(the commit before proper parsers will replace everything™™)

1 files changed, 30 insertions, 23 deletions

diff --git a/code/src/syntax_highlight/grammar.rs b/code/src/syntax_highlight/grammar.rs
index 0878b33..0cecac9 100644
--- a/code/src/syntax_highlight/grammar.rs
+++ b/code/src/syntax_highlight/grammar.rs
@@ -1,3 +1,8 @@
+const CONST: &str = "[\\. \\(,=^]([A-Z][A-Z0-9_]*)[\\. ,\\)=$]";
+const TYPE: &str = "[\\. \\(,=^]([A-Z][a-zA-Z0-9_]*)[\\. ,\\)\\(=$]";
+const IDENT: &str = "[\\. \\(,=^]([a-z][a-zA-Z0-9_]*)[\\. ,\\)=$]";
+const FUNC: &str = "[\\. \\(,=^]([a-z][A-Z0-9_]*)\\(";
+
 pub fn grammar_for(syntax: &str) -> &'static [(&'static str, &'static [&'static str])] {
     match syntax {
         "rs" | "rust" => &[
@@ -11,41 +16,43 @@ pub fn grammar_for(syntax: &str) -> &'static [(&'static str, &'static [&'static
             (
                 "type",
                 &[
-                    "[A-Z][a-z0-9]*",
-                    "bool",
-                    "usize",
-                    "u8",
-                    "u16",
-                    "u32",
-                    "u64",
-                    "u128",
-                    "i8",
-                    "i16",
-                    "i32",
-                    "i64",
-                    "i128",
-                    "isize",
-                    "f32",
-                    "f64",
+                    TYPE, "bool", "usize", "u8", "u16", "u32", "u64", "u128", "i8", "i16", "i32",
+                    "i64", "i128", "isize", "f32", "f64",
                 ],
             ),
+            ("constant", &[CONST]),
             ("comment", &["(?m)(//.*)$", "(?ms)/\\*.*?\\*/"]),
-            ("macro", &["([a-z_][A-Za-z0-9_]*!)\\s*"]),
-            ("identifier", &["([a-z_][A-Za-z0-9_]*)\\s*\\("]),
-            ("literal", &["\".*?\"", "\\d", "true", "false"]),
+            ("macro", &["[\\. ,=^]([a-z_][A-Za-z0-9_]*!)[\\. ,=$]"]),
+            ("identifier", &[IDENT]),
+            ("function", &[FUNC]),
+            (
+                "literal",
+                &["\".*?\"", "(0x|0o|0b)[0-9a-f]+", "\\d+", "true", "false"],
+            ),
         ],
         "py" | "python" => &[
             ("type", &["bytes", "bool", "int", "str", "float"]),
             (
                 "keyword",
                 &[
-                    "from", "for", "def", "if", "else", "elif", "while", "with", "in", "as",
-                    "import",
+                    "from", "for", "def", "if", "else", "elif", "while", "with", "in", "assert",
+                    "global", "nonlocal", "as", "import",
                 ],
             ),
             ("comment", &["(?m)(#.*)$"]),
-            ("identifier", &["([a-z_][A-Za-z0-9_]*)\\s*\\("]),
-            ("literal", &["b?\".*?\"", "\\d", "true", "false"]),
+            ("constant", &[CONST]),
+            ("identifier", &[IDENT]),
+            (
+                "literal",
+                &[
+                    "b?f?\".*?\"",
+                    "b?f?'.*?'",
+                    "(0x|0o|0b)[0-9a-f]+",
+                    "\\d+",
+                    "true",
+                    "false",
+                ],
+            ),
         ],
         "tree" => &[("keyword", &["[├─└│]+"])],
         // makefile doesnt really match the token-kinds, i'll just use something that looks goo