def __init__(self, aggregator, root_dir, include_prefixes=None):
self.aggregator = aggregator
if root_dir.endswith("\\") or root_dir.endswith("//"):
self.root_dir = root_dir[0:-1]
else:
self.root_dir = root_dir
self.root_dir = file_utils.normalize_path(self.root_dir)
self.cdir = None
self.cabsdir = None
self.prefixes = [file_utils.normalize_path(p) for p in include_prefixes]
def discover(self):
for d, subdlist, flist in os.walk(self.root_dir):
normalized_d = file_utils.normalize_path(d)
if flist:
for f in flist:
full_path = file_utils.normalize_path(
os.path.abspath(os.path.join(normalized_d, f)))
consider = False
for prefix in self.prefixes:
if normalized_d.startswith(prefix):
consider = True
break
if not consider: continue
file_ext = file_utils.get_extension(full_path)
if file_ext.lower() not in set(['py']): continue
parent_dir = file_utils.normalize_path(
os.path.dirname(full_path))
pkg_parent_dir = None # os.path.commonpath()
if parent_dir == self.root_dir:
pkg_parent_dir = ""
else:
pkg_parent_dir = parent_dir[parent_dir.
index(self.root_dir) +
len(self.root_dir) + 1:]
file_ext = file_utils.get_extension(full_path)
df = DiscoveredFile()
df.set_attr(DiscoveredFileAttributeEnum.NAME,
file_utils.get_nonext_basename(full_path))
df.set_attr(DiscoveredFileAttributeEnum.EXTENSION,
file_ext)
df.set_attr(DiscoveredFileAttributeEnum.FULL_NAME, f)
df.set_attr(
DiscoveredFileAttributeEnum.PACKAGE_DOT_NOTATION,
pkg_parent_dir.replace("/", "."))
df.set_attr(
DiscoveredFileAttributeEnum.DIRECTORY_RELATIVE_PATH,
pkg_parent_dir)
df.set_attr(
DiscoveredFileAttributeEnum.DIRECTORY_ABSOLUTE_PATH,
parent_dir)
replaced = df.attr(DiscoveredFileAttributeEnum.
DIRECTORY_RELATIVE_PATH).replace(
"/", "|")
replaced = replaced.replace("\\", "|")
df.set_attr(
DiscoveredFileAttributeEnum.
COMMA_SEPATARED_RELATIVE_PATH,
",".join(replaced.split("|")))
# df.set_attr(DiscoveredFileAttributeEnum.CONTAINER, attr.NA_STRING)
# df.set_attr(DiscoveredFileAttributeEnum.CONTAINER_TYPE, attr.NA_STRING)
self.aggregator.add(df)
self.aggregator.freeze()