From 6d7e09bf1801c2fe183df17cc67017a6d3d8513b Mon Sep 17 00:00:00 2001
From: Bruce Hill <bruce@bruce-hill.com>
Date: Mon, 2 Sep 2024 23:07:08 -0400
Subject: Add Text:split() and use that with an empty pattern instead of
 Text:clusters()

---
 builtins/text.c | 23 +++++++++++++++++++++++
 builtins/text.h |  1 +
 2 files changed, 24 insertions(+)

(limited to 'builtins')

diff --git a/builtins/text.c b/builtins/text.c
index ff1f5ba2..a1a98a00 100644
--- a/builtins/text.c
+++ b/builtins/text.c
@@ -1426,6 +1426,29 @@ public Text_t Text$replace(Text_t text, Text_t pattern, Text_t replacement)
     return ret;
 }
 
+public array_t Text$split(Text_t text, Text_t pattern)
+{
+    if (pattern.length == 0) // special case
+        return Text$clusters(text);
+
+    array_t chunks = {};
+
+    Int_t i = I_small(1);
+    for (;;) {
+        int64_t len;
+        Int_t found = Text$find(text, pattern, i, &len);
+        if (I_is_zero(found)) break;
+        Text_t chunk = Text$slice(text, i, Int$minus(found, I_small(1)));
+        Array$insert(&chunks, &chunk, I_small(0), sizeof(Text_t));
+        i = Int$plus(found, Int64_to_Int(len));
+    }
+
+    Text_t last_chunk = Text$slice(text, i, Int64_to_Int(text.length));
+    Array$insert(&chunks, &last_chunk, I_small(0), sizeof(Text_t));
+
+    return chunks;
+}
+
 public Text_t Text$format(const char *fmt, ...)
 {
     va_list args;
diff --git a/builtins/text.h b/builtins/text.h
index a4697507..cca18a2c 100644
--- a/builtins/text.h
+++ b/builtins/text.h
@@ -60,6 +60,7 @@ Text_t Text$title(Text_t text);
 Text_t Text$as_text(const void *text, bool colorize, const TypeInfo *info);
 Text_t Text$quoted(Text_t str, bool colorize);
 Text_t Text$replace(Text_t str, Text_t pat, Text_t replacement);
+array_t Text$split(Text_t text, Text_t pattern);
 Int_t Text$find(Text_t text, Text_t pattern, Int_t i, int64_t *match_length);
 bool Text$has(Text_t text, Text_t pattern);
 const char *Text$as_c_string(Text_t text);
-- 
cgit v1.2.3