From 3ac36b0cbeac190120cb39ecf9db2c2bf7036069 Mon Sep 17 00:00:00 2001
From: Max Brunsfeld <maxbrunsfeld@gmail.com>
Date: Sat, 25 Jun 2022 17:13:11 -0700
Subject: [PATCH] Handle backslashes in token names when printing DOT debug
 graphs

---
 lib/src/language.h | 25 +++++++++++++++++++++++++
 lib/src/parser.c   |  2 +-
 lib/src/stack.c    |  6 +-----
 lib/src/subtree.c  | 14 +-------------
 4 files changed, 28 insertions(+), 19 deletions(-)

diff --git a/lib/src/language.h b/lib/src/language.h
index 1b65f25f..7234685e 100644
--- a/lib/src/language.h
+++ b/lib/src/language.h
@@ -283,6 +283,31 @@ static inline void ts_language_aliases_for_symbol(
   }
 }
 
+static inline void ts_language_write_symbol_as_dot_string(
+  const TSLanguage *self,
+  FILE *f,
+  TSSymbol symbol
+) {
+  const char *name = ts_language_symbol_name(self, symbol);
+  for (const char *c = name; *c; c++) {
+    switch (*c) {
+      case '"':
+      case '\\':
+        fputc('\\', f);
+        fputc(*c, f);
+        break;
+      case '\n':
+        fputs("\\n", f);
+        break;
+      case '\t':
+        fputs("\\n", f);
+        break;
+      default:
+        fputc(*c, f);
+        break;
+    }
+  }
+}
 
 #ifdef __cplusplus
 }
diff --git a/lib/src/parser.c b/lib/src/parser.c
index 18c91f8e..a45204e1 100644
--- a/lib/src/parser.c
+++ b/lib/src/parser.c
@@ -160,7 +160,7 @@ static void ts_parser__log(TSParser *self) {
   if (self->dot_graph_file) {
     fprintf(self->dot_graph_file, "graph {\nlabel=\"");
     for (char *c = &self->lexer.debug_buffer[0]; *c != 0; c++) {
-      if (*c == '"') fputc('\\', self->dot_graph_file);
+      if (*c == '"' || *c == '\\') fputc('\\', self->dot_graph_file);
       fputc(*c, self->dot_graph_file);
     }
     fprintf(self->dot_graph_file, "\"\n}\n\n");
diff --git a/lib/src/stack.c b/lib/src/stack.c
index 2a11abd8..6962a157 100644
--- a/lib/src/stack.c
+++ b/lib/src/stack.c
@@ -846,11 +846,7 @@ bool ts_stack_print_dot_graph(Stack *self, const TSLanguage *language, FILE *f)
           fprintf(f, "label=\"");
           bool quoted = ts_subtree_visible(link.subtree) && !ts_subtree_named(link.subtree);
           if (quoted) fprintf(f, "'");
-          const char *name = ts_language_symbol_name(language, ts_subtree_symbol(link.subtree));
-          for (const char *c = name; *c; c++) {
-            if (*c == '\"' || *c == '\\') fprintf(f, "\\");
-            fprintf(f, "%c", *c);
-          }
+          ts_language_write_symbol_as_dot_string(language, f, ts_subtree_symbol(link.subtree));
           if (quoted) fprintf(f, "'");
           fprintf(f, "\"");
           fprintf(
diff --git a/lib/src/subtree.c b/lib/src/subtree.c
index 1516c18d..d6cd2d71 100644
--- a/lib/src/subtree.c
+++ b/lib/src/subtree.c
@@ -834,18 +834,6 @@ static size_t ts_subtree__write_char_to_string(char *s, size_t n, int32_t c) {
     return snprintf(s, n, "%d", c);
 }
 
-static void ts_subtree__write_dot_string(FILE *f, const char *string) {
-  for (const char *c = string; *c; c++) {
-    if (*c == '"') {
-      fputs("\\\"", f);
-    } else if (*c == '\n') {
-      fputs("\\n", f);
-    } else {
-      fputc(*c, f);
-    }
-  }
-}
-
 static const char *ROOT_FIELD = "__ROOT__";
 
 static size_t ts_subtree__write_to_string(
@@ -975,7 +963,7 @@ void ts_subtree__print_dot_graph(const Subtree *self, uint32_t start_offset,
   TSSymbol symbol = alias_symbol ? alias_symbol : subtree_symbol;
   uint32_t end_offset = start_offset + ts_subtree_total_bytes(*self);
   fprintf(f, "tree_%p [label=\"", (void *)self);
-  ts_subtree__write_dot_string(f, ts_language_symbol_name(language, symbol));
+  ts_language_write_symbol_as_dot_string(language, f, symbol);
   fprintf(f, "\"");
 
   if (ts_subtree_child_count(*self) == 0) fprintf(f, ", shape=plaintext");