Add Text:lines()

4 files changed, 34 insertions, 0 deletions

diff --git a/builtins/text.c b/builtins/text.c
index 1145fb08..ff1f5ba2 100644
--- a/builtins/text.c
+++ b/builtins/text.c
@@ -1535,6 +1535,29 @@ public Text_t Text$from_bytes(array_t bytes)
     return Text$from_str(bytes.data);
 }
 
+public array_t Text$lines(Text_t text)
+{
+    array_t lines = {};
+    iteration_state_t state = {0, 0};
+    for (int64_t i = 0, line_start = 0; i < text.length; i++) {
+        int32_t grapheme = _next_grapheme(text, &state, i);
+        if (grapheme == '\r' && _next_grapheme(text, &state, i + 1) == '\n') { // CRLF
+            Text_t line = Text$slice(text, Int64_to_Int(line_start+1), Int64_to_Int(i));
+            Array$insert(&lines, &line, I_small(0), sizeof(Text_t));
+            i += 1; // skip one extra for CR
+            line_start = i + 1;
+        } else if (grapheme == '\n') { // newline
+            Text_t line = Text$slice(text, Int64_to_Int(line_start+1), Int64_to_Int(i));
+            Array$insert(&lines, &line, I_small(0), sizeof(Text_t));
+            line_start = i + 1;
+        } else if (i == text.length-1 && line_start != i) { // last line
+            Text_t line = Text$slice(text, Int64_to_Int(line_start+1), Int64_to_Int(i+1));
+            Array$insert(&lines, &line, I_small(0), sizeof(Text_t));
+        }
+    }
+    return lines;
+}
+
 public const TypeInfo $Text = {
     .size=sizeof(Text_t),
     .align=__alignof__(Text_t),
diff --git a/builtins/text.h b/builtins/text.h
index 7cef834d..a4697507 100644
--- a/builtins/text.h
+++ b/builtins/text.h
@@ -71,6 +71,7 @@ array_t Text$codepoint_names(Text_t text);
 Text_t Text$from_codepoints(array_t codepoints);
 Text_t Text$from_codepoint_names(array_t codepoint_names);
 Text_t Text$from_bytes(array_t bytes);
+array_t Text$lines(Text_t text);
 
 extern const TypeInfo $Text;
 
diff --git a/environment.c b/environment.c
index 100cfcc2..2385455a 100644
--- a/environment.c
+++ b/environment.c
@@ -257,6 +257,7 @@ env_t *new_compilation_unit(CORD *libname)
             {"from_codepoints", "Text$from_codepoints", "func(codepoints:[Int32])->Text"},
             {"has", "Text$has", "func(text:Text, pattern:Text)->Bool"},
             {"join", "Text$join", "func(glue:Text, pieces:[Text])->Text"},
+            {"lines", "Text$lines", "func(text:Text)->[Text]"},
             {"lower", "Text$lower", "func(text:Text)->Text"},
             {"quoted", "Text$quoted", "func(text:Text, color=no)->Text"},
             {"replace", "Text$replace", "func(text:Text, pattern:Text, replacement:Text)->Text"},
diff --git a/test/text.tm b/test/text.tm
index 2d386048..dc342659 100644
--- a/test/text.tm
+++ b/test/text.tm
@@ -119,3 +119,12 @@ func main():
 	>> c == Text.from_bytes(c:utf8_bytes())
 	= yes
 
+
+	>> "one$(\n)two$(\n)three":lines()
+	= ["one", "two", "three"]
+	>> "one$(\n)two$(\n)three$(\n)":lines()
+	= ["one", "two", "three"]
+	>> "one$(\n)two$(\n)three$(\n\n)":lines()
+	= ["one", "two", "three", ""]
+	>> "one$(\r\n)two$(\r\n)three$(\r\n)":lines()
+	= ["one", "two", "three"]