feat: add support for even more languages

scrape zed-industries/extensions

extension.toml

@@ -1,6 +1,6 @@
 id = "discord_presence"
 name = "Discord Presence"
-version = "0.1.0"
+version = "0.2.0"
 schema_version = 1
 authors = ["Jozef Steinhübl <contact@xhyrom.dev>"]
 description = "Presence for your beautiful discord account :)"
@@ -10,7 +10,92 @@ repository = "https://git.sr.ht/~hyro/zed-discord-presence"
 name = "Discord Presence LSP"
 # Generated by scripts/get_languages.sh
 languages = [
+    "AsciiDoc",
+    "Assembly",
     "Shell Script",
+    "Beancount",
+    "Bend",
+    "Blade",
+    "BQN",
+    "Brainfuck",
+    "Cadence",
+    "Cairo",
+    "Cap'n Proto",
+    "Cedar",
+    "CFEngine",
+    "CSV",
+    "Gherkin",
+    "CUE",
+    "D",
+    "DBML",
+    "Earthfile",
+    "Elisp",
+    "Handlebars",
+    "Exograph",
+    "Fish",
+    "FSharp",
+    "GDScript",
+    "Godot Resource",
+    "Gemini",
+    "Diff",
+    "Git Attributes",
+    "Git Commit",
+    "Git Config",
+    "Git Ignore",
+    "Git Rebase",
+    "GraphQL",
+    "Groovy",
+    "GROQ",
+    "ini",
+    "Jsonnet",
+    "Julia",
+    "Just",
+    "LaTeX",
+    "Ledger",
+    "LESS",
+    "LilyPond",
+    "LilyPond Scheme",
+    "Liquid",
+    "Lox",
+    "Luau",
+    "Make",
+    "MATLAB",
+    "Meson",
+    "Navi",
+    "Navi Stream",
+    "Nim",
+    "Noir",
+    "Nu",
+    "Odin",
+    "Pact",
+    "Pest",
+    "pica200",
+    "Pkl",
+    "QML",
+    "R",
+    "Rainbow CSV",
+    "Rainbow TSV",
+    "rego",
+    "ReScript",
+    "Roc",
+    "RON",
+    "Scala",
+    "SCSS",
+    "Smithy",
+    "Solidity",
+    "bazelrc",
+    "Starlark",
+    "Templ",
+    "Twig",
+    "Typst",
+    "Unison",
+    "V",
+    "Vala",
+    "Verilog",
+    "VHS",
+    "Wgsl",
+    "WIT",
+    "XML",
     "C",
     "C++",
     "CSS",
@@ -21,6 +106,7 @@ languages = [
     "JSDoc",
     "JSON",
     "JSONC",
+    "Markdown-Inline",
     "Markdown",
     "proto",
     "Python",
@@ -61,4 +147,4 @@ languages = [
     "Vue.js",
     "Zig",
 ]
-language_ids = { "Shell Script" = "bash", "C" = "c", "C++" = "cpp", "CSS" = "css", "Go" = "go", "Go Mod" = "gomod", "Go Work" = "gowork", "JavaScript" = "javascript", "JSDoc" = "jsdoc", "JSON" = "json", "JSONC" = "jsonc", "Markdown" = "markdown", "proto" = "proto", "Python" = "python", "Regex" = "regex", "Rust" = "rust", "TSX" = "tsx", "TypeScript" = "typescript", "YAML" = "yaml", "Astro" = "astro", "Clojure" = "clojure", "CSharp" = "csharp", "Dart" = "dart", "Elixir" = "elixir", "HEEX" = "heex", "Elm" = "elm", "Erlang" = "erlang", "Gleam" = "gleam", "GLSL" = "glsl", "Haskell" = "haskell", "HTML" = "html", "Lua" = "lua", "OCaml Interface" = "ocaml-interface", "OCaml" = "ocaml", "PHP" = "php", "PHPDoc" = "phpdoc", "Prisma" = "prisma", "PureScript" = "purescript", "Racket" = "racket", "ERB" = "erb", "Ruby" = "ruby", "Scheme" = "scheme", "Svelte" = "svelte", "HCL" = "hcl", "Terraform Vars" = "terraform-vars", "Terraform" = "terraform", "TOML" = "toml", "Uiua" = "uiua", "Vue.js" = "vue", "Zig" = "zig"}
+language_ids = { "AsciiDoc" = "asciidoc", "Assembly" = "asm", "Shell Script" = "bash", "Beancount" = "beancount", "Bend" = "bend", "Blade" = "blade", "BQN" = "BQN", "Brainfuck" = "brainfuck", "Cadence" = "cadence", "Cairo" = "cairo", "Cap'n Proto" = "capnp", "Cedar" = "cedar", "CFEngine" = "cfengine", "CSV" = "csv", "Gherkin" = "gherkin", "CUE" = "cue", "D" = "d", "DBML" = "dbml", "Earthfile" = "earthfile", "Elisp" = "elisp", "Handlebars" = "glimmer", "Exograph" = "exograph", "Fish" = "fish", "FSharp" = "fsharp", "GDScript" = "gdscript", "Godot Resource" = "godot_resource", "Gemini" = "gemini", "Diff" = "diff", "Git Attributes" = "gitattributes", "Git Commit" = "gitcommit", "Git Config" = "gitconfig", "Git Ignore" = "gitignore", "Git Rebase" = "gitrebase", "GraphQL" = "graphql", "Groovy" = "groovy", "GROQ" = "groq", "ini" = "ini", "Jsonnet" = "jsonnet", "Julia" = "julia", "Just" = "just", "LaTeX" = "latex", "Ledger" = "ledger", "LESS" = "less", "LilyPond" = "lilypond", "LilyPond Scheme" = "lilypond_scheme", "Liquid" = "liquid", "Lox" = "lox", "Luau" = "luau", "Make" = "make", "MATLAB" = "matlab", "Meson" = "meson", "Navi" = "navi", "Navi Stream" = "navi_stream", "Nim" = "nim", "Noir" = "noir", "Nu" = "nu", "Odin" = "odin", "Pact" = "pact", "Pest" = "pest", "pica200" = "pica200", "Pkl" = "pkl", "QML" = "qml", "R" = "r", "Rainbow CSV" = "csv", "Rainbow TSV" = "tsv", "rego" = "rego", "ReScript" = "rescript", "Roc" = "roc", "RON" = "ron", "Scala" = "scala", "SCSS" = "scss", "Smithy" = "smithy", "Solidity" = "solidity", "bazelrc" = "bazelrc", "Starlark" = "starlark", "Templ" = "templ", "Twig" = "twig", "Typst" = "typst", "Unison" = "unison", "V" = "v", "Vala" = "vala", "Verilog" = "verilog", "VHS" = "vhs", "Wgsl" = "wgsl", "WIT" = "wit", "XML" = "xml", "C" = "c", "C++" = "cpp", "CSS" = "css", "Go" = "go", "Go Mod" = "gomod", "Go Work" = "gowork", "JavaScript" = "javascript", "JSDoc" = "jsdoc", "JSON" = "json", "JSONC" = "jsonc", "Markdown-Inline" = "markdown-inline", "Markdown" = "markdown", "proto" = "proto", "Python" = "python", "Regex" = "regex", "Rust" = "rust", "TSX" = "tsx", "TypeScript" = "typescript", "YAML" = "yaml", "Astro" = "astro", "Clojure" = "clojure", "CSharp" = "csharp", "Dart" = "dart", "Elixir" = "elixir", "HEEX" = "heex", "Elm" = "elm", "Erlang" = "erlang", "Gleam" = "gleam", "GLSL" = "glsl", "Haskell" = "haskell", "HTML" = "html", "Lua" = "lua", "OCaml Interface" = "ocaml-interface", "OCaml" = "ocaml", "PHP" = "php", "PHPDoc" = "phpdoc", "Prisma" = "prisma", "PureScript" = "purescript", "Racket" = "racket", "ERB" = "erb", "Ruby" = "ruby", "Scheme" = "scheme", "Svelte" = "svelte", "HCL" = "hcl", "Terraform Vars" = "terraform-vars", "Terraform" = "terraform", "TOML" = "toml", "Uiua" = "uiua", "Vue.js" = "vue", "Zig" = "zig" }

scripts/get_languages.sh

@@ -1,30 +1,64 @@
 #!/bin/bash
 
+fetch_files() {
+    local repo_owner=$1
+    local repo_name=$2
+    curl -s "https://api.github.com/repos/$repo_owner/$repo_name/git/trees/main?recursive=1" -H "Authorization: Bearer ${GITHUB_TOKEN}" | jq -c -r '.tree[] | .path'
+}
+
+process_language_files() {
+    local files=$1
+    local repo_owner=$2
+    local repo_name=$3
+    local file_patterns=("${@:4}")
+
+    for file in $files; do
+        local match=false
+        for pattern in "${file_patterns[@]}"; do
+            if [[ $file == $pattern ]]; then
+                match=true
+                break
+            fi
+        done
+
+        if ! $match; then
+            continue
+        fi
+
+        name=$(curl -s "https://raw.githubusercontent.com/$repo_owner/$repo_name/main/${file}" | grep -oP '^name = "\K[^"]*')
+        id="${file%/config.toml}"
+        id="${id##*/}"
+
+        if [[ $LANGUAGES == *"$name\","* && $LANGUAGE_IDS == *"$name\" = \"$id\","* ]]; then
+            echo "Skipping language $name with id $id (already added)"
+        else
+            echo "Adding language $name with id $id"
+            LANGUAGES+="    \"$name\",\n"
+            LANGUAGE_IDS+=" \"$name\" = \"$id\","
+        fi
+    done
+}
+
 RESPONSE=$(curl -s "https://api.github.com/repos/zed-industries/zed/git/trees/main?recursive=1" -H "Authorization: Bearer ${GITHUB_TOKEN}")
+EXTENSIONS_RESPONSE=$(curl -s "https://raw.githubusercontent.com/zed-industries/extensions/main/.gitmodules")
+REPOSITORIES=($(echo "$EXTENSIONS_RESPONSE" | grep -oP 'url = \K.*'))
 FILES=$(echo "$RESPONSE" | jq -c -r '.tree[] | .path')
 
 LANGUAGES="languages = [\n"
 LANGUAGE_IDS="language_ids = {"
 
-for file in $FILES; do
+for repository in "${REPOSITORIES[@]}"; do
-    if ! [[ $file == crates/languages/src/**/config.toml || $file == extensions/**/languages/**/config.toml ]]; then
+    repo_owner=$(echo "$repository" | cut -d'/' -f4)
-        continue
+    repo_name=$(echo "$repository" | cut -d'/' -f5 | sed 's/.git$//')
-    fi
 
-    if [[ $file == extensions/test-extension/* ]]; then
+    echo "Checking $repository ($repo_owner/$repo_name)"
-        continue
+    files=$(fetch_files "$repo_owner" "$repo_name")
-    fi
+    process_language_files "$files" "$repo_owner" "$repo_name" "languages/**/config.toml"
-
-    name=$(curl -s "https://raw.githubusercontent.com/zed-industries/zed/main/${file}" | head -n 1 | sed 's/^name = "\(.*\)"$/\1/')
-    id="${file%/config.toml}"
-    id="${id##*/}"
-
-    echo "Adding language $name with id $id"
-
-    LANGUAGES+="    \"$name\",\n"
-    LANGUAGE_IDS+=" \"$name\" = \"$id\","
 done
 
+echo "Processing zed-industries/zed"
+process_language_files "$FILES" "zed-industries" "zed" "crates/languages/src/**/config.toml" "extensions/**/languages/**/config.toml"
+
 LANGUAGES+="]"
 LANGUAGE_IDS+="}"