omartawfik / vscat Goto Github PK

The vscode-textmate library we use does not just export foreground color, it also supports a bunch of other metadata, like underlines, and background color. We should be able to extract this info and pass it along to the colorization engine.

https://github.com/microsoft/vscode-textmate/blob/9e3c5941668cbfcee5095eaec0e58090fda8f316/src/grammar.ts#L1081-L1099

Here is a comparison between vscat and cat explaining the problem:

Check these first few output lines from executing vscode ./package.json | less -R:

  1 │ {
  2 │ ··"name":·"vscat",
  3 │ ··"description":·"Bringing·VS·Code·syntax·highlighting·to·the·term…
  4 │ ··"version":·"0.0.0",
  5 │ ··"private":·false,
  6 │ ··"repository":·"https://github.com/OmarTawfik/vscat",
  7 │ ··"author":·"Omar·Tawfik·<[email protected]…
  8 │ ··"license":·"MIT",
  9 │ ··"main":·"out/src/index.js",
 10 │ ··"bin":·{
 11 │ ····"vscat":·"out/src/cli.js"
 12 │ ··},
 13 │ ··"scripts":·{
 14 │ ····"generate-metadata":·"ts-node-transpile-only·./scripts/metadat…
 15 │ ····"generate-readme":·"ts-node·./scripts/readme",
 16 │ ····"lint":·"jest·&&·prettier·.·--list-different·&&·tsc"

• Spaces are displayed as dots by default.
• Lines that are longer than a certain number of characters are truncated with an ellipses (sounds like a bug)
• Syntax highlighting doesn't show up with less -R.

Suggestions:

• An option to omit characters related to coloring and/or showing whitespaces.
• An option to force ANSI color escape sequences and OSC 8 hyperlink sequences, even if the output is piped.

When provided a file name, we try to do our best to match it to a specific language, according to VSCode's own configuration files. We use a library called micromatch (link) for glob-matching.

Using it, we can match the expression Dockerfile.* correctly to foo/Dockerfile.Production. But currently it does not match nested globs, like **/.git/config. So we would fail to deduce that foo/.git/config is a properties file.

One of the metadata pieces we collect for each grammar is the injectTo and embeddedLanguages properties. These can be passed to vscode-textmate so that it can support nested languages. Currently we do not. We should support that and add tests accourdingly.