From 5916411e9e07985ab211f1e773d059451f5a34f1 Mon Sep 17 00:00:00 2001
From: You Guang <yokurang@gmail.com>
Date: Wed, 14 Aug 2024 16:32:30 +0800
Subject: [PATCH] tried something

---
 lib/WordDiff.ml | 138 ++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 134 insertions(+), 4 deletions(-)

diff --git a/lib/WordDiff.ml b/lib/WordDiff.ml
index 600a162..25e93cf 100644
--- a/lib/WordDiff.ml
+++ b/lib/WordDiff.ml
@@ -81,7 +81,137 @@ let compute (block : string Block.t) : line_content Block.t =
   match block with
   | Block.Common line -> Block.Common [ Unchanged line ]
   | Block.Changed { mine; their; order } ->
-      let mine_str = String.concat " " mine in
-      let their_str = String.concat " " their in
-      let mine_words, their_words = diff_words mine_str their_str in
-      Block.Changed { mine = [ mine_words ]; their = [ their_words ]; order }
+      (* Helper function to convert a string to an array of words *)
+      let string_to_word_array s =
+        s |> String.split_on_char ' ' |> Array.of_list
+      in
+
+      (* Use Levenshtein distance to find the best pairing of lines *)
+      let pair_lines lines1 lines2 =
+        let distances =
+          Array.make_matrix (Array.length lines1) (Array.length lines2) 0
+        in
+        for i = 0 to Array.length lines1 - 1 do
+          for j = 0 to Array.length lines2 - 1 do
+            distances.(i).(j) <-
+              edit_distance ( = )
+                (string_to_word_array lines1.(i))
+                (string_to_word_array lines2.(j))
+          done
+        done;
+
+        (* Use a greedy approach to pair lines based on minimum distance *)
+        let paired = ref [] in
+        let used1 = Array.make (Array.length lines1) false in
+        let used2 = Array.make (Array.length lines2) false in
+
+        for _ = 1 to min (Array.length lines1) (Array.length lines2) do
+          let min_dist = ref max_int in
+          let min_i = ref (-1) in
+          let min_j = ref (-1) in
+
+          for i = 0 to Array.length lines1 - 1 do
+            for j = 0 to Array.length lines2 - 1 do
+              if
+                (not used1.(i))
+                && (not used2.(j))
+                && distances.(i).(j) < !min_dist
+              then (
+                min_dist := distances.(i).(j);
+                min_i := i;
+                min_j := j)
+            done
+          done;
+
+          if !min_i <> -1 && !min_j <> -1 then (
+            paired := (!min_i, !min_j) :: !paired;
+            used1.(!min_i) <- true;
+            used2.(!min_j) <- true)
+        done;
+
+        (* Add unpaired lines *)
+        let final_pairs = ref !paired in
+        for i = 0 to Array.length lines1 - 1 do
+          if not used1.(i) then final_pairs := (i, -1) :: !final_pairs
+        done;
+        for j = 0 to Array.length lines2 - 1 do
+          if not used2.(j) then final_pairs := (-1, j) :: !final_pairs
+        done;
+
+        List.sort compare !final_pairs
+      in
+
+      let mine_array = Array.of_list mine in
+      let their_array = Array.of_list their in
+      let pairs = pair_lines mine_array their_array in
+
+      let result_mine = ref [] in
+      let result_their = ref [] in
+
+      (* Helper function to trim leading/trailing whitespace and collapse multiple spaces *)
+      let normalize_whitespace s =
+        let s = String.trim s in
+        let buf = Buffer.create (String.length s) in
+        let space_seen = ref false in
+        String.iter
+          (fun c ->
+            match c with
+            | ' ' | '\t' | '\n' | '\r' ->
+                if not !space_seen then (
+                  Buffer.add_char buf ' ';
+                  space_seen := true)
+            | _ ->
+                Buffer.add_char buf c;
+                space_seen := false)
+          s;
+        Buffer.contents buf
+      in
+
+      List.iter
+        (fun (i, j) ->
+          match (i, j) with
+          | -1, j ->
+              let their_content = normalize_whitespace their_array.(j) in
+              if their_content <> "" then
+                result_their :=
+                  (diff_words "" their_content |> snd) :: !result_their
+          | i, -1 ->
+              let mine_content = normalize_whitespace mine_array.(i) in
+              if mine_content <> "" then
+                result_mine :=
+                  (diff_words mine_content "" |> fst) :: !result_mine
+          | i, j ->
+              let mine_content = normalize_whitespace mine_array.(i) in
+              let their_content = normalize_whitespace their_array.(j) in
+              if mine_content <> "" || their_content <> "" then (
+                let mine_diff, their_diff =
+                  diff_words mine_content their_content
+                in
+                result_mine := mine_diff :: !result_mine;
+                result_their := their_diff :: !result_their))
+        pairs;
+
+      (* Remove any empty lines at the beginning and end of the results *)
+      let trim_empty_lines lines =
+        let is_empty line =
+          List.for_all (function Unchanged "" -> true | _ -> false) line
+        in
+        let rec trim_start = function
+          | [] -> []
+          | hd :: tl when is_empty hd -> trim_start tl
+          | lines -> lines
+        in
+        let rec trim_end = function
+          | [] -> []
+          | hd :: tl ->
+              let trimmed_tail = trim_end tl in
+              if trimmed_tail = [] && is_empty hd then []
+              else hd :: trimmed_tail
+        in
+        lines |> trim_start |> trim_end
+      in
+
+      let trimmed_mine = trim_empty_lines (List.rev !result_mine) in
+      let trimmed_their = trim_empty_lines (List.rev !result_their) in
+
+      Block.Changed { mine = trimmed_mine; their = trimmed_their; order }