From 7babd66ee01e86de4263a4ca28d496e77287e8ea Mon Sep 17 00:00:00 2001
From: Roy Nieterau <roy_nieterau@hotmail.com>
Date: Wed, 18 Jan 2023 00:11:02 +0100
Subject: [PATCH] Add global unique subsets validator

---
 .../publish/help/validate_unique_subsets.xml  | 17 +++++
 .../publish/validate_unique_subsets.py        | 76 +++++++++++++++++++
 2 files changed, 93 insertions(+)
 create mode 100644 openpype/plugins/publish/help/validate_unique_subsets.xml
 create mode 100644 openpype/plugins/publish/validate_unique_subsets.py

diff --git a/openpype/plugins/publish/help/validate_unique_subsets.xml b/openpype/plugins/publish/help/validate_unique_subsets.xml
new file mode 100644
index 0000000000..b18f046f84
--- /dev/null
+++ b/openpype/plugins/publish/help/validate_unique_subsets.xml
@@ -0,0 +1,17 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<root>
+<error id="main">
+<title>Subset not unique</title>
+<description>
+## Clashing subset names found
+
+Multiples instances from your scene are set to publish into the same asset > subset.
+
+    Non unique subset names: '{non_unique}'
+
+### How to repair?
+
+Remove the offending instances or rename to have a unique name.
+</description>
+</error>
+</root>
\ No newline at end of file
diff --git a/openpype/plugins/publish/validate_unique_subsets.py b/openpype/plugins/publish/validate_unique_subsets.py
new file mode 100644
index 0000000000..11fb827770
--- /dev/null
+++ b/openpype/plugins/publish/validate_unique_subsets.py
@@ -0,0 +1,76 @@
+from collections import defaultdict
+import pyblish.api
+from openpype.pipeline.publish import (
+    PublishXmlValidationError,
+)
+
+
+class ValidateSubsetUniqueness(pyblish.api.ContextPlugin):
+    """Validate all subset names are unique.
+
+    This only validates whether the instances currently set to publish from
+    the workfile overlap one another for the asset + subset they are publishing
+    to.
+
+    This does not perform any check against existing publishes in the database
+    since it is allowed to publish into existing subsets resulting in
+    versioning.
+
+    A subset may appear twice to publish from the workfile if one
+    of them is set to publish to another asset than the other.
+
+    """
+
+    label = "Validate Subset Uniqueness"
+    order = pyblish.api.ValidatorOrder
+    families = ["*"]
+
+    def process(self, context):
+
+        # Find instance per (asset,subset)
+        instance_per_asset_subset = defaultdict(list)
+        for instance in context:
+
+            # Ignore disabled instances
+            if not instance.data.get('publish', True):
+                continue
+
+            # Ignore instance without asset data
+            asset = instance.data.get("asset")
+            if asset is None:
+                self.log.warning("Instance found without `asset` data: "
+                                 "{}".format(instance.name))
+                continue
+
+            # Ignore instance without subset data
+            subset = instance.data.get("subset")
+            if subset is None:
+                self.log.warning("Instance found without `subset` data: "
+                                 "{}".format(instance.name))
+                continue
+
+            instance_per_asset_subset[(asset, subset)].append(instance)
+
+        non_unique = []
+        for (asset, subset), instances in instance_per_asset_subset.items():
+
+            # A single instance per asset, subset is fine
+            if len(instances) < 2:
+                continue
+
+            non_unique.append("{asset} > {subset}".format(asset=asset,
+                                                          subset=subset))
+
+        if not non_unique:
+            # All is ok
+            return
+
+        msg = ("Instance subset names {} are not unique. ".format(non_unique) +
+               "Please remove or rename duplicates.")
+        formatting_data = {
+            "non_unique": ",".join(non_unique)
+        }
+
+        if non_unique:
+            raise PublishXmlValidationError(self, msg,
+                                            formatting_data=formatting_data)