Optionally use sed extended regex syntax (amber-lang#453)

* Optionally use sed extended regex syntax.

* Update src/std/text.ab

Co-authored-by: Phoenix Himself <pkaras.it@gmail.com>

* Add comment stating that sed extended regex may fail on some older Linux variants.

* Change regex used to detect GNU sed.

* Remove duplicate unsafe from stdlib function.

---------

Co-authored-by: Phoenix Himself <pkaras.it@gmail.com>

src/std/text.ab

@@ -11,11 +11,22 @@ pub fun replace(source, pattern, replacement) {
 /// Replaces all occurences of a regex pattern in the content with the provided replacement text.
 ///
 /// Function uses `sed`
-pub fun replace_regex(source: Text, pattern: Text, replacement: Text): Text {
- return unsafe $echo "{source}" | sed -e "s/{pattern}/{replacement}/g"$
+pub fun replace_regex(source: Text, pattern: Text, replacement: Text, extended: Bool = false): Text {
+ unsafe {
+ if extended {
+ // GNU sed versions 4.0 through 4.2 support extended regex syntax,
+ // but only via the "-r" option; use that if the version information
+ // contains "GNU sed".
+ $re='\bCopyright\b.+\bFree Software Foundation\b'; [[ \$(sed --version 2>/dev/null) =~ \$re ]]$
+ let flag = status == 0 then "-r" else "-E"
+ return $echo "{source}" | sed {flag} -e "s/{pattern}/{replacement}/g"$
+ } else {
+ return $echo "{source}" | sed -e "s/{pattern}/{replacement}/g"$
+ }
+ }
 }
 
-/// Splits the input `text` into an array of substrings using the specified `delimiter`. 
+/// Splits the input `text` into an array of substrings using the specified `delimiter`.
 pub fun split(text: Text, delimiter: Text): [Text] {
  let result = [Text]
  unsafe $IFS="{delimiter}" read -rd '' -a {nameof result} < <(printf %s "\${nameof text}")$
@@ -29,7 +40,7 @@ pub fun lines(text: Text): [Text] {
 
 /// Splits a `text` into an array of substrings based on space character.
 pub fun words(text: Text): [Text] {
- return split(text, " ")
+  return split(text, " ")
 }
 
 /// Merges text using the delimeter specified.

src/tests/stdlib/replace_regex_basic.ab

@@ -0,0 +1,8 @@
+import * from "std/text"
+
+// Output
+// abc[123]def
+
+main {
+ echo replace_regex("abc123def", "\([0-9][0-9]*\)", "[\1]")
+}

src/tests/stdlib/replace_regex_ext.ab

@@ -0,0 +1,10 @@
+import * from "std/text"
+
+// Output
+// abc[123]def
+
+main {
+ // This will fail on any system where sed does not support extended
+ // regex syntax, via "-r" on GNU sed and "-E" on all other versions.
+ echo replace_regex("abc123def", "([0-9]+)", "[\1]", true)
+}