This PR roughly implements issue #165. I am sure there is room for some improvements on my code changes so feel free to comment on everything I changed. One thing I dislike that I did is this weird if let bundle I created to check for a permission error at one place: https://github.com/Veykril/tokei/blob/0c7e18d1b0b45832c40c3a991be4bd57111ff5e5/src/utils/fs.rs#L52-L61 I couldnt come up with a better way since the nested errors are boxed and box patterns arent stable yet.

Example Output:

C:\Workspace\Rust\tokei (error_output -> origin)                                                                       
λ cargo run -- C:\Workspace\Rust\test                                                                                  
    Finished dev [unoptimized + debuginfo] target(s) in 0.18s                                                          
     Running `target\debug\tokei.exe C:\Workspace\Rust\test`                                                           
ERROR 2018-10-01T16:28:56Z: tokei::utils::fs: Permission denied for reading C:\Workspace\Rust\test\Cargo.toml          
Note: results can be inaccurate for languages marked with '(!)'                                                        
-----------------------------------------------------------------------------------------------------------------------
 Language                                                    Files        Lines         Code     Comments       Blanks 
-----------------------------------------------------------------------------------------------------------------------
 Markdown                                                        4         1380         1380            0            0 
 Rust                                                            1          123           99            1           23 
 Shell                                                           1           40           29            1           10 
 TOML (!)                                                        1           84           69            0           15 
 YAML                                                            1           80           60            8           12 
-----------------------------------------------------------------------------------------------------------------------
 Total (!)                                                        8         1707         1637           10           60
-----------------------------------------------------------------------------------------------------------------------

Skimming through the new comparison between tools, I noticed how all of them use the visually noisy hyphen ----- lines. The box drawing characters could be used to have clean, uninterrupted lines. These improve aesthetics and improve readability. Of course, this requires unicode support. I wasn't sure if the choice of hyphens was deliberate to support legacy terminals.

Some comparisons:

-----------------------------------------------------------------------------------
 Language                Files        Lines         Code     Comments       Blanks
-----------------------------------------------------------------------------------
 Dockerfile                 14          251          219            6           26
 Markdown                    3          180          180            0            0
 Rust                      102        41073        28770         8630         3673
 Shell                       6          310          195           63           52
 TOML                        8          141          120            3           18
 XML                         1       134861       133755            0         1106
-----------------------------------------------------------------------------------
 Total                     134       176816       163239         8702         4875
-----------------------------------------------------------------------------------

───────────────────────────────────────────────────────────────────────────────────
 Language                Files        Lines         Code     Comments       Blanks
───────────────────────────────────────────────────────────────────────────────────
 Dockerfile                 14          251          219            6           26
 Markdown                    3          180          180            0            0
 Rust                      102        41073        28770         8630         3673
 Shell                       6          310          195           63           52
 TOML                        8          141          120            3           18
 XML                         1       134861       133755            0         1106
───────────────────────────────────────────────────────────────────────────────────
 Total                     134       176816       163239         8702         4875
───────────────────────────────────────────────────────────────────────────────────

Following LaTeX booktabs style:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  Language                Files        Lines         Code     Comments       Blanks
───────────────────────────────────────────────────────────────────────────────────
 Dockerfile                 14          251          219            6           26
 Markdown                    3          180          180            0            0
 Rust                      102        41073        28770         8630         3673
 Shell                       6          310          195           63           52
 TOML                        8          141          120            3           18
 XML                         1       134861       133755            0         1106
───────────────────────────────────────────────────────────────────────────────────
 Total                     134       176816       163239         8702         4875
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

It seems piping output around doesn't update the file total, plus, only the file types from the final command are being tallied.

~$  tokei usercorn-unstable/
-------------------------------------------------------------------------------
 Language            Files        Lines         Code     Comments       Blanks
-------------------------------------------------------------------------------
 Go                    149         9163         7886          298          979
 Makefile                1           26           18            0            8
 Markdown                3          254          254            0            0
 Python                  2          154          125            2           27
 YAML                    1           20           17            0            3
-------------------------------------------------------------------------------
 Total                 156         9617         8300          300         1017
-------------------------------------------------------------------------------


~$  tokei go/src/github.com/
-------------------------------------------------------------------------------
 Language            Files        Lines         Code     Comments       Blanks
-------------------------------------------------------------------------------
 BASH                    1            8            6            1            1
 C Header                6         3435         2939          400           96
 Go                    170        29590        25755         1492         2343
 Markdown               13         1090         1090            0            0
 Python                  2          130          101            9           20
 YAML                    6           69           65            0            4
-------------------------------------------------------------------------------
 Total                 198        34322        29956         1902         2464
-------------------------------------------------------------------------------

~$  tokei -o json go/src/github.com/ | tokei -i stdin usercorn-unstable/
-------------------------------------------------------------------------------
 Language            Files        Lines         Code     Comments       Blanks
-------------------------------------------------------------------------------
 Go                    149        38753        33641         1790         3322
 Makefile                1           26           18            0            8
 Markdown                3         1344         1344            0            0
 Python                  2          284          226           11           47
 YAML                    1           89           82            0            7
-------------------------------------------------------------------------------
 Total                 163        43939        38256         2202         3481
-------------------------------------------------------------------------------

I am installing and testing master and 12.0.4 using the current stable compiler 1.45.2. All tests are passing. But running tokei results in STATUS_ACCESS_VIOLATION (segfault). However the debug build seems to run fine (So it could be caused through optimizations?). When debugging it (release mode + debug info), it seems that it is crashing inside of the ignore crate, in the walk.rs file, in the visit method, on line 1293, where all the thread handles are joined. [link] This seems to be called by your utils/fs.rs, in the get_all_files method, on line 57, where you build and run the parallel file walkers. [link]

This was an issue in the past: #543

I was trying to use tokei to analyze older versions of the Linux kernel and noticed that it had issues with files containing comments which were encoded in iso-8859-1.

Because DecodeReaderBytes wasn't actually configured to use any encoding it fell back to using UTF-8, unless a BOM was present. BOMs are exceedingly rare when dealing with code files in my experience, but it's still something that we should re-incorporate here before merging to maintain the old behavior. ~~My suggestion would be to perform an explicit check of the two first bytes in the stream~~ (see TODO).

Unfortunately file encodings tend to be determined by build systems rather than the files themselves. But we can do other things like look for vim comments (e.g. # vim: set fileencoding=).

While reading through the code, I noticed that the analysis performed by tokei didn't use a ton of string APIs, so I built the Bytes abstraction to handle all the cases that were needed to operate directly on top of a &[u8] slice.

Any char inspection that we do now tries to decode a line using a method called utf8_chars_lossy(), which replaces any illegal utf-8 sequences with a replacement character U+FFFD.

TODO

• [x] Check BOM.
• [x] Test for binary files
  • this was done before because DecodeReaderBytes always attempted to decode at least as UTF-8.

If a file don't have either an extension or a shebang, tokei should count it as plain text. If a file has an extension or a shebang, but it is not known by tokei, it should count it as Unsupported, and use the same rules as used with plain text (every line is code).

Clojure - Removed #
Forth   - ( Comment) style comments need a space after the opening paren
Haskell - Added multiline comment style {- Block comment -}
Jai     - Has nested block comments
Julia   - Has nested block comments
Kotlin  - Has nested block comments
Pascal  - Pascal should be multiline from { or (* to } or *)
Perl    - Perl5 and earlier for multiline comments need =pod to =cut.
            Perl6 will be different but as it is not finished I'm not going
            to add it.
Python  - Python can use """ or ''' for comments
Swift   - Has nested block comments

The is_fortran code was editted by rustfmt.

Now with the inclusion handling comments while in quotes, and out of them. All existing languages need to be updated with their string type as currently the default is (", ") which obviously doesn't apply to all languages. All that is needed to go to src/lib/language/languages.rs and use the set_quotes() method.

• [ ] ActionScript
• [ ] Ada
• [x] Assembly
• [ ] Autoconf
• [x] BASH
• [x] Batch
• [x] C
• [x] C Header
• [x] Clojure
• [x] CoffeeScript
• [x] ColdFusion
• [x] ColdFusion CFScript
• [ ] Coq
• [x] C++
• [x] C++ Header
• [ ] C#
• [ ] C Shell
• [x] CSS
• [ ] D
• [x] Dart
• [ ] Device Tree
• [ ] Erlang
• [ ] Forth
• [ ] FORTRAN Legacy
• [ ] FORTRAN Modern
• [ ] Go
• [ ] Handlebars
• [x] Haskell
• [x] HTML
• [x] Idris
• [ ] Isabelle
• [x] JAI
• [x] Java
• [x] JavaScript
• [x] Julia
• [x] JSON
• [ ] JSX
• [x] Kotlin
• [ ] LESS
• [ ] LD Script
• [ ] LISP
• [ ] Lua
• [ ] Makefile
• [x] Markdown
• [ ] Mustache
• [ ] Nim
• [ ] Objective C
• [ ] Objective C++
• [ ] OCaml
• [ ] Oz
• [ ] Pascal
• [ ] Perl
• [ ] Polly
• [x] PHP
• [ ] Protocol Buffers
• [ ] Prolog
• [ ] Python
• [ ] QCL
• [ ] R
• [ ] Ruby
• [ ] Ruby HTML
• [x] Rust
• [ ] Sass
• [ ] Scala
• [ ] Standard ML
• [ ] SQL
• [ ] Swift
• [ ] TeX
• [x] Plain Text
• [ ] TOML
• [ ] TypeScript
• [ ] Vim Script
• [ ] Unreal Script
• [ ] Wolfram
• [ ] XML
• [ ] YAML
• [x] Zsh

This pull adds support for Terra, a strict-typed systems language that is metaprogrammable in Lua, as well as testing support for both Terra and Lua. I also took the liberty of adding more of the blockquote and block-comment variants Lua supports (you can have arbitrarily many =s between the square brackets to produce different levels of nesting), and moved the blockquotes to verbatim, as they can include literal "s.

Closes #256.

Add a new option, -n/--num-format to specify whether the output should separated by commas, dots, or nothing (plain).

This adds a new import, num-format, to perform the formatting. This adds a total of three new dependencies: arrayvec version 0.4, ndodrop, and num-format.

If we like the way this looks, I can add a configuration option. I would assume that most users would prefer to put their preferred value in the configuration option so they don't have to specify it every time.

I'd like to analyze files that are in tarballs and bare git checkouts, on the fly, without copying these files to disk.

Could you expose an interface that would make it possible? e.g.

let mut tokei = Tokei::new();
for imaginary_path in fake_filesystem {
    tokei.add_file(imaginary_path, bytes_of_file_body); // e.g. "file.rs", &[u8]
}
let counts = tokei.totals();

This adds the jq language

• comments: https://github.com/stedolan/jq/wiki/jq-Language-Description#jq-program-structure-and-basic-syntax
• strings: https://github.com/stedolan/jq/wiki/jq-Language-Description#data-types

when working on gradle projects, gradlew scripts are vendored together. it would be great if tokei could default ignore following files.

gradlew
gradlew.bat

Visual programming language for music and multimedia, for details see:

What is Max? | Cycling '74 https://cycling74.com/products/max

Max (software) - Wikipedia https://en.wikipedia.org/wiki/Max_(software)

Note: maxpat's are json code, which do not support comments, thus I skipped adding a test file

Thanks a lot for this great piece of software!

I think having languages colored is a good way of quickly distinguishing them. So I went ahead and tried to implemented it.

I gave my best to implement it in a way compatible with the project structure.

I already added colors for JavaScript, HTML and Svelte files. If the code looks good I would also add colors for all the existing languages before merging, I just wanted to confirm the functionality and implementation with a maintainer. HTML and JavaScript make it possible to manually test if the code is working.

It works by adding a key in the languages.json: "color": {"r": number, "g": number, "b": number} and if none is defined it just does not apply any color.

This PR should also close #835.