What's the correct regexp pattern to match a VMS filename?

Question

The documentation at http://h71000.www7.hp.com/doc/731final/documentation/pdf/ovms_731_file_app.pdf (section 5-1) says the filename should look like this:

node::device:[root.][directory-name]filename.type;version

Most of them are optional (like node, device, version) - not sure which ones and how to correctly write this in a regexp, (including the directory name):

DISK1:[MYROOT.][MYDIR]FILE.DAT

DISK1:[MYDIR]FILE.DAT

[MYDIR]FILE.DAT

FILE.DAT;10

NODE::DISK5:[REMOTE.ACCESS]FILE.DAT

Warrantable

Is the text in the "[]" brackets in the example above optional as well? Are "MYROOT.", "MYDIR" and "REMOTE.ACCESS" aliases for something or are those string literals? – Vitrescence

As far as I can tell, the square brackets are litterals and not optional. I haven't found the exact specifications for the directory names and characters that are allowed. My understanding from the spec is that the directory names are identified with STRING dot STRING – Warrantable

Also note that each directory name and the filename are limited to 9 characters or less. Node and device have length limits, too (but I don't remember what they are). – Outage

@Loadmaster, I don't believe the limit is as small as that; I can remember creating files with 31-character names. – Misfortune

DECnet Phase IV node names are limited, but Phase V node names can be FQDN's. On an ODS-2 volume, filenames and extensions are limited to 39 characters (upper case and limited special characters). But on an ODS-5 volume (extended filename support), the file names are viritualy unlimited, case preserving and allow special charcters, including whitespace... – Barbey

You don't plan on supporting UIC format directories: [100,377]? – Bakelite

Answer 1

See the documentation and source for the VMS::Filespec Perl module.

Angi

Thanks - this looks pretty good ([^:]*::)?([^:]*:)?([^>]]*[>]])?([^.;]*)(\.?[^.;]*)([.;]?\d*) – Warrantable

Answer 2

From wikipedia, the full form is actually a bit more than that:

NODE"accountname password"::device:[directory.subdirectory]filename.type;ver

This one took a while, but here is an expression that should accept all valid variations, and place the components into capture groups.

(?:(?:(?:([^\s:\[\]]+)(?:"([^\s"]+) ([^\s"]+)")?::)?([^\s:\[\]]+):)?\[([^\s:\[\]]+)\])?([^\s:\[\]\.]+)(\.[^\s:\[\];]+)?(;\d+)?

Also, from what I can tell, your example of

DISK1:[MYROOT.][MYDIR]FILE.DAT

is not a valid name. I believe only one pair of brackets are allowed. I hope this helps!

Splendiferous

I suspect that is an example of a rooted directory. – Braunite

Answer 3

You could probably come up with a single complicated regex for this, but it will be much easier to read your code if you work your way from left to right stripping off each section if it is there. The following is some Python code that does just that:

lines = ["DISK1:[MYROOT.][MYDIR]FILE.DAT", "DISK1:[MYDIR]FILE.DAT", "[MYDIR]FILE.DAT", "FILE.DAT;10", "NODE::DISK5:[REMOTE.ACCESS]FILE.DAT"]
node_re = "(\w+)::"
device_re = "(\w+):"
root_re = "\[(\w+)\.]"
dir_re = "\[(\w+)]"
file_re = "(\w+)\."
type_re = "(\w+)"
version_re = ";(.*)"
re_dict = {"node": node_re, "device": device_re, "root": root_re, "directory": dir_re, "file": file_re, "type": type_re, "version": version_re}
order = ["node", "device", "root", "directory", "file", "type", "version"]
for line in lines:
    i = 0
    print line
    for item in order:
        m = re.search(re_dict[item], line[i:])
        if m is not None:
            print "  " + item + ": " + m.group(1)
            i += len(m.group(0))

and the output is

DISK1:[MYROOT.][MYDIR]FILE.DAT
  device: DISK1
  root: MYROOT
  directory: MYDIR
  file: FILE
  type: DAT
DISK1:[MYDIR]FILE.DAT
  device: DISK1
  directory: MYDIR
  file: FILE
  type: DAT
[MYDIR]FILE.DAT
  directory: MYDIR
  file: FILE
  type: DAT
FILE.DAT;10
  file: FILE
  type: DAT
  version: 10
NODE::DISK5:[REMOTE.ACCESS]FILE.DAT
  node: NODE
  device: DISK5
  directory: REMOTE.ACCESS
  file: FILE
  type: DAT

Anatropous

sorry for not being clear - I am using the regexp in a lex file – Warrantable

No problem, my answer was less about the language and more about the strategy. I would hate having ([^:]*::)?([^:]*:)?([^>]]*[>]])?([^.;]*)(\.?[^.;]*)([.;]?\d*) anywhere in my code, sometimes clarity is better than cleverness. – Anatropous