Improve codeowners lookup performance (#12058)

feat: improve codeowners lookup performance

fix: keep trailing slash for linear scan

Co-authored-by: daniel.mohedano <daniel.mohedano@datadoghq.com>

Author: daniel-mohedano

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/CodeownersImpl.java

@@ -4,11 +4,9 @@
 import java.io.BufferedReader;
 import java.io.IOException;
 import java.io.Reader;
-import java.util.ArrayDeque;
 import java.util.ArrayList;
 import java.util.Collection;
 import java.util.Collections;
-import java.util.Deque;
 import java.util.LinkedHashMap;
 import java.util.LinkedHashSet;
 import java.util.List;
@@ -20,9 +18,9 @@
 
 public class CodeownersImpl implements Codeowners {
 
-  private final Collection<Section> sections;
+  private final List<Section> sections;
 
-  private CodeownersImpl(Collection<Section> sections) {
+  private CodeownersImpl(List<Section> sections) {
     this.sections = sections;
   }
 
@@ -32,17 +30,25 @@ private CodeownersImpl(Collection<Section> sections) {
    */
   @Override
   public @Nullable Collection<String> getOwners(@Nonnull String path) {
-    char[] pathCharacters = path.toCharArray();
+    if (sections.size() == 1) {
+      Section section = sections.get(0);
+      if (section.isExcluded(path)) {
+        return new ArrayList<>();
+      }
+      Entry entry = section.findMatchingEntry(path);
+      return entry != null ? new ArrayList<>(entry.getOwners()) : null;
+    }
+
     Set<String> owners = null;
     for (Section section : sections) {
-      if (section.isExcluded(pathCharacters)) {
+      if (section.isExcluded(path)) {
         if (owners == null) {
           owners = new LinkedHashSet<>();
         }
         continue;
       }
 
-      Entry entry = section.findMatchingEntry(pathCharacters);
+      Entry entry = section.findMatchingEntry(path);
       if (entry != null) {
         if (owners == null) {
           owners = new LinkedHashSet<>();
@@ -90,36 +96,4 @@ public static Codeowners parse(Reader r) throws IOException {
     sections.addAll(namedSections.values());
     return new CodeownersImpl(sections);
   }
-
-  private static final class Section {
-
-    private final Deque<Entry> entries = new ArrayDeque<>();
-    private final Collection<Entry> exclusions = new ArrayList<>();
-
-    private void add(Entry entry) {
-      if (entry.isExclusion()) {
-        exclusions.add(entry);
-      } else {
-        entries.offerFirst(entry);
-      }
-    }
-
-    private boolean isExcluded(char[] path) {
-      for (Entry exclusion : exclusions) {
-        if (exclusion.getMatcher().consume(path, 0) >= 0) {
-          return true;
-        }
-      }
-      return false;
-    }
-
-    private @Nullable Entry findMatchingEntry(char[] path) {
-      for (Entry entry : entries) {
-        if (entry.getMatcher().consume(path, 0) >= 0) {
-          return entry;
-        }
-      }
-      return null;
-    }
-  }
 }

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/Entry.java

@@ -1,18 +1,24 @@
 package datadog.trace.civisibility.codeowners;
 
 import datadog.trace.civisibility.codeowners.matcher.Matcher;
+import java.util.ArrayList;
 import java.util.Collection;
+import java.util.LinkedHashSet;
+import javax.annotation.Nullable;
 
 public class Entry {
 
   private final Matcher matcher;
   private final Collection<String> owners;
   private final boolean exclusion;
+  private final @Nullable String indexKey;
 
-  public Entry(Matcher matcher, Collection<String> owners, boolean exclusion) {
+  public Entry(
+      Matcher matcher, Collection<String> owners, boolean exclusion, @Nullable String indexKey) {
     this.matcher = matcher;
-    this.owners = owners;
+    this.owners = owners.size() > 1 ? new ArrayList<>(new LinkedHashSet<>(owners)) : owners;
     this.exclusion = exclusion;
+    this.indexKey = indexKey;
   }
 
   public Matcher getMatcher() {
@@ -26,4 +32,8 @@ public Collection<String> getOwners() {
   public boolean isExclusion() {
     return exclusion;
   }
+
+  public @Nullable String getIndexKey() {
+    return indexKey;
+  }
 }

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/EntryBuilder.java

@@ -73,12 +73,13 @@ public EntryBuilder(CharacterMatcher.Factory characterMatcherFactory, String s)
         offset++;
       }
 
+      String indexKey = parseIndexKey();
       Matcher matcher = parseMatcher();
       Collection<String> owners = exclusion ? Collections.emptyList() : parseOwners();
       if (!exclusion && owners.isEmpty()) {
         owners = sectionDefaultOwners;
       }
-      return new Entry(matcher, owners, exclusion);
+      return new Entry(matcher, owners, exclusion, indexKey);
 
     } catch (Exception e) {
       log.warn("Skipping malformed CODEOWNERS entry: {}", new String(c), e);
@@ -224,6 +225,46 @@ private Matcher parseMatcher() {
     return new CompositeMatcher(characterMatchers.toArray(new Matcher[0]));
   }
 
+  private @Nullable String parseIndexKey() {
+    // Index by the first two fixed path segments; patterns without a safe prefix stay linear.
+    int position = offset;
+    int patternEnd = position;
+    while (patternEnd < c.length && !isPatternTerminator(c[patternEnd])) {
+      patternEnd++;
+    }
+    // Trailing-slash matchers can match beyond a segment boundary, making their prefix unsafe.
+    if (patternEnd > position && c[patternEnd - 1] == '/') {
+      return null;
+    }
+    boolean patternContainsSlashes = c[position] == '/';
+    if (patternContainsSlashes) {
+      position++;
+    }
+    int prefixStart = position;
+    int firstSeparator = -1;
+    for (; position < c.length && !isPatternTerminator(c[position]); position++) {
+      char character = c[position];
+      if (character == '*' || character == '?' || character == '[') {
+        return null;
+      }
+      if (character == '\\') {
+        return null;
+      }
+      if (character == '/') {
+        patternContainsSlashes = true;
+        if (firstSeparator >= 0) {
+          return new String(c, prefixStart, position - prefixStart);
+        }
+        firstSeparator = position;
+      }
+    }
+
+    if (!patternContainsSlashes || firstSeparator < 0 || firstSeparator == position - 1) {
+      return null;
+    }
+    return new String(c, prefixStart, position - prefixStart);
+  }
+
   private boolean consumeDoubleAsterisk() {
     int position = offset;
 

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/EntryIndex.java

@@ -0,0 +1,105 @@
+package datadog.trace.civisibility.codeowners;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import javax.annotation.Nullable;
+
+/**
+ * Resolves entries by rule priority, switching large rule sets from linear scans to fixed-prefix
+ * buckets while retaining a fallback for patterns that cannot be indexed.
+ */
+final class EntryIndex {
+
+  private static final int MIN_INDEX_SIZE = 512;
+
+  private List<IndexedEntry> entries = new ArrayList<>();
+  private Map<String, List<IndexedEntry>> entriesByKey;
+  private List<IndexedEntry> unindexedEntries;
+  private int indexableEntryCount;
+
+  void add(Entry entry, int order) {
+    IndexedEntry indexedEntry = new IndexedEntry(entry, order);
+    if (entriesByKey != null) {
+      index(indexedEntry);
+    } else {
+      entries.add(indexedEntry);
+      if (entry.getIndexKey() != null) {
+        indexableEntryCount++;
+      }
+      if (entries.size() >= MIN_INDEX_SIZE && indexableEntryCount * 2 >= entries.size()) {
+        entriesByKey = new HashMap<>();
+        unindexedEntries = new ArrayList<>();
+        for (IndexedEntry existingEntry : entries) {
+          index(existingEntry);
+        }
+        entries = Collections.emptyList();
+      }
+    }
+  }
+
+  @Nullable
+  Entry find(String path) {
+    IndexedEntry entry = findIndexedEntry(path);
+    return entry != null ? entry.entry : null;
+  }
+
+  private @Nullable IndexedEntry findIndexedEntry(String path) {
+    if (entriesByKey == null) {
+      return findFirstMatch(entries, path);
+    }
+    IndexedEntry unindexedMatch = findFirstMatch(unindexedEntries, path);
+
+    int firstSeparator = path.indexOf('/');
+    if (firstSeparator < 0) {
+      return unindexedMatch;
+    }
+    int secondSeparator = path.indexOf('/', firstSeparator + 1);
+    int keyEnd = secondSeparator >= 0 ? secondSeparator : path.length();
+    List<IndexedEntry> indexedEntries = entriesByKey.get(path.substring(0, keyEnd));
+    IndexedEntry indexedMatch = findFirstMatch(indexedEntries, path);
+
+    if (unindexedMatch == null) {
+      return indexedMatch;
+    }
+    if (indexedMatch == null) {
+      return unindexedMatch;
+    }
+    return indexedMatch.order > unindexedMatch.order ? indexedMatch : unindexedMatch;
+  }
+
+  private void index(IndexedEntry indexedEntry) {
+    String indexKey = indexedEntry.entry.getIndexKey();
+    if (indexKey != null) {
+      entriesByKey.computeIfAbsent(indexKey, key -> new ArrayList<>()).add(indexedEntry);
+    } else {
+      unindexedEntries.add(indexedEntry);
+    }
+  }
+
+  private static @Nullable IndexedEntry findFirstMatch(
+      @Nullable List<IndexedEntry> entries, String path) {
+    if (entries != null) {
+      for (int i = entries.size() - 1; i >= 0; i--) {
+        IndexedEntry entry = entries.get(i);
+        if (entry.entry.getMatcher().consume(path, 0) >= 0) {
+          return entry;
+        }
+      }
+    }
+    return null;
+  }
+
+  private static final class IndexedEntry {
+
+    private final Entry entry;
+    private final int order;
+
+    private IndexedEntry(Entry entry, int order) {
+      this.entry = entry;
+      this.order = order;
+    }
+  }
+}

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/Section.java

@@ -0,0 +1,28 @@
+package datadog.trace.civisibility.codeowners;
+
+import javax.annotation.Nullable;
+
+/** Groups ownership and exclusion rules for a CODEOWNERS section while preserving rule order. */
+final class Section {
+
+  private final EntryIndex entries = new EntryIndex();
+  private final EntryIndex exclusions = new EntryIndex();
+  private int entryOrder;
+
+  void add(Entry entry) {
+    if (entry.isExclusion()) {
+      exclusions.add(entry, entryOrder++);
+    } else {
+      entries.add(entry, entryOrder++);
+    }
+  }
+
+  boolean isExcluded(String path) {
+    return exclusions.find(path) != null;
+  }
+
+  @Nullable
+  Entry findMatchingEntry(String path) {
+    return entries.find(path);
+  }
+}

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/matcher/AsteriskMatcher.java

@@ -7,8 +7,8 @@ public class AsteriskMatcher implements Matcher {
   private AsteriskMatcher() {}
 
   @Override
-  public int consume(char[] line, int offset) {
-    return offset < line.length && line[offset] != '/' ? 1 : -1;
+  public int consume(String line, int offset) {
+    return offset < line.length() && line.charAt(offset) != '/' ? 1 : -1;
   }
 
   @Override

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/matcher/CharacterMatcher.java

@@ -12,8 +12,8 @@ private CharacterMatcher(char character) {
   }
 
   @Override
-  public int consume(char[] line, int offset) {
-    return offset < line.length && line[offset] == character ? 1 : -1;
+  public int consume(String line, int offset) {
+    return offset < line.length() && line.charAt(offset) == character ? 1 : -1;
   }
 
   @Override

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/matcher/CompositeMatcher.java

@@ -9,11 +9,11 @@ public CompositeMatcher(Matcher[] delegates) {
   }
 
   @Override
-  public int consume(char[] line, int offset) {
+  public int consume(String line, int offset) {
     return consume(line, offset, 0);
   }
 
-  private int consume(char[] line, int offset, int matcherOffset) {
+  private int consume(String line, int offset, int matcherOffset) {
     int position = offset;
     while (matcherOffset < delegates.length) {
       Matcher delegate = delegates[matcherOffset];

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/matcher/DoubleAsteriskMatcher.java

@@ -7,13 +7,13 @@ public class DoubleAsteriskMatcher implements Matcher {
   private DoubleAsteriskMatcher() {}
 
   @Override
-  public int consume(char[] line, int offset) {
-    if (offset == line.length) {
+  public int consume(String line, int offset) {
+    if (offset == line.length()) {
       return -1;
     }
 
     int position = offset;
-    while (position < line.length && line[position++] != '/') {}
+    while (position < line.length() && line.charAt(position++) != '/') {}
     return position - offset;
   }
 

dd-java-agent/agent-ci-visibility/src/main/java/datadog/trace/civisibility/codeowners/matcher/EndOfLineMatcher.java

@@ -5,8 +5,8 @@ public class EndOfLineMatcher implements Matcher {
   public static final Matcher INSTANCE = new EndOfLineMatcher();
 
   @Override
-  public int consume(char[] line, int offset) {
-    return offset == line.length ? 0 : -1;
+  public int consume(String line, int offset) {
+    return offset == line.length() ? 0 : -1;
   }
 
   @Override