libbe/diff.py

   1 # Copyright (C) 2005-2009 Aaron Bentley and Panometrics, Inc.
   2 #                         Gianluca Montecchi <gian@grys.it>
   3 #                         W. Trevor King <wking@drexel.edu>
   4 #
   5 # This program 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 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18
  19 """Compare two bug trees."""
  20
  21 import difflib
  22 import doctest
  23
  24 from libbe import bugdir, bug, settings_object, tree
  25 from libbe.utility import time_to_str
  26
  27
  28 class DiffTree (tree.Tree):
  29     """
  30     A tree holding difference data for easy report generation.
  31     >>> bugdir = DiffTree("bugdir")
  32     >>> bdsettings = DiffTree("settings", data="target: None -> 1.0")
  33     >>> bugdir.append(bdsettings)
  34     >>> bugs = DiffTree("bugs", "bug-count: 5 -> 6")
  35     >>> bugdir.append(bugs)
  36     >>> new = DiffTree("new", "new bugs: ABC, DEF")
  37     >>> bugs.append(new)
  38     >>> rem = DiffTree("rem", "removed bugs: RST, UVW")
  39     >>> bugs.append(rem)
  40     >>> print bugdir.report_string()
  41     target: None -> 1.0
  42     bug-count: 5 -> 6
  43       new bugs: ABC, DEF
  44       removed bugs: RST, UVW
  45     >>> print "\\n".join(bugdir.paths())
  46     bugdir
  47     bugdir/settings
  48     bugdir/bugs
  49     bugdir/bugs/new
  50     bugdir/bugs/rem
  51     >>> bugdir.child_by_path("/") == bugdir
  52     True
  53     >>> bugdir.child_by_path("/bugs") == bugs
  54     True
  55     >>> bugdir.child_by_path("/bugs/rem") == rem
  56     True
  57     >>> bugdir.child_by_path("bugdir") == bugdir
  58     True
  59     >>> bugdir.child_by_path("bugdir/") == bugdir
  60     True
  61     >>> bugdir.child_by_path("bugdir/bugs") == bugs
  62     True
  63     >>> bugdir.child_by_path("/bugs").masked = True
  64     >>> print bugdir.report_string()
  65     target: None -> 1.0
  66     """
  67     def __init__(self, name, data=None, data_part_fn=str,
  68                  requires_children=False, masked=False):
  69         tree.Tree.__init__(self)
  70         self.name = name
  71         self.data = data
  72         self.data_part_fn = data_part_fn
  73         self.requires_children = requires_children
  74         self.masked = masked
  75     def paths(self, parent_path=None):
  76         paths = []
  77         if parent_path == None:
  78             path = self.name
  79         else:
  80             path = "%s/%s" % (parent_path, self.name)
  81         paths.append(path)
  82         for child in self:
  83             paths.extend(child.paths(path))
  84         return paths
  85     def child_by_path(self, path):
  86         if hasattr(path, "split"): # convert string path to a list of names
  87             names = path.split("/")
  88             if names[0] == "":
  89                 names[0] = self.name # replace root with self
  90             if len(names) > 1 and names[-1] == "":
  91                 names = names[:-1] # strip empty tail
  92         else: # it was already an array
  93             names = path
  94         assert len(names) > 0, path
  95         if names[0] == self.name:
  96             if len(names) == 1:
  97                 return self
  98             for child in self:
  99                 if names[1] == child.name:
 100                     return child.child_by_path(names[1:])
 101         if len(names) == 1:
 102             raise KeyError, "%s doesn't match '%s'" % (names, self.name)
 103         raise KeyError, "%s points to child not in %s" % (names, [c.name for c in self])
 104     def report_string(self):
 105         return "\n".join(self.report())
 106     def report(self, root=None, parent=None, depth=0):
 107         if root == None:
 108             root = self.make_root()
 109         if self.masked == True:
 110             return None
 111         data_part = self.data_part(depth)
 112         if self.requires_children == True and len(self) == 0:
 113             pass
 114         else:
 115             self.join(root, parent, data_part)
 116             if data_part != None:
 117                 depth += 1
 118         for child in self:
 119             child.report(root, self, depth)
 120         return root
 121     def make_root(self):
 122         return []
 123     def join(self, root, parent, data_part):
 124         if data_part != None:
 125             root.append(data_part)
 126     def data_part(self, depth, indent=True):
 127         if self.data == None:
 128             return None
 129         if hasattr(self, "_cached_data_part"):
 130             return self._cached_data_part
 131         data_part = self.data_part_fn(self.data)
 132         if indent == True:
 133             data_part_lines = data_part.splitlines()
 134             indent = "  "*(depth)
 135             line_sep = "\n"+indent
 136             data_part = indent+line_sep.join(data_part_lines)
 137         self._cached_data_part = data_part
 138         return data_part
 139
 140 class Diff (object):
 141     """
 142     Difference tree generator for BugDirs.
 143     >>> import copy
 144     >>> bd = bugdir.SimpleBugDir(sync_with_disk=False)
 145     >>> bd.user_id = "John Doe <j@doe.com>"
 146     >>> bd_new = copy.deepcopy(bd)
 147     >>> bd_new.target = "1.0"
 148     >>> a = bd_new.bug_from_uuid("a")
 149     >>> rep = a.comment_root.new_reply("I'm closing this bug")
 150     >>> rep.uuid = "acom"
 151     >>> rep.date = "Thu, 01 Jan 1970 00:00:00 +0000"
 152     >>> a.status = "closed"
 153     >>> b = bd_new.bug_from_uuid("b")
 154     >>> bd_new.remove_bug(b)
 155     >>> c = bd_new.new_bug("c", "Bug C")
 156     >>> d = Diff(bd, bd_new)
 157     >>> r = d.report_tree()
 158     >>> print "\\n".join(r.paths())
 159     bugdir
 160     bugdir/settings
 161     bugdir/bugs
 162     bugdir/bugs/new
 163     bugdir/bugs/new/c
 164     bugdir/bugs/rem
 165     bugdir/bugs/rem/b
 166     bugdir/bugs/mod
 167     bugdir/bugs/mod/a
 168     bugdir/bugs/mod/a/settings
 169     bugdir/bugs/mod/a/comments
 170     bugdir/bugs/mod/a/comments/new
 171     bugdir/bugs/mod/a/comments/new/acom
 172     bugdir/bugs/mod/a/comments/rem
 173     bugdir/bugs/mod/a/comments/mod
 174     >>> print r.report_string()
 175     Changed bug directory settings:
 176       target: None -> 1.0
 177     New bugs:
 178       c:om: Bug C
 179     Removed bugs:
 180       b:cm: Bug B
 181     Modified bugs:
 182       a:cm: Bug A
 183         Changed bug settings:
 184           status: open -> closed
 185         New comments:
 186           from John Doe <j@doe.com> on Thu, 01 Jan 1970 00:00:00 +0000
 187             I'm closing this bug...
 188     >>> bd.cleanup()
 189     """
 190     def __init__(self, old_bugdir, new_bugdir):
 191         self.old_bugdir = old_bugdir
 192         self.new_bugdir = new_bugdir
 193
 194     # data assembly methods
 195
 196     def _changed_bugs(self):
 197         """
 198         Search for differences in all bugs between .old_bugdir and
 199         .new_bugdir.  Returns
 200           (added_bugs, modified_bugs, removed_bugs)
 201         where added_bugs and removed_bugs are lists of added and
 202         removed bugs respectively.  modified_bugs is a list of
 203         (old_bug,new_bug) pairs.
 204         """
 205         if hasattr(self, "__changed_bugs"):
 206             return self.__changed_bugs
 207         added = []
 208         removed = []
 209         modified = []
 210         for uuid in self.new_bugdir.list_uuids():
 211             new_bug = self.new_bugdir.bug_from_uuid(uuid)
 212             try:
 213                 old_bug = self.old_bugdir.bug_from_uuid(uuid)
 214             except KeyError:
 215                 added.append(new_bug)
 216             else:
 217                 if old_bug.sync_with_disk == True:
 218                     old_bug.load_comments()
 219                 if new_bug.sync_with_disk == True:
 220                     new_bug.load_comments()
 221                 if old_bug != new_bug:
 222                     modified.append((old_bug, new_bug))
 223         for uuid in self.old_bugdir.list_uuids():
 224             if not self.new_bugdir.has_bug(uuid):
 225                 old_bug = self.old_bugdir.bug_from_uuid(uuid)
 226                 removed.append(old_bug)
 227         added.sort()
 228         removed.sort()
 229         modified.sort(self._bug_modified_cmp)
 230         self.__changed_bugs = (added, modified, removed)
 231         return self.__changed_bugs
 232     def _bug_modified_cmp(self, left, right):
 233         return cmp(left[1], right[1])
 234     def _changed_comments(self, old, new):
 235         """
 236         Search for differences in all loaded comments between the bugs
 237         old and new.  Returns
 238           (added_comments, modified_comments, removed_comments)
 239         analogous to ._changed_bugs.
 240         """
 241         if hasattr(self, "__changed_comments"):
 242             if new.uuid in self.__changed_comments:
 243                 return self.__changed_comments[new.uuid]
 244         else:
 245             self.__changed_comments = {}
 246         added = []
 247         removed = []
 248         modified = []
 249         old.comment_root.sort(key=lambda comm : comm.time)
 250         new.comment_root.sort(key=lambda comm : comm.time)
 251         old_comment_ids = [c.uuid for c in old.comments()]
 252         new_comment_ids = [c.uuid for c in new.comments()]
 253         for uuid in new_comment_ids:
 254             new_comment = new.comment_from_uuid(uuid)
 255             try:
 256                 old_comment = old.comment_from_uuid(uuid)
 257             except KeyError:
 258                 added.append(new_comment)
 259             else:
 260                 if old_comment != new_comment:
 261                     modified.append((old_comment, new_comment))
 262         for uuid in old_comment_ids:
 263             if uuid not in new_comment_ids:
 264                 new_comment = new.comment_from_uuid(uuid)
 265                 removed.append(new_comment)
 266         self.__changed_comments[new.uuid] = (added, modified, removed)
 267         return self.__changed_comments[new.uuid]
 268     def _attribute_changes(self, old, new, attributes):
 269         """
 270         Take two objects old and new, and compare the value of *.attr
 271         for attr in the list attribute names.  Returns a list of
 272           (attr_name, old_value, new_value)
 273         tuples.
 274         """
 275         change_list = []
 276         for attr in attributes:
 277             old_value = getattr(old, attr)
 278             new_value = getattr(new, attr)
 279             if old_value != new_value:
 280                 change_list.append((attr, old_value, new_value))
 281         if len(change_list) >= 0:
 282             return change_list
 283         return None
 284     def _settings_properties_attribute_changes(self, old, new,
 285                                               hidden_properties=[]):
 286         properties = sorted(new.settings_properties)
 287         for p in hidden_properties:
 288             properties.remove(p)
 289         attributes = [settings_object.setting_name_to_attr_name(None, p)
 290                       for p in properties]
 291         return self._attribute_changes(old, new, attributes)
 292     def _bugdir_attribute_changes(self):
 293         return self._settings_properties_attribute_changes( \
 294             self.old_bugdir, self.new_bugdir,
 295             ["vcs_name"]) # tweaked by bugdir.duplicate_bugdir
 296     def _bug_attribute_changes(self, old, new):
 297         return self._settings_properties_attribute_changes(old, new)
 298     def _comment_attribute_changes(self, old, new):
 299         return self._settings_properties_attribute_changes(old, new)
 300
 301     # report generation methods
 302
 303     def report_tree(self, diff_tree=DiffTree):
 304         """
 305         Pretty bare to make it easy to adjust to specific cases.  You
 306         can pass in a DiffTree subclass via diff_tree to override the
 307         default report assembly process.
 308         """
 309         if hasattr(self, "__report_tree"):
 310             return self.__report_tree
 311         bugdir_settings = sorted(self.new_bugdir.settings_properties)
 312         bugdir_settings.remove("vcs_name") # tweaked by bugdir.duplicate_bugdir
 313         root = diff_tree("bugdir")
 314         bugdir_attribute_changes = self._bugdir_attribute_changes()
 315         if len(bugdir_attribute_changes) > 0:
 316             bugdir = diff_tree("settings", bugdir_attribute_changes,
 317                                self.bugdir_attribute_change_string)
 318             root.append(bugdir)
 319         bug_root = diff_tree("bugs")
 320         root.append(bug_root)
 321         add,mod,rem = self._changed_bugs()
 322         bnew = diff_tree("new", "New bugs:", requires_children=True)
 323         bug_root.append(bnew)
 324         for bug in add:
 325             b = diff_tree(bug.uuid, bug, self.bug_add_string)
 326             bnew.append(b)
 327         brem = diff_tree("rem", "Removed bugs:", requires_children=True)
 328         bug_root.append(brem)
 329         for bug in rem:
 330             b = diff_tree(bug.uuid, bug, self.bug_rem_string)
 331             brem.append(b)
 332         bmod = diff_tree("mod", "Modified bugs:", requires_children=True)
 333         bug_root.append(bmod)
 334         for old,new in mod:
 335             b = diff_tree(new.uuid, (old,new), self.bug_mod_string)
 336             bmod.append(b)
 337             bug_attribute_changes = self._bug_attribute_changes(old, new)
 338             if len(bug_attribute_changes) > 0:
 339                 bset = diff_tree("settings", bug_attribute_changes,
 340                                  self.bug_attribute_change_string)
 341                 b.append(bset)
 342             if old.summary != new.summary:
 343                 data = (old.summary, new.summary)
 344                 bsum = diff_tree("summary", data, self.bug_summary_change_string)
 345                 b.append(bsum)
 346             cr = diff_tree("comments")
 347             b.append(cr)
 348             a,m,d = self._changed_comments(old, new)
 349             cnew = diff_tree("new", "New comments:", requires_children=True)
 350             for comment in a:
 351                 c = diff_tree(comment.uuid, comment, self.comment_add_string)
 352                 cnew.append(c)
 353             crem = diff_tree("rem", "Removed comments:",requires_children=True)
 354             for comment in d:
 355                 c = diff_tree(comment.uuid, comment, self.comment_rem_string)
 356                 crem.append(c)
 357             cmod = diff_tree("mod","Modified comments:",requires_children=True)
 358             for o,n in m:
 359                 c = diff_tree(n.uuid, (o,n), self.comment_mod_string)
 360                 cmod.append(c)
 361                 comm_attribute_changes = self._comment_attribute_changes(o, n)
 362                 if len(comm_attribute_changes) > 0:
 363                     cset = diff_tree("settings", comm_attribute_changes,
 364                                      self.comment_attribute_change_string)
 365                 if o.body != n.body:
 366                     data = (o.body, n.body)
 367                     cbody = diff_tree("cbody", data,
 368                                       self.comment_body_change_string)
 369                     c.append(cbody)
 370             cr.extend([cnew, crem, cmod])
 371         self.__report_tree = root
 372         return self.__report_tree
 373
 374     # change data -> string methods.
 375     # Feel free to play with these in subclasses.
 376
 377     def attribute_change_string(self, attribute_changes, indent=0):
 378         indent_string = "  "*indent
 379         change_strings = [u"%s: %s -> %s" % f for f in attribute_changes]
 380         for i,change_string in enumerate(change_strings):
 381             change_strings[i] = indent_string+change_string
 382         return u"\n".join(change_strings)
 383     def bugdir_attribute_change_string(self, attribute_changes):
 384         return "Changed bug directory settings:\n%s" % \
 385             self.attribute_change_string(attribute_changes, indent=1)
 386     def bug_attribute_change_string(self, attribute_changes):
 387         return "Changed bug settings:\n%s" % \
 388             self.attribute_change_string(attribute_changes, indent=1)
 389     def comment_attribute_change_string(self, attribute_changes):
 390         return "Changed comment settings:\n%s" % \
 391             self.attribute_change_string(attribute_changes, indent=1)
 392     def bug_add_string(self, bug):
 393         return bug.string(shortlist=True)
 394     def bug_rem_string(self, bug):
 395         return bug.string(shortlist=True)
 396     def bug_mod_string(self, bugs):
 397         old_bug,new_bug = bugs
 398         return new_bug.string(shortlist=True)
 399     def bug_summary_change_string(self, summaries):
 400         old_summary,new_summary = summaries
 401         return "summary changed:\n  %s\n  %s" % (old_summary, new_summary)
 402     def _comment_summary_string(self, comment):
 403         return "from %s on %s" % (comment.author, time_to_str(comment.time))
 404     def comment_add_string(self, comment):
 405         summary = self._comment_summary_string(comment)
 406         first_line = comment.body.splitlines()[0]
 407         return "%s\n  %s..." % (summary, first_line)
 408     def comment_rem_string(self, comment):
 409         summary = self._comment_summary_string(comment)
 410         first_line = comment.body.splitlines()[0]
 411         return "%s\n  %s..." % (summary, first_line)
 412     def comment_mod_string(self, comments):
 413         old_comment,new_comment = comments
 414         return self._comment_summary_string(new_comment)
 415     def comment_body_change_string(self, bodies):
 416         old_body,new_body = bodies
 417         return difflib.unified_diff(old_body, new_body)
 418
 419
 420 suite = doctest.DocTestSuite()