multiarchanalyze: make it easily consumable by tracker.d.o

Many thanks to Paul Wise for his detailed feedback on the data format.

1 files changed, 70 insertions, 10 deletions

diff --git a/multiarchanalyze.py b/multiarchanalyze.py
index 5cd5de7..201ffc3 100755
--- a/multiarchanalyze.py
+++ b/multiarchanalyze.py
@@ -1,7 +1,11 @@
 #!/usr/bin/python
 
 import argparse
+import os.path
 import sqlite3
+import sys
+
+import yaml
 
 from dedup.utils import fetchiter
 
@@ -61,13 +65,16 @@ def show_combinations(combinations):
             return "%d bipartitions of %s" % (len(partitions),
                                               show_archset(edges.keys()))
     if all(len(outedges) == len(edges) - 1 for outedges in edges.values()):
-        return "any two of " + show_archset(edges.keys(), limit=4)
+        return "any two of " + show_archset(edges.keys(), limit=5)
     return "%s with %d combinations" % (show_archset(edges.keys()),
                                         len(combinations))
 
 def show_files(filenames):
     if len(filenames) == 1:
         return next(iter(filenames))
+    prefix = os.path.commonprefix(filenames)
+    if prefix not in ('', '/'):
+        return "%d files starting with %s" % (len(filenames), prefix)
     return "%d files" % len(filenames)
 
 def main():
@@ -76,39 +83,92 @@ def main():
                         default="test.sqlite3",
                         help="path to the sqlite3 database file")
     args = parser.parse_args()
+    hints = []
     db = sqlite3.connect(args.database)
     cur = db.cursor()
     cur.execute("SELECT name1, architecture1, architecture2, filename FROM masame_conflict;")
     same_conflicts = {}
     for package, arch1, arch2, filename in fetchiter(cur):
+        if filename.startswith("./"):
+            filename = filename[1:]
         conflicts = same_conflicts.setdefault(package, {})
         conflicts.setdefault(filename, set()).add((arch1, arch2))
     for package, conflicts in sorted(same_conflicts.items()):
         all_combinations = set()
         for combinations in conflicts.values():
             all_combinations.update(combinations)
-        print("%s conflicts on %s on %s" %
-              (package, show_files(conflicts.keys()),
-               show_combinations(all_combinations)))
+        desc = "%s conflicts on %s on %s" % \
+               (package, show_files(conflicts.keys()),
+                show_combinations(all_combinations))
+        hints.append(dict(
+            binary=package,
+            description=desc,
+            link="https://wiki.debian.org/MultiArch/Hints#file-conflict",
+            severity="high"))
 
     cur.execute("SELECT name FROM maforeign_candidate ORDER BY name;")
     for name, in fetchiter(cur):
-        print("%s could be Multi-Arch: foreign" % name)
+        hints.append(dict(
+            binary=name,
+            description="%s could be marked Multi-Arch: foreign" % name,
+            link="https://wiki.debian.org/MultiArch/Hints#ma-foreign",
+            severity="rdeps"))
 
     cur.execute("SELECT name FROM archall_candidate ORDER BY name;")
     archall_suggests = set()
     for name, in fetchiter(cur):
         archall_suggests.add(name)
-        print("%s could be converted to Architecture: all + Multi-Arch: foreign" % name)
+        desc = "%s could be converted to Architecture: all and marked Multi-Arch: foreign" % \
+               name
+        hints.append(dict(
+            binary=name,
+            description=desc,
+            link="https://wiki.debian.org/MultiArch/Hints#arch-all",
+            severity="low"))
 
     cur.execute("SELECT name FROM masame_candidate ORDER BY name;")
     for name, in fetchiter(cur):
-        if name not in archall_suggests:
-            print("%s could be Multi-Arch: same" % name)
+        if name in archall_suggests:
+            continue
+        hints.append(dict(
+            binary=name,
+            description="%s could be marked Multi-Arch: same" % name,
+            link="https://wiki.debian.org/MultiArch/Hints#ma-same",
+            severity="normal"))
+
     cur.execute("SELECT depender, dependee FROM colonany_candidate ORDER BY depender;")
     for depender, dependee in fetchiter(cur):
-        print("%s could have its %s dependency annotated with :any" %
-              (depender, dependee))
+        desc = "%s could have its dependency on %s annotated with :any" % \
+               (depender, dependee)
+        hints.append(dict(
+            binary=depender,
+            description=desc,
+            link="https://wiki.debian.org/MultiArch/Hints#dep-any",
+            severity="normal"))
+
+    for hint in hints:
+        if "source" not in hint:
+            cur.execute("SELECT distinct(source) FROM package WHERE name = ?;",
+                        (hint["binary"],))
+            row = cur.fetchone()
+            if row and row[0] and not cur.fetchone():
+                hint["source"] = row[0]
+
+        if hint.get("severity") == "rdeps":
+            cur.execute("SELECT count(*) FROM depends WHERE dependee = ?;",
+                        (hint["binary"],))
+            rdeps = cur.fetchone()[0]
+            if rdeps >= 50:
+                hint["severity"] = "high"
+            elif rdeps:
+                hint["severity"] = "normal"
+            else:
+                hint["severity"] = "low"
+
+    yaml.safe_dump(dict(format="multiarch-hints-1.0", hints=hints),
+                   default_flow_style=False,
+                   width=1000,
+                   stream=sys.stdout)
 
 if __name__ == "__main__":
     main()