Awesome

diagnostic-languageserver

General purpose Language Server that integrate with linter to support diagnostic features

Main features

diagnostic with linters
document format

screenshot with neovim and coc

Install

yarn global add diagnostic-languageserver

make sure your yarn's global bin path is include in PATH

for example export PATH="$(yarn global bin):$PATH"

Config & Document

languageserver config:

{
  "languageserver": {
    "dls": {
      "command": "diagnostic-languageserver",
      "args": ["--stdio", "--log-level", "2"],
      "filetypes": [ "sh", "email" ], // filetypes that you want to enable this lsp
      "initializationOptions": {
        "linters": {
          ...
        },
        "filetypes": {
          ...
        },
        "formatters": {
          ...
        },
        "formatFiletypes": {
          ...
        }
      }
    }
  }
}

linters field:

{
  "linterName": {                                    // linter name, for example: vint
    "command": "shellcheck",                         // linter command
    "rootPatterns": [],                              // root patterns, default empty array
    "isStdout": true,                                // use stdout output, default true
    "isStderr": false,                               // use stderr output, default false
    "debounce": 100,                                 // debounce time
    "onSaveOnly": false,                             // linter is triggered only when file is saved
    "args": [ "--format=gcc", "-"],                  // args
    "offsetLine": 0,                                 // offsetline
    "offsetColumn": 0,                               // offsetColumn
    "sourceName": "shellcheck",                      // source name
    "ignore": [".git", "dist/"]                      // ignore pattern same as `.gitignore`
                                                     // don't forget to add `rootPatterns` when using `ignore`
                                                     // it need workspace to filter

    // Using regular expressions:
    "formatLines": 1,                                // how much lines for formatPattern[0] to match
    "formatPattern": [
      "^([^:]+):(\\d+):(\\d+):\\s+([^:]+):\\s+(.*)$",  // line match pattern (javascript regex)
      {
        "sourceName": 1,                             // diagnostic file use match group 1. Will default to the file being linted.
        "sourceNameFilter": true,                    // Display diagnostics only for the current file.
                                                     // Only works when sourceName is defined and when it contains either an absolute
                                                     // or relative path to the file being linted. Defaults to false.
        "line": 2,                                   // diagnostic line use match group 2
        "column": 3,                                 // diagnostic column use match group 3
        "endLine": 2,                                // diagnostic end line use match group 2. Will default to group from `line`
        "endColumn": 3,                              // diagnostic end column use match group 3. Will default to group from `column`
        "message": [5],                              // message to display use match group 5
        "security": 4                                // security to use match group 4, ignore if linter do not support security
      }
    ],

    // Using JSON:
    "parseJson": {
      "errorsRoot": "[0].messages",                  // dot separated path. Will default to whatever JSON is output
                                                     // for more information see examples at https://lodash.com/docs/#get
                                                     // %filepath will be replaced with full path to the file (like in `args`)

      // All of these support lodash.get syntax.
      "sourceName": "file",                          // propert that contains the `file`. Will default to the file being linted.
      "sourceNameFilter": true,                      // Display diagnostics only for the current file.
                                                     // Only works when sourceName is defined and when it contains either an absolute
                                                     // or relative path to the file being linted. Defaults to false.
      "line": "line",                                // property that contains the `line`
      "column": "column",                            // property that contains the `column`
      "endLine": "endLine",                          // property that contains the `endLine`. Will default to `line`
      "endColumn": "endColumn",                      // property that contains the `endColumn`. Will default to `column`
      "security": "severity",                        // property that contains the `security`
      "message": "${message} [${code}]",             // message to display
    },

    "securities": {                                  // security keys, ignore if linter do not support security
      "error": "error",                              // [key: string]?: "error" | "warning" | "info" | "hint"
      "warning": "warning",
      "note": "info"
    },
    "requiredFiles": [                               // only run linter if any of these files exist. option
      ".shellcheckrc",
      "shellcheckrc"
    ]
  }
}

filetypes field:

{
  "sh": "linterName",                          // filetype: linterName or linterName[]
  "*": "linterName"                            // `*` is for all filetypes
}

formatters field:

  "dartfmt": {                                 // formatter name
    "command": "dartfmt",                      // format command
    "args": [ "--fix" ],                       // args
    "rootPatterns": [],                        // root patterns, default empty array
    "isStdout": true,                          // use stdout output, default true
    "isStderr": false,                         // use stderr output, default false
    "doesWriteToFile": false,                  // use if formatter doesn't support stdio. should be paired with `%file`
    "requiredFiles": [                         // only run formatter if any of these files exist. optional
      ".run_dartfmt",
    ],
    "ignoreExitCode": false,                   // ignore exit code. default false: exit code > 0 will not change the file.
                                               // some formatter may exit with code > 0 so you need set it to true or number[]
                                               // exit code array that you want to ignore.
    "ignore": [".git", "dist/"]                // ignore pattern same as `.gitignore`
                                               // don't forget to add `rootPatterns` when using `ignore`
                                               // it need workspace to filter
  }

formatFiletypes field:

{
  "dart": "dartfmt",                          // filetype: formatterName or formatterName[]
  "*": "linterName"                           // `*` is for all filetypes
}

Args additional syntax

args: ["%text", "%filename", "%relativepath", "%file", "%filepath", "%dirname", "%tempfile"]

%filename will replace with basename of file
%text will replace with file content
%file will replace with full path to the file and not use stdio
%filepath will replace with full path to the file
%relativepath will replace with relative path of file
%dirname will replace with dirname of file
%tempfile will replace with the full path to a temporary file written with the contents of the document and not use stdio; this file will automatically be deleted when the command completes

How to config a new linter

shellcheck for example:

file test.sh:

#!/usr/bin/env bash

echo `ls -al`

then:

shellcheck --format=gcc test.sh

output:

t.sh:3:6: warning: Quote this to prevent word splitting. [SC2046]
t.sh:3:6: note: Useless echo? Instead of 'echo $(cmd)', just use 'cmd'. [SC2005]
t.sh:3:6: note: Use $(...) notation instead of legacy backticked `...`. [SC2006]

write pattern to match the line for line column message security:

const line = "t.sh:3:6: warning: Quote this to prevent word splitting. [SC2046]"
const formatPattern = "^[^:]+:(\\d+):(\\d+):\\s+([^:]+):\\s+(.*)$"
const match = line.match(new RegExp(formatPattern))
console.log(match)

output:

{
  0: "t.sh:3:6: warning: Quote this to prevent word splitting. [SC2046]"
  1: "3"
  2: "6"
  3: "warning"
  4: "Quote this to prevent word splitting. [SC2046]"
}

so you got:

line: match[1]
column: match[2]
message: match[4]
security: match[3]

and your formatPattern field will be:

"formatPattern": [
  "^[^:]+:(\\d+):(\\d+):\\s+([^:]+):\\s+(.*)$",    // line match pattern (javascript regex)
  {
    "line": 1,                                     // diagnostic line use match group 1
    "column": 2,                                   // diagnostic column use match group 2
    "message": [4],                                // message to display use match group 4
    "security": 3                                  // security to use match group 3, ignore if linter do not support security
  }
]

Notes if the linter's message for per issue more then one line, you have to set the formatLines to fill your pattern, and you can view the languagetool pattern for example which formatLines = 2

Example with coc.nvim

Each LSP client should support initializationOptions option, all you need for diagnostic-languageserver is put the config in initializationOptions option.

shellcheck for shell
languagetool for grammer check
more Linters config example.

coc-settings.json:

you can use this extension https://github.com/iamcco/coc-diagnostic

{
  "languageserver": {
    "dls": {
      "command": "diagnostic-languageserver",
      "args": ["--stdio"],
      "filetypes": [ "sh", "email", "dart" ],
      "initializationOptions": {
        "linters": {
          "shellcheck": {
            "command": "shellcheck",
            "debounce": 100,
            "args": [ "--format=gcc", "-"],
            "offsetLine": 0,
            "offsetColumn": 0,
            "sourceName": "shellcheck",
            "formatLines": 1,
            "formatPattern": [
              "^[^:]+:(\\d+):(\\d+):\\s+([^:]+):\\s+(.*)$",
              {
                "line": 1,
                "column": 2,
                "message": 4,
                "security": 3
              }
            ],
            "securities": {
              "error": "error",
              "warning": "warning",
              "note": "info"
            }
          },
          "languagetool": {
            "command": "languagetool",
            "debounce": 200,
            "args": ["-"],
            "offsetLine": 0,
            "offsetColumn": 0,
            "sourceName": "languagetool",
            "formatLines": 2,
            "formatPattern": [
              "^\\d+?\\.\\)\\s+Line\\s+(\\d+),\\s+column\\s+(\\d+),\\s+([^\\n]+)\nMessage:\\s+(.*)$",
              {
                "line": 1,
                "column": 2,
                "message": [4, 3]
              }
            ],
          }
        },
        "formatters": {
          "dartfmt": {
            "command": "dartfmt",
            "args": [ "--fix" ],
          }
        },
        "filetypes": {
          "sh": "shellcheck",
          "email": "languagetool"
        },
        "formatFiletypes": {
          "dart": "dartfmt"
        }
      }
    }
  }
}

TODO

local node_modules linter support like eslint or textlint
diagnostic severity
root pattern
document format

References

inspired by efm-langserver

Buy Me A Coffee ☕️

btc