From 36c8a7effc83b9d9e8cb889eda076f5caade3ec6 Mon Sep 17 00:00:00 2001
From: Louis Gesbert <louis.gesbert@ocamlpro.com>
Date: Mon, 18 Mar 2024 17:38:10 +0100
Subject: [PATCH] Sort strings in natural order when they contain numbers

Seeing results sorted as 1 10 11 2 doesn't look nice.
---
 compiler/catala_utils/string.ml  | 32 ++++++++++++++++++++++++++++++++
 compiler/catala_utils/string.mli |  3 +++
 2 files changed, 35 insertions(+)

diff --git a/compiler/catala_utils/string.ml b/compiler/catala_utils/string.ml
index 57311837..44dc5e6f 100644
--- a/compiler/catala_utils/string.ml
+++ b/compiler/catala_utils/string.ml
@@ -68,7 +68,39 @@ module Arg = struct
   include Stdlib.String
 
   let format = format
+
+  let compare s1 s2 =
+    let len1 = length s1 in
+    let len2 = length s2 in
+    let int c = int_of_char c - int_of_char '0' in
+    let rec readnum acc s i =
+      if i >= length s then acc, i
+      else
+        match get s i with
+        | '0' .. '9' as c -> readnum ((acc * 10) + int c) s (i + 1)
+        | _ -> acc, i
+    in
+    let rec aux i1 i2 =
+      if i1 >= len1 then if i2 >= len2 then 0 else -1
+      else if i2 >= len2 then 1
+      else
+        match get s1 i1, get s2 i2 with
+        | ('0' .. '9' as c1), ('0' .. '9' as c2) -> (
+          let x1, i1' = readnum (int c1) s1 (i1 + 1) in
+          let x2, i2' = readnum (int c2) s2 (i2 + 1) in
+          match Int.compare x1 x2 with
+          | 0 -> (
+            match Int.compare (i1' - i1) (i2' - i2) with
+            | 0 -> aux i1' i2'
+            | n -> n)
+          | n -> n)
+        | c1, c2 -> (
+          match Char.compare c1 c2 with 0 -> aux (i1 + 1) (i2 + 1) | n -> n)
+    in
+    aux 0 0
 end
 
+let compare = Arg.compare
+
 module Set = Set.Make (Arg)
 module Map = Map.Make (Arg)
diff --git a/compiler/catala_utils/string.mli b/compiler/catala_utils/string.mli
index 7c9abf74..ee7248be 100644
--- a/compiler/catala_utils/string.mli
+++ b/compiler/catala_utils/string.mli
@@ -20,6 +20,9 @@ module Map : Map.S with type key = string
 
 (** Helper functions used for string manipulation. *)
 
+val compare : string -> string -> int
+(** String comparison with natural ordering of numbers within strings *)
+
 val to_ascii : string -> string
 (** Removes all non-ASCII diacritics from a string by converting them to their
     base letter in the Latin alphabet. *)