Merge pull request #2105 from Omikhleia/refactor-collated-sort

core/utilities/sorting.lua

@@ -1,21 +1,30 @@
 --- Table sorting with language-dependent collation.
 -- @module SU.sorting
+--
 
 local icu = require("justenoughicu")
 
 local collatedSort = {
    -- No ICU for language "und", fallback to 'natural' table.sort
-   und = function (t, _)
-      table.sort(t)
+   und = function (t, _, comparator)
+      if comparator then
+         table.sort(t, function (e1, e2)
+            return comparator(e1, e2, function (s1, s2)
+               return s1 < s2 and -1 or s1 > s2 and 1 or 0
+            end)
+         end)
+      else
+         table.sort(t)
+      end
    end,
 }
 
 setmetatable(collatedSort, {
-   __call = function (self, t, options)
+   __call = function (self, t, options, comparator)
       local lang = SILE.settings:get("document.language")
       if self[lang] and type(self[lang]) == "function" then
          -- Allow overriding ICU for some languages, typically "und"
-         return self[lang](t, options)
+         return self[lang](t, options, comparator)
       end
 
       if self[lang] and type(self[lang]) == "table" then
@@ -25,8 +34,17 @@ setmetatable(collatedSort, {
       -- Be efficient: create the collator once before sorting.
       -- I don't think we need to cache it, still.
       local collator = icu.collation_create(lang, options or {})
-      table.sort(t, function (s1, s2)
+
+      local stringCompareClosure = function (s1, s2)
          return icu.compare(collator, s1, s2)
+      end
+      table.sort(t, function (e1, e2)
+         -- Allow custom comparison function, notably for complex objects
+         -- Pass the stringCompare function so that it can be used.
+         if comparator then
+            return comparator(e1, e2, stringCompareClosure)
+         end
+         return stringCompareClosure(e1, e2) < 0
       end)
       icu.collation_destroy(collator)
    end,

justenough/justenoughicu.c

@@ -483,7 +483,7 @@ int je_icu_compare(lua_State *L) {
     return luaL_error(L, "Internal failure to perform comparison");
   }
 
-  lua_pushboolean(L, result == UCOL_LESS);
+  lua_pushinteger(L, result); // UCOL_EQUAL(0), UCOL_GREATER(1), UCOL_LESS(-1)
   return 1;
   // IMPLEMENTATION NOTE FOR PORTABILITY
   // Good news, ucol_strcollUTF8 was introduced in ICU 50.

spec/utilities_spec.lua

@@ -307,6 +307,44 @@ describe("SILE.utilities", function ()
                "Jean-Paul",
             }, sortme)
          end)
+         it("should sort complex tables with callback comparison function", function ()
+            local sortme = {
+               { name = "Jean", age = 30 },
+               { name = "Charlie", age = 25 },
+               { name = "Bob", age = 30 },
+               { name = "Alice", age = 25 },
+            }
+            SU.collatedSort(sortme, nil, function (a, b, stringCompare)
+               -- Sort by ascending age then ascending name
+               if a.age < b.age then return true end
+               if a.age > b.age then return false end
+               return stringCompare(a.name, b.name) < 0
+            end)
+            assert.is.same({
+               { name = "Alice", age = 25 },
+               { name = "Charlie", age = 25 },
+               { name = "Bob", age = 30 },
+               { name = "Jean", age = 30 },
+            }, sortme)
+            local namesAndYears = {
+               { name = "Alice", year = 2005 },
+               { name = "Charlie", year = 1995 },
+               { name = "Bob", year = 1990 },
+               { name = "Alice", year = 1995 }
+            }
+            SU.collatedSort(namesAndYears, nil, function (a, b, stringCompare)
+               local nameCompare = stringCompare(a.name, b.name)
+               if nameCompare < 0 then return true end
+               if nameCompare > 0 then return false end
+               return a.year < b.year
+            end)
+            assert.is.same({
+               { name = "Alice", year = 1995 },
+               { name = "Alice", year = 2005 },
+               { name = "Bob", year = 1990 },
+               { name = "Charlie", year = 1995 },
+            }, namesAndYears)
+         end)
       end)
    end)
 end)