Java: add diagnostic query indicating low database quality

Author: smowton

java/ql/src/Telemetry/DatabaseQuality.qll

@@ -0,0 +1,65 @@
+/**
+ * Provides database quality statistics that are reported by java/telemetry/extractor-information
+ * and perhaps warned about by java/diagnostics/database-quality.
+ */
+
+import java
+
+signature module StatsSig {
+  int getNumberOfOk();
+
+  int getNumberOfNotOk();
+
+  string getOkText();
+
+  string getNotOkText();
+}
+
+module ReportStats<StatsSig Stats> {
+  predicate numberOfOk(string key, int value) {
+    value = Stats::getNumberOfOk() and
+    key = "Number of " + Stats::getOkText()
+  }
+
+  predicate numberOfNotOk(string key, int value) {
+    value = Stats::getNumberOfNotOk() and
+    key = "Number of " + Stats::getNotOkText()
+  }
+
+  predicate percentageOfOk(string key, float value) {
+    value = Stats::getNumberOfOk() * 100.0 / (Stats::getNumberOfOk() + Stats::getNumberOfNotOk()) and
+    key = "Percentage of " + Stats::getOkText()
+  }
+}
+
+module CallTargetStats implements StatsSig {
+  int getNumberOfOk() { result = count(Call c | exists(c.getCallee())) }
+
+  int getNumberOfNotOk() { result = count(Call c | not exists(c.getCallee())) }
+
+  string getOkText() { result = "calls with call target" }
+
+  string getNotOkText() { result = "calls with missing call target" }
+}
+
+private class SourceExpr extends Expr {
+  SourceExpr() { this.getFile().isSourceFile() }
+}
+
+private predicate hasGoodType(Expr e) {
+  exists(e.getType()) and not e.getType() instanceof ErrorType
+}
+
+module ExprTypeStats implements StatsSig {
+  int getNumberOfOk() { result = count(SourceExpr e | hasGoodType(e)) }
+
+  int getNumberOfNotOk() { result = count(SourceExpr e | not hasGoodType(e)) }
+
+  string getOkText() { result = "expressions with known type" }
+
+  string getNotOkText() { result = "expressions with unknown type" }
+}
+
+module CallTargetStatsReport = ReportStats<CallTargetStats>;
+
+module ExprTypeStatsReport = ReportStats<ExprTypeStats>;

java/ql/src/Telemetry/DatabaseQualityDiagnostics.ql

@@ -0,0 +1,40 @@
+/**
+ * @name Low Java analysis quality
+ * @description Low Java analysis quality
+ * @kind diagnostic
+ * @id java/diagnostic/database-quality
+ */
+
+import java
+import DatabaseQuality
+
+private newtype TDbQualityDiagnostic =
+  TTheDbQualityDiagnostic() {
+    exists(float percentageGood |
+      CallTargetStatsReport::percentageOfOk(_, percentageGood)
+      or
+      ExprTypeStatsReport::percentageOfOk(_, percentageGood)
+    |
+      percentageGood < 95
+    )
+  }
+
+class DbQualityDiagnostic extends TDbQualityDiagnostic {
+  string toString() {
+    result =
+      "There were significant issues scanning Java code. " +
+        "This may be caused by problems identifying dependencies or use of generated source code, among other reasons -- "
+        + "see other CodeQL diagnostics reported here for more details of possible causes. " +
+        "This may lead to false-positive or missing results. Consider analysing Java using the `autobuild` or `manual` build modes."
+  }
+}
+
+query predicate diagnosticAttributes(DbQualityDiagnostic e, string key, string value) {
+  e = e and // Quieten warning about unconstrained 'e'
+  key = ["visibilityCliSummaryTable", "visibilityTelemetry", "visibilityStatusPage"] and
+  value = "true"
+}
+
+from DbQualityDiagnostic d
+select d, d.toString(), 1
+/* Warning severity */

java/ql/src/Telemetry/ExtractorInformation.ql

@@ -8,6 +8,7 @@
 
 import java
 import semmle.code.java.Diagnostics
+import DatabaseQuality
 
 extensible predicate extractorInformationSkipKey(string key);
 
@@ -85,65 +86,6 @@ predicate extractorTotalDiagnostics(string key, int value) {
   )
 }
 
-signature module StatsSig {
-  int getNumberOfOk();
-
-  int getNumberOfNotOk();
-
-  string getOkText();
-
-  string getNotOkText();
-}
-
-module ReportStats<StatsSig Stats> {
-  predicate numberOfOk(string key, int value) {
-    value = Stats::getNumberOfOk() and
-    key = "Number of " + Stats::getOkText()
-  }
-
-  predicate numberOfNotOk(string key, int value) {
-    value = Stats::getNumberOfNotOk() and
-    key = "Number of " + Stats::getNotOkText()
-  }
-
-  predicate percentageOfOk(string key, float value) {
-    value = Stats::getNumberOfOk() * 100.0 / (Stats::getNumberOfOk() + Stats::getNumberOfNotOk()) and
-    key = "Percentage of " + Stats::getOkText()
-  }
-}
-
-module CallTargetStats implements StatsSig {
-  int getNumberOfOk() { result = count(Call c | exists(c.getCallee())) }
-
-  int getNumberOfNotOk() { result = count(Call c | not exists(c.getCallee())) }
-
-  string getOkText() { result = "calls with call target" }
-
-  string getNotOkText() { result = "calls with missing call target" }
-}
-
-private class SourceExpr extends Expr {
-  SourceExpr() { this.getFile().isSourceFile() }
-}
-
-private predicate hasGoodType(Expr e) {
-  exists(e.getType()) and not e.getType() instanceof ErrorType
-}
-
-module ExprTypeStats implements StatsSig {
-  int getNumberOfOk() { result = count(SourceExpr e | hasGoodType(e)) }
-
-  int getNumberOfNotOk() { result = count(SourceExpr e | not hasGoodType(e)) }
-
-  string getOkText() { result = "expressions with known type" }
-
-  string getNotOkText() { result = "expressions with unknown type" }
-}
-
-module CallTargetStatsReport = ReportStats<CallTargetStats>;
-
-module ExprTypeStatsReport = ReportStats<ExprTypeStats>;
-
 from string key, int value
 where
   not exists(string pattern | extractorInformationSkipKey(pattern) and key.matches(pattern)) and