From 3cd000d9b30e3bb6507bf7f7e2b4e9b58d487d93 Mon Sep 17 00:00:00 2001
From: Rudi Grinberg <me@rgrinberg.com>
Date: Sun, 27 Oct 2024 00:01:23 +0100
Subject: [PATCH] test: more automata tests

---
 lib_test/expect/test_automata.ml | 94 ++++++++++++++++++++++++++++----
 1 file changed, 84 insertions(+), 10 deletions(-)

diff --git a/lib_test/expect/test_automata.ml b/lib_test/expect/test_automata.ml
index 44948c9..18cd6aa 100644
--- a/lib_test/expect/test_automata.ml
+++ b/lib_test/expect/test_automata.ml
@@ -13,6 +13,7 @@ include struct
   let eps = eps
   let cst = cst
   let seq = seq
+  let rep = rep
 end
 
 let pp_state state = print_dyn (State.to_dyn state)
@@ -30,14 +31,20 @@ let str ids sem str =
   loop (String.to_seq str)
 ;;
 
-let rec loop wa d c =
-  print_dyn (State.to_dyn d);
-  match State.status d with
-  | Failed -> Format.printf "> failed@."
-  | Match _ -> Format.printf "> matched@."
-  | Running ->
-    let d = Automata.delta wa cat (Cset.of_char c) d in
-    loop wa d c
+let loop ?(max = 100) wa d c =
+  let cset = Cset.of_char c in
+  let rec loop d n =
+    if n > 0
+    then (
+      print_dyn (State.to_dyn d);
+      match State.status d with
+      | Failed -> Format.printf "> failed@."
+      | Match _ -> Format.printf "> matched@."
+      | Running ->
+        let d = Automata.delta wa cat cset d in
+        loop d (n - 1))
+  in
+  loop d max
 ;;
 
 let%expect_test "string" =
@@ -71,8 +78,8 @@ let%expect_test "string" =
 ;;
 
 let%expect_test "alternation" =
-  let ids = Ids.create () in
   let re =
+    let ids = Ids.create () in
     let n = 4 in
     let s =
       let c = 'a' in
@@ -112,8 +119,8 @@ let%expect_test "alternation" =
 
 let%expect_test "alternation shared prefix" =
   let n = 4 in
-  let ids = Ids.create () in
   let re =
+    let ids = Ids.create () in
     let prefix =
       let s =
         let c = 'a' in
@@ -142,3 +149,70 @@ let%expect_test "alternation shared prefix" =
     > failed
     |}]
 ;;
+
+let%expect_test "kleene star" =
+  let re =
+    let ids = Ids.create () in
+    rep ids `Greedy `First (cst ids (Cset.csingle 'z'))
+  in
+  let wa = Working_area.create () in
+  loop ~max:4 wa (State.create cat re) 'z';
+  [%expect
+    {|
+    ((TExp (first (Rep 122))))
+    ((TExp (first (Rep 122))) (TMarks ()))
+    ((TExp (first (Rep 122))) (TMarks ()))
+    ((TExp (first (Rep 122))) (TMarks ()))
+    |}];
+  loop ~max:3 wa (State.create cat re) 'a';
+  [%expect {|
+    ((TExp (first (Rep 122))))
+    ((TMarks ()))
+    > matched
+    |}]
+;;
+
+let%expect_test "derivative recomputation" =
+  let sem = `Longest in
+  let re =
+    let ids = Ids.create () in
+    let lhs = rep ids `Non_greedy sem (cst ids Cset.cany) in
+    let rhs =
+      seq
+        ids
+        sem
+        (Automata.mark ids Automata.Mark.start)
+        (Automata.alt ids [ cst ids (Cset.csingle 'z'); cst ids (Cset.csingle 'b') ])
+    in
+    seq ids sem lhs rhs
+  in
+  let wa = Working_area.create () in
+  loop ~max:7 wa (State.create cat re) 'z';
+  [%expect
+    {|
+    ((TExp (long (Seq (Rep ((0 255))) (Seq (Mark 0) (Alt 122 98))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 0)))) Eps))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 1)))) Eps) (TMarks ((marks ((0 0))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 0)))) Eps) (TMarks ((marks ((0 1))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 1)))) Eps) (TMarks ((marks ((0 0))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 0)))) Eps) (TMarks ((marks ((0 1))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98))))
+     (TExp ((marks ((0 1)))) Eps) (TMarks ((marks ((0 0))))))
+    |}];
+  loop ~max:7 wa (State.create cat re) 'a';
+  [%expect
+    {|
+    ((TExp (long (Seq (Rep ((0 255))) (Seq (Mark 0) (Alt 122 98))))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    ((long (TSeq ((TExp (Rep ((0 255))))) (Seq (Mark 0) (Alt 122 98)))))
+    |}]
+;;