dfVFS was developed to provide a generic means for Plaso to access files via the operating system and those contained in storage media image formats.
In dfVFS terminology:
a data stream encapsulates the contents of a file, an NTFS ADS or HFS resource fork.
a file entry encapsulates the file system metadata related to files, directories, symbolic links or equivalent file system related structures.
a file system encapsulates the hierarchy of file entries within a single volume.
a volume is part of a volume system and often contains a file system (e.g. NTFS) or a sub volume system (e.g. VSS).
a volume system encapsulates one or more volumes.
dfVFS separates the concerns relating:
addressing of file entries, such as the file path, with type specific path specification;
file-based Input/Output (IO), with type specific basic file-like objects;
traversing file entries and metadata, with type specific virtual file system and/or file entry;
format detection, with a format analyzer with type specific helpers.
The path specification
The path specification is equivalent to the path on most operating systems. Though the actual format of the path can be operating systems specific, e.g. Linux uses /home/myuser/myfile.txt and Windows C:\Users\MyUser\MyFile.txt.
In dfVFS the path specification is defined as a container with a type indicator with zero or more addressing attributes. One of the more common addressing attributes is e.g. location with for the operating system path specification would contain the operating system specific path, e.g. on Windows location=C:\Users\MyUser\MyFile.txt.
Path specifications can be stacked to access files in storage media images and other container formats e.g. archive files like ZIP. For this the path specification can define a parent path specification. Note that parent here refers to the path specification and not the parent in the path segment e.g. defined by the location addressing attribute.
Next follows a pseudo code implementation of a stacked path specification that defines the location of C:\Users\MyUser\MyFile.txt on the first partition of QCOW version 2 image.
type=OS, location=/home/myuser/myimages/image.qcow2 type=QCOW type=TSK_PARTITION, location=/p1 type=TSK, inode=128, location=/Users/MyUser/MyFile.txt
Note that the parent is not explicitly defined in the example but the parent is defined implicit by the preceding path specification. E.g. OS is the parent of the QCOW path specification type and QCOW the parent of the TSK_PARTITION path specification type.
Also note that though the location of the file within the Windows operating system would be C:\Users\MyUser\MyFile.txt it is defined in the the TSK path specification type as /Users/MyUser/MyFile.txt. dfVFS defines a Windows path resolver helper to map Windows paths to a location within images.
The inode addressing attribute in the TSK path specification type is used for faster access to the file within the TSK virtual file system implementation.
The path specification resolver
The path specification resolver is used to resolve a path specification to a file-like object or virtual file system object.
The path specification resolver uses a resolver context to cache previously resolved file-like object or virtual file system object. This cache uses a weak reference so code interacting with the resolver should maintain a reference if such an object is still used.
The mount point manager
The mount point manager can be used to globally (within the same process space) define mount points used by path specifications of type MOUNT. The same can be accomplished with a resolver context, but note that specified mount points are local to the specific context.
The credentials manager
Besides the previously mentioned objects dfVFS also provides several helper objects.