MAINT-7081: Make packages-formatter.py handle multi-line copyrights.

The nghttp2 autobuild package has copyright information that embeds a newline.
autobuild install --copyrights correctly produces that information onto two
lines. But that means packages-formatter.py must process any lines that do not
match its expected 'packagename: copyright' pattern as the continuation of the
preceding package's copyright information.

Since the processing for autobuild install --versions is so very similar, fold
both into the same outer loop.

Also report all duplicates for any package, instead of stopping at the first.

1 files changed, 56 insertions, 27 deletions

diff --git a/scripts/packages-formatter.py b/scripts/packages-formatter.py
index f91f5819b7..b1eef3c721 100755
--- a/scripts/packages-formatter.py
+++ b/scripts/packages-formatter.py
@@ -62,39 +62,68 @@ def autobuild(*args):
     # no exceptions yet, let caller read stdout
     return child.stdout
 
-version={}
-versions=autobuild('install', '--versions')
-for line in versions:
-    pkg_info = pkg_line.match(line)
-    if pkg_info:
-        pkg = pkg_info.group(1)
-        if pkg not in version:
-            version[pkg] = pkg_info.group(2).strip()
-        else:
-            sys.exit("Duplicate version for %s" % pkg)
+info=dict(versions={},    copyrights={})
+dups=dict(versions=set(), copyrights=set())
+
+def add_info(key, pkg, lines):
+    if pkg not in info[key]:
+        info[key][pkg] = '\n'.join(lines)
     else:
-        sys.exit("Unrecognized --versions output: %s" % line)
+        dups[key].add(pkg)
 
-copyright={}
+versions=autobuild('install', '--versions')
 copyrights=autobuild('install', '--copyrights')
 viewer_copyright = copyrights.readline() # first line is the copyright for the viewer itself
-for line in copyrights:
-    pkg_info = pkg_line.match(line)
-    if pkg_info:
-        pkg = pkg_info.group(1)
-        if pkg not in copyright:
-            copyright[pkg] = pkg_info.group(2).strip()
-        else:
-            sys.exit("Duplicate copyright for %s" % pkg)
+
+# Two different autobuild outputs, but we treat them essentially the same way:
+# populating each into a dict; each a subdict of 'info'.
+for key, rawdata in ("versions", versions), ("copyrights", copyrights):
+    lines = iter(rawdata)
+    try:
+        line = next(lines)
+    except StopIteration:
+        # rawdata is completely empty? okay...
+        pass
     else:
-        sys.exit("Unrecognized --copyrights output: %s" % line)
+        pkg_info = pkg_line.match(line)
+        # The first line for each package must match pkg_line.
+        if not pkg_info:
+            sys.exit("Unrecognized --%s output: %r" % (key, line))
+        # Only the very first line in rawdata MUST match; for the rest of
+        # rawdata, matching the regexp is how we recognize the start of the
+        # next package.
+        while True:                     # iterate over packages in rawdata
+            pkg = pkg_info.group(1)
+            pkg_lines = [pkg_info.group(2).strip()]
+            for line in lines:
+                pkg_info = pkg_line.match(line)
+                if pkg_info:
+                    # we hit the start of the next package data
+                    add_info(key, pkg, pkg_lines)
+                    break
+                else:
+                    # no package prefix: additional line for same package
+                    pkg_lines.append(line.rstrip())
+            else:
+                # last package in the output -- finished 'lines'
+                add_info(key, pkg, pkg_lines)
+                break
+
+# Now that we've run through all of both outputs -- are there duplicates?
+if any(pkgs for pkgs in dups.values()):
+    for key, pkgs in dups.items():
+        if pkgs:
+            print >>sys.stderr, "Duplicate %s for %s" % (key, ", ".join(pkgs))
+    sys.exit(1)
 
 print "%s %s" % (args.channel, args.version)
 print viewer_copyright
-for pkg in sorted(version):
-    print ': '.join([pkg, version[pkg]])
-    if pkg in copyright:
-        print copyright[pkg]
-    else:
+version = list(info['versions'].items())
+version.sort()
+for pkg, pkg_version in version:
+    print ': '.join([pkg, pkg_version])
+    try:
+        print info['copyrights'][pkg]
+    except KeyError:
         sys.exit("No copyright for %s" % pkg)
-    print ''
+    print