repos: build_report

This is the initial version of cross-repository report
that tries to give an approximate answer to the question:
"What other packages we should build with this one, and
why?"

port_stats/repos.py

@@ -52,6 +52,18 @@ class Dependency(collections.namedtuple(
             self.name, self.version, self.flags,
             other.name, other.version, other.flags))
 
+    def is_setversion(self):
+        return self.version and self.version.startswith(b'set:')
+
+    def pretty_str(self):
+        if not self.flags and not self.version:
+            return '{}[ANY]'.format(self.name.decode())
+        if self.is_setversion():
+            return '{}[set:<>]'.format(self.name.decode())
+        return '{}[{} {}]'.format(
+            self.name.decode(), self.flags,
+            self.version.decode() if self.version else None)
+
 
 class Source:
 
@@ -162,7 +174,93 @@ class Repository:
         return result
 
 
+_SPECIAL_DEPS = (
+    # ABI parts:
+    b'(GCC_', b'(CXXABI_', b'(GLIBC_',
+    # elf loaders:
+    b'/lib64/ld', b'/lib/ld', b'ld-linux')
+
+
+def _from_64bit_dep(dep):
+    if dep.name.endswith(b'()(64bit)'):
+        new_name = dep.name[:-9]
+    elif dep.name.endswith(b'(64bit)'):
+        new_name = dep.name[:-7]
+    elif b'/lib64/' in dep.name:
+        new_name = dep.name.replace(b'/lib64/', b'/lib/')
+    else:
+        return dep
+    return Dependency(new_name, dep.flags, dep.version)
+
+
+def have_same_source(repoA, repoB, source_name):
+    sourceA = repoA.sources.get(source_name)
+    sourceB = repoB.sources.get(source_name)
+    return (sourceA and sourceB and
+            sourceA.epoch == sourceB.epoch and
+            sourceA.version == sourceB.version and
+            sourceA.release == sourceB.release)
+
+
+def build_report(from_repo, to_repo, source_name):
+    result = set()
+    srpm = from_repo.sources[source_name]
+    # XXX: assumes from_repo is x86_64
+    translate = '64' not in to_repo.name
+
+    missing_buidreqs = (dep for dep in srpm.requires
+                        if not any(to_repo.providers(dep)))
+    for dep in missing_buidreqs:
+        for _dep, provider in from_repo.providers(dep):
+            result.add(('0000-BR', dep, provider))
+
+    for b in from_repo.binaries.values():
+        if b.source_name != source_name:
+            continue
+        # now, we have a binary that was built from the srpm
+        for dep in b.requires:
+            # skip some platform-specific stuff
+            if any(x in dep.name for x in _SPECIAL_DEPS):
+                continue
+            # if needed, try to translate from 64 to 32 bits
+            the_dep = _from_64bit_dep(dep) if translate else dep
+            # skip dependencies already present in to_repo
+            if any(to_repo.providers(the_dep)):
+                continue
+            # skip inter-sub-package dependencies
+            if any(p.source_name == source_name
+                   for _d, p in from_repo.providers(dep)):
+                continue
+            # set-versions may be platform-dependent.
+            # double-check that if have the same source
+            if dep.is_setversion():
+                if any(have_same_source(from_repo, to_repo, p.source_name)
+                       for _d, p in from_repo.providers(dep)):
+                    continue
+            # ok, it's true missing dependency
+            for _dep, provider in from_repo.providers(dep):
+                result.add((b.name.decode(), dep, provider))
+
+    # format and the result
+    by_source = collections.defaultdict(list)
+    lines = ['\n== %s ==\n' % source_name.decode()]
+    for item in result:
+        by_source[item[2].source_rpm].append(item)
+    for that_source in sorted(by_source):
+        lines.append(that_source.decode())
+        by_knp = collections.defaultdict(list)
+        for kind, dep, provider in by_source[that_source]:
+            by_knp[(kind, provider.name.decode())].append(dep.pretty_str())
+        for (kind, pname), pdep in sorted(by_knp.items()):
+            lines.append('\t {} {}  {}'.format(kind, pname, '  '.join(pdep)))
+        lines.append('')
+    return '\n'.join(lines)
+
+
 if __name__ == '__main__':
     from port_stats.utils import interactive_setup
+    from pydoc import pager
     CONFIG = interactive_setup()
-    repo = Repository.load_from_config('sisyphus_riscv64', CONFIG)
+    riscv64 = Repository.load_from_config('sisyphus_riscv64', CONFIG)
+    x86_64 = Repository.load_from_config('sisyphus', CONFIG)
+    mipsel = Repository.load_from_config('sisyphus_mipsel', CONFIG)