From dc1ae1664e835073ad7f046d935053f5cd9f4fe6 Mon Sep 17 00:00:00 2001
From: Bertrand Lorentz <bertrand.lorentz@publications.europa.eu>
Date: Mon, 12 Aug 2024 10:35:09 +0200
Subject: [PATCH] schematron: Move calculation of missing diagnostics to Java
 code

Doing this in Drools causes very high memory usage (over 5GB) when run
on the SDK, probably because of the high number of asserts.
---
 .../sdk/analysis/fact/SchematronFileFact.java   | 17 +++++++++++++++++
 .../sdk/analysis/drools/schematronRules.drl     |  5 ++---
 2 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/src/main/java/eu/europa/ted/eforms/sdk/analysis/fact/SchematronFileFact.java b/src/main/java/eu/europa/ted/eforms/sdk/analysis/fact/SchematronFileFact.java
index 76be4ad..97140a7 100644
--- a/src/main/java/eu/europa/ted/eforms/sdk/analysis/fact/SchematronFileFact.java
+++ b/src/main/java/eu/europa/ted/eforms/sdk/analysis/fact/SchematronFileFact.java
@@ -46,6 +46,23 @@ public List<String> getDuplicateAssertIds() {
         .collect(Collectors.toList());
   }
 
+  /**
+   * Returns the list of diagnostic identifiers that are referenced in an assert but not defined by
+   * a diagnostic element.
+   * 
+   * @return List of diagnostic identifiers missing a definition.
+   */
+  public List<String> getMissingDiagnostics() {
+    Set<String> definedDiagnosticIds = getDiagnostics().stream()
+        .map(SchematronDiagnostic::getId)
+        .collect(Collectors.toSet());
+
+    return getAsserts().stream()
+        .map(SchematronAssert::getDiagnostics)
+        .filter(id -> !definedDiagnosticIds.contains(id))
+        .collect(Collectors.toList());
+  }
+
   @Override
   public String getId() {
       return schematronFile.getPath().toString();
diff --git a/src/main/resources/eu/europa/ted/eforms/sdk/analysis/drools/schematronRules.drl b/src/main/resources/eu/europa/ted/eforms/sdk/analysis/drools/schematronRules.drl
index 900e180..47d4c54 100644
--- a/src/main/resources/eu/europa/ted/eforms/sdk/analysis/drools/schematronRules.drl
+++ b/src/main/resources/eu/europa/ted/eforms/sdk/analysis/drools/schematronRules.drl
@@ -35,8 +35,7 @@ end
 
 rule "Every assert diagnostics is defined in the schematron file"
 when
-  /schematrons[ $sch: this, $schId: id ]/asserts[ $diagnosticId: diagnostics, diagnostics != null ]
-  not (exists /schematrons[ id == $schId ]/diagnostics[ id == $diagnosticId ])
+  $missing : /schematrons[ $sch: this]/missingDiagnostics
 then
-  results.add(new ValidationResult($sch, "Schematron is missing diagnostic definitions for: " + $diagnosticId, ValidationStatusEnum.ERROR));
+  results.add(new ValidationResult($sch, "Schematron is missing diagnostic definitions for: " + $missing, ValidationStatusEnum.ERROR));
 end