Refactor Variant to work with char. Add function to select variant.

src/mfvi.rs

@@ -14,22 +14,22 @@ struct Translation<'a> {
 
 #[derive(Clone)]
 /// Holds the set of delimiters that should be removed from the MML to select a variant.
-pub struct Variant<'v> {
-    keep_delim: &'v str,
-    remove_delims: HashSet<&'v str>,
+pub struct Variant {
+    keep_delim: char,
+    remove_delims: HashSet<char>,
 }
 
-impl<'v> Variant<'v> {
+impl Variant {
     /// Get the set of character delimiters to remove from MML for this variant.
-    pub fn ignores(&self) -> &HashSet<&str> {
+    pub fn ignores(&self) -> &HashSet<char> {
         &self.remove_delims
     }
 }
 
 /// The list of all variants defined in an MML file.
 pub struct VariantList<'v> {
-    all_delims: HashSet<&'v str>,
-    variants: BTreeMap<&'v str, Variant<'v>>,
+    all_delims: HashSet<char>,
+    variants: BTreeMap<&'v str, Variant>,
 }
 
 impl<'v> VariantList<'v> {
@@ -43,14 +43,16 @@ impl<'v> VariantList<'v> {
     fn finish(mut self) -> Self {
         for (_, v) in self.variants.iter_mut() {
             v.remove_delims = self.all_delims.clone();
-            v.remove_delims.remove(v.keep_delim);
+            v.remove_delims.remove(&v.keep_delim);
         }
 
         if self.variants.is_empty() {
             self.variants.insert(
                 "_default_",
                 Variant {
-                    keep_delim: "",
+                    // Not that we expect to find NUL bytes, but effectively we will remove all
+                    // delimiters with this variant.
+                    keep_delim: '\0',
                     remove_delims: self.all_delims.clone(),
                 },
             );
@@ -115,17 +117,17 @@ pub fn get_variants<'a>(input: &'a str, sfx_mode: SfxMode) -> VariantList<'a> {
                 if let Some((first, second)) = l.trim().rsplit_once(char::is_whitespace) {
                     use SfxMode::*;
                     match sfx_mode {
-                        Off => state.all_delims.insert(first),
-                        On => state.all_delims.insert(second),
+                        Off => state.all_delims.insert(first_char(first)),
+                        On => state.all_delims.insert(first_char(second)),
                     };
                 }
             } else if let Some(l) = line.strip_prefix("#VARIANT") {
                 let (first, second) = {
                     let rest = l.trim();
                     if let Some((f, s)) = rest.rsplit_once(' ') {
-                        (f, s)
+                        (first_char(f), s)
                     } else {
-                        (rest, "_default_")
+                        (first_char(rest), "_default_")
                     }
                 };
 
@@ -153,3 +155,42 @@ pub fn get_variants<'a>(input: &'a str, sfx_mode: SfxMode) -> VariantList<'a> {
         })
         .finish()
 }
+
+fn first_char(s: &str) -> char {
+    s.chars().next().unwrap_or('\0')
+}
+
+pub fn select_variant<'i>(input: &'i str, var: &'i Variant) -> String {
+    let mut output = input.to_string();
+    // Workspace for converting char to &str in the loop.
+    let mut buf = [0; 4];
+    // Set of chars to find in `input`.
+    let charlist: Vec<char> = var
+        .remove_delims
+        .iter()
+        .copied()
+        .chain([var.keep_delim])
+        .collect();
+    let mut needles = input.rmatch_indices(&charlist[..]).peekable();
+
+    while let Some((idx, c)) = needles.next() {
+        match c {
+            // Remove the char if it is the `keep_delim`.
+            d if var.keep_delim.encode_utf8(&mut buf) == d => {
+                // `replace_range` doesn't make a full copy of the string the way `remove` does.
+                output.replace_range(idx..=idx, "");
+            }
+            // Anything else will be in `remove_delim`, so find a matching pair and remove
+            // them and everything between them.
+            d => match needles.peek() {
+                Some(&(idx_1, d2)) if d == d2 => {
+                    needles.next();
+                    output.replace_range(idx_1..=idx, "");
+                }
+                _ => {}
+            },
+        }
+    }
+
+    output
+}