Add support for regexp character classes

Closes #21
Beaglefoot · Jan 14, 2024 · dcf4ac4 · dcf4ac4
1 parent 4b4b46c
commit dcf4ac4
Show file tree

Hide file tree

Showing 6 changed files with 55,258 additions and 54,698 deletions.
diff --git a/grammar.d.ts b/grammar.d.ts
@@ -57,7 +57,8 @@ declare interface IRules {
   regex: ($: IRules & IExternals) => any;
   _regex_char: ($: IRules & IExternals) => any;
   _regex_char_escaped: ($: IRules & IExternals) => any;
-  _regex_char_list: ($: IRules & IExternals) => any;
+  _regex_char_class: ($: IRules & IExternals) => any;
+  _regex_bracket_exp: ($: IRules & IExternals) => any;
   regex_pattern: ($: IRules & IExternals) => any;
   regex_flags: ($: IRules & IExternals) => any;
   regex_constant: ($: IRules & IExternals) => any;

diff --git a/grammar.js b/grammar.js
@@ -375,15 +375,37 @@ module.exports = grammar({
 
     _regex_char_escaped: $ => token.immediate(seq('\\', /./)),
 
-    _regex_char_list: $ =>
+    _regex_char_class: $ =>
       seq(
         token.immediate('['),
-        repeat1(choice($._regex_char_escaped, $._regex_char)),
+        token.immediate(':'),
+        choice(
+          'alnum',
+          'alpha',
+          'blank',
+          'cntrl',
+          'digit',
+          'graph',
+          'lower',
+          'print',
+          'punct',
+          'space',
+          'upper',
+          'xdigit'
+        ),
+        token.immediate(':'),
+        token.immediate(']')
+      ),
+
+    _regex_bracket_exp: $ =>
+      seq(
+        token.immediate('['),
+        repeat1(choice($._regex_char_escaped, $._regex_char, $._regex_char_class)),
         token.immediate(']')
       ),
 
     regex_pattern: $ => {
-      return repeat1(choice($._regex_char, $._regex_char_escaped, $._regex_char_list));
+      return repeat1(choice($._regex_char, $._regex_char_escaped, $._regex_bracket_exp));
     },
 
     regex_flags: $ => token.immediate(/[a-z]+/),

diff --git a/src/grammar.json b/src/grammar.json
@@ -2583,7 +2583,93 @@
         ]
       }
     },
-    "_regex_char_list": {
+    "_regex_char_class": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": "["
+          }
+        },
+        {
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": ":"
+          }
+        },
+        {
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "alnum"
+            },
+            {
+              "type": "STRING",
+              "value": "alpha"
+            },
+            {
+              "type": "STRING",
+              "value": "blank"
+            },
+            {
+              "type": "STRING",
+              "value": "cntrl"
+            },
+            {
+              "type": "STRING",
+              "value": "digit"
+            },
+            {
+              "type": "STRING",
+              "value": "graph"
+            },
+            {
+              "type": "STRING",
+              "value": "lower"
+            },
+            {
+              "type": "STRING",
+              "value": "print"
+            },
+            {
+              "type": "STRING",
+              "value": "punct"
+            },
+            {
+              "type": "STRING",
+              "value": "space"
+            },
+            {
+              "type": "STRING",
+              "value": "upper"
+            },
+            {
+              "type": "STRING",
+              "value": "xdigit"
+            }
+          ]
+        },
+        {
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": ":"
+          }
+        },
+        {
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": "]"
+          }
+        }
+      ]
+    },
+    "_regex_bracket_exp": {
       "type": "SEQ",
       "members": [
         {
@@ -2605,6 +2691,10 @@
               {
                 "type": "SYMBOL",
                 "name": "_regex_char"
+              },
+              {
+                "type": "SYMBOL",
+                "name": "_regex_char_class"
               }
             ]
           }
@@ -2633,7 +2723,7 @@
           },
           {
             "type": "SYMBOL",
-            "name": "_regex_char_list"
+            "name": "_regex_bracket_exp"
           }
         ]
       }

diff --git a/src/node-types.json b/src/node-types.json
@@ -5165,6 +5165,18 @@
     "type": "^=",
     "named": false
   },
+  {
+    "type": "alnum",
+    "named": false
+  },
+  {
+    "type": "alpha",
+    "named": false
+  },
+  {
+    "type": "blank",
+    "named": false
+  },
   {
     "type": "break_statement",
     "named": true
@@ -5173,6 +5185,10 @@
     "type": "case",
     "named": false
   },
+  {
+    "type": "cntrl",
+    "named": false
+  },
   {
     "type": "concatenating_space",
     "named": true
@@ -5189,6 +5205,10 @@
     "type": "delete",
     "named": false
   },
+  {
+    "type": "digit",
+    "named": false
+  },
   {
     "type": "do",
     "named": false
@@ -5221,6 +5241,10 @@
     "type": "getline",
     "named": false
   },
+  {
+    "type": "graph",
+    "named": false
+  },
   {
     "type": "identifier",
     "named": true
@@ -5233,6 +5257,10 @@
     "type": "in",
     "named": false
   },
+  {
+    "type": "lower",
+    "named": false
+  },
   {
     "type": "namespace",
     "named": false
@@ -5253,6 +5281,10 @@
     "type": "printf",
     "named": false
   },
+  {
+    "type": "punct",
+    "named": false
+  },
   {
     "type": "regex_flags",
     "named": true
@@ -5261,14 +5293,26 @@
     "type": "return",
     "named": false
   },
+  {
+    "type": "space",
+    "named": false
+  },
   {
     "type": "switch",
     "named": false
   },
+  {
+    "type": "upper",
+    "named": false
+  },
   {
     "type": "while",
     "named": false
   },
+  {
+    "type": "xdigit",
+    "named": false
+  },
   {
     "type": "{",
     "named": false