From 6ded6738ff5004fce014e08748a39fd1ad4908ad Mon Sep 17 00:00:00 2001
From: Justin Lin <justin.sdk@gmail.com>
Date: Fri, 5 Mar 2021 20:07:14 +0800
Subject: [PATCH] performance improved

---
 src/util/_impl/_dedup_impl.scad | 43 +++++++++++++++------------------
 src/util/dedup.scad             | 19 +++++++++++++--
 2 files changed, 36 insertions(+), 26 deletions(-)

diff --git a/src/util/_impl/_dedup_impl.scad b/src/util/_impl/_dedup_impl.scad
index 35c35d89..595b56df 100644
--- a/src/util/_impl/_dedup_impl.scad
+++ b/src/util/_impl/_dedup_impl.scad
@@ -1,27 +1,22 @@
-use <../some.scad> ;
 
-function _dedup_sorted(lt, leng, eq) =
-    leng == 0 ? lt : 
-        is_function(eq) ?      
-            concat(
-                [lt[0]],
-                [for(i = [1:leng - 1]) if(!eq(lt[i], lt[i - 1])) lt[i]]
-            ) :
-            concat(
-                [lt[0]],
-                [for(i = [1:leng - 1]) if(lt[i] != lt[i - 1]) lt[i]]
-            );  
+use <../slice.scad>;
+use <../some.scad>;
 
-function _dedup_vt(src, dest, leng, i = 0) = 
-    i == leng ? dest :
-    // src[i] in dest ?
-    search([src[i]], dest) != [[]] ? _dedup_vt(src, dest, leng, i + 1) : 
-                                     _dedup_vt(src, concat(dest, [src[i]]), leng, i + 1);
+function _dedup(elems, leng, buckets, eq, hash, bucket_numbers, i = 0) = 
+    i == leng ? buckets :
+	_dedup(elems, leng, _dedup_add(buckets, [i, elems[i]], eq, hash, bucket_numbers), eq, hash, bucket_numbers, i + 1);
 
-function _dedup_eq(src, dest, eq, leng, i = 0) = 
-    i == leng ? dest :
-    some(dest, function(el) eq(el, src[i])) ? _dedup_eq(src, dest, eq, leng, i + 1) : 
-                                                 _dedup_eq(src, concat(dest, [src[i]]), eq, leng, i + 1);
-
-function _dedup(src, dest, leng, eq) = 
-    is_function(eq) ? _dedup_eq(src, dest, eq, leng) : _dedup_vt(src, dest, leng);
\ No newline at end of file
+function _dedup_add(buckets, i_elem, eq, hash, bucket_numbers) =
+    let(
+		i =  i_elem[0],
+		elem = i_elem[1],
+	    b_idx = hash(elem) % bucket_numbers,
+		bucket = buckets[b_idx]
+	)
+	len(bucket) == 0 ? concat(slice(buckets, 0, b_idx), [[i_elem]], slice(buckets, b_idx + 1)) :
+	some(bucket, function(i_e) eq(i_e[1], elem)) ? buckets :
+	concat(
+	    slice(buckets, 0, b_idx), 
+		[concat(bucket, [i_elem])], 
+		slice(buckets, b_idx + 1)
+	);
\ No newline at end of file
diff --git a/src/util/dedup.scad b/src/util/dedup.scad
index bfb9a50e..b87a7252 100644
--- a/src/util/dedup.scad
+++ b/src/util/dedup.scad
@@ -8,7 +8,22 @@
 *
 **/ 
 
+use <../__comm__/_str_hash.scad>;
 use <_impl/_dedup_impl.scad>;
+use <sort.scad>;
 
-function dedup(lt, sorted = false, eq) = 
-    sorted ? _dedup_sorted(lt, len(lt), eq) : _dedup(lt, [], len(lt), eq);
\ No newline at end of file
+function dedup(lt, eq = function(e1, e2) e1 == e2, hash = function(e) _str_hash(e)) =
+    lt == [] ? [] :
+    let(
+		leng_lt = len(lt),
+		bucket_numbers = ceil(sqrt(leng_lt)),
+	    buckets = [for(i = [0:bucket_numbers - 1]) []],
+		deduped = _dedup(lt, leng_lt, buckets, eq, hash, bucket_numbers),
+		i_elem_lt = [
+			for(bucket = deduped) 
+				for(i_elem = bucket)
+					i_elem		
+		],
+		sorted = sort(i_elem_lt, by = function(e1, e2) e1[0] - e2[0])
+	)
+	[for(i_elem = sorted) i_elem[1]];