[game-data-packager] 44/51: make_template: Reuse groups from the base if possible

[Simon McVittie]

This is an automated email from the git hooks/post-receive script.

smcv pushed a commit to branch master
in repository game-data-packager.

commit e3551178fd937ffe985cf211fc39d0cada4760da
Author: Simon McVittie <smcv at debian.org>
Date:   Thu Dec 28 21:52:09 2017 +0000

    make_template: Reuse groups from the base if possible
    
    Signed-off-by: Simon McVittie <smcv at debian.org>
---
 game_data_packager/make_template.py | 65 ++++++++++++++++++++++++++++++++-----
 1 file changed, 56 insertions(+), 9 deletions(-)

diff --git a/game_data_packager/make_template.py b/game_data_packager/make_template.py
index f89c957..7d18680 100644
--- a/game_data_packager/make_template.py
+++ b/game_data_packager/make_template.py
@@ -208,29 +208,76 @@ class Template:
 
     def reconcile_groups(self, files,
             stem=None, provider=None, package=None):
+        names = set()
         remaining = set()
-        groups = {}
+        groups_found = {}
         optional = set()
 
         for f in files:
+            names.add(f.name)
             remaining.add(f.name)
 
+        # If there are pre-existing groups that are strict subsets of
+        # what's left to classify, then take them out, largest first.
         if remaining:
-            main_group = self.new_group('contents of %s' % stem)
-            opt_group = self.new_group('contents of %s - optional' % stem)
+            weighted = []
+
+            for name, g in self.game.groups.items():
+                weighted.append((len(g.group_members), g.name, g))
+
+            weighted.sort(reverse=True)
+
+            for _, _, g in weighted:
+                if g.group_members <= remaining:
+                    remaining -= g.group_members
+                    groups_found[g.name] = g
+
+        # If there are pre-existing groups with a non-trivial
+        # intersection with what's left to classify, then take those,
+        # largest intersection first.
+        while remaining:
+            weighted = []
+
+            for name, g in self.game.groups.items():
+                intersection = g.group_members & remaining
+
+                # Arbitrarily decide that overlaps of less than 5 are
+                # uninteresting
+                if len(intersection) >= 5:
+                    weighted.append((len(intersection), g.name, g))
+
+            if not weighted:
+                break
+
+            weighted.sort(reverse=True)
+
+            for _, _, g in weighted:
+                intersecting_group = self.new_group('%s in %s' % (g.name, stem))
+                intersecting_group.group_members = g.group_members & names
+                disjoint_group = self.new_group('%s not in %s' % (g.name, stem))
+                disjoint_group.group_members = g.group_members - names
+
+                remaining -= intersecting_group.group_members
+
+        # If there's anything left, create new groups.
+        if remaining:
+            prefix = 'remaining ' if groups_found else ''
+
+            main_group = self.new_group('%scontents of %s' % (prefix, stem))
+            opt_group = self.new_group('%scontents of %s - optional' % (prefix, stem))
             optional.add(opt_group.name)
-            doc_group = self.new_group('contents of %s - documentation' % stem)
+            doc_group = self.new_group('%scontents of %s - documentation' % (prefix, stem))
             doc_group.doc = True
-            license_group = self.new_group('contents of %s - licenses' % stem)
+            license_group = self.new_group('%scontents of %s - licenses' % (prefix, stem))
             license_group.license = True
             # only from .deb
-            abs_group = self.new_group('contents of %s - absolute paths' % stem)
+            abs_group = self.new_group('%scontents of %s - absolute paths' % (prefix, stem))
             abs_group.install_to = '.'
             optional.add(abs_group.name)
 
             for g in (
                     main_group, opt_group, doc_group, license_group, abs_group):
-                groups[g.name] = g
+                groups_found[g.name] = g
 
             for n in remaining:
                 f = self.game.files[n]
@@ -252,12 +299,12 @@ class Template:
         if provider:
             provider.provides = set()
 
-            for name, g in groups.items():
+            for name, g in groups_found.items():
                 if g.group_members:
                     provider.provides.add(name)
 
         if package:
-            for name, g in groups.items():
+            for name, g in groups_found.items():
                 if g.license or g.doc or g.ignorable or name in optional:
                     package.optional.add(name)
                 else:

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-games/game-data-packager.git