Merge pull request #1048 from SpecLad:xls-conf

modules/ts/misc/testlog_parser.py

@@ -1,6 +1,9 @@
 #!/usr/bin/env python
 
-import sys, re, os.path
+import collections
+import re
+import os.path
+import sys
 from xml.dom.minidom import parse
 
 class TestInfo(object):
@@ -159,12 +162,31 @@ class TestInfo(object):
                 return 1
         return 0
 
+# This is a Sequence for compatibility with old scripts,
+# which treat parseLogFile's return value as a list.
+class TestRunInfo(collections.Sequence):
+    def __init__(self, properties, tests):
+        self.properties = properties
+        self.tests = tests
+
+    def __len__(self):
+        return len(self.tests)
+
+    def __getitem__(self, key):
+        return self.tests[key]
+
 def parseLogFile(filename):
-    tests = []
     log = parse(filename)
-    for case in log.getElementsByTagName("testcase"):
+
-        tests.append(TestInfo(case))
+    properties = {
-    return tests
+        attr_name[3:]: attr_value
+        for (attr_name, attr_value) in log.documentElement.attributes.items()
+        if attr_name.startswith('cv_')
+    }
+
+    tests = map(TestInfo, log.getElementsByTagName("testcase"))
+
+    return TestRunInfo(properties, tests)
 
 
 if __name__ == "__main__":
@@ -173,8 +195,18 @@ if __name__ == "__main__":
         exit(0)
 
     for arg in sys.argv[1:]:
-        print "Tests found in", arg
+        print "Processing {}...".format(arg)
-        tests = parseLogFile(arg)
+
-        for t in sorted(tests):
+        run = parseLogFile(arg)
+
+        print "Properties:"
+
+        for (prop_name, prop_value) in run.properties.items():
+          print "\t{} = {}".format(prop_name, prop_value)
+
+        print "Tests:"
+
+        for t in sorted(run.tests):
             t.dump()
+
         print

modules/ts/misc/xls-report.py

@@ -1,8 +1,73 @@
 #!/usr/bin/env python
 
+"""
+    This script can generate XLS reports from OpenCV tests' XML output files.
+
+    To use it, first, create a directory for each machine you ran tests on.
+    Each such directory will become a sheet in the report. Put each XML file
+    into the corresponding directory.
+
+    Then, create your configuration file(s). You can have a global configuration
+    file (specified with the -c option), and per-sheet configuration files, which
+    must be called sheet.conf and placed in the directory corresponding to the sheet.
+    The settings in the per-sheet configuration file will override those in the
+    global configuration file, if both are present.
+
+    A configuration file must consist of a Python dictionary. The following keys
+    will be recognized:
+
+    * 'comparisons': [{'from': string, 'to': string}]
+        List of configurations to compare performance between. For each item,
+        the sheet will have a column showing speedup from configuration named
+        'from' to configuration named "to".
+
+    * 'configuration_matchers': [{'properties': {string: object}, 'name': string}]
+        Instructions for matching test run property sets to configuration names.
+
+        For each found XML file:
+
+        1) All attributes of the root element starting with the prefix 'cv_' are
+           placed in a dictionary, with the cv_ prefix stripped and the cv_module_name
+           element deleted.
+
+        2) The first matcher for which the XML's file property set contains the same
+           keys with equal values as its 'properties' dictionary is searched for.
+           A missing property can be matched by using None as the value.
+
+           Corollary 1: you should place more specific matchers before less specific
+           ones.
+
+           Corollary 2: an empty 'properties' dictionary matches every property set.
+
+        3) If a matching matcher is found, its 'name' string is presumed to be the name
+           of the configuration the XML file corresponds to. Otherwise, a warning is
+           printed. A warning is also printed if two different property sets match to the
+           same configuration name.
+
+    * 'configurations': [string]
+        List of names for compile-time and runtime configurations of OpenCV.
+        Each item will correspond to a column of the sheet.
+
+    * 'module_colors': {string: string}
+        Mapping from module name to color name. In the sheet, cells containing module
+        names from this mapping will be colored with the corresponding color. You can
+        find the list of available colors here:
+        <http://www.simplistix.co.uk/presentations/python-excel.pdf>.
+
+    * 'sheet_name': string
+        Name for the sheet. If this parameter is missing, the name of sheet's directory
+        will be used.
+
+    Note that all keys are optional, although to get useful results, you'll want to
+    specify at least 'configurations' and 'configuration_matchers'.
+
+    Finally, run the script. Use the --help option for usage information.
+"""
+
 from __future__ import division
 
 import ast
+import fnmatch
 import logging
 import numbers
 import os, os.path
@@ -17,21 +82,6 @@ import xlwt
 
 from testlog_parser import parseLogFile
 
-# To build XLS report you neet to put your xmls (OpenCV tests output) in the
-# following way:
-#
-# "root" --- folder, representing the whole XLS document. It contains several
-# subfolders --- sheet-paths of the XLS document. Each sheet-path contains it's
-# subfolders --- config-paths. Config-paths are columns of the sheet and
-# they contains xmls files --- output of OpenCV modules testing.
-# Config-path means OpenCV build configuration, including different
-# options such as NEON, TBB, GPU enabling/disabling.
-#
-# root
-# root\sheet_path
-# root\sheet_path\configuration1 (column 1)
-# root\sheet_path\configuration2 (column 2)
-
 re_image_size = re.compile(r'^ \d+ x \d+$', re.VERBOSE)
 re_data_type = re.compile(r'^ (?: 8 | 16 | 32 | 64 ) [USF] C [1234] $', re.VERBOSE)
 
@@ -45,16 +95,65 @@ no_speedup_style = no_time_style
 error_speedup_style = xlwt.easyxf('pattern: pattern solid, fore_color orange')
 header_style = xlwt.easyxf('font: bold true; alignment: horizontal centre, vertical top, wrap True')
 
-def collect_xml(collection, configuration, xml_fullname):
+class Collector(object):
-    xml_fname = os.path.split(xml_fullname)[1]
+    def __init__(self, config_match_func):
-    module = xml_fname[:xml_fname.index('_')]
+        self.__config_cache = {}
+        self.config_match_func = config_match_func
+        self.tests = {}
 
-    module_tests = collection.setdefault(module, OrderedDict())
+    # Format a sorted sequence of pairs as if it was a dictionary.
+    # We can't just use a dictionary instead, since we want to preserve the sorted order of the keys.
+    @staticmethod
+    def __format_config_cache_key(pairs):
+        return '{' + ', '.join(repr(k) + ': ' + repr(v) for (k, v) in pairs) + '}'
 
-    for test in sorted(parseLogFile(xml_fullname)):
+    def collect_from(self, xml_path):
+        run = parseLogFile(xml_path)
+
+        module = run.properties['module_name']
+
+        properties = run.properties.copy()
+        del properties['module_name']
+
+        props_key = tuple(sorted(properties.iteritems())) # dicts can't be keys
+
+        if props_key in self.__config_cache:
+            configuration = self.__config_cache[props_key]
+        else:
+            configuration = self.config_match_func(properties)
+
+            if configuration is None:
+                logging.warning('failed to match properties to a configuration: %s',
+                    Collector.__format_config_cache_key(props_key))
+            else:
+                same_config_props = [it[0] for it in self.__config_cache.iteritems() if it[1] == configuration]
+                if len(same_config_props) > 0:
+                    logging.warning('property set %s matches the same configuration %r as property set %s',
+                        Collector.__format_config_cache_key(props_key),
+                        configuration,
+                        Collector.__format_config_cache_key(same_config_props[0]))
+
+            self.__config_cache[props_key] = configuration
+
+        if configuration is None: return
+
+        module_tests = self.tests.setdefault(module, OrderedDict())
+
+        for test in run.tests:
             test_results = module_tests.setdefault((test.shortName(), test.param()), {})
             test_results[configuration] = test.get("gmean") if test.status == 'run' else test.status
 
+def make_match_func(matchers):
+    def match_func(properties):
+        for matcher in matchers:
+            if all(properties.get(name) == value
+                   for (name, value) in matcher['properties'].iteritems()):
+                return matcher['name']
+
+        return None
+
+    return match_func
+
 def main():
     arg_parser = ArgumentParser(description='Build an XLS performance report.')
     arg_parser.add_argument('sheet_dirs', nargs='+', metavar='DIR', help='directory containing perf test logs')
@@ -83,23 +182,15 @@ def main():
 
         sheet_conf = dict(global_conf.items() + sheet_conf.items())
 
-        if 'configurations' in sheet_conf:
+        config_names = sheet_conf.get('configurations', [])
-            config_names = sheet_conf['configurations']
+        config_matchers = sheet_conf.get('configuration_matchers', [])
-        else:
-            try:
-                config_names = [p for p in os.listdir(sheet_path)
-                    if os.path.isdir(os.path.join(sheet_path, p))]
-            except Exception as e:
-                logging.warning('error while determining configuration names for %s: %s', sheet_path, e)
-                continue
 
-        collection = {}
+        collector = Collector(make_match_func(config_matchers))
 
-        for configuration, configuration_path in \
+        for root, _, filenames in os.walk(sheet_path):
-                [(c, os.path.join(sheet_path, c))  for c in config_names]:
+            logging.info('looking in %s', root)
-            logging.info('processing %s', configuration_path)
+            for filename in fnmatch.filter(filenames, '*.xml'):
-            for xml_fullname in glob(os.path.join(configuration_path, '*.xml')):
+                collector.collect_from(os.path.join(root, filename))
-                collect_xml(collection, configuration, xml_fullname)
 
         sheet = wb.add_sheet(sheet_conf.get('sheet_name', os.path.basename(os.path.abspath(sheet_path))))
 
@@ -126,7 +217,7 @@ def main():
         module_styles = {module: xlwt.easyxf('pattern: pattern solid, fore_color {}'.format(color))
                          for module, color in module_colors.iteritems()}
 
-        for module, tests in sorted(collection.iteritems()):
+        for module, tests in sorted(collector.tests.iteritems()):
             for ((test, param), configs) in tests.iteritems():
                 sheet.write(row, 0, module, module_styles.get(module, xlwt.Style.default_style))
                 sheet.write(row, 1, test)