scanner: speed up token position -> location using a cache

Signed-off-by: Ran Benita <[email protected]>

src/scanner-utils.c

@@ -31,16 +31,32 @@ scanner_token_location(struct scanner *s)
      * https://lemire.me/blog/2017/02/14/how-fast-can-you-count-lines/
      * https://github.com/lemire/Code-used-on-Daniel-Lemire-s-blog/blob/master/2017/02/14/newlines.c
      * This is the fastest portable one for me, and it's simple.
+     *
+     * To avoid rescanning on each call, cache the previously found position
+     * and start from that on the next call. This is effective as long as the
+     * tokens go forward.
      */
-    size_t line = 1, column;
+    size_t line, column;
     size_t line_pos = 0;
-    const char *ptr = s->s;
+    if (s->cached_pos > s->token_pos) {
+        s->cached_pos = 0;
+        s->cached_loc.line = s->cached_loc.column = 1;
+    }
+    line = s->cached_loc.line;
+    const char *ptr = s->s + s->cached_pos;
     const char *last = s->s + s->token_pos;
     while ((ptr = memchr(ptr, '\n', last - ptr))) {
         line++;
         ptr++;
         line_pos = ptr - s->s;
     }
-    column = s->token_pos - line_pos + 1;
-    return (struct scanner_loc){.line = line, .column = column};
+    if (line == s->cached_loc.line) {
+        column = s->cached_loc.column + (s->token_pos - s->cached_pos);
+    } else {
+        column = s->token_pos - line_pos + 1;
+    }
+    struct scanner_loc loc = {.line = line, .column = column};
+    s->cached_pos = s->token_pos;
+    s->cached_loc = loc;
+    return loc;
 }

src/scanner-utils.h

@@ -67,6 +67,9 @@ struct scanner {
     size_t buf_pos;
     /* The position of the start of the current token. */
     size_t token_pos;
+    /* Cached values used by scanner_token_location. */
+    size_t cached_pos;
+    struct scanner_loc cached_loc;
     const char *file_name;
     struct xkb_context *ctx;
     void *priv;
@@ -106,6 +109,8 @@ scanner_init(struct scanner *s, struct xkb_context *ctx,
     s->len = len;
     s->pos = 0;
     s->token_pos = 0;
+    s->cached_pos = 0;
+    s->cached_loc.line = s->cached_loc.column = 1;
     s->file_name = file_name;
     s->ctx = ctx;
     s->priv = priv;