Geografic-Information-System/lib/python/gunittest/reporters.py

# -*- coding: utf-8 -*-
"""GRASS Python testing framework module for report generation

Copyright (C) 2014 by the GRASS Development Team
This program is free software under the GNU General Public
License (>=v2). Read the file COPYING that comes with GRASS GIS
for details.

:authors: Vaclav Petras
"""

import os
import datetime
import xml.sax.saxutils as saxutils
import xml.etree.ElementTree as et
import subprocess
import sys
import collections
import re

from .utils import ensure_dir
from .checkers import text_to_keyvalue


if sys.version_info[0] == 2:
    from StringIO import StringIO
else:
    from io import StringIO


# TODO: change text_to_keyvalue to same sep as here
# TODO: create keyvalue file and move it there together with things from checkers
def keyvalue_to_text(keyvalue, sep='=', vsep='\n', isep=',',
                     last_vertical=None):
    if not last_vertical:
        last_vertical = vsep == '\n'
    items = []
    for key, value in keyvalue.items():
        # TODO: use isep for iterables other than strings
        if (not isinstance(value, str)
                and isinstance(value, collections.Iterable)):
            # TODO: this does not work for list of non-strings
            value = isep.join(value)
        items.append('{key}{sep}{value}'.format(
            key=key, sep=sep, value=value))
    text = vsep.join(items)
    if last_vertical:
        text = text + vsep
    return text


def replace_in_file(file_path, pattern, repl):
    """

    :param repl: a repl parameter of ``re.sub()`` function
    """
    # using tmp file to store the replaced content
    tmp_file_path = file_path + '.tmp'
    old_file = open(file_path, 'r')
    new_file = open(tmp_file_path, 'w')
    for line in old_file:
        new_file.write(re.sub(pattern=pattern, string=line, repl=repl))
    new_file.close()
    old_file.close()
    # remove old file since it must not exist for rename/move
    os.remove(file_path)
    # replace old file by new file
    # TODO: this can fail in some (random) cases on MS Windows
    os.rename(tmp_file_path, file_path)


class NoopFileAnonymizer(object):
    def anonymize(self, filenames):
        pass


# TODO: why not remove GISDBASE by default?
class FileAnonymizer(object):
    def __init__(self, paths_to_remove, remove_gisbase=True,
                 remove_gisdbase=False):
        self._paths_to_remove = []
        if remove_gisbase:
            gisbase = os.environ['GISBASE']
            self._paths_to_remove.append(gisbase)
        if remove_gisdbase:
            # import only when really needed to avoid problems with
            # translations when environment is not set properly
            import grass.script as gscript
            gisdbase = gscript.gisenv()['GISDBASE']
            self._paths_to_remove.append(gisdbase)
        if paths_to_remove:
            self._paths_to_remove.extend(paths_to_remove)

    def anonymize(self, filenames):
        # besides GISBASE and test recursion start directory (which is
        # supposed to be source root directory or similar) we can also try
        # to remove user home directory and GISDBASE
        # we suppuse that we run in standard grass session
        # TODO: provide more effective implementation
        for path in self._paths_to_remove:
            for filename in filenames:
                path_end = r'[\\/]?'
                replace_in_file(filename, path + path_end, '')


def get_source_url(path, revision, line=None):
    """

    :param path: directory or file path relative to remote repository root
    :param revision: SVN revision (should be a number)
    :param line: line in the file (should be None for directories)
    """
    tracurl = 'http://trac.osgeo.org/grass/browser/'
    if line:
        return '{tracurl}{path}?rev={revision}#L{line}'.format(**locals())
    else:
        return '{tracurl}{path}?rev={revision}'.format(**locals())


def html_escape(text):
    """Escape ``'&'``, ``'<'``, and ``'>'`` in a string of data."""
    return saxutils.escape(text)


def html_unescape(text):
    """Unescape ``'&amp;'``, ``'&lt;'``, and ``'&gt;'`` in a string of data."""
    return saxutils.unescape(text)


def color_error_line(line):
    if line.startswith('ERROR: '):
        # TODO: use CSS class
        # ignoring the issue with \n at the end, HTML don't mind
        line = '<span style="color: red">' + line + "</span>"
    if line.startswith('FAIL: '):
        # TODO: use CSS class
        # ignoring the issue with \n at the end, HTML don't mind
        line = '<span style="color: red">' + line + "</span>"
    if line.startswith('WARNING: '):
        # TODO: use CSS class
        # ignoring the issue with \n at the end, HTML don't mind
        line = '<span style="color: blue">' + line + "</span>"
    #if line.startswith('Traceback ('):
    #    line = '<span style="color: red">' + line + "</span>"
    return line


def to_web_path(path):
    """Replace OS dependent path separator with slash.

    Path on MS Windows are not usable in links on web. For MS Windows,
    this replaces backslash with (forward) slash.
    """
    if os.path.sep != '/':
        return path.replace(os.path.sep, '/')
    else:
        return path


def get_svn_revision():
    """Get SVN revision number

    :returns: SVN revision number as string or None if it is
        not possible to get
    """
    # TODO: here should be starting directory
    # but now we are using current as starting
    p = subprocess.Popen(['svnversion', '.'],
                         stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    stdout, stderr = p.communicate()
    rc = p.poll()
    if not rc:
        stdout = stdout.strip()
        if stdout.endswith('M'):
            stdout = stdout[:-1]
        if ':' in stdout:
            # the first one is the one of source code
            stdout = stdout.split(':')[0]
        return stdout
    else:
        return None


def get_svn_info():
    """Get important information from ``svn info``

    :returns: SVN info as dictionary or None
        if it is not possible to obtain it
    """
    try:
        # TODO: introduce directory, not only current
        p = subprocess.Popen(['svn', 'info', '.', '--xml'],
                             stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        stdout, stderr = p.communicate()
        rc = p.poll()
        info = {}
        if not rc:
            root = et.fromstring(stdout)
            # TODO: get also date if this make sense
            # expecting only one <entry> element
            entry = root.find('entry')
            info['revision'] = entry.get('revision')
            info['url'] = entry.find('url').text
            relurl = entry.find('relative-url')
            # element which is not found is None
            # empty element would be bool(el) == False
            if relurl is not None:
                relurl = relurl.text
                # relative path has ^ at the beginning in SVN version 1.8.8
                if relurl.startswith('^'):
                    relurl = relurl[1:]
            else:
                # SVN version 1.8.8 supports relative-url but older do not
                # so, get relative part from absolute URL
                const_url_part = 'https://svn.osgeo.org/grass/'
                relurl = info['url'][len(const_url_part):]
            info['relative-url'] = relurl
            return info
    # TODO: add this to svnversion function
    except OSError as e:
        import errno
        # ignore No such file or directory
        if e.errno != errno.ENOENT:
            raise
    return None


def years_ago(date, years):
    # dateutil relative delte would be better but this is more portable
    return date - datetime.timedelta(weeks=years * 52)


# TODO: these functions should be called only if we know that svn is installed
# this will simplify the functions, caller must handle it anyway
def get_svn_path_authors(path, from_date=None):
    """

    :returns: a set of authors
    """
    if from_date is None:
        # this is the SVN default for local copies
        revision_range = 'BASE:1'
    else:
        revision_range = 'BASE:{%s}' % from_date
    try:
        # TODO: allow also usage of --limit
        p = subprocess.Popen(['svn', 'log', '--xml',
                              '--revision', revision_range, path],
                             stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        stdout, stderr = p.communicate()
        rc = p.poll()
        if not rc:
            root = et.fromstring(stdout)
            # TODO: get also date if this make sense
            # expecting only one <entry> element
            author_nodes = root.iterfind('*/author')
            authors = [n.text for n in author_nodes]
            return set(authors)
    except OSError as e:
        import errno
        # ignore No such file or directory
        if e.errno != errno.ENOENT:
            raise
    return None


def get_html_test_authors_table(directory, tests_authors):
    # SVN gives us authors of code together with authors of tests
    # so test code authors list also contains authors of tests only
    # TODO: don't do this for the top level directories?
    tests_authors = set(tests_authors)
    no_svn_text = ('<span style="font-size: 60%">'
                   'Test file authors were not obtained.'
                   '</span>')
    if (not tests_authors
            or (len(tests_authors) == 1 and list(tests_authors)[0] == '')):
        return '<h3>Code and test authors</h3>' + no_svn_text
    from_date = years_ago(datetime.date.today(), years=1)
    tested_dir_authors = get_svn_path_authors(directory, from_date)
    if tested_dir_authors is not None:
        not_testing_authors = tested_dir_authors - tests_authors
    else:
        no_svn_text = ('<span style="font-size: 60%">'
                       'Authors cannot be obtained using SVN.'
                       '</span>')
        not_testing_authors = tested_dir_authors = [no_svn_text]
    if not not_testing_authors:
        not_testing_authors = ['all recent authors contributed tests']

    test_authors = (
        '<h3>Code and test authors</h3>'
        '<p style="font-size: 60%"><em>'
        'Note that determination of authors is approximate and only'
        ' recent code authors are considered.'
        '</em></p>'
        '<table><tbody>'
        '<tr><td>Test authors:</td><td>{file_authors}</td></tr>'
        '<tr><td>Authors of tested code:</td><td>{code_authors}</td></tr>'
        '<tr><td>Authors owing tests:</td><td>{not_testing}</td></tr>'
        '</tbody></table>'
        .format(
            file_authors=', '.join(sorted(tests_authors)),
            code_authors=', '.join(sorted(tested_dir_authors)),
            not_testing=', '.join(sorted(not_testing_authors))
        ))
    return test_authors


class GrassTestFilesMultiReporter(object):

    def __init__(self, reporters, forgiving=False):
        self.reporters = reporters
        self.forgiving = forgiving

    def start(self, results_dir):
        # TODO: no directory cleaning (self.clean_before)? now cleaned by caller
        # TODO: perhaps only those whoe need it should do it (even multiple times)
        # and there is also the delet problem
        ensure_dir(os.path.abspath(results_dir))
        for reporter in self.reporters:
            try:
                reporter.start(results_dir)
            except AttributeError:
                if self.forgiving:
                    pass
                else:
                    raise

    def finish(self):
        for reporter in self.reporters:
            try:
                reporter.finish()
            except AttributeError:
                if self.forgiving:
                    pass
                else:
                    raise

    def start_file_test(self, module):
        for reporter in self.reporters:
            try:
                reporter.start_file_test(module)
            except AttributeError:
                if self.forgiving:
                    pass
                else:
                    raise

    def end_file_test(self, **kwargs):
        for reporter in self.reporters:
            try:
                reporter.end_file_test(**kwargs)
            except AttributeError:
                if self.forgiving:
                    pass
                else:
                    raise


class GrassTestFilesCountingReporter(object):
    def __init__(self):
        self.test_files = None
        self.files_fail = None
        self.files_pass = None

        self.file_pass_per = None
        self.file_fail_per = None

        self.main_start_time = None
        self.main_end_time = None
        self.main_time = None

        self.file_start_time = None
        self.file_end_time = None
        self.file_time = None
        self._start_file_test_called = False

    def start(self, results_dir):
        self.test_files = 0
        self.files_fail = 0
        self.files_pass = 0

        # this might be moved to some report start method
        self.main_start_time = datetime.datetime.now()

    def finish(self):
        self.main_end_time = datetime.datetime.now()
        self.main_time = self.main_end_time - self.main_start_time

        assert self.test_files == self.files_fail + self.files_pass
        if self.test_files:
            self.file_pass_per = 100 * float(self.files_pass) / self.test_files
            self.file_fail_per = 100 * float(self.files_fail) / self.test_files
        else:
            # if no tests were executed, probably something bad happened
            # try to report at least something
            self.file_pass_per = None
            self.file_fail_per = None

    def start_file_test(self, module):
        self.file_start_time = datetime.datetime.now()
        self._start_file_test_called = True
        self.test_files += 1

    def end_file_test(self, returncode, **kwargs):
        assert self._start_file_test_called
        self.file_end_time = datetime.datetime.now()
        self.file_time = self.file_end_time - self.file_start_time
        if returncode:
            self.files_fail += 1
        else:
            self.files_pass += 1
        self._start_file_test_called = False


def percent_to_html(percent):
    if percent is None:
        return '<span style="color: {color}">unknown percentage</span>'
    elif percent > 100 or percent < 0:
        return "? {:.2f}% ?".format(percent)
    elif percent < 40:
        color = 'red'
    elif percent < 70:
        color = 'orange'
    else:
        color = 'green'
    return '<span style="color: {color}">{percent:.0f}%</span>'.format(
        percent=percent, color=color)


def wrap_stdstream_to_html(infile, outfile, module, stream):
    before = '<html><body><h1>%s</h1><pre>' % (module.name + ' ' + stream)
    after = '</pre></body></html>'
    html = open(outfile, 'w')
    html.write(before)
    with open(infile) as text:
        for line in text:
            html.write(color_error_line(html_escape(line)))
    html.write(after)
    html.close()


def html_file_preview(filename):
    before = '<pre>'
    after = '</pre>'
    if not os.path.isfile(filename):
        return '<p style="color: red>File %s does not exist<p>' % filename
    size = os.path.getsize(filename)
    if not size:
        return '<p style="color: red>File %s is empty<p>' % filename
    max_size = 10000
    html = StringIO()
    html.write(before)
    if size < max_size:
        with open(filename) as text:
            for line in text:
                html.write(color_error_line(html_escape(line)))
    elif size < 10 * max_size:
        def tail(filename, n):
            return collections.deque(open(filename), n)
        html.write('... (lines omitted)\n')
        for line in tail(filename, 50):
            html.write(color_error_line(html_escape(line)))
    else:
        return '<p style="color: red>File %s is too large to show<p>' % filename
    html.write(after)
    return html.getvalue()


def returncode_to_html_text(returncode):
    if returncode:
        return '<span style="color: red">FAILED</span>'
    else:
        # alternatives: SUCCEEDED, passed, OK
        return '<span style="color: green">succeeded</span>'


# not used
def returncode_to_html_sentence(returncode):
    if returncode:
        return ('<span style="color: red">&#x274c;</span>'
                ' Test failed (return code %d)' % (returncode))
    else:
        return ('<span style="color: green">&#x2713;</span>'
                ' Test succeeded (return code %d)' % (returncode))


def returncode_to_success_html_par(returncode):
    if returncode:
        return ('<p> <span style="color: red">&#x274c;</span>'
                ' Test failed</p>')
    else:
        return ('<p> <span style="color: green">&#x2713;</span>'
                ' Test succeeded</p>')


def success_to_html_text(total, successes):
    if successes < total:
        return '<span style="color: red">FAILED</span>'
    elif successes == total:
        # alternatives: SUCCEEDED, passed, OK
        return '<span style="color: green">succeeded</span>'
    else:
        return ('<span style="color: red; font-size: 60%">'
                '? more successes than total ?</span>')


UNKNOWN_NUMBER_HTML = '<span style="font-size: 60%">unknown</span>'


def success_to_html_percent(total, successes):
    if total:
        pass_per = 100 * (float(successes) / total)
        pass_per = percent_to_html(pass_per)
    else:
        pass_per = UNKNOWN_NUMBER_HTML
    return pass_per


class GrassTestFilesHtmlReporter(GrassTestFilesCountingReporter):

    unknown_number = UNKNOWN_NUMBER_HTML

    def __init__(self, file_anonymizer, main_page_name='index.html'):
        super(GrassTestFilesHtmlReporter, self).__init__()
        self.main_index = None
        self._file_anonymizer = file_anonymizer
        self._main_page_name = main_page_name

    def start(self, results_dir):
        super(GrassTestFilesHtmlReporter, self).start(results_dir)
        # having all variables public although not really part of API
        main_page_name = os.path.join(results_dir, self._main_page_name)
        self.main_index = open(main_page_name, 'w')

        # TODO: this can be moved to the counter class
        self.failures = 0
        self.errors = 0
        self.skipped = 0
        self.successes = 0
        self.expected_failures = 0
        self.unexpected_success = 0
        self.total = 0

        svn_info = get_svn_info()
        if not svn_info:
            svn_text = ('<span style="font-size: 60%">'
                        'SVN revision cannot be obtained'
                        '</span>')
        else:
            url = get_source_url(path=svn_info['relative-url'],
                                 revision=svn_info['revision'])
            svn_text = ('SVN revision'
                        ' <a href="{url}">'
                        '{rev}</a>'
                        ).format(url=url, rev=svn_info['revision'])
        self.main_index.write('<html><body>'
                              '<h1>Test results</h1>'
                              '{time:%Y-%m-%d %H:%M:%S}'
                              ' ({svn})'
                              '<table>'
                              '<thead><tr>'
                              '<th>Tested directory</th>'
                              '<th>Test file</th>'
                              '<th>Status</th>'
                              '<th>Tests</th><th>Successful</td>'
                              '<th>Failed</th><th>Percent successful</th>'
                              '</tr></thead><tbody>'.format(
                                  time=self.main_start_time,
                                  svn=svn_text))

    def finish(self):
        super(GrassTestFilesHtmlReporter, self).finish()

        pass_per = success_to_html_percent(total=self.total,
                                           successes=self.successes)
        tfoot = ('<tfoot>'
                 '<tr>'
                 '<td>Summary</td>'
                 '<td>{nfiles} test files</td>'
                 '<td>{nsper}</td>'
                 '<td>{total}</td><td>{st}</td><td>{ft}</td><td>{pt}</td>'
                 '</tr>'
                 '</tfoot>'.format(
                     nfiles=self.test_files,
                     nsper=percent_to_html(self.file_pass_per),
                     st=self.successes, ft=self.failures + self.errors,
                     total=self.total, pt=pass_per
                     ))

        # this is the second place with this function
        # TODO: provide one implementation
        def format_percentage(percentage):
            if percentage is not None:
                return "{nsper:.0f}%".format(nsper=percentage)
            else:
                return "unknown percentage"

        summary_sentence = ('\nExecuted {nfiles} test files in {time:}.'
                            '\nFrom them'
                            ' {nsfiles} files ({nsper}) were successful'
                            ' and {nffiles} files ({nfper}) failed.\n'
                            .format(
                                nfiles=self.test_files,
                                time=self.main_time,
                                nsfiles=self.files_pass,
                                nffiles=self.files_fail,
                                nsper=format_percentage(self.file_pass_per),
                                nfper=format_percentage(self.file_fail_per)))

        self.main_index.write('<tbody>{tfoot}</table>'
                              '<p>{summary}</p>'
                              '</body></html>'
                              .format(
                                  tfoot=tfoot,
                                  summary=summary_sentence))
        self.main_index.close()

    def start_file_test(self, module):
        super(GrassTestFilesHtmlReporter, self).start_file_test(module)
        self.main_index.flush()  # to get previous lines to the report

    def end_file_test(self, module, cwd, returncode, stdout, stderr,
                      test_summary):
        super(GrassTestFilesHtmlReporter, self).end_file_test(
            module=module, cwd=cwd, returncode=returncode,
            stdout=stdout, stderr=stderr)
        # considering others according to total is OK when we more or less
        # know that input data make sense (total >= errors + failures)
        total = test_summary.get('total', None)
        failures = test_summary.get('failures', 0)
        errors = test_summary.get('errors', 0)
        # Python unittest TestResult class is reporting success for no
        # errors or failures, so skipped, expected failures and unexpected
        # success are ignored
        # but successful tests are only total - the others
        # TODO: add success counter to GrassTestResult base class
        skipped = test_summary.get('skipped', 0)
        expected_failures = test_summary.get('expected_failures', 0)
        unexpected_successes = test_summary.get('unexpected_successes', 0)
        successes = test_summary.get('successes', 0)

        self.failures += failures
        self.errors += errors
        self.skipped += skipped
        self.expected_failures += expected_failures
        self.unexpected_success += unexpected_successes

        # zero would be valid here
        if total is not None:
            # success are only the clear ones
            # percentage is influenced by all
            # but putting only failures to table
            self.successes += successes
            self.total += total
            # this will handle zero
            pass_per = success_to_html_percent(total=total,
                                               successes=successes)
        else:
            total = successes = pass_per = self.unknown_number
        bad_ones = failures + errors
        self.main_index.write(
            '<tr><td>{d}</td>'
            '<td><a href="{d}/{m}/index.html">{m}</a></td>'
            '<td>{status}</td>'
            '<td>{ntests}</td><td>{stests}</td>'
            '<td>{ftests}</td><td>{ptests}</td>'
            '<tr>'.format(
                d=to_web_path(module.tested_dir), m=module.name,
                status=returncode_to_html_text(returncode),
                stests=successes, ftests=bad_ones, ntests=total,
                ptests=pass_per))
        wrap_stdstream_to_html(infile=stdout,
                               outfile=os.path.join(cwd, 'stdout.html'),
                               module=module, stream='stdout')
        wrap_stdstream_to_html(infile=stderr,
                               outfile=os.path.join(cwd, 'stderr.html'),
                               module=module, stream='stderr')

        file_index_path = os.path.join(cwd, 'index.html')
        file_index = open(file_index_path, 'w')
        file_index.write(
            '<!DOCTYPE html><html><head><meta charset="utf-8"></head><body>'
            '<h1>{m.name}</h1>'
            '<h2>{m.tested_dir} &ndash; {m.name}</h2>'
            '{status}'
            .format(
                m=module,
                status=returncode_to_success_html_par(returncode),
                ))

        # TODO: include optionally hyper link to test suite
        # TODO: file_path is reconstucted in a naive way
        # file_path should be stored in the module/test file object and just used here
        summary_section = (
            '<table><tbody>'
            '<tr><td>Test</td><td>{m}</td></tr>'
            '<tr><td>Testsuite</td><td>{d}</td></tr>'
            '<tr><td>Test file</td><td>{file_path}</td></tr>'
            '<tr><td>Status</td><td>{status}</td></tr>'
            '<tr><td>Return code</td><td>{rc}</td></tr>'
            '<tr><td>Number of tests</td><td>{ntests}</td></tr>'
            '<tr><td>Successful tests</td><td>{stests}</td></tr>'
            '<tr><td>Failed tests</td><td>{ftests}</td></tr>'
            '<tr><td>Percent successful</td><td>{ptests}</td></tr>'
            '<tr><td>Test duration</td><td>{dur}</td></tr>'
            .format(
                d=module.tested_dir, m=module.name,
                file_path=os.path.join(module.tested_dir, 'testsuite', module.name + '.' + module.file_type),
                status=returncode_to_html_text(returncode),
                stests=successes, ftests=bad_ones, ntests=total,
                ptests=pass_per, rc=returncode,
                dur=self.file_time))
        file_index.write(summary_section)

        modules = test_summary.get('tested_modules', None)
        if modules:
            # TODO: replace by better handling of potential lists when parsing
            # TODO: create link to module if running in grass or in addons
            # alternatively a link to module test summary
            if type(modules) is not list:
                modules = [modules]
            file_index.write(
                '<tr><td>Tested modules</td><td>{0}</td></tr>'.format(
                    ', '.join(sorted(set(modules)))))
        file_index.write('</tbody></table>')

        # here we would have also links to coverage, profiling, ...
        #'<li><a href="testcodecoverage/index.html">code coverage</a></li>'
        files_section = (
            '<h3>Supplementary files</h3>'
            '<ul>'
            '<li><a href="stdout.html">standard output (stdout)</a></li>'
            '<li><a href="stderr.html">standard error output (stderr)</a></li>'
            )
        file_index.write(files_section)

        supplementary_files = test_summary.get('supplementary_files', None)
        if supplementary_files:
            # this is something we might want to do once for all and not
            # risk that it will be done twice or rely that somebody else
            # will do it for use
            # the solution is perhaps do the multi reporter more grass-specific
            # and do all common things, so that other can rely on it and
            # moreover something can be shared with other explicitly
            # using constructors as seems advantageous for counting
            self._file_anonymizer.anonymize(supplementary_files)
            for f in supplementary_files:
                file_index.write('<li><a href="{f}">{f}</a></li>'.format(f=f))

        file_index.write('</ul>')

        if returncode:
            file_index.write('<h3>Standard error output (stderr)</h3>')
            file_index.write(html_file_preview(stderr))

        file_index.write('</body></html>')
        file_index.close()

        if returncode:
            pass
            # TODO: here we don't have opportunity to write error file
            # to stream (stdout/stderr)
            # a stream can be added and if not none, we could write


# TODO: document info: additional information to be stored type: dict
# allows overwriting what was collected
class GrassTestFilesKeyValueReporter(GrassTestFilesCountingReporter):

    def __init__(self, info=None):
        super(GrassTestFilesKeyValueReporter, self).__init__()
        self.result_dir = None
        self._info = info

    def start(self, results_dir):
        super(GrassTestFilesKeyValueReporter, self).start(results_dir)
        # having all variables public although not really part of API
        self.result_dir = results_dir

        # TODO: this can be moved to the counter class
        self.failures = 0
        self.errors = 0
        self.skipped = 0
        self.successes = 0
        self.expected_failures = 0
        self.unexpected_success = 0
        self.total = 0

        # TODO: document: tested_dirs is a list and it should fit with names
        self.names = []
        self.tested_dirs = []
        self.files_returncodes = []

        # sets (no size specified)
        self.modules = set()
        self.test_files_authors = set()

    def finish(self):
        super(GrassTestFilesKeyValueReporter, self).finish()

        # this shoul be moved to some additional meta passed in constructor
        svn_info = get_svn_info()
        if not svn_info:
            svn_revision = ''
        else:
            svn_revision = svn_info['revision']

        summary = {}
        summary['files_total'] = self.test_files
        summary['files_successes'] = self.files_pass
        summary['files_failures'] = self.files_fail

        summary['names'] = self.names
        summary['tested_dirs'] = self.tested_dirs
        # TODO: we don't have a general mechanism for storing any type in text
        summary['files_returncodes'] = [str(item)
                                        for item in self.files_returncodes]

        # let's use seconds as a universal time delta format
        # (there is no standard way how to store time delta as string)
        summary['time'] = self.main_time.total_seconds()

        status = 'failed' if self.files_fail else 'succeeded'
        summary['status'] = status

        summary['total'] = self.total
        summary['successes'] = self.successes
        summary['failures'] = self.failures
        summary['errors'] = self.errors
        summary['skipped'] = self.skipped
        summary['expected_failures'] = self.expected_failures
        summary['unexpected_successes'] = self.unexpected_success

        summary['test_files_authors'] = self.test_files_authors
        summary['tested_modules'] = self.modules
        summary['svn_revision'] = svn_revision
        # ignoring issues with time zones
        summary['timestamp'] = self.main_start_time.strftime('%Y-%m-%d %H:%M:%S')
        # TODO: add some general metadata here (passed in constructor)

        # add additional information
        for key, value in self._info.items():
            summary[key] = value

        summary_filename = os.path.join(self.result_dir,
                                        'test_keyvalue_result.txt')
        with open(summary_filename, 'w') as summary_file:
            text = keyvalue_to_text(summary, sep='=', vsep='\n', isep=',')
            summary_file.write(text)

    def end_file_test(self, module, cwd, returncode, stdout, stderr,
                      test_summary):
        super(GrassTestFilesKeyValueReporter, self).end_file_test(
            module=module, cwd=cwd, returncode=returncode,
            stdout=stdout, stderr=stderr)
        # TODO: considering others according to total, OK?
        # here we are using 0 for total but HTML reporter is using None
        total = test_summary.get('total', 0)
        failures = test_summary.get('failures', 0)
        errors = test_summary.get('errors', 0)
        # Python unittest TestResult class is reporting success for no
        # errors or failures, so skipped, expected failures and unexpected
        # success are ignored
        # but successful tests are only total - the others
        skipped = test_summary.get('skipped', 0)
        expected_failures = test_summary.get('expected_failures', 0)
        unexpected_successes = test_summary.get('unexpected_successes', 0)
        successes = test_summary.get('successes', 0)

        # TODO: move this to counter class and perhaps use aggregation
        # rather then inheritance
        self.failures += failures
        self.errors += errors
        self.skipped += skipped
        self.expected_failures += expected_failures
        self.unexpected_success += unexpected_successes

        # TODO: should we test for zero?
        if total is not None:
            # success are only the clear ones
            # percentage is influenced by all
            # but putting only failures to table
            self.successes += successes
            self.total += total

        self.files_returncodes.append(returncode)

        self.tested_dirs.append(module.tested_dir)
        self.names.append(module.name)

        modules = test_summary.get('tested_modules', None)
        if modules:
            # TODO: replace by better handling of potential lists when parsing
            # TODO: create link to module if running in grass or in addons
            # alternatively a link to module test summary
            if type(modules) not in [list, set]:
                modules = [modules]
            self.modules.update(modules)

        test_file_authors = test_summary['test_file_authors']
        if type(test_file_authors) not in [list, set]:
            test_file_authors = [test_file_authors]
        self.test_files_authors.update(test_file_authors)


class GrassTestFilesTextReporter(GrassTestFilesCountingReporter):

    def __init__(self, stream):
        super(GrassTestFilesTextReporter, self).__init__()
        self._stream = stream

    def start(self, results_dir):
        super(GrassTestFilesTextReporter, self).start(results_dir)

    def finish(self):
        super(GrassTestFilesTextReporter, self).finish()

        def format_percentage(percentage):
            if percentage is not None:
                return "{nsper:.0f}%".format(nsper=percentage)
            else:
                return "unknown percentage"

        summary_sentence = ('\nExecuted {nfiles} test files in {time:}.'
                            '\nFrom them'
                            ' {nsfiles} files ({nsper}) were successful'
                            ' and {nffiles} files ({nfper}) failed.\n'
                            .format(
                                nfiles=self.test_files,
                                time=self.main_time,
                                nsfiles=self.files_pass,
                                nffiles=self.files_fail,
                                nsper=format_percentage(self.file_pass_per),
                                nfper=format_percentage(self.file_fail_per)))
        self._stream.write(summary_sentence)

    def start_file_test(self, module):
        super(GrassTestFilesTextReporter, self).start_file_test(module)
        self._stream.flush()  # to get previous lines to the report

    def end_file_test(self, module, cwd, returncode, stdout, stderr,
                      test_summary):
        super(GrassTestFilesTextReporter, self).end_file_test(
            module=module, cwd=cwd, returncode=returncode,
            stdout=stdout, stderr=stderr)

        if returncode:
            self._stream.write(
                '{m} from {d} failed'
                .format(
                    d=module.tested_dir,
                    m=module.name))
            num_failed = test_summary.get('failures', 0)
            num_failed += test_summary.get('errors', 0)
            if num_failed:
                if num_failed > 1:
                    text = ' ({f} tests failed)'
                else:
                    text = ' ({f} test failed)'
                self._stream.write(text.format(f=num_failed))
            self._stream.write('\n')
            # TODO: here we lost the possibility to include also file name
            # of the appropriate report


# TODO: there is a quite a lot duplication between this class and html reporter
# TODO: document: do not use it for two reports, it accumulates the results
# TODO: add also keyvalue summary generation?
# wouldn't this conflict with collecting data from report afterwards?
class TestsuiteDirReporter(object):
    def __init__(self, main_page_name, testsuite_page_name='index.html',
                 top_level_testsuite_page_name=None):
        self.main_page_name = main_page_name
        self.testsuite_page_name = testsuite_page_name
        self.top_level_testsuite_page_name = top_level_testsuite_page_name

        # TODO: this might be even a object which could add and validate
        self.failures = 0
        self.errors = 0
        self.skipped = 0
        self.successes = 0
        self.expected_failures = 0
        self.unexpected_successes = 0
        self.total = 0

        self.testsuites = 0
        self.testsuites_successes = 0
        self.files = 0
        self.files_successes = 0

    def report_for_dir(self, root, directory, test_files):
        # TODO: create object from this, so that it can be passed from
        # one function to another
        # TODO: put the inside of for loop to another function
        dir_failures = 0
        dir_errors = 0
        dir_skipped = 0
        dir_successes = 0
        dir_expected_failures = 0
        dir_unexpected_success = 0
        dir_total = 0
        test_files_authors = []

        file_total = 0
        file_successes = 0

        page_name = os.path.join(root, directory, self.testsuite_page_name)
        if (self.top_level_testsuite_page_name and
                os.path.abspath(os.path.join(root, directory))
                == os.path.abspath(root)):
            page_name = os.path.join(root, self.top_level_testsuite_page_name)
        page = open(page_name, 'w')
        # TODO: should we use forward slashes also for the HTML because
        # it is simpler are more consistent with the rest on MS Windows?
        head = (
            '<html><body>'
            '<h1>{name} testsuite results</h1>'
            .format(name=directory))
        tests_table_head = (
            '<h3>Test files results</h3>'
            '<table>'
            '<thead><tr>'
            '<th>Test file</th><th>Status</th>'
            '<th>Tests</th><th>Successful</td>'
            '<th>Failed</th><th>Percent successful</th>'
            '</tr></thead><tbody>'
            )
        page.write(head)
        page.write(tests_table_head)
        for test_file_name in test_files:
            # TODO: put keyvalue fine name to constant
            summary_filename = os.path.join(root, directory, test_file_name,
                                            'test_keyvalue_result.txt')
            #if os.path.exists(summary_filename):
            with open(summary_filename, 'r') as keyval_file:
                summary = text_to_keyvalue(keyval_file.read(), sep='=')
            #else:
                # TODO: write else here
            #    summary = None

            if 'total' not in summary:
                bad_ones = successes = UNKNOWN_NUMBER_HTML
                total = None
            else:
                bad_ones = summary['failures'] + summary['errors']
                successes = summary['successes']
                total = summary['total']

                self.failures += summary['failures']
                self.errors += summary['errors']
                self.skipped += summary['skipped']
                self.successes += summary['successes']
                self.expected_failures += summary['expected_failures']
                self.unexpected_successes += summary['unexpected_successes']
                self.total += summary['total']

                dir_failures += summary['failures']
                dir_errors += summary['failures']
                dir_skipped += summary['skipped']
                dir_successes += summary['successes']
                dir_expected_failures += summary['expected_failures']
                dir_unexpected_success += summary['unexpected_successes']
                dir_total += summary['total']

            # TODO: keyvalue method should have types for keys function
            # perhaps just the current post processing function is enough
            test_file_authors = summary['test_file_authors']
            if type(test_file_authors) is not list:
                test_file_authors = [test_file_authors]
            test_files_authors.extend(test_file_authors)

            file_total += 1
            file_successes += 0 if summary['returncode'] else 1

            pass_per = success_to_html_percent(total=total,
                                               successes=successes)
            row = (
                '<tr>'
                '<td><a href="{f}/index.html">{f}</a></td>'
                '<td>{status}</td>'
                '<td>{ntests}</td><td>{stests}</td>'
                '<td>{ftests}</td><td>{ptests}</td>'
                '<tr>'
                .format(
                    f=test_file_name,
                    status=returncode_to_html_text(summary['returncode']),
                    stests=successes, ftests=bad_ones, ntests=total,
                    ptests=pass_per))
            page.write(row)

        self.testsuites += 1
        self.testsuites_successes += 1 if file_successes == file_total else 0
        self.files += file_total
        self.files_successes += file_successes

        dir_pass_per = success_to_html_percent(total=dir_total,
                                               successes=dir_successes)
        file_pass_per = success_to_html_percent(total=file_total,
                                                successes=file_successes)
        tests_table_foot = (
            '</tbody><tfoot><tr>'
            '<td>Summary</td>'
            '<td>{status}</td>'
            '<td>{ntests}</td><td>{stests}</td>'
            '<td>{ftests}</td><td>{ptests}</td>'
            '</tr></tfoot></table>'
            .format(
                status=file_pass_per,
                stests=dir_successes, ftests=dir_failures + dir_errors,
                ntests=dir_total, ptests=dir_pass_per))
        page.write(tests_table_foot)
        test_authors = get_html_test_authors_table(
            directory=directory, tests_authors=test_files_authors)
        page.write(test_authors)
        page.write('</body></html>')
        page.close()

        status = success_to_html_text(total=file_total, successes=file_successes)
        row = (
            '<tr>'
            '<td><a href="{d}/{page}">{d}</a></td><td>{status}</td>'
            '<td>{nfiles}</td><td>{sfiles}</td><td>{pfiles}</td>'
            '<td>{ntests}</td><td>{stests}</td>'
            '<td>{ftests}</td><td>{ptests}</td>'
            '<tr>'
            .format(
                d=to_web_path(directory), page=self.testsuite_page_name,
                status=status,
                nfiles=file_total, sfiles=file_successes, pfiles=file_pass_per,
                stests=dir_successes, ftests=dir_failures + dir_errors,
                ntests=dir_total, ptests=dir_pass_per))
        return row

    def report_for_dirs(self, root, directories):
        # TODO: this will need chanages according to potential changes in absolute/relative paths

        page_name = os.path.join(root, self.main_page_name)
        page = open(page_name, 'w')
        head = (
            '<html><body>'
            '<h1>Testsuites results</h1>'
            )
        tests_table_head = (
            '<table>'
            '<thead><tr>'
            '<th>Testsuite</th>'
            '<th>Status</th>'
            '<th>Test files</th><th>Successful</td>'
            '<th>Percent successful</th>'
            '<th>Tests</th><th>Successful</td>'
            '<th>Failed</th><th>Percent successful</th>'
            '</tr></thead><tbody>'
            )
        page.write(head)
        page.write(tests_table_head)

        for directory, test_files in directories.items():
            row = self.report_for_dir(root=root, directory=directory,
                                      test_files=test_files)
            page.write(row)

        pass_per = success_to_html_percent(total=self.total,
                                           successes=self.successes)
        file_pass_per = success_to_html_percent(total=self.files,
                                                successes=self.files_successes)
        testsuites_pass_per = success_to_html_percent(
            total=self.testsuites, successes=self.testsuites_successes)
        tests_table_foot = (
            '<tfoot>'
            '<tr>'
            '<td>Summary</td><td>{status}</td>'
            '<td>{nfiles}</td><td>{sfiles}</td><td>{pfiles}</td>'
            '<td>{ntests}</td><td>{stests}</td>'
            '<td>{ftests}</td><td>{ptests}</td>'
            '</tr>'
            '</tfoot>'
            .format(
                status=testsuites_pass_per, nfiles=self.files,
                sfiles=self.files_successes, pfiles=file_pass_per,
                stests=self.successes, ftests=self.failures + self.errors,
                ntests=self.total, ptests=pass_per))
        page.write(tests_table_foot)
        page.write('</body></html>')