From 4061c088f2654f2fc1e1eaad22d949658bc7a7b3 Mon Sep 17 00:00:00 2001
From: Roland Grunberg <rgrunber@redhat.com>
Date: Mon, 23 Aug 2021 10:33:23 -0400
Subject: [PATCH] Report grammar identifier to didOpen telemetry events.

- Fixes #1105
- Do not report local grammars
- When exactly 1 grammar reference exists per file, simply send the
  relevant values (not a list of size 1). Otherwise, send the values as
  a list

Signed-off-by: Roland Grunberg <rgrunber@redhat.com>
---
 USAGE_DATA.md                                 |  1 +
 .../telemetry/DocumentTelemetryInfo.java      | 30 ++++++++++++++++++-
 2 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/USAGE_DATA.md b/USAGE_DATA.md
index 2c3a147a38..6038da7859 100644
--- a/USAGE_DATA.md
+++ b/USAGE_DATA.md
@@ -17,6 +17,7 @@ When telemetry events are enabled, the following information is emitted when the
    * When a document is opened :
       * The file extension (eg. `xml`, `xsd`, `dtd`)
       * The associated grammar types (eg. `none`, `doctype`, `xml-model`, `xsi:schemaLocation`, `xsi:noNamespaceSchemaLocation`)
+      * The grammar identifiers for an XML document (eg. `http://maven.apache.org/xsd/maven-4.0.0.xsd`)
       * The resolver used to resolve the grammar identifier (eg. `catalog`, `file association`, `embedded catalog.xsd`, `embedded xml.xsd`, `embedded xslt.xsd`)
  * Note: Does NOT include the `JAVA_HOME` environment variable for privacy reasons
 
diff --git a/org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/telemetry/DocumentTelemetryInfo.java b/org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/telemetry/DocumentTelemetryInfo.java
index 6fd42e4379..b11e7b79f0 100644
--- a/org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/telemetry/DocumentTelemetryInfo.java
+++ b/org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/telemetry/DocumentTelemetryInfo.java
@@ -11,6 +11,8 @@
 *******************************************************************************/
 package org.eclipse.lemminx.telemetry;
 
+import java.net.URI;
+import java.net.URISyntaxException;
 import java.util.ArrayList;
 import java.util.HashMap;
 import java.util.List;
@@ -20,11 +22,13 @@
 import org.eclipse.lemminx.dom.DOMDocument;
 import org.eclipse.lemminx.extensions.contentmodel.model.ContentModelManager;
 import org.eclipse.lemminx.extensions.contentmodel.model.ReferencedGrammarInfo;
+import org.eclipse.lemminx.utils.URIUtils;
 
 public class DocumentTelemetryInfo {
 
 	private static final String DOC_PROP_EXT = "file.extension";
 	private static final String DOC_PROP_RESOLVER = "file.resolver";
+	private static final String DOC_PROP_IDENTIFIER = "file.identifier";
 	private static final String DOC_PROP_GRAMMAR_NONE = "file.grammar.none";
 	private static final String DOC_PROP_GRAMMAR_DOCTYPE = "file.grammar.doctype";
 	private static final String DOC_PROP_GRAMMAR_XMLMODEL = "file.grammar.xmlmodel";
@@ -45,9 +49,11 @@ public static Map<String, Object> getDocumentTelemetryInfo (DOMDocument doc, Con
 			}
 		} else {
 			List<String> resolvers = new ArrayList<String>();
+			List<String> identifiers = new ArrayList<String>();
 			for (ReferencedGrammarInfo info : referencedGrammarInfos) {
 				String kind = info.getBindingKind() == null ? "none" : info.getBindingKind();
 				String resolver = info.getResolvedBy();
+				String identifier = info.getIdentifierURI();
 				if ("xml".equals(fileExtension)) {
 					switch (kind) {
 					case "none":
@@ -67,10 +73,32 @@ public static Map<String, Object> getDocumentTelemetryInfo (DOMDocument doc, Con
 						break;
 					}
 				}
+				if (identifier != null) {
+					try {
+						// non-local identifiers only
+						if (new URI(identifier).getScheme() != null && !URIUtils.isFileResource(identifier)) {
+							int limit = Math.min(identifier.length(), 200); // 200 char limit
+							identifiers.add(identifier.substring(0, limit));
+						}
+					} catch (URISyntaxException e) {
+						// continue
+					}
+				}
 				resolvers.add(resolver == null ? "default" : resolver);
 			}
 
-			props.put(DOC_PROP_RESOLVER, resolvers);
+			if (resolvers.size() == 1) {
+				props.put(DOC_PROP_RESOLVER, resolvers.iterator().next());
+			} else {
+				props.put(DOC_PROP_RESOLVER, resolvers);
+			}
+			if (!identifiers.isEmpty()) {
+				if (identifiers.size() == 1) {
+					props.put(DOC_PROP_IDENTIFIER, identifiers.iterator().next());
+				} else {
+					props.put(DOC_PROP_IDENTIFIER, identifiers);
+				}
+			}
 		}
 		return props;
 	}