Fix a bug in uambiguous prefixes.

Specifically, given the strings ABCX, CDAX and BCX, it was reporting
the unambiguous set as A, BCX and CDAX, which is wrong since A is a
substring of CDAX.

unambiguous_prefixes is now quite a bit of a mess, but so is the rest
of the literal extraction code. The only thing it has going for it is
a massive test suite.

Fixes rust-lang#289

regex-syntax/src/literals.rs

@@ -216,14 +216,17 @@ impl Literals {
         if self.lits.is_empty() {
             return self.to_empty();
         }
+        let mut old: Vec<Lit> = self.lits.iter().cloned().collect();
         let mut new = self.to_empty();
     'OUTER:
-        for lit1 in &self.lits {
+        while let Some(mut candidate) = old.pop() {
+            if candidate.is_empty() {
+                continue;
+            }
             if new.lits.is_empty() {
-                new.lits.push(lit1.clone());
+                new.lits.push(candidate);
                 continue;
             }
-            let mut candidate = lit1.clone();
             for lit2 in &mut new.lits {
                 if lit2.is_empty() {
                     continue;
@@ -236,11 +239,14 @@ impl Literals {
                     lit2.cut = candidate.cut;
                     continue 'OUTER;
                 }
-                if candidate.len() <= lit2.len() {
+                if candidate.len() < lit2.len() {
                     if let Some(i) = position(&candidate, &lit2) {
-                        lit2.truncate(i);
-                        lit2.cut();
                         candidate.cut();
+                        let mut lit3 = lit2.clone();
+                        lit3.truncate(i);
+                        lit3.cut();
+                        old.push(lit3);
+                        lit2.clear();
                     }
                 } else {
                     if let Some(i) = position(&lit2, &candidate) {
@@ -1381,6 +1387,9 @@ mod tests {
     test_unamb!(unambiguous11,
                 vec![M("zazb"), M("azb")], vec![C("azb"), C("z")]);
     test_unamb!(unambiguous12, vec![M("foo"), C("foo")], vec![C("foo")]);
+    test_unamb!(unambiguous13,
+                vec![M("ABCX"), M("CDAX"), M("BCX")],
+                vec![C("A"), C("BCX"), C("CD")]);
 
     // ************************************************************************
     // Tests for suffix trimming.

tests/regression.rs

@@ -68,3 +68,6 @@ mat!(ascii_boundary_capture, u!(r"(?-u)(\B)"), "\u{28f3e}", Some((0, 0)));
 // See: https://github.com/rust-lang-nursery/regex/issues/280
 ismatch!(partial_anchor_alternate_begin, u!(r"^a|z"), "yyyyya", false);
 ismatch!(partial_anchor_alternate_end, u!(r"a$|z"), "ayyyyy", false);
+
+// See: https://github.com/rust-lang-nursery/regex/issues/289
+mat!(lits_unambiguous, u!(r"(ABC|CDA|BC)X"), "CDAX", Some((0, 4)));