Improve JS/HTML syntax highlighting

Mozilla-Ocho · Nov 4, 2024 · 9b96502 · 9b96502
1 parent 14713b5
commit 9b96502
Show file tree

Hide file tree

Showing 7 changed files with 65 additions and 18 deletions.
diff --git a/llamafile/highlight.h b/llamafile/highlight.h
@@ -239,7 +239,9 @@ class HighlightJs : public Highlight {
     int u_ = 0;
     int t_ = 0;
     int expect_;
+    int nesti_ = 0;
     std::string word_;
+    unsigned char nest_[16];
 };
 
 class HighlightTypescript : public Highlight {

diff --git a/llamafile/highlight_html.cpp b/llamafile/highlight_html.cpp
@@ -206,7 +206,7 @@ void HighlightHtml::feed(std::string *r, std::string_view input) {
             if (name_ == "script") {
                 pending_.clear();
                 closer_ = "</script>";
-                highlighter_ = new HighlightC(is_keyword_js);
+                highlighter_ = new HighlightJs;
                 t_ = RELAY;
                 i_ = 0;
             } else if (name_ == "style") {

diff --git a/llamafile/highlight_js.cpp b/llamafile/highlight_js.cpp
@@ -33,6 +33,7 @@ enum {
     SLASH_STAR_STAR,
     TICK,
     TICK_BACKSLASH,
+    TICK_DOLLAR,
     REGEX,
     REGEX_BACKSLASH,
     REGEX_SQUARE,
@@ -99,18 +100,26 @@ void HighlightJs::feed(std::string *r, std::string_view input) {
             } else if (c == '\'') {
                 t_ = QUOTE;
                 *r += HI_STRING;
-                lf::append_wchar(r, c);
+                *r += '\'';
                 expect_ = EXPECT_OPERATOR;
             } else if (c == '"') {
                 t_ = DQUOTE;
                 *r += HI_STRING;
-                lf::append_wchar(r, c);
+                *r += '"';
                 expect_ = EXPECT_OPERATOR;
             } else if (c == '`') {
                 t_ = TICK;
                 *r += HI_STRING;
-                lf::append_wchar(r, c);
+                *r += '`';
                 expect_ = EXPECT_OPERATOR;
+            } else if (c == '{' && nesti_ && nesti_ < sizeof(nest_)) {
+                expect_ = EXPECT_VALUE;
+                *r += '{';
+                nest_[nesti_++] = NORMAL;
+            } else if (c == '}' && nesti_) {
+                if ((t_ = nest_[--nesti_]) != NORMAL)
+                    *r += HI_STRING;
+                *r += '}';
             } else if (c == ')' || c == '}' || c == ']') {
                 expect_ = EXPECT_OPERATOR;
                 lf::append_wchar(r, c);
@@ -240,13 +249,19 @@ void HighlightJs::feed(std::string *r, std::string_view input) {
             t_ = DQUOTE;
             break;
 
+        Tick:
         case TICK:
-            lf::append_wchar(r, c);
             if (c == '`') {
+                *r += '`';
                 *r += HI_RESET;
                 t_ = NORMAL;
+            } else if (c == '$') {
+                t_ = TICK_DOLLAR;
             } else if (c == '\\') {
+                *r += '\\';
                 t_ = TICK_BACKSLASH;
+            } else {
+                lf::append_wchar(r, c);
             }
             break;
 
@@ -255,6 +270,27 @@ void HighlightJs::feed(std::string *r, std::string_view input) {
             t_ = TICK;
             break;
 
+        case TICK_DOLLAR:
+            if (c == '{' && nesti_ < sizeof(nest_)) {
+                *r += HI_BOLD;
+                *r += '$';
+                *r += HI_UNBOLD;
+                *r += HI_STRING;
+                *r += '{';
+                *r += HI_RESET;
+                expect_ = EXPECT_VALUE;
+                nest_[nesti_++] = TICK;
+                t_ = NORMAL;
+            } else {
+                *r += HI_WARNING;
+                *r += '$';
+                *r += HI_UNBOLD;
+                *r += HI_STRING;
+                t_ = TICK;
+                goto Tick;
+            }
+            break;
+
         case REGEX:
             lf::append_wchar(r, c);
             if (c == '/') {
@@ -316,6 +352,10 @@ void HighlightJs::flush(std::string *r) {
     case SLASH:
         *r += '/';
         break;
+    case TICK_DOLLAR:
+        *r += '$';
+        *r += HI_RESET;
+        break;
     case TICK:
     case TICK_BACKSLASH:
     case QUOTE:
@@ -337,4 +377,5 @@ void HighlightJs::flush(std::string *r) {
     c_ = 0;
     u_ = 0;
     t_ = NORMAL;
+    nesti_ = 0;
 }
diff --git a/llamafile/highlight_kotlin.cpp b/llamafile/highlight_kotlin.cpp
@@ -30,20 +30,20 @@ enum {
     SLASH_SLASH,
     SLASH_STAR,
     SLASH_STAR_STAR,
-    DQUOTE, // "
-    DQUOTE_DOLLAR, // "
-    DQUOTE_VAR, // "
-    DQUOTESTR, // "...
-    DQUOTESTR_BACKSLASH, // "...
-    DQUOTESTR_DOLLAR, // "...
-    DQUOTESTR_VAR, // "...
-    DQUOTE2, // ""
-    DQUOTE3, // """...
+    DQUOTE,
+    DQUOTE_DOLLAR,
+    DQUOTE_VAR,
+    DQUOTESTR,
+    DQUOTESTR_BACKSLASH,
+    DQUOTESTR_DOLLAR,
+    DQUOTESTR_VAR,
+    DQUOTE2,
+    DQUOTE3,
     DQUOTE3_BACKSLASH,
-    DQUOTE3_DOLLAR, // """...
-    DQUOTE3_VAR, // """...
-    DQUOTE31, // """..."
-    DQUOTE32, // """...""
+    DQUOTE3_DOLLAR,
+    DQUOTE3_VAR,
+    DQUOTE31,
+    DQUOTE32,
 };
 
 HighlightKotlin::HighlightKotlin() {
@@ -384,4 +384,5 @@ void HighlightKotlin::flush(std::string *r) {
         break;
     }
     t_ = NORMAL;
+    nesti_ = 0;
 }
diff --git a/llamafile/highlight_scala.cpp b/llamafile/highlight_scala.cpp
@@ -384,4 +384,5 @@ void HighlightScala::flush(std::string *r) {
         break;
     }
     t_ = NORMAL;
+    nesti_ = 0;
 }
diff --git a/llamafile/highlight_swift.cpp b/llamafile/highlight_swift.cpp
@@ -452,4 +452,5 @@ void HighlightSwift::flush(std::string *r) {
     c_ = 0;
     u_ = 0;
     t_ = NORMAL;
+    nesti_ = 0;
 }
diff --git a/llamafile/highlight_typescript.cpp b/llamafile/highlight_typescript.cpp
@@ -368,4 +368,5 @@ void HighlightTypescript::flush(std::string *r) {
     c_ = 0;
     u_ = 0;
     t_ = NORMAL;
+    nesti_ = 0;
 }
-Original file line number
+Diff line change
@@ Expand Up / @@ -384,4 +384,5 @@ void HighlightScala::flush(std::string *r) { @@
             break;
         }
         t_ = NORMAL;
+        nesti_ = 0;
     }