job '{0}' ...".
- format(job))
- builds_data = dict()
for build in builds:
- logging.info(" Processing the build '{0}'".
- format(build["build"]))
- self._cfg.set_input_state(job, build["build"], "failed")
- if not download_and_unzip_data_file(self._cfg, job, build):
- logging.error("It is not possible to download the input "
- "data file from the job '{job}', build "
- "'{build}', or it is damaged. Skipped.".
- format(job=job, build=build["build"]))
- continue
+ work_queue.put((job, build, repeat))
- logging.info(" Processing data from the build '{0}' ...".
- format(build["build"]))
- data = InputData._parse_tests(job, build)
- if data is None:
- logging.error("Input data file from the job '{job}', build "
- "'{build}' is damaged. Skipped.".
- format(job=job, build=build["build"]))
- continue
+ work_queue.join()
- self._cfg.set_input_state(job, build["build"], "processed")
+ logging.info("Done.")
- try:
- remove(build["file-name"])
- except OSError as err:
- logging.error("Cannot remove the file '{0}': {1}".
- format(build["file-name"], err))
+ while not data_queue.empty():
+ result = data_queue.get()
+ job = result["job"]
+ build_nr = result["build"]["build"]
+
+ if result["data"]:
+ data = result["data"]
build_data = pd.Series({
"metadata": pd.Series(data["metadata"].values(),
index=data["metadata"].keys()),
@@ -825,15 +903,35 @@ class InputData(object):
index=data["suites"].keys()),
"tests": pd.Series(data["tests"].values(),
index=data["tests"].keys())})
- builds_data[str(build["build"])] = build_data
- build["status"] = "processed"
- logging.info(" Done.")
- job_data[job] = pd.Series(builds_data.values(),
- index=builds_data.keys())
- logging.info(" Done.")
+ 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["build"]["file-name"])
+
+ self._cfg.set_input_state(job, build_nr, result["state"])
+
+ for item in result["logs"]:
+ if item[0] == "INFO":
+ logging.info(item[1])
+ elif item[0] == "ERROR":
+ logging.error(item[1])
+ elif item[0] == "DEBUG":
+ logging.debug(item[1])
+ elif item[0] == "CRITICAL":
+ logging.critical(item[1])
+ elif item[0] == "WARNING":
+ logging.warning(item[1])
+
+ del data_queue
+
+ # Terminate all workers
+ for worker in workers:
+ worker.terminate()
+ worker.join()
- self._input_data = pd.Series(job_data.values(), index=job_data.keys())
logging.info("Done.")
@staticmethod
@@ -997,3 +1095,44 @@ class InputData(object):
merged_data[ID] = item_data
return merged_data
+
+
+class Worker(multiprocessing.Process):
+ """Worker class used to download and process input files in separate
+ parallel processes.
+ """
+
+ def __init__(self, work_queue, data_queue, func):
+ """Initialization.
+
+ :param work_queue: Queue with items to process.
+ :param data_queue: Shared memory between processes. Queue which keeps
+ the result data. This data is then read by the main process and used
+ in further processing.
+ :param func: Function which is executed by the worker.
+ :type work_queue: multiprocessing.JoinableQueue
+ :type data_queue: multiprocessing.Manager().Queue()
+ :type func: Callable object
+ """
+ super(Worker, self).__init__()
+ self._work_queue = work_queue
+ self._data_queue = data_queue
+ self._func = func
+
+ def run(self):
+ """Method representing the process's activity.
+ """
+
+ while True:
+ try:
+ self.process(self._work_queue.get())
+ finally:
+ self._work_queue.task_done()
+
+ def process(self, item_to_process):
+ """Method executed by the runner.
+
+ :param item_to_process: Data to be processed by the function.
+ :type item_to_process: tuple
+ """
+ self._func(self.pid, self._data_queue, *item_to_process)
diff --git a/resources/tools/presentation/pal.py b/resources/tools/presentation/pal.py
index 1ccefd3b43..013c921124 100644
--- a/resources/tools/presentation/pal.py
+++ b/resources/tools/presentation/pal.py
@@ -96,7 +96,7 @@ def main():
prepare_static_content(spec)
data = InputData(spec)
- data.download_and_parse_data()
+ data.download_and_parse_data(repeat=2)
generate_tables(spec, data)
generate_plots(spec, data)