restructure and integrate code

README

@@ -8,7 +8,7 @@ $ ./update-portdir.sh
 
 At this point, you can already run the following script to get the list of
 Debian source packages that can be mapped to Gentoo packages and also have
-dependencies that can be dropped in Gentoo:
+build dependencies that can be dropped in Gentoo:
 
 $ ./find_reduced.py
 
@@ -17,37 +17,11 @@ It will print:
   - a list of Debian packages that cannot be mapped to Gentoo packages
   - a list of Gentoo packages that cannot be mapped to Debian packages
 
-Above script depends on the following data files to be present:
+Overview:
 
 portage               - portage sourcecode
                           * retrieved as a git submodule
 portdir               - database of ebuild files
                           * retrieved by ./update-portdir.sh
-reduced_gen_deps.list - Gentoo packages with reduced dependencies
-                          * generated by ./gentoo_dep_list.py
-gen_source_pkgs.list  - list of Gentoo source packages
-                          * generated by ./gentoo_pkg_list_raw.sh
-deb_source_pkgs.list  - list of Debian source packages
-                          * supplied by yourself
-deb2gen_mapping.list  - mapping of Debian to Gentoo packages
-                          * created manually and/or with the help of
-                            ./find_pkg_matches.py
-
-deb_source_pkgs.list is filled by the user. All other *.list files can be
-regenerated by the following scripts:
-
-$ ./gentoo_dep_list.py
-
-   It writes all source packages with reduced build dependencies plus the build
-   dependencies that can be dropped to reduced_gen_deps.list. Execution takes
-   about 2.5 hours.  
-
-$ ./gentoo_pkg_list_raw.sh
-
-   It writes to gen_source_pkgs.list a list of all available Gentoo packages.
-
-$ ./find_pkg_matches.py
-
-   Inspects the current content of deb2gen_mapping.list and prints a list of
-   Debian packages that miss a mapping to Gentoo packages as well as a list of
-   suggestions of Gentoo packages that sound similar.
+deb_source_pkgs.list  - list of interesting Debian source packages
+deb2gen_mapping.list  - mapping of Debian source packages to Gentoo packages

find_pkg_matches.py

@@ -1,46 +0,0 @@
-#!/usr/bin/env python
-
-import sys
-
-deb2gen = dict()
-gen2deb = dict()
-with open("./deb2gen_mapping.list") as f:
-    for line in f:
-        d, g = line.strip().split('\t')
-        deb2gen[d] = g
-        gen2deb[g] = d
-
-with open("./deb_source_pkgs.list") as f:
-    debian_names = [p[4:].strip() for p in f]
-
-with open("./gen_source_pkgs.list") as f:
-    gentoo_names = [p.strip() for p in f]
-
-for p in debian_names:
-    # if mapping exists, continue
-    orig = p
-    if deb2gen.get(p):
-        continue
-    # first try exact match
-    matches = [g for g in gentoo_names if p == g.split('/')[-1]]
-    if matches:
-        print "%s\t%s"%(orig, matches[0])
-        continue
-    # then try substring matches
-    # remove '-perl' prefix
-    if p.endswith("-perl"):
-        p = p[:-5]
-    # remove 'python-' suffix
-    if p.startswith("python-"):
-        p = p[7:]
-    # remove 'lib' suffix
-    if p.startswith("lib"):
-        p = p[3:]
-    # remove dashes, dots and numbers from the end of the package name
-    p = p.rstrip(".-1234567890")
-    matches = [g for g in gentoo_names if p in g.split('/')[-1].lower()]
-    if matches:
-        print "%s\t%s"%(orig, str(matches))
-        continue
-    # we give up and print the debian source package name that was not matched
-    print orig

find_reduced.py

@@ -1,13 +1,48 @@
 #!/usr/bin/env python
 
 import sys
+
+if len(sys.argv) != 2:
+    print "usage: %s /path/to/Sources.bz"%sys.argv[0]
+
+import os
 from os import path
+from subprocess import Popen, PIPE
+from functools import cmp_to_key
+from datetime import datetime
+import apt_pkg
 
 pym_path = path.join(path.dirname(path.realpath(__file__)), "portage", "pym")
 sys.path.insert(0, pym_path)
 
-from portage.versions import pkgsplit
+from portage.versions import pkgcmp, pkgsplit
+from portage.dep import use_reduce, Atom
+from portage.exception import InvalidAtom
 
+# given the starting time, the overall count and the current progress,
+# return a string that shows the remaining time in hours/minutes/seconds
+def estimate_remaining_time(before, count, i):
+    now = datetime.now()
+    delta = now-before
+    # delta in seconds
+    delta = delta.days*60*60*24+delta.seconds
+    if delta == 0 or i == 0:
+        return "n.a."
+    speed = float(i)/delta
+    remaining = (count - i)/speed
+    remaining_hours = remaining/3600
+    remaining_minutes = (remaining%3600)/60
+    remaining_seconds = remaining%60
+    return "%02d:%02d:%02d"%(remaining_hours, remaining_minutes, remaining_seconds)
+
+# flatten an arbitrarily nested list
+def flatten(l):
+    while l:
+        while l and isinstance(l[0], list):
+            l[0:1] = l[0]
+        if l: yield l.pop(0)
+
+# TODO: what about mapping to multiple packets?
 deb2gen = dict()
 gen2deb = dict()
 with open("./deb2gen_mapping.list") as f:
@@ -19,32 +54,143 @@ with open("./deb2gen_mapping.list") as f:
 with open("./deb_source_pkgs.list") as f:
     debian_names = [p[4:].strip() for p in f]
 
-gentoo_deps = dict()
-with open("./reduced_gen_deps.list") as f:
-    for line in f:
-        p = line.strip().split()
-        gentoo_deps[pkgsplit(p[0])[0]] = p[1:]
+portdir = "./portdir"
 
-missing_deb_mappings = list()
+pkgnames = list()
+
+# traverse ./portdir to find .ebuild files that interest us
+for cat in os.listdir(portdir):
+    catpath = os.path.join(portdir, cat)
+
+    if not os.path.isdir(catpath):
+        continue
+
+    for pkg in os.listdir(catpath):
+        pkgpath = os.path.join(catpath, pkg)
+
+        if not os.path.isdir(pkgpath):
+            continue
+
+        # get all .ebuild files
+        pkgs = [pkgsplit(ver[:-7]) for ver in os.listdir(pkgpath) if ver.endswith(".ebuild")]
+
+        if not pkgs:
+            continue
+
+        # only grab newest package
+        pkg, ver, rev = sorted(pkgs, key=cmp_to_key(pkgcmp), reverse=True)[0]
+        deb = gen2deb.get("%s/%s"%(cat, pkg))
+
+        # assure that the package is relevant
+        if deb not in debian_names:
+            continue
+
+        pkgnames.append((cat, pkg, ver, rev))
+        sys.stderr.write("\rGenerating list of packages... %d"%len(pkgnames))
+
+sys.stderr.write("\rGenerating list of packages... Done.\n")
+
+count = 0
+
+before = datetime.now()
+
+debian_deps = dict()
 missing_gen_mappings = list()
 
-for d in debian_names:
-    g = deb2gen.get(d)
-    if g:
-        deps = gentoo_deps.get(g)
-        if deps:
+# look for reduced build dependencies
+for i, (cat, pkg, ver, rev) in enumerate(pkgnames):
+    if rev == "r0":
+        pkgname = "%s/%s-%s"%(cat, pkg, ver)
+    else:
+        pkgname = "%s/%s-%s-%s"%(cat, pkg, ver, rev)
+    p = Popen(["./portage/bin/portageq", "metadata", "/", "ebuild", pkgname, "DEPEND"], stderr=PIPE, stdout=PIPE, env={"PORTDIR": portdir})
+    r = p.communicate()
+    if p.returncode != 0:
+        sys.stderr.write("cannot parse %s. Output: %s\n"%(pkgname, r[1]))
+    depend = r[0]
+    all_use = use_reduce(depend, matchall=True)
+    no_use = use_reduce(depend, matchnone=True)
+    l = list()
+    for a in all_use:
+        # filter out all dependencies that are also present without USE flags enabled
+        if a in no_use:
+            continue
+
+        if not isinstance(a, list) :
+            a = [a]
+        else:
+            # there are disjunctions but we better collect to many build dependencies than too few
+            a = flatten(a)
+
+        for p in a:
+            if p == "||":
+                continue
+
+            try:
+                l.append(Atom(p).cp)
+            except InvalidAtom:
+                sys.stderr.write("Invalid Atom for %s: %s\n"%(pkgname, str(p)))
+            except TypeError:
+                sys.stderr.write("TypeError for %s: %s\n"%(pkgname, str(p)))
+
+    if l:
         ddeps = []
-            for p in deps:
+        for p in l:
             r = gen2deb.get(p)
-                if r:
-                    ddeps.append(r)
-                else:
+
+            if not r:
                 missing_gen_mappings.append(p)
-            ddeps = list(set(ddeps))
-            if ddeps:
-                print d, ddeps
-    else:
+                continue
+
+            ddeps.append(r)
+        debian_deps["%s/%s"%(cat,pkg)] = list(set(ddeps))
+        count +=1
+
+    sys.stderr.write("\rFinding reduced dependencies: %d/%d, found %d. Estimated time left: %s h"%(i, len(pkgnames), count, estimate_remaining_time(before, len(pkgnames), i)))
+    sys.stdout.flush()
+
+sys.stderr.write("\rFinding reduced dependencies: %d/%d, found %d. Estimated time left: %s h\n"%(i, len(pkgnames), count, estimate_remaining_time(before, len(pkgnames), i)))
+
+missing_deb_mappings = list()
+
+src2bin = dict()
+src_deps = dict()
+
+for pkg in apt_pkg.TagFile(sys.argv[1]):
+    p = pkg['Package']
+    try:
+        src_deps[p] = [name for (name,ver,rel) in flatten(apt_pkg.parse_src_depends(pkg.get('Build-Depends', '')))]
+    except ValueError:
+        print "cannot parse depends line for %s: %s"%(pkg['Package'], pkg.get('Build-Depends'))
+        src_deps[p] = []
+    src2bin[p] = [b.strip() for b in pkg['Binary'].split(',')]
+
+# process all debian source packages
+for d in debian_names:
+    # find gentoo package name
+    g = deb2gen.get(d)
+    if not g:
         missing_deb_mappings.append(d)
+        continue
+
+    # get the droppable dependencies
+    deps = debian_deps.get(g)
+
+    if not deps:
+        continue
+
+    bindeps = list()
+    # get the debian build dependencies
+    pdeps = src_deps[d]
+
+    # for each droppable dependency
+    for r in deps:
+        # convert source package to binary packages
+        # only retain those binary packages that are a build dependency
+        bindeps.extend([p for p in src2bin[r] if p in pdeps])
+
+    if bindeps:
+        print d, bindeps
 
 print list(set(missing_deb_mappings))
 print list(set(missing_gen_mappings))

gentoo_dep_list.py

@@ -1,96 +0,0 @@
-#!/usr/bin/env python
-
-import sys
-import os
-from os import path
-from subprocess import Popen, PIPE
-from functools import cmp_to_key
-from datetime import datetime
-
-pym_path = path.join(path.dirname(path.realpath(__file__)), "portage", "pym")
-sys.path.insert(0, pym_path)
-
-from portage.versions import pkgcmp, pkgsplit
-from portage.dep import use_reduce, Atom
-from portage.exception import InvalidAtom
-
-def format_timedelta(delta):
-    secs = delta.days*60*60*24+delta.seconds
-    hours = secs/3600
-    minutes = (secs%3600)/60
-    return "%02d:%02d"%(hours, minutes)
-
-def estimate_remaining_time(before, count, i):
-    now = datetime.now()
-    delta = now-before
-    # delta in seconds
-    delta = delta.days*60*60*24+delta.seconds
-    if delta == 0:
-        return "n.a."
-    speed = float(i)/delta
-    remaining = (count - i)/speed
-    remaining_hours = remaining/3600
-    remaining_minutes = (remaining%3600)/60
-    return "%02d:%02d"%(remaining_hours, remaining_minutes)
-
-portdir = "./portdir"
-
-pkgnames = list()
-
-for cat in os.listdir(portdir):
-    catpath = os.path.join(portdir, cat)
-    if not os.path.isdir(catpath):
-        continue
-    for pkg in os.listdir(catpath):
-        pkgpath = os.path.join(catpath, pkg)
-        if not os.path.isdir(pkgpath):
-            continue
-        pkgs = [pkgsplit(ver[:-7]) for ver in os.listdir(pkgpath) if ver.endswith(".ebuild")]
-        if len(pkgs) > 0:
-            # only grab newest package
-            pkg, ver, rev = sorted(pkgs, key=cmp_to_key(pkgcmp), reverse=True)[0]
-            if rev == "r0":
-                pkgname = "%s/%s-%s"%(cat, pkg, ver)
-            else:
-                pkgname = "%s/%s-%s-%s"%(cat, pkg, ver, rev)
-            pkgnames.append(pkgname)
-            sys.stderr.write("\rGenerating list of packages... %d"%len(pkgnames))
-
-sys.stderr.write("\rGenerating list of packages... Done.\n")
-
-count = 0
-
-deplist = open("./out", "wb")
-
-before = datetime.now()
-
-for i, pkgname in enumerate(pkgnames):
-    p = Popen(["./portage/bin/portageq", "metadata", "/", "ebuild", pkgname, "DEPEND"], stderr=PIPE, stdout=PIPE, env={"PORTDIR": portdir})
-    r = p.communicate()
-    if p.returncode != 0:
-        sys.stderr.write("cannot parse %s. Output: %s\n"%(pkgname, r[1]))
-    depend = r[0]
-    all_use = use_reduce(depend, matchall=True)
-    no_use = use_reduce(depend, matchnone=True)
-    # cannot use sets because of possible sublists which are not hashable
-    l = []
-    for a in all_use:
-        # FIXME: do not discard disjunctions
-        if a not in no_use and a != "||" and not isinstance(a, list):
-            try:
-                l.append(Atom(a).cp)
-            except InvalidAtom:
-                sys.stderr.write("Invalid Atom for %s: %s\n"%(pkgname, str(a)))
-            except TypeError:
-                sys.stderr.write("TypeError for %s: %s\n"%(pkgname, str(a)))
-
-    if len(l) > 0:
-        deplist.write("%s %s\n"%(pkgname, " ".join(l)))
-        count +=1
-
-    sys.stderr.write("\rFinding reduced dependencies: %d/%d, found %d. Estimated time left: %s h"%(i, len(pkgnames), count, estimate_remaining_time(before, len(pkgnames), i)))
-    sys.stdout.flush()
-
-sys.stderr.write("\rFinding reduced dependencies: %d/%d, found %d. Estimated time left: %s h\n"%(i, len(pkgnames), count, estimate_remaining_time(before, len(pkgnames), i)))
-
-sys.stderr.write("Done. Took %s h\n"%format_timedelta(datetime.now()-before))

gentoo_pkg_list_raw.sh

@@ -1,5 +0,0 @@
-#!/bin/bash
-
-for f in `find "./portdir/" -mindepth 2 -maxdepth 2 -type d`; do
-	echo ${f#./portdir/}
-done > gen_source_pkgs.list