Fixing license match post-filtering calculation with traling contained range. (#1489)

Author: isoos

CHANGELOG.md

@@ -3,6 +3,7 @@
 - Fixed small ambiguity in the order of license match evaluation.
 - Resolve `analysis_options.yaml` dev dependency to expose transitive formatter options.
 - Fixed license coverage calculation.
+- Fixed license match post-filtering.
 
 ## 0.22.23
 

lib/src/license_detection/license_detector.dart

@@ -268,25 +268,33 @@ double calculateUnclaimedTokenPercentage(
 /// Returns the number of tokens in the longest unclaimed token
 /// sequence.
 int findLongestUnclaimedTokenRange(List<LicenseMatch> matches, int end) {
-  var ranges = <Range>[];
   if (matches.isEmpty) return end;
 
-  for (var match in matches) {
-    ranges.add(Range(match.tokenRange.start, match.tokenRange.end));
-  }
-
-  ranges.sort(sortRangeOnStartValue);
-  var maxTokenSequence = ranges.first.start - 0;
+  var maxUnclaimed = 0;
+
+  // create an end-of-range marker at the end of ranges for easier computation
+  final ranges = [...matches.map((m) => m.tokenRange), Range(end, end)];
+
+  // move position from the start up to the end, checking uncovered ranges
+  for (var pos = 0; pos < end;) {
+    // check if inside a range
+    final skipToEnd = ranges
+        .where((r) => r.containsIndex(pos))
+        .map((r) => r.end)
+        .fold(-1, max);
+    if (skipToEnd > -1) {
+      assert(pos < skipToEnd);
+      pos = skipToEnd;
+      continue;
+    }
 
-  for (var i = 1; i < ranges.length; i++) {
-    maxTokenSequence = max(
-      ranges[i].start - ranges[i - 1].end,
-      maxTokenSequence,
-    );
+    // jump to next range
+    final nextRange = ranges
+        .where((r) => r.start > pos)
+        .reduce((a, b) => a.start.compareTo(b.start) <= 0 ? a : b);
+    maxUnclaimed = max(maxUnclaimed, nextRange.start - pos);
+    pos = nextRange.end;
   }
 
-  maxTokenSequence = max(maxTokenSequence, end - ranges.last.end);
-  return maxTokenSequence;
+  return maxUnclaimed;
 }
-
-int sortRangeOnStartValue(Range a, Range b) => a.start - b.start;

lib/src/license_detection/token_matcher.dart

@@ -54,6 +54,8 @@ class Range {
         (end >= other.start && end <= other.end);
   }
 
+  bool containsIndex(int value) => start <= value && value < end;
+
   @override
   String toString() => 'Range($start-$end)';
 }

test/license_detector_test.dart

@@ -258,6 +258,16 @@ void main() {
       160,
       1000,
     );
+
+    _testLongestUnclaimedTokenRange(
+      'Trailing contained range',
+      [
+        _dummyLicenseMatchInstance(0.9, '', start: 30, end: 940),
+        _dummyLicenseMatchInstance(0.9, '', start: 50, end: 70),
+      ],
+      60,
+      1000,
+    );
   });
 }
 

test/licenses/license_coverage_test.dart

@@ -16,10 +16,6 @@ void main() {
           license.content,
           relativePath: 'LICENSE',
         );
-        // TODO: fix detection and return at least one match (https://github.com/dart-lang/pana/issues/1484)
-        if (license.identifier == 'SSH-OpenSSH') {
-          continue;
-        }
         final match = detected
             .where((l) => l.spdxIdentifier == license.identifier)
             .single;