From b5576e2f34f4446411d7994abc139b4b40bf04d4 Mon Sep 17 00:00:00 2001
From: Carl Meyer <carl@oddbird.net>
Date: Sun, 6 Dec 2020 09:31:33 -0700
Subject: [PATCH] Re-add per-sign maximal subsets for matching.

---
 beancount_import/matching.py      | 34 ++++++++++------
 beancount_import/matching_test.py | 64 ++++++++++++++++++++++++++++---
 2 files changed, 82 insertions(+), 16 deletions(-)

diff --git a/beancount_import/matching.py b/beancount_import/matching.py
index 5157fd12..c01f451f 100644
--- a/beancount_import/matching.py
+++ b/beancount_import/matching.py
@@ -704,7 +704,7 @@ def get_aggregate_posting_candidates(
     5. Subsets may not sum to zero, or contain any sub-subsets that sum to zero.
 
     6. To limit the computational cost, subsets are limited to at most 4
-       elements, except that all maximal subsets are also returned.
+       elements, except that all same-sign maximal subsets are also returned.
 
     The returned subsets are not, in general, disjoint.
 
@@ -730,15 +730,26 @@ def get_aggregate_posting_candidates(
     def posting_set_id(postings):
         return tuple(id(x) for x in postings)
 
-    def add_subset(account, currency, subset):
+    def partition(predicate, postings):
+        t = []
+        f = []
+        for p in postings:
+            if predicate(p):
+                t.append(p)
+            else:
+                f.append(p)
+        return t, f
+
+    def add_subset(account, currency, subset, check_zero=True):
         total = sum(x.units.number for x in subset)
-        if total == ZERO:
-            sum_to_zero.add(posting_set_id(subset))
-            return
-        for subsubset_size in range(2, len(subset)):
-            for subsubset in itertools.combinations(subset, subsubset_size):
-                if posting_set_id(subsubset) in sum_to_zero:
-                    return
+        if check_zero:
+            if total == ZERO:
+                sum_to_zero.add(posting_set_id(subset))
+                return
+            for subsubset_size in range(2, len(subset)):
+                for subsubset in itertools.combinations(subset, subsubset_size):
+                    if posting_set_id(subsubset) in sum_to_zero:
+                        return
         aggregate_posting = Posting(
             account=account,
             units=Amount(currency=currency, number=total),
@@ -751,8 +762,9 @@ def add_subset(account, currency, subset):
     for (account, currency), posting_list in possible_sets.items():
         if len(posting_list) == 1:
             continue
-        if len(posting_list) > max_subset_size:
-            add_subset(account, currency, posting_list)
+        for samesign_list in partition(lambda p: p.units.number > ZERO, posting_list):
+            if len(samesign_list) > max_subset_size:
+                add_subset(account, currency, samesign_list, check_zero=False)
         for subset_size in range(
                 2, min(len(posting_list) + 1, max_subset_size + 1)):
             for subset in itertools.combinations(posting_list, subset_size):
diff --git a/beancount_import/matching_test.py b/beancount_import/matching_test.py
index ee008f1b..57bf370b 100755
--- a/beancount_import/matching_test.py
+++ b/beancount_import/matching_test.py
@@ -786,11 +786,11 @@ def test_match_grouped_differing_signs_sum_zero():
           Expenses:FIXME 1.35 USD
             note1: "B"
           Expenses:FIXME 2.90 USD
-            note1: "D"
+            note1: "C"
           Expenses:FIXME -1.35 USD
-            note1: "F"
+            note1: "D"
           Expenses:FIXME -2.90 USD
-            note1: "G"
+            note1: "E"
         """,
         journal="""
         2020-12-05 * "Narration"
@@ -807,14 +807,68 @@ def test_match_grouped_differing_signs_sum_zero():
           note3: "A"
           Assets:Bank -1.35 USD
             cleared: TRUE
-            note1: "F"
+            note1: "D"
             note2: "A"
           Expenses:Foo 1.35 USD
             note1: "B"
             note3: "B"
           Expenses:FIXME 2.90 USD
-            note1: "D"
+            note1: "C"
           Expenses:FIXME -2.90 USD
+            note1: "E"
+        """,
+    )
+
+def test_match_grouped_maximal_differing_signs():
+    # Maximal matching groups are still per-sign.
+    assert_match(
+        pending_candidate="""
+        2020-12-05 * "Narration"
+          note1: "A"
+          Expenses:FIXME 1 USD
+            note1: "B"
+          Expenses:FIXME 2 USD
+            note1: "C"
+          Expenses:FIXME 3 USD
+            note1: "D"
+          Expenses:FIXME 4 USD
+            note1: "E"
+          Expenses:FIXME 5 USD
+            note1: "F"
+          Expenses:FIXME -15 USD
             note1: "G"
         """,
+        journal="""
+        2020-12-05 * "Narration"
+          note2: "A"
+          Assets:Bank -15 USD
+            cleared: TRUE
+            note2: "B"
+          Expenses:Foo 15 USD
+            note2: "C"
+        """,
+        matches="""
+        2020-12-05 * "Narration"
+          note1: "A"
+          note2: "A"
+          Assets:Bank -15 USD
+            cleared: TRUE
+            note1: "G"
+            note2: "B"
+          Expenses:Foo 1 USD
+            note1: "B"
+            note2: "C"
+          Expenses:Foo 2 USD
+            note1: "C"
+            note2: "C"
+          Expenses:Foo 3 USD
+            note1: "D"
+            note2: "C"
+          Expenses:Foo 4 USD
+            note1: "E"
+            note2: "C"
+          Expenses:Foo 5 USD
+            note1: "F"
+            note2: "C"
+        """,
     )