From e27408c6622559fb7725fe340e012d32e42e9ff5 Mon Sep 17 00:00:00 2001
From: Dustin Wilson <dustin@dustinwilson.com>
Date: Mon, 28 Jun 2021 16:52:38 -0500
Subject: [PATCH] Improvements to run command, starting on Grammar

---
 data/lua/tree-sitter-lua.json | 136 ----------------------------------
 lib/Grammar.php               |  16 ++++
 lib/Grammar/Registry.php      |  12 +--
 run                           |  68 +++++++++++------
 4 files changed, 66 insertions(+), 166 deletions(-)
 delete mode 100644 data/lua/tree-sitter-lua.json
 create mode 100644 lib/Grammar.php

diff --git a/data/lua/tree-sitter-lua.json b/data/lua/tree-sitter-lua.json
deleted file mode 100644
index 8a375d8..0000000
--- a/data/lua/tree-sitter-lua.json
+++ /dev/null
@@ -1,136 +0,0 @@
-{
-  "name": "Lua",
-  "scopeName": "source.lua",
-  "type": "tree-sitter",
-  "parser": "tree-sitter-lua",
-
-  "fileTypes": [ "lua" ],
-
-  "folds": [
-    {
-      "type": [
-        "do_statement",
-        "while_statement",
-        "repeat_statement",
-        "for_statement",
-        "for_in_statement",
-        "function",
-        "local_function",
-        "function_definition",
-        "table"
-      ],
-      "end": { "index": -1 }
-    },
-    {
-      "type": ["if_statement","elseif"],
-      "start": { "index": 1 },
-      "end": { "type": [ "else", "elseif" ] }
-    },
-    {
-      "type": "if_statement",
-      "start": { "index": 1 },
-      "end": { "index": -1 }
-    },
-    {
-      "type": "elseif",
-      "start": { "index": 1 }
-    },
-    {
-      "type": "else",
-      "start": { "index": 0 }
-    }
-  ],
-
-  "comments": {
-    "start": "-- "
-  },
-
-  "scopes": {
-    "program": "source.lua",
-
-    "comment": "comment",
-    "string": "string",
-    "number": "constant.numeric",
-
-    "function > function_name > identifier": "entity.name.function",
-    "function > function_name > property_identifier": "entity.name.function",
-    "function > function_name > method": "entity.name.function",
-    "function > parameters > identifier": "variable.parameter",
-
-    "local_function > identifier": "entity.name.function",
-    "local_function > parameters > identifier": "variable.parameter",
-
-    "function_call > identifier": "entity.name.function",
-    "function_call > property_identifier": "entity.name.function",
-    "function_call > method": "entity.name.function",
-
-    "function_definition > parameters > identifier": "variable.parameter",
-
-    "self": "variable.language",
-    "next": "entity.name.function",
-    "spread": "variable.parameter",
-    "global_variable": "variable.language",
-
-    "nil": "constant.language.nil",
-    "true": "constant.language.boolean",
-    "false": "constant.language.boolean",
-
-    "'local'": "storage.modifier",
-    "'function'": "storage.type.function",
-    "label_statement": "storage.type.label",
-
-    "'{'": "punctuation.definition.table.begin",
-    "'}'": "punctuation.definition.table.end",
-    "'('": "punctuation.definition.parameters.begin",
-    "')'": "punctuation.definition.parameters.end",
-
-    "'do'": "keyword.control",
-    "'return'": "keyword.control",
-    "'if'": "keyword.control",
-    "'then'": "keyword.control",
-    "'elseif'": "keyword.control",
-    "'else'": "keyword.control",
-    "'while'": "keyword.control",
-    "'repeat'": "keyword.control",
-    "'until'": "keyword.control",
-    "'for'": "keyword.control",
-    "'in'": "keyword.control",
-    "'goto'": "keyword.control",
-    "'end'": "keyword.control",
-
-    "'or'": "keyword.operator.logical",
-    "'and'": "keyword.operator.logical",
-    "'not'": "keyword.operator.logical",
-    "'|'": "keyword.operator.logical",
-    "'&'": "keyword.operator.logical",
-    "'~'": "keyword.operator.logical",
-
-    "'='": "keyword.operator.assignment",
-
-    "'<'": "keyword.operator.comparison",
-    "'<='": "keyword.operator.comparison",
-    "'=='": "keyword.operator.comparison",
-    "'~='": "keyword.operator.comparison",
-    "'>='": "keyword.operator.comparison",
-    "'>'": "keyword.operator.comparison",
-
-    "'<<'": "keyword.operator.arithmetic.bitwise",
-    "'>>'": "keyword.operator.arithmetic.bitwise",
-    "'+'": "keyword.operator.arithmetic",
-    "'-'": "keyword.operator.arithmetic",
-    "'*'": "keyword.operator.arithmetic",
-    "'/'": "keyword.operator.arithmetic",
-    "'//'": "keyword.operator.arithmetic",
-    "'%'": "keyword.operator.arithmetic",
-
-    "'..'": "keyword.operator.arithmetic",
-    "'^'": "keyword.operator.arithmetic",
-
-    "'#'": "keyword.operator.arithmetic",
-
-    "'.'": "punctuation.separator.object",
-    "','": "punctuation.separator.object",
-    "':'": "punctuation.separator.method",
-    "';'": "punctuation.separator.statement"
-  }
-}
diff --git a/lib/Grammar.php b/lib/Grammar.php
new file mode 100644
index 0000000..f45c8ee
--- /dev/null
+++ b/lib/Grammar.php
@@ -0,0 +1,16 @@
+<?php
+/** @license MIT
+ * Copyright 2021 Dustin Wilson et al.
+ * See LICENSE and AUTHORS files for details */
+
+declare(strict_types=1);
+namespace dW\Lit;
+
+interface Grammar {
+    public static array $fileTypes;
+    public static string $firstLineMatch;
+    public static string $name;
+    public static GrammarPattern $patterns;
+    public static GrammarRepository $repository;
+    public static string $scopeName;
+}
\ No newline at end of file
diff --git a/lib/Grammar/Registry.php b/lib/Grammar/Registry.php
index 805bc3f..97451af 100644
--- a/lib/Grammar/Registry.php
+++ b/lib/Grammar/Registry.php
@@ -34,18 +34,18 @@ class Registry {
         return false;
     }
 
-    public static function set(string $grammarPath, bool $force = false): bool {
-        if (!file_exists($grammarPath)) {
-            throw new \Exception("Path \"$grammarPath\" either does not exist or you do not have permission to view the file.");
+    public static function import(string $jsonPath, bool $force = false): bool {
+        if (!file_exists($jsonPath)) {
+            throw new \Exception("Path \"$jsonPath\" either does not exist or you do not have permission to view the file.");
         }
 
-        $grammar = json_decode(file_get_contents($grammarPath), true);
+        $grammar = json_decode(file_get_contents($jsonPath), true);
         if ($grammar === null) {
-            throw new \Exception("\"$grammarPath\" is not a valid grammar file.");
+            throw new \Exception("\"$jsonPath\" is not a valid grammar JSON file.");
         }
 
         if (!isset($grammar['scopeName'])) {
-            throw new \Exception("\"$grammarPath\" is missing the required scopeName property.");
+            throw new \Exception("\"$jsonPath\" is missing the required scopeName property.");
         }
 
         if (!$force && isset(self::$grammars[$grammar['scopeName']])) {
diff --git a/run b/run
index 27d27d5..2392404 100755
--- a/run
+++ b/run
@@ -19,30 +19,30 @@
 ##?         * bash >=5.0
 ##?         * composer >=2.0.6
 ##?         * docopts >=0.6.3
-##?         * gnu awk >=5.1.0
-##?         * gnu coreutils >=8.0
-##?         * gnu parallel >=20180922
+##?         * GNU awk >=5.1.0
+##?         * GNU coreutils >=8.0
+##?         * GNU parallel >=20180922
 ##?         * yarn >=1.22.10
 ##?
 ##?     Arch Linux:
-##?     ```
-##?     pacman -S composer docopts parallel yarn
-##?     ```
+##?         ```
+##?         pacman -S composer docopts parallel yarn
+##?         ```
 ##?
 ##?     macOS:
-##?     ```
-##?     brew install bash composer coreutils docopts gawk parallel yarn
-##?     echo \"/usr/local/bin/bash\" >> /etc/shells
-##?     chsh -s /usr/local/bin/bash
-##?     ```
+##?         ```
+##?         brew install bash composer coreutils docopts gawk parallel yarn
+##?         echo \"/usr/local/bin/bash\" >> /etc/shells
+##?         chsh -s /usr/local/bin/bash
+##?         ```
 ##?
 ##?     Building the grammars is then a matter of running the following commands
 ##?     from the project folder:
 ##?
-##?     ```
-##?     ./run init
-##?     ./run build
-##?     ```
+##?         ```
+##?         ./run init
+##?         ./run build
+##?         ```
 ##?
 ##?
 ##? Commands:
@@ -57,29 +57,41 @@
 ##?     needed
 ##?         List the language grammars that are included in the current grammars
 ##?         but are not in the data/ folder.
+##?
 
+set -e
 cwd=$(pwd)
 PATH="$cwd/node_modules/.bin:$PATH"
 help=$(grep "^##?" "$0" | cut -c 5-)
-eval "$(docopts -h "$help" : "$@")"
+eval "$(docopts -A args -h "$help" : "$@")"
 
 fail() {
     echo $help
     exit 1
 }
 
-# Check for bash 4+
-shopt -s globstar extglob || help
+# Check for bash 4+ and set extra glob options
+shopt -s globstar extglob || fail
 
+# macOS uses the BSD version of awk which isn't compatible with the linux
+# version, so this script requires Mac users to install the GNU version; it
+# therefore has a different name on macOS.
 awkcmd="awk"
 if [ "$(uname)" == "Darwin" ]; then
     awkcmd="gawk"
 fi
 
 # Check for presence of dependencies
-if (( $(which $awkcmd composer parallel yarn | wc -l) < 4 )); then help; fi
+if (( $(which $awkcmd composer parallel yarn | wc -l) < 4 )); then fail; fi
+
+# Process the subcommand
+subcmd="build"
+subcmd=$([ "${args[init]}" == "true" ] && echo "init" || echo "$subcmd")
+subcmd=$([ "${args[needed]}" == "true" ] && echo "needed" || echo "$subcmd")
 
-read -r -d '' grammarList <<'EOF'
+# Grammar table of the github repository location followed by its local folder
+# name
+read -r -d '' grammarTable <<'EOF'
 franzheidl/atom-applescript applescript
 atom/language-c c
 atom/language-coffee-script coffeescript
@@ -102,14 +114,14 @@ atom/language-typescript typescript
 atom/language-xml xml
 EOF
 
-case $1 in
+case $subcmd in
 init)
     yarn install
     mkdir -p "$cwd"/deps
     rm -rf "$cwd"/deps/*
 
     cd "$cwd"/deps
-    printf "$grammarList" | parallel --colsep ' ' "git clone https://github.com/{1}.git"
+    printf "$grammarTable" | parallel --colsep ' ' "git clone https://github.com/{1}.git"
     cd "$cwd"
     ;;
 
@@ -122,14 +134,22 @@ needed)
 build | *)
     rm -rf "$cwd"/data/*
 
+    # Copy grammars folders from all the local language grammar repositories to a
+    # temporary folder with the names provided in the table above
     temp=$(mktemp -d)
-    printf "$grammarList" | parallel --colsep ' ' "mkdir -p $temp/{2} &&
+    printf "$grammarTable" | parallel --colsep ' ' "mkdir -p $temp/{2} &&
         cp -r \"$cwd\"/deps/\$(basename {1})/grammars/* $temp/{2}/"
 
+    # Convert all relevant CSON files to a format that can actually be used :)
     ls -1 $temp/**/!(tree-sitter-*).cson | parallel "csonc --output={.}.json {} && rm {}"
 
-    rm -rf $temp/**/*.cson
+    # Remove any remaining tree sitter files; will also remove JSON ones from
+    # packages that didn't use CSON
+    rm $temp/**/tree-sitter-*
 
+    # Lastly, move everything over to the project's data folder and delete the
+    # temporary folder
     mv $temp/* "$cwd"/data/
+    rmdir $temp
     ;;
 esac
\ No newline at end of file