1#!/usr/bin/python
2#
3# Copyright (C) 2014 Free Software Foundation, Inc.
4#
5# This script is free software; you can redistribute it and/or modify
6# it under the terms of the GNU General Public License as published by
7# the Free Software Foundation; either version 3, or (at your option)
8# any later version.
9
10import sys
11import getopt
12import re
13import io
14from datetime import datetime
15from operator import attrgetter
16
17# True if unrecognised lines should cause a fatal error.  Might want to turn
18# this on by default later.
19strict = False
20
21# True if the order of .log segments should match the .sum file, false if
22# they should keep the original order.
23sort_logs = True
24
25# A version of open() that is safe against whatever binary output
26# might be added to the log.
27def safe_open (filename):
28    if sys.version_info >= (3, 0):
29        return open (filename, 'r', errors = 'surrogateescape')
30    return open (filename, 'r')
31
32# Force stdout to handle escape sequences from a safe_open file.
33if sys.version_info >= (3, 0):
34    sys.stdout = io.TextIOWrapper (sys.stdout.buffer,
35                                   errors = 'surrogateescape')
36
37class Named:
38    def __init__ (self, name):
39        self.name = name
40
41class ToolRun (Named):
42    def __init__ (self, name):
43        Named.__init__ (self, name)
44        # The variations run for this tool, mapped by --target_board name.
45        self.variations = dict()
46
47    # Return the VariationRun for variation NAME.
48    def get_variation (self, name):
49        if name not in self.variations:
50            self.variations[name] = VariationRun (name)
51        return self.variations[name]
52
53class VariationRun (Named):
54    def __init__ (self, name):
55        Named.__init__ (self, name)
56        # A segment of text before the harness runs start, describing which
57        # baseboard files were loaded for the target.
58        self.header = None
59        # The harnesses run for this variation, mapped by filename.
60        self.harnesses = dict()
61        # A list giving the number of times each type of result has
62        # been seen.
63        self.counts = []
64
65    # Return the HarnessRun for harness NAME.
66    def get_harness (self, name):
67        if name not in self.harnesses:
68            self.harnesses[name] = HarnessRun (name)
69        return self.harnesses[name]
70
71class HarnessRun (Named):
72    def __init__ (self, name):
73        Named.__init__ (self, name)
74        # Segments of text that make up the harness run, mapped by a test-based
75        # key that can be used to order them.
76        self.segments = dict()
77        # Segments of text that make up the harness run but which have
78        # no recognized test results.  These are typically harnesses that
79        # are completely skipped for the target.
80        self.empty = []
81        # A list of results.  Each entry is a pair in which the first element
82        # is a unique sorting key and in which the second is the full
83        # PASS/FAIL line.
84        self.results = []
85
86    # Add a segment of text to the harness run.  If the segment includes
87    # test results, KEY is an example of one of them, and can be used to
88    # combine the individual segments in order.  If the segment has no
89    # test results (e.g. because the harness doesn't do anything for the
90    # current configuration) then KEY is None instead.  In that case
91    # just collect the segments in the order that we see them.
92    def add_segment (self, key, segment):
93        if key:
94            assert key not in self.segments
95            self.segments[key] = segment
96        else:
97            self.empty.append (segment)
98
99class Segment:
100    def __init__ (self, filename, start):
101        self.filename = filename
102        self.start = start
103        self.lines = 0
104
105class Prog:
106    def __init__ (self):
107        # The variations specified on the command line.
108        self.variations = []
109        # The variations seen in the input files.
110        self.known_variations = set()
111        # The tools specified on the command line.
112        self.tools = []
113        # Whether to create .sum rather than .log output.
114        self.do_sum = True
115        # Regexps used while parsing.
116        self.test_run_re = re.compile (r'^Test Run By (\S+) on (.*)$')
117        self.tool_re = re.compile (r'^\t\t=== (.*) tests ===$')
118        self.result_re = re.compile (r'^(PASS|XPASS|FAIL|XFAIL|UNRESOLVED'
119                                     r'|WARNING|ERROR|UNSUPPORTED|UNTESTED'
120                                     r'|KFAIL):\s*(.+)')
121        self.completed_re = re.compile (r'.* completed at (.*)')
122        # Pieces of text to write at the head of the output.
123        # start_line is a pair in which the first element is a datetime
124        # and in which the second is the associated 'Test Run By' line.
125        self.start_line = None
126        self.native_line = ''
127        self.target_line = ''
128        self.host_line = ''
129        self.acats_premable = ''
130        # Pieces of text to write at the end of the output.
131        # end_line is like start_line but for the 'runtest completed' line.
132        self.acats_failures = []
133        self.version_output = ''
134        self.end_line = None
135        # Known summary types.
136        self.count_names = [
137            '# of DejaGnu errors\t\t',
138            '# of expected passes\t\t',
139            '# of unexpected failures\t',
140            '# of unexpected successes\t',
141            '# of expected failures\t\t',
142            '# of unknown successes\t\t',
143            '# of known failures\t\t',
144            '# of untested testcases\t\t',
145            '# of unresolved testcases\t',
146            '# of unsupported tests\t\t'
147        ]
148        self.runs = dict()
149
150    def usage (self):
151        name = sys.argv[0]
152        sys.stderr.write ('Usage: ' + name
153                          + ''' [-t tool] [-l variant-list] [-L] log-or-sum-file ...
154
155    tool           The tool (e.g. g++, libffi) for which to create a
156                   new test summary file.  If not specified then output
157                   is created for all tools.
158    variant-list   One or more test variant names.  If the list is
159                   not specified then one is constructed from all
160                   variants in the files for <tool>.
161    sum-file       A test summary file with the format of those
162                   created by runtest from DejaGnu.
163    If -L is used, merge *.log files instead of *.sum.  In this
164    mode the exact order of lines may not be preserved, just different
165    Running *.exp chunks should be in correct order.
166''')
167        sys.exit (1)
168
169    def fatal (self, what, string):
170        if not what:
171            what = sys.argv[0]
172        sys.stderr.write (what + ': ' + string + '\n')
173        sys.exit (1)
174
175    # Parse the command-line arguments.
176    def parse_cmdline (self):
177        try:
178            (options, self.files) = getopt.getopt (sys.argv[1:], 'l:t:L')
179            if len (self.files) == 0:
180                self.usage()
181            for (option, value) in options:
182                if option == '-l':
183                    self.variations.append (value)
184                elif option == '-t':
185                    self.tools.append (value)
186                else:
187                    self.do_sum = False
188        except getopt.GetoptError as e:
189            self.fatal (None, e.msg)
190
191    # Try to parse time string TIME, returning an arbitrary time on failure.
192    # Getting this right is just a nice-to-have so failures should be silent.
193    def parse_time (self, time):
194        try:
195            return datetime.strptime (time, '%c')
196        except ValueError:
197            return datetime.now()
198
199    # Parse an integer and abort on failure.
200    def parse_int (self, filename, value):
201        try:
202            return int (value)
203        except ValueError:
204            self.fatal (filename, 'expected an integer, got: ' + value)
205
206    # Return a list that represents no test results.
207    def zero_counts (self):
208        return [0 for x in self.count_names]
209
210    # Return the ToolRun for tool NAME.
211    def get_tool (self, name):
212        if name not in self.runs:
213            self.runs[name] = ToolRun (name)
214        return self.runs[name]
215
216    # Add the result counts in list FROMC to TOC.
217    def accumulate_counts (self, toc, fromc):
218        for i in range (len (self.count_names)):
219            toc[i] += fromc[i]
220
221    # Parse the list of variations after 'Schedule of variations:'.
222    # Return the number seen.
223    def parse_variations (self, filename, file):
224        num_variations = 0
225        while True:
226            line = file.readline()
227            if line == '':
228                self.fatal (filename, 'could not parse variation list')
229            if line == '\n':
230                break
231            self.known_variations.add (line.strip())
232            num_variations += 1
233        return num_variations
234
235    # Parse from the first line after 'Running target ...' to the end
236    # of the run's summary.
237    def parse_run (self, filename, file, tool, variation, num_variations):
238        header = None
239        harness = None
240        segment = None
241        final_using = 0
242
243        # If this is the first run for this variation, add any text before
244        # the first harness to the header.
245        if not variation.header:
246            segment = Segment (filename, file.tell())
247            variation.header = segment
248
249        # Parse the rest of the summary (the '# of ' lines).
250        if len (variation.counts) == 0:
251            variation.counts = self.zero_counts()
252
253        # Parse up until the first line of the summary.
254        if num_variations == 1:
255            end = '\t\t=== ' + tool.name + ' Summary ===\n'
256        else:
257            end = ('\t\t=== ' + tool.name + ' Summary for '
258                   + variation.name + ' ===\n')
259        while True:
260            line = file.readline()
261            if line == '':
262                self.fatal (filename, 'no recognised summary line')
263            if line == end:
264                break
265
266            # Look for the start of a new harness.
267            if line.startswith ('Running ') and line.endswith (' ...\n'):
268                # Close off the current harness segment, if any.
269                if harness:
270                    segment.lines -= final_using
271                    harness.add_segment (first_key, segment)
272                name = line[len ('Running '):-len(' ...\n')]
273                harness = variation.get_harness (name)
274                segment = Segment (filename, file.tell())
275                first_key = None
276                final_using = 0
277                continue
278
279            # Record test results.  Associate the first test result with
280            # the harness segment, so that if a run for a particular harness
281            # has been split up, we can reassemble the individual segments
282            # in a sensible order.
283            #
284            # dejagnu sometimes issues warnings about the testing environment
285            # before running any tests.  Treat them as part of the header
286            # rather than as a test result.
287            match = self.result_re.match (line)
288            if match and (harness or not line.startswith ('WARNING:')):
289                if not harness:
290                    self.fatal (filename, 'saw test result before harness name')
291                name = match.group (2)
292                # Ugly hack to get the right order for gfortran.
293                if name.startswith ('gfortran.dg/g77/'):
294                    name = 'h' + name
295                key = (name, len (harness.results))
296                harness.results.append ((key, line))
297                if not first_key and sort_logs:
298                    first_key = key
299                if line.startswith ('ERROR: (DejaGnu)'):
300                    for i in range (len (self.count_names)):
301                        if 'DejaGnu errors' in self.count_names[i]:
302                            variation.counts[i] += 1
303                            break
304
305            # 'Using ...' lines are only interesting in a header.  Splitting
306            # the test up into parallel runs leads to more 'Using ...' lines
307            # than there would be in a single log.
308            if line.startswith ('Using '):
309                final_using += 1
310            else:
311                final_using = 0
312
313            # Add other text to the current segment, if any.
314            if segment:
315                segment.lines += 1
316
317        # Close off the final harness segment, if any.
318        if harness:
319            segment.lines -= final_using
320            harness.add_segment (first_key, segment)
321
322        while True:
323            before = file.tell()
324            line = file.readline()
325            if line == '':
326                break
327            if line == '\n':
328                continue
329            if not line.startswith ('# '):
330                file.seek (before)
331                break
332            found = False
333            for i in range (len (self.count_names)):
334                if line.startswith (self.count_names[i]):
335                    count = line[len (self.count_names[i]):-1].strip()
336                    variation.counts[i] += self.parse_int (filename, count)
337                    found = True
338                    break
339            if not found:
340                self.fatal (filename, 'unknown test result: ' + line[:-1])
341
342    # Parse an acats run, which uses a different format from dejagnu.
343    # We have just skipped over '=== acats configuration ==='.
344    def parse_acats_run (self, filename, file):
345        # Parse the preamble, which describes the configuration and logs
346        # the creation of support files.
347        record = (self.acats_premable == '')
348        if record:
349            self.acats_premable = '\t\t=== acats configuration ===\n'
350        while True:
351            line = file.readline()
352            if line == '':
353                self.fatal (filename, 'could not parse acats preamble')
354            if line == '\t\t=== acats tests ===\n':
355                break
356            if record:
357                self.acats_premable += line
358
359        # Parse the test results themselves, using a dummy variation name.
360        tool = self.get_tool ('acats')
361        variation = tool.get_variation ('none')
362        self.parse_run (filename, file, tool, variation, 1)
363
364        # Parse the failure list.
365        while True:
366            before = file.tell()
367            line = file.readline()
368            if line.startswith ('*** FAILURES: '):
369                self.acats_failures.append (line[len ('*** FAILURES: '):-1])
370                continue
371            file.seek (before)
372            break
373
374    # Parse the final summary at the end of a log in order to capture
375    # the version output that follows it.
376    def parse_final_summary (self, filename, file):
377        record = (self.version_output == '')
378        while True:
379            line = file.readline()
380            if line == '':
381                break
382            if line.startswith ('# of '):
383                continue
384            if record:
385                self.version_output += line
386            if line == '\n':
387                break
388
389    # Parse a .log or .sum file.
390    def parse_file (self, filename, file):
391        tool = None
392        target = None
393        num_variations = 1
394        while True:
395            line = file.readline()
396            if line == '':
397                return
398
399            # Parse the list of variations, which comes before the test
400            # runs themselves.
401            if line.startswith ('Schedule of variations:'):
402                num_variations = self.parse_variations (filename, file)
403                continue
404
405            # Parse a testsuite run for one tool/variation combination.
406            if line.startswith ('Running target '):
407                name = line[len ('Running target '):-1]
408                if not tool:
409                    self.fatal (filename, 'could not parse tool name')
410                if name not in self.known_variations:
411                    self.fatal (filename, 'unknown target: ' + name)
412                self.parse_run (filename, file, tool,
413                                tool.get_variation (name),
414                                num_variations)
415                # If there is only one variation then there is no separate
416                # summary for it.  Record any following version output.
417                if num_variations == 1:
418                    self.parse_final_summary (filename, file)
419                continue
420
421            # Parse the start line.  In the case where several files are being
422            # parsed, pick the one with the earliest time.
423            match = self.test_run_re.match (line)
424            if match:
425                time = self.parse_time (match.group (2))
426                if not self.start_line or self.start_line[0] > time:
427                    self.start_line = (time, line)
428                continue
429
430            # Parse the form used for native testing.
431            if line.startswith ('Native configuration is '):
432                self.native_line = line
433                continue
434
435            # Parse the target triplet.
436            if line.startswith ('Target is '):
437                self.target_line = line
438                continue
439
440            # Parse the host triplet.
441            if line.startswith ('Host   is '):
442                self.host_line = line
443                continue
444
445            # Parse the acats premable.
446            if line == '\t\t=== acats configuration ===\n':
447                self.parse_acats_run (filename, file)
448                continue
449
450            # Parse the tool name.
451            match = self.tool_re.match (line)
452            if match:
453                tool = self.get_tool (match.group (1))
454                continue
455
456            # Skip over the final summary (which we instead create from
457            # individual runs) and parse the version output.
458            if tool and line == '\t\t=== ' + tool.name + ' Summary ===\n':
459                if file.readline() != '\n':
460                    self.fatal (filename, 'expected blank line after summary')
461                self.parse_final_summary (filename, file)
462                continue
463
464            # Parse the completion line.  In the case where several files
465            # are being parsed, pick the one with the latest time.
466            match = self.completed_re.match (line)
467            if match:
468                time = self.parse_time (match.group (1))
469                if not self.end_line or self.end_line[0] < time:
470                    self.end_line = (time, line)
471                continue
472
473            # Sanity check to make sure that important text doesn't get
474            # dropped accidentally.
475            if strict and line.strip() != '':
476                self.fatal (filename, 'unrecognised line: ' + line[:-1])
477
478    # Output a segment of text.
479    def output_segment (self, segment):
480        with safe_open (segment.filename) as file:
481            file.seek (segment.start)
482            for i in range (segment.lines):
483                sys.stdout.write (file.readline())
484
485    # Output a summary giving the number of times each type of result has
486    # been seen.
487    def output_summary (self, tool, counts):
488        for i in range (len (self.count_names)):
489            name = self.count_names[i]
490            # dejagnu only prints result types that were seen at least once,
491            # but acats always prints a number of unexpected failures.
492            if (counts[i] > 0
493                or (tool.name == 'acats'
494                    and name.startswith ('# of unexpected failures'))):
495                sys.stdout.write ('%s%d\n' % (name, counts[i]))
496
497    # Output unified .log or .sum information for a particular variation,
498    # with a summary at the end.
499    def output_variation (self, tool, variation):
500        self.output_segment (variation.header)
501        for harness in sorted (variation.harnesses.values(),
502                               key = attrgetter ('name')):
503            sys.stdout.write ('Running ' + harness.name + ' ...\n')
504            if self.do_sum:
505                harness.results.sort()
506                for (key, line) in harness.results:
507                    sys.stdout.write (line)
508            else:
509                # Rearrange the log segments into test order (but without
510                # rearranging text within those segments).
511                for key in sorted (harness.segments.keys()):
512                    self.output_segment (harness.segments[key])
513                for segment in harness.empty:
514                    self.output_segment (segment)
515        if len (self.variations) > 1:
516            sys.stdout.write ('\t\t=== ' + tool.name + ' Summary for '
517                              + variation.name + ' ===\n\n')
518            self.output_summary (tool, variation.counts)
519
520    # Output unified .log or .sum information for a particular tool,
521    # with a summary at the end.
522    def output_tool (self, tool):
523        counts = self.zero_counts()
524        if tool.name == 'acats':
525            # acats doesn't use variations, so just output everything.
526            # It also has a different approach to whitespace.
527            sys.stdout.write ('\t\t=== ' + tool.name + ' tests ===\n')
528            for variation in tool.variations.values():
529                self.output_variation (tool, variation)
530                self.accumulate_counts (counts, variation.counts)
531            sys.stdout.write ('\t\t=== ' + tool.name + ' Summary ===\n')
532        else:
533            # Output the results in the usual dejagnu runtest format.
534            sys.stdout.write ('\n\t\t=== ' + tool.name + ' tests ===\n\n'
535                              'Schedule of variations:\n')
536            for name in self.variations:
537                if name in tool.variations:
538                    sys.stdout.write ('    ' + name + '\n')
539            sys.stdout.write ('\n')
540            for name in self.variations:
541                if name in tool.variations:
542                    variation = tool.variations[name]
543                    sys.stdout.write ('Running target '
544                                      + variation.name + '\n')
545                    self.output_variation (tool, variation)
546                    self.accumulate_counts (counts, variation.counts)
547            sys.stdout.write ('\n\t\t=== ' + tool.name + ' Summary ===\n\n')
548        self.output_summary (tool, counts)
549
550    def main (self):
551        self.parse_cmdline()
552        try:
553            # Parse the input files.
554            for filename in self.files:
555                with safe_open (filename) as file:
556                    self.parse_file (filename, file)
557
558            # Decide what to output.
559            if len (self.variations) == 0:
560                self.variations = sorted (self.known_variations)
561            else:
562                for name in self.variations:
563                    if name not in self.known_variations:
564                        self.fatal (None, 'no results for ' + name)
565            if len (self.tools) == 0:
566                self.tools = sorted (self.runs.keys())
567
568            # Output the header.
569            if self.start_line:
570                sys.stdout.write (self.start_line[1])
571            sys.stdout.write (self.native_line)
572            sys.stdout.write (self.target_line)
573            sys.stdout.write (self.host_line)
574            sys.stdout.write (self.acats_premable)
575
576            # Output the main body.
577            for name in self.tools:
578                if name not in self.runs:
579                    self.fatal (None, 'no results for ' + name)
580                self.output_tool (self.runs[name])
581
582            # Output the footer.
583            if len (self.acats_failures) > 0:
584                sys.stdout.write ('*** FAILURES: '
585                                  + ' '.join (self.acats_failures) + '\n')
586            sys.stdout.write (self.version_output)
587            if self.end_line:
588                sys.stdout.write (self.end_line[1])
589        except IOError as e:
590            self.fatal (e.filename, e.strerror)
591
592Prog().main()
593