Personal emacs config
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

197 lines
6.6 KiB

  1. ;;; language-id.el --- Library to work with programming language identifiers -*- lexical-binding: t -*-
  2. ;;
  3. ;; Author: Lassi Kortela <lassi@lassi.io>
  4. ;; URL: https://github.com/lassik/emacs-language-id
  5. ;; Package-Version: 0.10
  6. ;; Package-Commit: 3f0ad28202207c266bd8fc7904b224db69ceccf6
  7. ;; Version: 0.10
  8. ;; Package-Requires: ((emacs "24") (cl-lib "0.5"))
  9. ;; Keywords: languages util
  10. ;; SPDX-License-Identifier: ISC
  11. ;;
  12. ;; This file is not part of GNU Emacs.
  13. ;;
  14. ;;; Commentary:
  15. ;;
  16. ;; language-id is a small, focused library that helps other Emacs
  17. ;; packages identify the programming languages and markup languages
  18. ;; used in Emacs buffers. The main point is that it contains an
  19. ;; evolving table of language definitions that doesn't need to be
  20. ;; replicated in other packages.
  21. ;;
  22. ;; Right now there is only one public function, `language-id-buffer'.
  23. ;; It looks at the major mode and other variables and returns the
  24. ;; language's GitHub Linguist identifier. We can add support for
  25. ;; other kinds of identifiers if there is demand.
  26. ;;
  27. ;; This library does not do any statistical text matching to guess the
  28. ;; language.
  29. ;;
  30. ;;; Code:
  31. (defvar language-id--file-name-extension nil
  32. "Internal variable for file name extension during lookup.")
  33. ;; <https://raw.githubusercontent.com/github/linguist/master/lib/linguist/languages.yml>
  34. (defconst language-id--definitions
  35. '(
  36. ;;; Definitions that need special attention to precedence order.
  37. ;; json-mode is derived from javascript-mode.
  38. ("JSON"
  39. json-mode
  40. (web-mode (web-mode-content-type "json") (web-mode-engine "none")))
  41. ;; php-mode is derived from c-mode.
  42. ("PHP" php-mode)
  43. ;; scss-mode is derived from css-mode.
  44. ("SCSS" scss-mode)
  45. ;; terraform-mode is derived from hcl-mode.
  46. ("Terraform" terraform-mode)
  47. ;; TypeScript/TSX need to come before JavaScript/JSX because in
  48. ;; web-mode we can tell them apart by file name extension only.
  49. ;;
  50. ;; This implies that we inconsistently classify unsaved temp
  51. ;; buffers using TypeScript/TSX as JavaScript/JSX.
  52. ("TSX"
  53. typescript-tsx-mode
  54. (web-mode
  55. (web-mode-content-type "jsx")
  56. (web-mode-engine "none")
  57. (language-id--file-name-extension ".tsx")))
  58. ("TypeScript"
  59. typescript-mode
  60. (web-mode
  61. (web-mode-content-type "javascript")
  62. (web-mode-engine "none")
  63. (language-id--file-name-extension ".ts")))
  64. ;; vue-html-mode is derived from html-mode.
  65. ("Vue"
  66. vue-mode
  67. vue-html-mode
  68. (web-mode (web-mode-content-type "html") (web-mode-engine "vue")))
  69. ;;; The rest of the definitions are in alphabetical order.
  70. ("Assembly" asm-mode nasm-mode)
  71. ("ATS" ats-mode)
  72. ("Bazel" bazel-mode)
  73. ("BibTeX" bibtex-mode)
  74. ("C" c-mode)
  75. ("C++" c++-mode)
  76. ("Cabal Config" haskell-cabal-mode)
  77. ("Clojure" clojurescript-mode clojurec-mode clojure-mode)
  78. ("CMake" cmake-mode)
  79. ("Common Lisp" lisp-mode)
  80. ("Crystal" crystal-mode)
  81. ("CSS"
  82. css-mode
  83. (web-mode (web-mode-content-type "css") (web-mode-engine "none")))
  84. ("D" d-mode)
  85. ("Dart" dart-mode)
  86. ("Dhall" dhall-mode)
  87. ("Dockerfile" dockerfile-mode)
  88. ("Elixir" elixir-mode)
  89. ("Elm" elm-mode)
  90. ("Emacs Lisp" emacs-lisp-mode)
  91. ("Fish" fish-mode)
  92. ("GLSL" glsl-mode)
  93. ("Go" go-mode)
  94. ("GraphQL" graphql-mode)
  95. ("Haskell" haskell-mode)
  96. ("HCL" hcl-mode)
  97. ("HTML"
  98. html-helper-mode mhtml-mode html-mode nxhtml-mode
  99. (web-mode (web-mode-content-type "html") (web-mode-engine "none")))
  100. ("Java" java-mode)
  101. ("JavaScript"
  102. (js-mode (flow-minor-mode nil))
  103. (js2-mode (flow-minor-mode nil))
  104. (js3-mode (flow-minor-mode nil))
  105. (web-mode (web-mode-content-type "javascript") (web-mode-engine "none")))
  106. ("JSON"
  107. json-mode
  108. (web-mode (web-mode-content-type "json") (web-mode-engine "none")))
  109. ("Jsonnet" jsonnet-mode)
  110. ("JSX"
  111. js2-jsx-mode jsx-mode rjsx-mode react-mode
  112. (web-mode (web-mode-content-type "jsx") (web-mode-engine "none")))
  113. ("Kotlin" kotlin-mode)
  114. ("LaTeX" latex-mode)
  115. ("Less" less-css-mode)
  116. ("Literate Haskell" literate-haskell-mode)
  117. ("Lua" lua-mode)
  118. ("Markdown" gfm-mode markdown-mode)
  119. ("Nix" nix-mode)
  120. ("Objective-C" objc-mode)
  121. ("OCaml" caml-mode tuareg-mode)
  122. ("Perl" cperl-mode perl-mode)
  123. ("Protocol Buffer" protobuf-mode)
  124. ("PureScript" purescript-mode)
  125. ("Python" python-mode)
  126. ("R" ess-r-mode (ess-mode (ess-dialect "R")))
  127. ("Racket" racket-mode)
  128. ("Reason" reason-mode)
  129. ("Ruby" enh-ruby-mode ruby-mode)
  130. ("Rust" rust-mode rustic-mode)
  131. ("Scala" scala-mode)
  132. ("Scheme" scheme-mode)
  133. ("Shell" sh-mode)
  134. ("Solidity" solidity-mode)
  135. ("SQL" sql-mode)
  136. ("Swift" swift-mode swift3-mode)
  137. ("TOML" toml-mode conf-toml-mode)
  138. ("Verilog" verilog-mode)
  139. ("XML"
  140. nxml-mode xml-mode
  141. (web-mode (web-mode-content-type "xml") (web-mode-engine "none")))
  142. ("YAML" yaml-mode))
  143. "Internal table of programming language definitions.")
  144. (defun language-id--mode-match-p (mode)
  145. "Interal helper to match current buffer against MODE."
  146. (let ((mode (if (listp mode) mode (list mode))))
  147. (cl-destructuring-bind (wanted-major-mode . variables) mode
  148. (and (derived-mode-p wanted-major-mode)
  149. (cl-every
  150. (lambda (variable)
  151. (cl-destructuring-bind (symbol wanted-value) variable
  152. (equal wanted-value
  153. (if (boundp symbol) (symbol-value symbol) nil))))
  154. variables)))))
  155. (defun language-id-buffer ()
  156. "Get GitHub Linguist language name for current buffer.
  157. Return the name of the programming language or markup language
  158. used in the current buffer. The name is a string from the GitHub
  159. Linguist language list. The language is determined by looking at
  160. the active `major-mode'. Some major modes support more than one
  161. language. In that case minor modes and possibly other variables
  162. are consulted to disambiguate the language.
  163. In addition to the modes bundled with GNU Emacs, many third-party
  164. modes are recognized. No statistical text matching or other
  165. heuristics are used in detecting the language.
  166. The language definitions live inside the language-id library and
  167. are updated in new releases of the library.
  168. If the language is not unambiguously recognized, the function
  169. returns nil."
  170. (let ((language-id--file-name-extension
  171. (downcase (file-name-extension (or (buffer-file-name) "") t))))
  172. (cl-some (lambda (definition)
  173. (cl-destructuring-bind (language-id . modes) definition
  174. (when (cl-some #'language-id--mode-match-p modes)
  175. language-id)))
  176. language-id--definitions)))
  177. (provide 'language-id)
  178. ;;; language-id.el ends here