Am 25.11.2014 20:30 schrieb "Wolfgang Mauerer" <wm@xxxxxxxxxxxxxxxx>: > > > > Am 25/11/2014 19:28, schrieb Matthias Dittrich: > > > > On 25.11.2014 16:38, Mitchell Joblin wrote: > >> On Wed, Nov 19, 2014 at 9:40 PM, Matthias Dittrich > >> <matthi.d@xxxxxxxxxxxxxx> wrote: > >>> - The _get_feature_lines function calculates the feature sets in a > >>> similar way how _getFunctionsLines > >>> calculates the current functions for all source code lines. > >>> - Added some fields to codeface/fileCommit.py to save those results. > >>> - parse_feature_line function to parse a single line of cppstats output. > >>> - Added FileDict class to encapsulate information about the features > >>> of a source code file. > >>> - parse_feature_lines function to generate a FileDict instance from > >>> all lines of cppstats output. > >>> - We expect cppstats to be in path and check that the executable > >>> works before starting the analysis. > >>> - Update documentation. > >>> > >>> Signed-off-by: Matthias Dittrich <matthi.d@xxxxxxxxx> > >>> Reviewed-by: Wolfgang Mauerer <wolfgang.mauerer@xxxxxxxxxxx> > >>> --- > >>> README.md | 25 ++++++ > >>> codeface/VCS.py | 230 > >>> +++++++++++++++++++++++++++++++++++++++++++++++++ > >>> codeface/fileCommit.py | 70 +++++++++++++++ > >>> codeface/project.py | 24 +++++- > >>> codeface/util.py | 22 +++++ > >>> 5 files changed, 367 insertions(+), 4 deletions(-) > >>> > >>> diff --git a/README.md b/README.md > >>> index 277a67a..eced236 100644 > >>> --- a/README.md > >>> +++ b/README.md > >>> @@ -59,6 +59,31 @@ in the step "Database Setup", and modify > >>> codeface.conf accordingly. > >>> # Devel packages required for python packages > >>> sudo apt-get install libyaml-dev > >>> > >>> +* When using the feature or feature_file analysis you need to have a > >>> working > >>> + "cppstats" in your path. > >>> + One way to get it is: > >>> + > >>> + cd ~ > >>> + git clone https://github.com/clhunsen/cppstats.git > >>> + > >>> + Now create a file like ~/scripts/cppstats and add ~/scripts to > >>> your PATH. > >>> + It should have something along the lines of: > >>> + > >>> + #!/bin/bash > >>> + > >>> + cd ~/cppstats > >>> + PYTHONPATH="~/cppstats/lib" ~/cppstats/cppstats.py "$@" > >>> + > >>> + Note that the script has to be executable: > >>> + > >>> + chmod +x ~/scripts/cppstats > >>> + > >>> + and then add ~/scripts to your PATH. > >>> + (maybe you have to replace ~ with the full path (/home/$user) if > >>> it doesn't work). > >>> + > >>> + You can test this script by running "~/scripts/cppstats --help" and > >>> + validate that you get an help message > >>> + > >>> ## Preparing the R installation > >>> > >>> * Run `sudo R CMD javareconf`; make sure that the tool reports > >>> success in > >>> diff --git a/codeface/VCS.py b/codeface/VCS.py > >>> index c1060f2..079be83 100644 > >>> --- a/codeface/VCS.py > >>> +++ b/codeface/VCS.py > >>> @@ -34,15 +34,19 @@ > >>> # VCS-specific. > >>> # TODO: Unify range handling. Either a range is always a list, or > >>> always > >>> # represented by two parameters. > >>> +import itertools > >>> +import readline > >>> > >>> import commit > >>> import fileCommit > >>> import re > >>> import os > >>> +import bisect > >>> import ctags > >>> import tempfile > >>> import source_analysis > >>> import shutil > >>> +from fileCommit import FileDict > >>> from progressbar import ProgressBar, Percentage, Bar, ETA > >>> from ctags import CTags, TagEntry > >>> from logging import getLogger; log = getLogger(__name__) > >>> @@ -182,6 +186,181 @@ class VCS: > >>> return subsys=="__main__" or subsys in > >>> self.subsys_description.keys() > >>> > >>> > >>> +def parse_sep_line(line): > >>> + if not line.startswith("\"sep="): > >>> + raise ParseError( > >>> + ("expected that the csv file header starts with '\"sep=' " > >>> + "but it started with '{}'") > >>> + .format(line), 'CSVFile') > >>> + stripped = line.rstrip() > >>> + if not stripped.endswith("\""): > >>> + raise ParseError( > >>> + ("expected that the csv file header ends with '\"' " > >>> + "but the line was '{}'") > >>> + .format(line), 'CSVFile') > >>> + return stripped[5:-1] > >>> + > >>> + > >>> +def parse_line(sep, line): > >>> + """ > >>> + Parses a line from a csv file > >>> + :param sep: > >>> + :param line: > >>> + :return: > >>> + """ > >>> + # TODO: Handle escaping: sep is escaped with quotes, quotes are > >>> escaped with quotes > >>> + # 'test,test' will be '"test,test"' in the csv file > >>> + # 'test"this,"test' will be '"test""this,""test"' in the csv file > >>> + return [l.strip() for l in line.split(sep)] > >>> + > >>> + > >>> +class LineType: > >>> + IF = "#if" > >>> + ELSE = "#else" > >>> + ELIF = "#elif" > >>> + > >>> + > >>> +def parse_feature_line(sep, line): > >>> + """ > >>> + parse the current line which is something like: > >>> + FILENAME,LINE_START,LINE_END,TYPE,EXPRESSION,CONSTANTS > >>> + :param line: the line to parse > >>> + :return: start_line, end_line, line_type, feature_list > >>> + """ > >>> + parsed_line = parse_line(sep, line) > >>> + # FILENAME,LINE_START,LINE_END,TYPE,EXPRESSION,CONSTANTS > >>> + try: > >>> + start_line = int(parsed_line[1]) > >>> + end_line = int(parsed_line[2]) > >>> + line_type_raw = parsed_line[3] > >>> + if line_type_raw not in (LineType.IF, LineType.ELSE, > >>> LineType.ELIF): > >>> + raise ParseError( > >>> + ("could not parse feature line (because we could" > >>> + "not parse the line_type): \"{}\"") > >>> + .format(line), 'CSVFile') > >>> + line_type = line_type_raw > >>> + feature_list = parsed_line[5].split(';') > >>> + return start_line, end_line, line_type, feature_list > >>> + except ValueError: > >>> + raise ParseError( > >>> + ("could not parse feature line (most likely because we " > >>> + "could not parse the start- or end-line which should " > >>> + "be on index 2 and 3): \"{}\"") > >>> + .format(line), 'CSVFile') > >>> + > >>> + > >>> +def get_feature_lines(parsed_lines, filename): > >>> + """ > >>> + calculates an dictionary representing the feature sets for any line > >>> + of the given file. > >>> + :param parsed_lines: a list of tuples with > >>> + (start_line, end_line, line_type, feature_list) elements > >>> + :param filename: the name or the analysed files > >>> + (only used for descriptive error messages if the calculation fails) > >>> + :return: > >>> + feature_lines: a FileDict object to access the feature sets on > >>> any line > >>> + """ > >>> + # mapping line -> feature list, we only add changing elements > >>> + feature_lines = FileDict() > >>> + feature_lines.add_line(0, []) > >>> + > >>> + # we want a format like (is_start, features) for every line with an > >>> + # #ifdef (ie. line that changes the feature set) > >>> + annotated_lines = {} > >>> + > >>> + def check_line(line): > >>> + if line in annotated_lines: > >>> + raise ParseError( > >>> + ("every line index can be used at most once " > >>> + "(problematic line was {0} in file {1})") > >>> + .format(line, filename), filename) > >>> + > >>> + # We now transform the cppstats output in another output which will > >>> + # help to implement the algorithm below in a simple and fast way. > >>> + # The old format is a list of > >>> + # (start_line, end_line, line_type, feature_list) tuples for every > >>> + # #ifdef/#else. > >>> + # The new format is a list of (is_start, feature_set) > >>> + # for every #ifdef(/#else)/#endif > >>> + # We try to ignore #else wherever possible or handle > >>> + # the #else like a nested #if. > >>> + for start_line, end_line, line_type, feature_list in parsed_lines: > >>> + if start_line >= end_line: > >>> + raise ParseError( > >>> + ("start_line can't be greater or equal to end_line " > >>> + "(problematic line was {0} in file {1})") > >>> + .format(start_line, filename), filename) > >>> + > >>> + if line_type == LineType.IF: > >>> + # ifs start on their own line, however the end_line could > >>> + # already be used by the start of an else/elif > >>> + # (#else is the end of the previous #if > >>> + # and the start of another '#if') > >>> + check_line(start_line) > >>> + if end_line in annotated_lines: > >>> + # in that case we just say the #else line belongs to > >>> the > >>> + # virtual starting '#if' > >>> + end_line -= 1 > >>> + # Now end_line should be unused > >>> + check_line(end_line) > >>> + annotated_lines[start_line] = (True, feature_list) > >>> + annotated_lines[end_line] = (False, feature_list) > >>> + else: > >>> + # we try to mostly ignore else and elif if the feature_ > >>> + # list doesn't change > >>> + is_start, old_feature_list = annotated_lines[start_line] > >>> + if (not is_start) and old_feature_list == feature_list: > >>> + # We are on an ELSE, however the feature list did not > >>> + # change so we just delete the current line and move > >>> the > >>> + # list to the new end > >>> + del annotated_lines[start_line] > >>> + annotated_lines[end_line] = is_start, old_feature_list > >>> + elif is_start: > >>> + raise ParseError( > >>> + ("line {0} appeared twice as start line " > >>> + "(problematic file was {1})") > >>> + .format(start_line, filename), filename) > >>> + else: > >>> + # So we have a elif with different features, > >>> + # so we start more features now end add them to the > >>> ending > >>> + # later > >>> + # (so we handle this as if there was a new #ifdef > >>> started) > >>> + del annotated_lines[start_line] > >>> + annotated_lines[start_line] = (True, feature_list) > >>> + annotated_lines[end_line] = \ > >>> + (False, old_feature_list + feature_list) > >>> + > >>> + # Now that we have calculated the annotated_lines we just > >>> calculate the > >>> + # feature sets on those lines and save them in a FileDict instance. > >>> + # We can always access the last feature_list with the FileDict > >>> + # (because we sorted the lines) > >>> + for line in sorted(annotated_lines): > >>> + is_start, features = annotated_lines[line] > >>> + # Get last info > >>> + last_feature_list = feature_lines.get_line_info_raw(line) > >>> + # Copy last list and create new list for current line > >>> + new_feature_list = list(last_feature_list) > >>> + if is_start: > >>> + # if the current line starts a new list of features, > >>> + # we just need to add those to > >>> + # the new list (note that order matters in this case). > >>> + for r in features: > >>> + new_feature_list.insert(0, r) > >>> + else: > >>> + # if the current line ends a list of features, > >>> + # we remove them from the list > >>> + # (reverse order as adding). > >>> + for r in reversed(features): > >>> + item = new_feature_list.pop(0) > >>> + assert(item == r) > >>> + # Remove in next line > >>> + # (because we want to count the current #endif line as > >>> well). > >>> + line += 1 > >>> + > >>> + feature_lines.add_line(line, new_feature_list) > >>> + return feature_lines > >>> + > >>> + > >>> class gitVCS (VCS): > >>> def __init__(self): > >>> VCS.__init__(self) # Python OOP braindamage > >>> @@ -1067,6 +1246,57 @@ class gitVCS (VCS): > >>> src_line_rmv = re.sub(rmv_char, ' ', src_line.strip()) > >>> file_commit.addFuncImplLine(line_num, src_line_rmv) > >>> > >>> + @staticmethod > >>> + def _get_feature_lines(file_layout_src, file_commit): > >>> + """ > >>> + similar to _getFunctionLines but computes the line numbers > >>> of each > >>> + feature in the file. > >>> + """ > >>> + ''' > >>> + - Input - > >>> + file_layout_src: > >>> + dictionary with 'key=line number' and 'value=line of code' > >>> + file_commit: fileCommit instance where the results will be > >>> stored > >>> + > >>> + - Description - > >>> + The file_layout is used to construct a source code file that > >>> can be > >>> + parsed by cppstats to generate a cppstats csv file. > >>> + The cppstats csv file is then accessed to extract the > >>> feature sets > >>> + and line numbers to be saved in the fileCommit object > >>> + ''' > >>> + > >>> + # grab the file extension to determine the language of the file > >>> + fileExt = os.path.splitext(file_commit.filename)[1] > >>> + > >>> + # temporary file where we write transient data needed for ctags > >>> + srcFile = tempfile.NamedTemporaryFile(suffix=fileExt) > >>> + featurefile = tempfile.NamedTemporaryFile(suffix=".csv") > >>> + # generate a source code file from the file_layout_src > >>> dictionary > >>> + # and save it to a temporary location > >>> + for line in file_layout_src: > >>> + srcFile.write(line) > >>> + srcFile.flush() > >>> + > >>> + # run cppstats analysis on the file to get the feature > >>> locations > >>> + cmd = "/usr/bin/env cppstats --kind featurelocations --file > >>> {0} {1}"\ > >>> + .format(srcFile.name, featurefile.name).split() > >>> + output = execute_command(cmd).splitlines() > >>> + > >>> + results_file = open(featurefile.name, 'r') > >>> + sep = parse_sep_line(next(results_file)) > >>> + headlines = parse_line(sep, next(results_file)) > >>> + feature_lines = \ > >>> + get_feature_lines( > >>> + [parse_feature_line(sep, line) for line in > >>> results_file], > >>> + file_commit.filename) > >>> + > >>> + # clean up temporary files > >>> + srcFile.close() > >>> + featurefile.close() > >>> + > >>> + # save result to the file commit instance > >>> + file_commit.set_feature_infos(feature_lines) > >>> + > >>> def cmtHash2CmtObj(self, cmtHash): > >>> ''' > >>> input: cmtHash > >>> diff --git a/codeface/fileCommit.py b/codeface/fileCommit.py > >>> index aa99e84..6474669 100644 > >>> --- a/codeface/fileCommit.py > >>> +++ b/codeface/fileCommit.py > >>> @@ -23,6 +23,67 @@ single file.''' > >>> import commit > >>> import bisect > >>> > >>> + > >>> +class FileDict: > >>> + """ > >>> + A generic dictionary for saving per-line information. > >>> + We assume that this information is available on any line, > >>> + and that the information only changes on some lines. > >>> + So we only save the information on lines that change that info > >>> + and use bisect to retrieve that information (for any line). > >>> + """ > >>> + def __init__(self, line_list, line_dict): > >>> + """ > >>> + :rtype : FileDict > >>> + """ > >>> + self.line_list = line_list > >>> + self.line_dict = line_dict > >>> + self.lastItem = line_list[-1] > >>> + > >>> + def __init__(self): > >>> + """ > >>> + :rtype : FileDict > >>> + """ > >>> + self.line_list = [] > >>> + self.line_dict = {} > >>> + self.lastItem = -1 > >>> + > >>> + def __iter__(self): > >>> + return self.line_dict.__iter__() > >>> + > >>> + def get_line_info_raw(self, line_nr): > >>> + """ > >>> + Returns the info for the given line > >>> + (if the line was never set, the info for the last set line > >>> + is returned) > >>> + :param line_nr: the line to retrieve the information for. > >>> + :return: the information for the given line. > >>> + """ > >>> + i = bisect.bisect_right(self.line_list, line_nr) > >>> + info_line = self.line_list[i-1] > >>> + return self.line_dict[info_line] > >>> + > >>> + def get_line_info(self, line_nr): > >>> + return set(self.get_line_info_raw(line_nr)) > >>> + > >>> + def add_line(self, line_nr, info): > >>> + """ > >>> + Add the given information to the current dictionary. > >>> + Note: while filling the dictionary the line_nr argument has to > >>> + be incremented (this is only to make sure the caller > >>> + gets the intended behavior)! > >>> + :param line_nr: the line number of the information > >>> + :param info: the information for the current line > >>> + """ > >>> + if line_nr < self.lastItem: > >>> + raise ValueError("can only incrementally add items") > >>> + self.line_list.append(line_nr) > >>> + self.line_dict[line_nr] = info > >>> + > >>> + def values(self): > >>> + return self.line_dict.values() > >>> + > >>> + > >>> class FileCommit: > >>> def __init__(self): > >>> > >>> @@ -57,6 +118,9 @@ class FileCommit: > >>> # meta data > >>> self._src_elem_list = [] > >>> > >>> + # dictionary with key = line number, value = feature list > >>> + self.feature_info = FileDict() > >>> + > >>> #Getter/Setters > >>> def getFileSnapShots(self): > >>> return self.fileSnapShots > >>> @@ -84,6 +148,9 @@ class FileCommit: > >>> def setSrcElems(self, src_elem_list): > >>> self._src_elem_list.extend(src_elem_list) > >>> > >>> + def set_feature_infos(self, feature_line_infos): > >>> + self.feature_info = feature_line_infos > >>> + > >>> #Methods > >>> def addFileSnapShot(self, key, dict): > >>> self.fileSnapShots[key] = dict > >>> @@ -116,3 +183,6 @@ class FileCommit: > >>> def addFuncImplLine(self, lineNum, srcLine): > >>> id = self.findFuncId(lineNum) > >>> self.functionImpl[id].append(srcLine) > >>> + > >>> + def findFeatureList(self, lineNum): > >>> + return self.feature_info.get_line_info(int(lineNum)) > >>> diff --git a/codeface/project.py b/codeface/project.py > >>> index a311fb7..1d8c0b6 100644 > >>> --- a/codeface/project.py > >>> +++ b/codeface/project.py > >>> @@ -23,7 +23,7 @@ from .configuration import Configuration > >>> from .cluster.cluster import doProjectAnalysis > >>> from .ts import dispatch_ts_analysis > >>> from .util import (execute_command, generate_reports, layout_graph, > >>> - check4ctags, BatchJobPool, generate_analysis_windows) > >>> + check4ctags, check4cppstats, BatchJobPool, > >>> generate_analysis_windows) > >>> > >>> def loginfo(msg): > >>> ''' Pickleable function for multiprocessing ''' > >>> @@ -54,7 +54,20 @@ def project_analyse(resdir, gitdir, codeface_conf, > >>> project_conf, > >>> no_report, loglevel, logfile, recreate, > >>> profile_r, n_jobs): > >>> pool = BatchJobPool(int(n_jobs)) > >>> conf = Configuration.load(codeface_conf, project_conf) > >>> - project, tagging = conf["project"], conf["tagging"] > >>> + tagging = conf["tagging"] > >>> + if collab_type is not "default": > >>> + # as collab_type is ignored on some tagging values we should > >>> either > >>> + # => throw an exception to tell the user he specified > >>> something weird > >>> + # => set tagging to something valid > >>> + if tagging is not "proximity": > >>> + log.warn("tagging value is overwritten to proximity > >>> because of --collaboration") > >>> + tagging = "proximity" > >>> + conf["tagging"] = tagging > >>> + else: > >>> + # default is function > >>> + collab_type = "function" > >>> + > >>> + project = conf["project"] > >>> repo = pathjoin(gitdir, conf["repo"], ".git") > >>> project_resdir = pathjoin(resdir, project, tagging) > >>> range_by_date = False > >>> @@ -67,8 +80,11 @@ def project_analyse(resdir, gitdir, codeface_conf, > >>> project_conf, > >>> range_by_date = True > >>> > >>> # TODO: Sanity checks (ensure that git repo dir exists) > >>> - if 'proximity' == conf["tagging"]: > >>> - check4ctags() > >>> + if 'proximity' == tagging: > >>> + if collab_type is 'function': > >>> + check4ctags() > >>> + else: > >>> + check4cppstats() > >>> > >>> project_id, dbm, all_range_ids = project_setup(conf, recreate) > >>> > >>> diff --git a/codeface/util.py b/codeface/util.py > >>> index 7b8e4ab..9de4b5b 100644 > >>> --- a/codeface/util.py > >>> +++ b/codeface/util.py > >>> @@ -381,6 +381,28 @@ def check4ctags(): > >>> log.error("Ctags version '{0}' not > >>> found".format(prog_version)) > >>> raise Exception("Incompatible ctags-exuberant version") > >>> > >>> + > >>> +def check4cppstats(): > >>> + """ > >>> + check if the appropriate cppstats is installed on the system. > >>> + """ > >>> + # We can not check the version directly as there is no version > >>> switch on cppstats > >>> + # We just check if the first two lines of --help are OK. > >>> + line_1 = "usage: cppstats.py [-h] [--kind <K> | -a] [--list > >>> [LIST] | --file IN OUT]" > >>> + line_2 = " [--nobak] [--stf] [-l] [-v] > >>> [--check CHECK] [--dall]" > >>> + cmd = "/usr/bin/env cppstats --help".split() > >>> + res = execute_command(cmd).splitlines() > >> The check for cppstats fails even though I have it installed. The > >> problem seems to be that the line breaking is not consistent with your > >> checks for line_1 and line_2. I added a log statement on the "res" > >> variable and the output is below. > >> > >> 2014-11-25 17:30:12 [codeface.util] MainProcess DEBUG: Running > >> command: /usr/bin/env cppstats --help > >> 2014-11-25 17:30:13 [codeface.util] MainProcess ERROR: ['usage: > >> cppstats.py [-h] [--kind <K> | -a] [--list [LIST] | --file IN OUT] > >> [--nobak] [--norewriteifdefs] [-l] [-v] [--check CHECK] [--dall]', '', > >> 'optional arguments:', ' -h, --help show this help message > >> and exit', ' --kind <K> the preparation to be performed > >> [default: general]', ' -a, --all perform all available kinds > >> of preparation [default: False]', ' --list [LIST] a file that > >> contains the list of input projects/folders [default: > >> cppstats_input.txt]', ' --file IN OUT a source file IN that is > >> prepared and analyzed, the analysis results are written to OUT', ' > >> (--list is the default)', ' --nobak do not > >> backup files during preparation [default: False]', '', 'POSSIBLE KINDS > >> OF ANALYSES <K>:', ' general, generalvalues, discipline, > >> featurelocations, derivative, interaction', '', "OPTIONS FOR ANALYSIS > >> 'GENERALVALUES':", ' --norewriteifdefs rewrite nested #ifdefs and > >> #elifs as a conjunction of inner and outer expressions > >> [default=True]', ' (exception are #else tags, > >> which ARE rewritten as negation of the #if branch! see also > >> --norewriteelse of analysis GENERALVALUES)', '', "OPTIONS FOR ANALYSIS > >> 'DISCIPLINE':", ' This analysis counts the number of the disciplined > >> CPP usage in software projects. ', ' To this end, it checks xml > >> representations of header and source files and returns the number of > >> disciplined ifdefs in those. ', '', ' -l, --log log to > >> stdout [default=True]', ' -v, --verbose verbose output > >> [default=False]', ' --check CHECK CHECK sets the patterns that > >> are checked [default=1].', ' Supply sum of wanted > >> patterns:', ' (1) check top level siblings > >> (compilation unit) ', ' (2) check sibling > >> (excludes check top level siblings; NOT CLASSIFIED) ', ' > >> (4) check if-then enframement (wrapper) ', ' > >> (8) check case enframement (conditional) ', ' (16) > >> check else-if enframement (conditional) ', ' (32) > >> check param/argument enframement (parameter) ', ' > >> (64) check expression enframement (expression) ', ' > >> (128) check else enframement (NOT CLASSIFIED) ', ' --dall > >> check all patterns [default=True] ', ' (overrides > >> --check)'] > >> 2014-11-25 17:30:13 [codeface.util] MainProcess ERROR: program > >> cppstats does not exist, or it is not working as expected (expected ' > >> [--nobak] [--stf] [-l] [-v] [--check CHECK] [--dall]' > >> in the second line but got '' > >> Traceback (most recent call last): > >> File "/home/au/.local/bin/codeface", line 9, in <module> > >> load_entry_point('codeface==0.2.0', 'console_scripts', 'codeface')() > >> File "/home/au/workspace/codeface/codeface/cli.py", line 197, in main > >> return run(sys.argv) > >> File "/home/au/workspace/codeface/codeface/cli.py", line 193, in run > >> return args.func(args) > >> File "/home/au/workspace/codeface/codeface/cli.py", line 112, in > >> cmd_run > >> args.profile_r, args.jobs, args.tagging) > >> File "/home/au/workspace/codeface/codeface/project.py", line 88, in > >> project_analyse > >> check4cppstats() > >> File "/home/au/workspace/codeface/codeface/util.py", line 411, in > >> check4cppstats > >> raise Exception("cppstats not found ({0})".format(error_message)) > >> Exception: cppstats not found (expected ' [--nobak] > >> [--stf] [-l] [-v] [--check CHECK] [--dall]' in the second line but got > >> '') > >> > >> I guess you never experienced this on your system? Maybe this check is > >> not very robust because it relies on having consistent line breaks. > >> Perhaps we need another solution. > > Yep it is working fine on my system, maybe the output depends on some > > system defined console width. > > Because cppstats has no --version switch I guess it's fine if we check > > if the output starts with "usage: cppstats.py [-h]". > > If you have an idea for a better check, please let me know. > > Instead of coming up with voodoo checks for certain help message > patterns (that would, for instance, also break if cppstats were ever > localised), how about fixing the issue upstream? You could extend > cppstats so that there are means of showing a version number, and then > see if this produces the expected output. Yes, I agree that this would be the best solution. Anyway cppstats is maintained by people in your group so that shouldn't be difficult to manage. --Mitchell > > Best regards, Wolfgang > > --Matthias > >> > >> --Mitchell > >> > >> > >>> + if not (res[0].startswith(line_1)): > >>> + error_message = "expected '{0}' in the first line but got > >>> '{1}'".format(line_1, res[0]) > >>> + log.error("program cppstats does not exist, or it is not > >>> working as expected ({0}".format(error_message)) > >>> + raise Exception("cppstats not found > >>> ({0})".format(error_message)) > >>> + > >>> + if not (res[1].startswith(line_2)): > >>> + error_message = "expected '{0}' in the second line but got > >>> '{1}'".format(line_2, res[1]) > >>> + log.error("program cppstats does not exist, or it is not > >>> working as expected ({0}".format(error_message)) > >>> + raise Exception("cppstats not found > >>> ({0})".format(error_message)) > >>> + > >>> + > >>> def generate_analysis_windows(repo, window_size_months): > >>> ''' > >>> Generates a list of revisions (commit hash) in increments of > >>> the window_size > >>> -- > >>> 1.8.5.5 > >>> > >>> > > > > >