def write_sitemap(self,outfile=None,capabilities=None,dump=None):
# Set up base_path->base_uri mappings, get inventory from disk
i = self.inventory
i.capabilities = capabilities
s=Sitemap(verbose=self.verbose, pretty_xml=True, allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.max_sitemap_entries is not None):
s.max_sitemap_entries = self.max_sitemap_entries
if (outfile is None):
print s.inventory_as_xml(i)
else:
s.write(i,basename=outfile)
self.write_dump_if_requested(i,dump)
def changeset_sitemap(self,
outfile=None,
ref_sitemap=None,
capabilities=None,
dump=None):
# 1. Get and parse reference sitemap
rs = Sitemap(verbose=self.verbose,
allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.verbose):
print "Reading sitemap(s) from %s ..." % (ref_sitemap)
ri = rs.read(ref_sitemap)
num_entries = len(ri)
print "Read reference sitemap with %d entries in %d sitemaps" % (
num_entries, rs.sitemaps_created)
if (self.verbose):
to_show = 100
override_str = ' (override with --max-sitemap-entries)'
if (self.max_sitemap_entries):
to_show = self.max_sitemap_entries
override_str = ''
if (num_entries > to_show):
print "Showing first %d entries sorted by URI%s..." % (
to_show, override_str)
n = 0
for r in ri.resource_uris():
print ri.resources[r]
n += 1
if (n >= to_show):
break
# 2. Set up base_path->base_uri mappings, get inventory from disk
disk_inventory = self.inventory
# 3. Calculate changeset
(num_same, updated, deleted, created) = ri.compare(disk_inventory)
changeset = Inventory()
changeset.capabilities = capabilities
changeset.add(disk_inventory.changeset(updated, changetype='updated'))
changeset.add(ri.changeset(deleted, changetype='deleted'))
changeset.add(disk_inventory.changeset(created, changetype='created'))
# 4. Write out changeset
s = Sitemap(verbose=self.verbose,
pretty_xml=True,
allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.max_sitemap_entries is not None):
s.max_sitemap_entries = self.max_sitemap_entries
if (outfile is None):
print s.inventory_as_xml(changeset)
else:
s.write(changeset, basename=outfile)
self.write_dump_if_requested(changeset, dump)
def write_sitemap(self, outfile=None, capabilities=None, dump=None):
# Set up base_path->base_uri mappings, get inventory from disk
i = self.inventory
i.capabilities = capabilities
s = Sitemap(verbose=self.verbose,
pretty_xml=True,
allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.max_sitemap_entries is not None):
s.max_sitemap_entries = self.max_sitemap_entries
if (outfile is None):
print s.inventory_as_xml(i)
else:
s.write(i, basename=outfile)
self.write_dump_if_requested(i, dump)
def test2_pretty_output(self):
ib = InventoryBuilder()
ib.mapper = Mapper(['http://example.org/t','resync/test/testdata/dir1'])
i = ib.from_disk()
s = Sitemap()
s.pretty_xml=True
self.assertEqual(s.inventory_as_xml(i),'<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://resourcesync.org/change/0.1">\n<url><loc>http://example.org/t/file_a</loc><lastmod>2012-03-14T17:46:04</lastmod><rs:size>20</rs:size></url>\n<url><loc>http://example.org/t/file_b</loc><lastmod>2012-03-14T17:46:25</lastmod><rs:size>45</rs:size></url>\n</urlset>' )
def test3_with_md5(self):
ib = InventoryBuilder(do_md5=True)
i = ib.from_disk('resync/test/testdata/dir1','http://example.org/t')
s = Sitemap()
xml = s.inventory_as_xml(i)
self.assertNotEqual( None, re.search('<loc>http://example.org/t/file_a</loc><lastmod>[\w\:\-]+</lastmod><rs:size>20</rs:size><rs:md5>6bf26fd66601b528d2e0b47eaa87edfd</rs:md5>',xml), 'size/checksum for file_a')
self.assertNotEqual( None, re.search('<loc>http://example.org/t/file_b</loc><lastmod>[\w\:\-]+</lastmod><rs:size>45</rs:size><rs:md5>452e54bdae1626ac5d6e7be81b39de21</rs:md5>',xml), 'size/checksum for file_b' )
def test2_pretty_output(self):
ib = InventoryBuilder()
i = ib.from_disk('resync/test/testdata/dir1', 'http://example.org/t')
s = Sitemap()
s.pretty_xml = True
self.assertEqual(
s.inventory_as_xml(i),
'<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://resourcesync.org/change/0.1">\n<url><loc>http://example.org/t/file_a</loc><lastmod>2012-03-14T17:46:04</lastmod><rs:size>20</rs:size></url>\n<url><loc>http://example.org/t/file_b</loc><lastmod>2012-03-14T17:46:25</lastmod><rs:size>45</rs:size></url>\n</urlset>'
)
def changeset_sitemap(self,outfile=None,ref_sitemap=None,capabilities=None,
dump=None):
# 1. Get and parse reference sitemap
rs = Sitemap(verbose=self.verbose, allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.verbose):
print "Reading sitemap(s) from %s ..." % (ref_sitemap)
ri = rs.read(ref_sitemap)
num_entries = len(ri)
print "Read reference sitemap with %d entries in %d sitemaps" % (num_entries,rs.sitemaps_created)
if (self.verbose):
to_show = 100
override_str = ' (override with --max-sitemap-entries)'
if (self.max_sitemap_entries):
to_show = self.max_sitemap_entries
override_str = ''
if (num_entries>to_show):
print "Showing first %d entries sorted by URI%s..." % (to_show,override_str)
n=0
for r in ri.resource_uris():
print ri.resources[r]
n+=1
if ( n >= to_show ):
break
# 2. Set up base_path->base_uri mappings, get inventory from disk
disk_inventory = self.inventory
# 3. Calculate changeset
(num_same,updated,deleted,created)=ri.compare(disk_inventory)
changeset = Inventory()
changeset.capabilities = capabilities
changeset.add( disk_inventory.changeset( updated, changetype='updated' ) )
changeset.add( ri.changeset( deleted, changetype='deleted' ) )
changeset.add( disk_inventory.changeset( created, changetype='created' ) )
# 4. Write out changeset
s = Sitemap(verbose=self.verbose, pretty_xml=True, allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.max_sitemap_entries is not None):
s.max_sitemap_entries = self.max_sitemap_entries
if (outfile is None):
print s.inventory_as_xml(changeset)
else:
s.write(changeset,basename=outfile)
self.write_dump_if_requested(changeset,dump)
def test3_with_md5(self):
ib = InventoryBuilder(do_md5=True)
i = ib.from_disk('resync/test/testdata/dir1', 'http://example.org/t')
s = Sitemap()
xml = s.inventory_as_xml(i)
self.assertNotEqual(
None,
re.search(
'<loc>http://example.org/t/file_a</loc><lastmod>[\w\:\-]+</lastmod><rs:size>20</rs:size><rs:md5>6bf26fd66601b528d2e0b47eaa87edfd</rs:md5>',
xml), 'size/checksum for file_a')
self.assertNotEqual(
None,
re.search(
'<loc>http://example.org/t/file_b</loc><lastmod>[\w\:\-]+</lastmod><rs:size>45</rs:size><rs:md5>452e54bdae1626ac5d6e7be81b39de21</rs:md5>',
xml), 'size/checksum for file_b')
