def setRDFFile(self, value, **kwargs):
"""Copy rdf file on filesystem"""
# Put value in buffer
buffer = StringIO(value)
copy_file(buffer, self.getRDFFilePath(**kwargs))
buffer.close()
def copyValueFile(self, **kwargs):
"""Duplicate file value on filesystem"""
src_path = self.getValueFilePath(uid=kwargs['src_uid'], name=kwargs['name'])
if not os.path.exists(src_path):
return
dst_path = self.getValueFilePath(**kwargs)
copy_file(src_path, dst_path)
def copyValueFile(self, **kwargs):
"""Duplicate file value on filesystem"""
src_path = self.getValueFilePath(uid=kwargs['src_uid'], path=kwargs['src_path'], name=kwargs['name'], title=kwargs.get('title', ''))
if not os.path.exists(src_path):
return
dst_path = self.getValueFilePath(**kwargs)
self.createSubDirectories(self.getValueDirectoryPath(**kwargs), self.storage_path)
copy_file(src_path, dst_path)
def copyValueFile(self, **kwargs):
"""Duplicate file value on filesystem"""
src_path = self.getValueFilePath(uid=kwargs['src_uid'], path=kwargs['src_path'], name=kwargs['name'], title=kwargs.get('title', ''))
if not os.path.exists(src_path):
return
name = kwargs['name']
path = self.getValueDirectoryPath(**kwargs)
# Process new filename
filename = self.getValueFilename(**kwargs)
# Update config file
self.createSubDirectories(path, self.storage_path)
self.setFilenameInConfigFile(path, name, filename)
# Then copy file
dst_path = self.getValueFilePath(**kwargs)
copy_file(src_path, dst_path)
def getData(self):
"""
Returns data (as a string) stored on filesystem.
"""
# Use StringIO for large files
virtual_file = StringIO()
value = ''
# Make sure file exists. If not returns empty string
if not os.path.exists(self.path) or os.path.getsize(self.path) == 0:
return ''
# Copy all data in one pass
try:
copy_file(self.path, virtual_file)
virtual_file.seek(0)
value = virtual_file.getvalue()
finally:
virtual_file.close()
return value
def build_fs_tree(src_path, dst_path, lib_path):
"""Build FS tree"""
# Rdf imports
sys.path.append(lib_path)
from rdf import RDFReader
from FileUtils import move_file, copy_file
print "Build filesystem data in %s from %s" % (src_path, dst_path)
sys_encoding = sys.getfilesystemencoding()
# Store rdf files
# List of dictionnary {'field': ..., 'uid': ...}
rdf_files = []
# Walk into filesystem
for root, dirs, files in os.walk(src_path):
if root == src_path:
# Loop on files
for item in files:
match = SEARCH_RDF_RE.match(item)
if match is None:
continue
# Get field name and content uid
uid = match.group('uid')
field = match.group('field')
rdf_files.append({'uid': uid, 'field': field})
# Processing collected rdf files
print "Processing %s rdf files" % str(len(rdf_files))
file_paths = []
for rdf_file in rdf_files:
uid = rdf_file['uid']
field = rdf_file['field']
# Get RDF file
rdf_filename = '%s_%s.rdf' % (uid, field)
rdf_path = os.path.join(src_path, rdf_filename)
rdf_file = StringIO()
rdf_text = ''
try:
copy_file(rdf_path, rdf_file)
rdf_file.seek(0)
rdf_text = rdf_file.getvalue()
finally:
rdf_file.close()
# Read RDF properties
try:
rdf_reader = RDFReader(rdf_text)
except:
try:
# XXX known bug to fix
rdf_text = rdf_text.replace('&', '&')
rdf_reader = RDFReader(rdf_text)
except:
print rdf_path
print rdf_text
raise
field_url = rdf_reader.getFieldUrl()
field_url = field_url.encode(sys_encoding, 'replace')
# Create tree directories
content_path_array = field_url.split('/')[:-2]
content_path = dst_path
for content_dir in content_path_array:
content_path = os.path.join(content_path, content_dir)
if os.path.exists(content_path):
continue
print "Create path: %s" % content_path
os.mkdir(content_path)
# Get source file
src_filename = '%s_%s' % (uid, field)
src_file_path = os.path.join(src_path, src_filename)
if not os.path.exists(src_file_path):
print "Source file doesn't exist, we continue: %s" % src_file_path
continue
# Get destination file
dst_filename = field
dst_filenames = rdf_reader.getFieldProperty('fss:filename')
if dst_filenames:
dst_filename = dst_filenames[0]
if not dst_filename:
dst_filename = field
else:
dst_filename = dst_filename.encode(sys_encoding, 'replace')
dst_file_path = os.path.join(content_path, dst_filename)
# In some cases, you can have a content having 2 fss fields with
# 2 files with the same name
orig_dst_filename = dst_filename
dst_file_path_ok = False
index = 0
while not dst_file_path_ok:
if dst_file_path not in file_paths:
dst_file_path_ok = True
file_paths.append(dst_file_path)
else:
index += 1
dst_filename = '%s-%s' % (str(index), orig_dst_filename)
print dst_filename
dst_file_path = os.path.join(content_path, dst_filename)
print "Create file: %s" % dst_file_path
copy_file(src_file_path, dst_file_path)
print "Create RDF file: %s.rdf" % dst_file_path
copy_file(rdf_path, dst_file_path + '.rdf')
print "Filesystem data built complete"