def convert_archive(archive):
# regarding the usage of del in this function... bear in mind these sets
# could easily have 10k -> 100k entries in extreme cases; thus the del
# usage, explicitly trying to ensure we don't keep refs long term.
# this one is a bit fun.
raw = list(archive_to_fsobj(archive))
# we use the data source as the unique key to get position.
files_ordering = list(enumerate(x for x in raw if x.is_reg))
files_ordering = {x.data: idx for idx, x in files_ordering}
t = contents.contentsSet(raw, mutable=True)
del raw, archive
# first rewrite affected syms.
raw_syms = t.links()
syms = contents.contentsSet(raw_syms)
while True:
for x in sorted(syms):
affected = syms.child_nodes(x.location)
if not affected:
continue
syms.difference_update(affected)
syms.update(affected.change_offset(x.location, x.resolved_target))
del affected
break
else:
break
t.difference_update(raw_syms)
t.update(syms)
del raw_syms
syms = sorted(syms, reverse=True)
# ok, syms are correct. now we get the rest.
# we shift the readds into a separate list so that we don't reinspect
# them on later runs; this slightly reduces the working set.
additions = []
for x in syms:
affected = t.child_nodes(x.location)
if not affected:
continue
t.difference_update(affected)
additions.extend(affected.change_offset(x.location, x.resolved_target))
t.update(additions)
t.add_missing_directories()
# finally... an insane sort.
def sort_func(x, y):
if x.is_dir:
if not y.is_dir:
return -1
return cmp(x, y)
elif y.is_dir:
return +1
elif x.is_reg:
if y.is_reg:
return cmp(files_ordering[x.data], files_ordering[y.data])
return +1
elif y.is_reg:
return -1
return cmp(x, y)
return contents.OrderedContentsSet(sorted_cmp(t, sort_func), mutable=False)
def convert_archive(archive):
# regarding the usage of del in this function... bear in mind these sets
# could easily have 10k -> 100k entries in extreme cases; thus the del
# usage, explicitly trying to ensure we don't keep refs long term.
# this one is a bit fun.
raw = list(archive_to_fsobj(archive))
# we use the data source as the unique key to get position.
files_ordering = list(enumerate(x for x in raw if x.is_reg))
files_ordering = {x.data: idx for idx, x in files_ordering}
t = contents.contentsSet(raw, mutable=True)
del raw, archive
# first rewrite affected syms.
raw_syms = t.links()
syms = contents.contentsSet(raw_syms)
while True:
for x in sorted(syms):
affected = syms.child_nodes(x.location)
if not affected:
continue
syms.difference_update(affected)
syms.update(affected.change_offset(x.location, x.resolved_target))
del affected
break
else:
break
t.difference_update(raw_syms)
t.update(syms)
del raw_syms
syms = sorted(syms, reverse=True)
# ok, syms are correct. now we get the rest.
# we shift the readds into a separate list so that we don't reinspect
# them on later runs; this slightly reduces the working set.
additions = []
for x in syms:
affected = t.child_nodes(x.location)
if not affected:
continue
t.difference_update(affected)
additions.extend(affected.change_offset(x.location, x.resolved_target))
t.update(additions)
t.add_missing_directories()
# finally... an insane sort.
def sort_func(x, y):
if x.is_dir:
if not y.is_dir:
return -1
return cmp(x, y)
elif y.is_dir:
return +1
elif x.is_reg:
if y.is_reg:
return cmp(files_ordering[x.data], files_ordering[y.data])
return +1
elif y.is_reg:
return -1
return cmp(x, y)
return contents.OrderedContentsSet(sorted_cmp(t, sort_func), mutable=False)