What is the naming standard for path components?

Question

I keep getting myself in knots when I am manipulating paths and file names because I don’t follow a naming standard for path components.

Consider the following toy problem (Windows example, but hopefully the answer should be platform independent). You have been given the path of a folder:

C:\users\OddThinking\Documents\My Source\

You want to walk the folders underneath and compile all the .src files to .obj files.

At some point you are looking at the following path:

C:\users\OddThinking\Documents\My Source\Widget\foo.src

How would you name the following path components?

A. foo
B. foo.src
C. src
D. .src
E. C:\users\OddThinking\Documents\My Source\ (i.e. the absolute path of the root)
F. Widget\foo.src (i.e. the relative path of the file)
G. Widget\
H. C:\users\OddThinking\Documents\My Source\Widget\
I. C:\users\OddThinking\Documents\My Source\Widget\foo.src

Here is my attempt:

A. Base name? Basename?

B. File name? Filename? The difference is important when choosing identifier names, and I am never consistent here.

C. Extension?

D. Extension? Wait, that is what I called C. Should I avoid storing the dot, and just put it in when required? What if there is no dot on a particular file?

E. ?

F. ?

G. Folder? But isn’t this a Windows-specific term?

H. Path name? Pathname? Path?

I. File name? Wait, that is what I called C. Path name? Wait, that is what I called H.

Solution 1

I think your search for a "standard" naming convention will be in vain. Here are my proposals, based on existing, well-known programs:

A) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                                   ---

Vim calls it file root (:help filename-modifiers)

B) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                                   -------

file name or base name

C) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                                       ___ (without dot)

file/name extension

D) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                                      ____ (with dot)

also file extension. Simply store without the dot, if there is no dot on a file, it has no extension

E) C:\users\OddThinking\Documents\My Source\Widget\foo.src
   -----------------------------------------

top of the tree
No convention, git calls it base directory

F) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                            --------------

path from top of the tree to the leaf
relative path

G) C:\users\OddThinking\Documents\My Source\Widget\foo.src
                                            ------

one node of the tree
no convention, maybe a simple directory

H) C:\users\OddThinking\Documents\My Source\Widget\foo.src
   ------------------------------------------------

dir name

I) C:\users\OddThinking\Documents\My Source\Widget\foo.src
   -------------------------------------------------------

full/absolute path

Solution 2

Good question first of all, my +1. This thing bugged me when I had to create a slew of functions in Utility class once. GetFileName? or GetFullName? GetApplicationPath means full path or the directory name? and so on. I come from .NET background, so I think I can add little more to otherwise excellent answer by @blinry.

Summary: (In italics is what I would not use as a programmer)

1. Path: Path specifies a unique location in the file system (unless its relative path). Path name is less often used, but I would stick with path - it pretty much explains what it is. Path can point to a file or a folder or even nothing (C:\). Path can be:

   1. Relative Path: My Source\Widget\ is relative path as well as Widget\foo.src. Self explanatory.
   2. Absolute Path or Full Path: Is the fully qualified path that points to the target. I tend to use the latter more often. C:\users\OddThinking\Documents\My Source\Widget\foo.src is hence full path. See at the end what I call full path that points to a file and that ends as a directory.

   The wiki page and .NET naming for path is consistent.

2. Root Path or Root Directory: Former is .NET convention while latter is more heard in UNIX circles. Though I like both I tend to use the former more. In windows, unlike UNIX, has many different root paths, one for each partition. Unix systems have one root directory which holds information on other directories and files. Eg. C:\ is root path.

3. Folder or Folder Name: Widget, OddThinking etc in your case. This might be a Windows only convention (in fact its my own odd thinking :)), nevertheless I strongly object to blinry`s answer "Directory". Though for a normal user directory means the same as a folder (like subfolders, subdirectories), I believe from a technical angle "directory" should sound like a qualified address to the target and not the target itself. More below.

   1. Sub Folders: With respect to users OddThinking and Documents are sub folders.
   2. Sub Directories: With respect to users OddThinking\, OddThinking\Documents\ and OddThinking\Documents\My Source\Widget\ are sub directories. But we do not often need to bother about it, do we?
   3. Child Folder: With respect to users OddThinking is a child folder (as well as sub folder)
   4. Parent Folder: For OddThinking users is its parent folder (Just mentioning different terminologies, no big deal).

4. Directory or Directory Name: The former to use generally in real life, the latter to be in code. This refers to the fully qualified path (or simply full path) till the target's parent folder. In your case, C:\users\OddThinking\Documents\My Source\Widget (Yes a directory is never meant to point to a file). I use directory name in my code since directory is a class in .NET and Directory Name is what the library itself calls it. Its quite consistent with dirname used in UNIX systems.

5. File Name or Basename: Name of the file along with extension. In your case: foo.src. I would say that for a non technical use I prefer file name (it is what it means for an end user) but for technical purposes I would strictly stick with basename. File Name is often used by MS, but I am surprised how they are not consistent not just in documentation but even in library. There filename could mean either basename or full path of the file. So I favour basename, that's what I call them in code. This page on wiki too says file name could mean either full path or the basename. Surprisingly even in .NET I can find the usage basename to mean the root name of the file.

6. Extension or Filename Extension or File Extension: I like the last one. All refers to the same thing but what is it is again a matter of debate! Wiki says it is src while back then I remember reading that many of the languages interprets it as .src. Note the dot. So once again my take is, for casual uses it doesn't matter what it is, but as a programmer I always see extension as .src.

   Ok I might have tried to fetch some standard usages, but here are two of my conventions I follow. And it is about full paths.

   1. I generally call a full path that point to a file as file path. To me file path is clear cut, it tells me what it is. Though with file name I find it as the name of the file, in my code I call it file name. It's also consistent with "directory name". From the technical side, name refers to the fully qualified name! Frustratingly .NET uses the term file name (so I have my case here) and sometimes file path for this.

   2. I call a full path that ends as a directory a directory. In fact one can call any piece of address that doesn't point to a file a directory. So C:\users\OddThinking\Documents\My Source\ is a directory, C:\users\OddThinking\ is a directory, or even OddThinking\Documents\My Source\ (better to call it sub directory or even better relative path - all that depends on the context you are dealing with it). Well above I mentioned something different about directory which is directory name. Here is my take on it: I'll get a new path to avoid confusion. What is this D:\Fruit\Apple\Pip\? A directory. But if the question is what is the directory or even better directory name of D:\Fruit\Apple\Pip\, the answer is D:\Fruit\Apple\. Hope its clear.

   I would say it's better not to worry about the final two terms as that is what create the most confusion (for me personally). Just use the term full path!

To answer you:

1. with respect to the path you have given

   A) No idea. Anyways I never needed to get that one alone.

   B) basename

   C) I would just call it file extension for time being, I am least worried since I never needed that alone to be named in my code.

   D) file extension surely.

   E) I do not think this is a general purpose requirement. No idea. In .NET base directory is the same as directory name.

   F) relative path

   G) folder (parent folder to basename foo.src)

   H) directory name

   I) full path (or even file name)

2. in general (sorry for being a bit verbose, just to drive the point home) but assuming foo.src is indeed a file

   A) NA

   B) basename

   C) NA

   D) extension

   E) directory or simply path

   F) relative path

   G) NA

   H) directory or simply path

   I) full path (or even file name)

Further driving with one example from my side:

1. Consider the path C:\Documents and Settings\All Users\Application Data\s.sql.

   1. C:\Documents and Settings\All Users\Application Data\s.sql is the full path (which is a file name)
   2. C:\Documents and Settings\All Users\Application Data\ is the directory name.

2. Now consider the path C:\Documents and Settings\All Users\Application Data

   1. C:\Documents and Settings\All Users\Application Data is the full path (which happens to be a directory)
   2. C:\Documents and Settings\All Users is the directory name.

Two tips of mine:

1. I follow this rule of thumb that when it comes to addressing a full address irrespective of its type, I almost always call it "full path". This not only eliminates the use of two terminologies for file path and folder path, it also avoids the potential confusion if you are going to name that of file as file name (which for most users right away translates to basename). But yes if you have to be specific about the type of path, its better to name then file name or directory instead of more generic "path".

2. Whatever it is you would have your own idea in mind, be consistent with it throughout. Have a consensus among team members that this means this and not that.

Now that just from the circle I have some practice. A new brand of terms would be what is used on OS X and android machines. And all these are just about physical paths in filesystem. A whole new set of terminologies would arise in case of web addresses. I expect someone to fill the void in this same thread :) I would be glad to hear the convention with which you have went ahead..

Solution 3

In C++, Boost.Filesystem has devised a nomenclature for the various parts of a path. See the path decomposition reference documentation for details, as well as this tutorial.

Here's a summary based on the tutorial. For:

• Windows path: c:\foo\bar\baa.txt
• Unix path: /foo/bar/baa.txt

you get:

Part            Windows          Posix
--------------  ---------------  ---------------
Root name       c:               <empty>
Root directory  \                /
Root path       c:\              /
Relative path   foo\bar\baa.txt  foo/bar/baa.txt
Parent path     c:\foo\bar       /foo/bar
Filename        baa.txt          baa.txt
Stem            baa              baa
Extension       .txt             .txt

C++ standard ISO/IEC 14882:2017

Moreover Boost.Filesystem terminology has been adopted by C++17 => See std::filesystem

Function name     Meaning
----------------  -------------------------------
root_name()       Root-name of the path
root_directory()  Root directory of the path
root_path()       Root path of the path
relative_path()   Path relative to the root path
parent_path()     Path of the parent path
filename()        Path without base directory (basename)
stem()            Filename without extension
extension()       Component after last dot

Solution 4

The Pathlib standard library in Python has a simple naming convention for path components:

A. /x/y/z/foo.tar.gz > stem.

B. /x/y/z/foo.tar.gz > name.

C. /x/y/z/foo.tar.gz (excluding dot) > N/A.

D. /x/y/z/foo.tar.gz (including dot) > suffix.

E. /x/y/z/foo.tar.gz > grand parent path.

F. /x/y/z/foo.tar.gz > relative path to grand parent path.

G. /x/y/z/foo.tar.gz > parent name.

H. /x/y/z/foo.tar.gz > parent path.

I. /x/y/z/foo.tar.gz > path.

Solution 5

No you're not crazy.

In Windows systems, sometimes the path of the directory containing the file is called path, which is how it was from the beginning. So, for example,

    x:\dir1\dir2\myfile.txt

    Windows:
    --------
        PATH:  x:\dir1\dir2
        FILE:  myfile.txt

    Unix/Linux:
    -----------
        PATH:  /dir1/dir2/myfile.txt
        FILE:  myfile.txt

The Unix/Linux approach is a lot more logical, and that's what everyone mentioned above: path including the file name itself. However, if you type "call /?" in the Windows command line, you get this:

    %~1         - expands %1 removing any surrounding quotes (")
    %~f1        - expands %1 to a fully qualified path name
    %~d1        - expands %1 to a drive letter only
    %~p1        - expands %1 to a path only
    %~n1        - expands %1 to a file name only
    %~x1        - expands %1 to a file extension only

So there it is, "path only" and "file name only". At the same time, they refer to the whole string as "fully qualified path name" which is understood as drive letter plus path plus file name. So there's no real truth. It's futile. You've been betrayed.

Anyway,

To answer your question

This is how I'd name your examples:

A: -
B: basename
C: extension
D: -
E: -
F: -
G: -
H: pathname (or dirname or containing path)
I: full name

A-D-E-F have no simple nicknames. And since php is probably the most widely known cross-platform language, everyone understands "basename" and "dirname" so I'd stick with that naming. Full name is also obvious; full path would be a bit ambiguous but most of the time it means the very same thing.

View more solutions

Comments

[Oddthinking]

Mike Pope, a technical editor at Microsoft, points out on his blog that that while the Microsoft style guide sticks consistently to two words: file name, folder name, volume name, the Apple Style Guide sometimes joins them: filename, pathname, volume name.

[wisbucky]

A) should definitely not be called basename because basename is already used in many places to mean the last item in a path (for a file, that would be the filename without dirpath). Some places call the filename without extension the stem.

[user117529]

Also, for files with multiple periods (e.g., foo.src.txt) , is there any standard way of identifying (and naming) the extension/s?

[Oddthinking]

It's getting off-topic, but be careful with the storage of the extension separate to the dot. You need to handle file names of "foo", "foo." and "foo.txt" (and even "foo.txt.bak".)

[Victor]

hi guys, great example. It would be more easy to read if you put the answer next to the question, instead of using references that forces to scroll up. I make an edit by the way in order to improve that. Grettings

[blinry]

Victor, since your edit got rejected (wtf guys, this is a very good improvement!) I just did it myself :-)

[toxalot]

Actually, I think the edit makes it much worse. I had to look at the source code to see what each point was referring to. The font used for the <code> element is already bold on my system, so I can't see which part of the of the absolute path is bolded. It would be helpful to include the A), B), etc. and to remove the code elements. I'll try suggesting an edit.

[polyvertex]

For 1. (file name only without extension), I decided to go with File Title a long ago due to the lack of a clear convention or at least a global consensus.

[wisbucky]

What do they call the entire thing then? path, fullpath?

[Emile Cormier]

@wisbucky The entire thing is called "path" in their nomenclature.

[wisbucky]

For A (filename without extension), you could use stem. References: doc.rust-lang.org/std/path/struct.Path.html#method.file_stem , llvm.org/docs/doxygen/html/… , boost.org/doc/libs/1_60_0/libs/filesystem/doc/…

[Emile Cormier]

@wisbucky Fixed the link. Thanks.

[GDS]

@blinry How about a "C:\users\OddThinking\Documents\My Source\Widget\" vs "C:\users\OddThinking\Documents\My Source\Widget" (final slash missing). Is there a difference in naming these two?

[Emile Cormier]

@olibre: Thanks for the C++17 update. But stem() is a part of the filename, not the path.

[oHo]

Oops, you are right! std::filesystem::path("/foo/bar.txt").stem() --> "bar" Thanks for pointing out this important detail ;-) Cheers

[john c. j.]

Best answer here.

[Emile Cormier]

@johnc.j. It's too bad Boost.Filesystem wasn't as well known when the question was first asked. I'd rather adopt the nomenclature of a peer-reviewed library than make something up on my own.

[Nate]

For a long time I've been using the word "pathname" to mean the entire absolute path including the full filename. Other answers here and resources elsewhere have changed my mind about that, and now i'll use the word "fullpath" for this, "path" for the location without filename, and "filename" or "name" for the filename itself.

[Nate]

For a long time I've been using the word "pathname" to mean the entire absolute path including the full filename. Your answer, others here, and resources elsewhere have changed my mind about that, and now i'll use the word "fullpath" for this, "path" for the location without filename, and "filename" or "name" for the filename itself.

[dkellner]

This "dot without extension" thing is strange, and it's a very good point. I think I've never seen a file with a dot at the end. That would mean "no extension" and I'm pretty sure someone will swallow that abandoned dot. It seems to me that it just means the same: "file." or "file" - but yes, in theory these are different. Maybe on Linux they are. MSDOS lists files with no extension if you say "dir *."

[run_the_race]

I want to adopt this but suffix instead of extension, where did that come? Hust seems so unintuitive.

[run_the_race]

Th docs say "PurePath.suffix The file extension of the final component, if any:" I could say I want to call it shortyendthing, and call it a synonym too. Path.extension does not exist. I get what its for, I don't get why they made up a new name for an existing concept.

[a2k42]

I'd suggest removing the leading slash for directories so that they can be appended.

[user90726]

I should note that the standard itself (fifth edition, 2017-12) doesn't use any of these words. "File name" is used only once and written as two words. On page 26: "The sequences in both forms of header-names are mapped in an implementation-defined manner to headers or to external source file names as specified in 19.2."

[Emile Cormier]

@jsv The words appear in the function names, as shown by olibre's edit.

[Inigo]

the use of bold to identify parts of the path made those parts hard to discern, at least on my system. I replaced with underlines and fixed font, which should be much easier for people to read on any system, regardless of CSS in use.