Mercurial > fife-parpg
view tests/analyzers/dep_analyzer.py @ 628:57938e939efc
* Removed unused files containing the unused GUIEventListener class.
author | prock@33b003aa-7bff-0310-803a-e67f0ece8222 |
---|---|
date | Mon, 04 Oct 2010 19:44:29 +0000 |
parents | 81641655bc38 |
children |
line wrap: on
line source
#!/usr/bin/env python # -*- coding: utf-8 -*- # #################################################################### # Copyright (C) 2005-2009 by the FIFE team # http://www.fifengine.de # This file is part of FIFE. # # FIFE is free software; you can redistribute it and/or # modify it under the terms of the GNU Lesser General Public # License as published by the Free Software Foundation; either # version 2.1 of the License, or (at your option) any later version. # # This library is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # Lesser General Public License for more details. # # You should have received a copy of the GNU Lesser General Public # License along with this library; if not, write to the # Free Software Foundation, Inc., # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA # #################################################################### import sys, re, os try: from os.shutil import copyfile except ImportError: def copyfile(src,dest): srcf = open(src) destf = open(dest,"w") destf.write( srcf.read() ) destf.close() srcf.close() if '.' not in sys.path: sys.path.append('.') from utils.util_scripts.path import path from _allowed_dependencies import ALLOWED_MODULE_DEPS _S = os.path.sep ROOTDIRNAME = 'engine%score' % _S FILE_DEPS_OUT = 'doc%sdependencies%sfiledeps' % (_S, _S) DIR_DEPS_OUT = 'doc%sdependencies%sdirdeps' % (_S, _S) MODULE_DEPS_OUT = 'doc%sdependencies%smoduledeps' % (_S, _S) SKIPPED_PROVIDERS = [] SKIPPED_USERS = [] reInc = re.compile('#include "(.*?)"') def add_to_dictset(d, key, val): try: d[key].add(val) except KeyError: d[key] = set([val]) # return error string in case of possible error unknown_modules = [] def check_dep_error(allowed_deps, user, provider): global unknown_modules msg = '' try: if not provider in allowed_deps[user]: msg = 'Illegal dependency between %s -> %s, % s can use only:\n' % (user, provider, user) for d in allowed_deps[user]: msg += ' %s\n' % d except KeyError: print unknown_modules if user not in unknown_modules: msg = 'Unknown module %s found in static check\n' % user msg += ' please adjust dep_analyzer script to match new structure' unknown_modules.append(user) return msg def get_file2inc(sources): file2inc = {} for f in sources: inComment = False for line in open(f): if not inComment and line.find('/*') != -1: inComment = True continue elif inComment: if line.find('*/') != -1: inComment = False continue elif line.strip().startswith('//'): continue m = reInc.search(line) if m: add_to_dictset(file2inc, f, m.group(1).replace( '/', _S )) return file2inc def fill_dep_infos(file2inc, fileUser2provider, dirUser2provider, moduleUser2provider, unknownIncludes, dirclusters): for f, incs in file2inc.items(): #if f.find('engine.cpp') != -1: # import pdb; pdb.set_trace() skip = False for user in SKIPPED_USERS: if f.find(user) != -1: skip = True break if skip: continue for i in incs: user = str(f.dirname()).replace(ROOTDIRNAME + _S, '') header = path((f.dirname() / path(i)).abspath().split(ROOTDIRNAME + _S)[1]) if not header.isfile(): header = path(ROOTDIRNAME) + _S + path(i) if not header.isfile(): add_to_dictset(unknownIncludes, str(f), str(i)) continue provider = str(header.dirname()).replace(ROOTDIRNAME + _S, '') skip = False for skipped in SKIPPED_PROVIDERS: if header.find(skipped) != -1: skip = True break if skip: continue add_to_dictset(dirUser2provider, user, provider) usermodule = user.split(_S)[0] providermodule = provider.split(_S)[0] userfile = user.split(_S)[-1].split('.')[0] providerfile = provider.split(_S)[-1].split('.')[0] add_to_dictset(dirclusters, usermodule, user) add_to_dictset(dirclusters, providermodule, provider) add_to_dictset(moduleUser2provider, usermodule, providermodule) add_to_dictset(fileUser2provider, userfile, providerfile) def write_dot_file(fname, contents): lines = [] a = lines.append a('digraph "source tree" {') a(' overlap=scale;') a(' size="8,10";') a(' ratio="fill";') a(' fontsize="16";') a(' fontname="Helvetica";') a(' clusterrank="local";') if type(contents) in (list, tuple): lines += contents else: lines.append(contents) a('}') open(fname, 'w').write('\n'.join(lines)) def get_cluster_str(ind, elements, label): lines = [] a = lines.append a('subgraph cluster_%d {' % ind) a(' style=filled;') a(' color=lightgrey;') a(' node [style=filled,color=white];') a(' %s' % '; '.join('"%s"' % d for d in elements)) a(' label = "%s";' % label) a('}') return '\n'.join(lines) def run_dot(basename,type): dotname = basename + ".dot" outname = basename + "." + type dotchanged = True try: olddot = open(dotname + "~").read() dotchanged = olddot != open(dotname).read() dotchanged = dotchanged or not os.path.exists(outname) except IOError: pass if not dotchanged: return print "Generating: ",outname cmd = 'dot -T%(type)s %(dotname)s > %(outname)s' % locals() os.system(cmd) copyfile(dotname,dotname + "~") def analyze(write_postscript=False): root = path(ROOTDIRNAME) headers = list(root.walkfiles('*.h')) sources = headers + list(root.walkfiles('*.cpp')) file2inc = get_file2inc(sources) moduleUser2provider = {} dirUser2provider = {} fileUser2provider = {} unknownIncludes = {} dirclusters = {} fill_dep_infos(file2inc, fileUser2provider, dirUser2provider, moduleUser2provider, unknownIncludes, dirclusters) # write module dep graph out = [] illegalModuleDeps = [] for user, providers in sorted(moduleUser2provider.items()): for provider in sorted(providers): if user != provider: out.append(' "' + user + '" -> "' + provider + '"') msg = check_dep_error(ALLOWED_MODULE_DEPS, user, provider) if msg: illegalModuleDeps.append(msg) write_dot_file('%s.dot' % MODULE_DEPS_OUT, out) if write_postscript: run_dot(MODULE_DEPS_OUT, "ps") run_dot(MODULE_DEPS_OUT,"png") # write dir dep graph out = [] for cluster, subdirs in sorted(dirclusters.items()): out.append(get_cluster_str(len(out), sorted(subdirs), cluster)) for user, providers in sorted(dirUser2provider.items()): for provider in sorted(providers): if user != provider: out.append(' "' + user + '" -> "' + provider + '"') write_dot_file('%s.dot' % DIR_DEPS_OUT, out) if write_postscript: run_dot(DIR_DEPS_OUT, "ps") # write file dep graph out = [] for user, providers in sorted(file2inc.items()): for provider in sorted(providers): if user != provider: out.append(' "' + user + '" -> "' + provider + '"') write_dot_file('%s.dot' % FILE_DEPS_OUT, out) # os.system('dot -Tps %s.dot > %s.ps' % (MODULE_DEPS_OUT, MODULE_DEPS_OUT)) # write raw dep info #out = [] #for f, file2inc result = '\n'.join(illegalModuleDeps) if result: print result else: print "no dependency analyzer errors found" return result _ANALYZE_FN_ = analyze if __name__ == '__main__': analyze(True)