LUCENE-9386 add case insensitive RegExp matching option (#1541)

Added case insensitive search option (currently only works with ASCII characters)
apache · Jul 8, 2020 · 887fe4c · 887fe4c
1 parent 00203c2
commit 887fe4c
Show file tree

Hide file tree

Showing 4 changed files with 224 additions and 82 deletions.
diff --git a/lucene/core/src/java/org/apache/lucene/search/RegexpQuery.java b/lucene/core/src/java/org/apache/lucene/search/RegexpQuery.java
@@ -83,7 +83,7 @@ public RegexpQuery(Term term, int flags) {
    * Constructs a query for terms matching <code>term</code>.
    * 
    * @param term regular expression.
-   * @param flags optional RegExp features from {@link RegExp}
+   * @param flags optional RegExp syntax features from {@link RegExp}
    * @param maxDeterminizedStates maximum number of states that compiling the
    *  automaton for the regexp can result in.  Set higher to allow more complex
    *  queries and lower to prevent memory exhaustion.
@@ -96,16 +96,46 @@ public RegexpQuery(Term term, int flags, int maxDeterminizedStates) {
    * Constructs a query for terms matching <code>term</code>.
    * 
    * @param term regular expression.
-   * @param flags optional RegExp features from {@link RegExp}
+   * @param syntax_flags optional RegExp syntax features from {@link RegExp}
+   *  automaton for the regexp can result in.  Set higher to allow more complex
+   *  queries and lower to prevent memory exhaustion.
+   * @param match_flags boolean 'or' of match behavior options such as case insensitivity
+   * @param maxDeterminizedStates maximum number of states that compiling the
+   */
+  public RegexpQuery(Term term, int syntax_flags, int match_flags, int maxDeterminizedStates) {
+    this(term, syntax_flags, match_flags, defaultProvider, maxDeterminizedStates);
+  }
+
+  /**
+   * Constructs a query for terms matching <code>term</code>.
+   * 
+   * @param term regular expression.
+   * @param syntax_flags optional RegExp features from {@link RegExp}
+   * @param provider custom AutomatonProvider for named automata
+   * @param maxDeterminizedStates maximum number of states that compiling the
+   *  automaton for the regexp can result in.  Set higher to allow more complex
+   *  queries and lower to prevent memory exhaustion.
+   */
+  public RegexpQuery(Term term, int syntax_flags, AutomatonProvider provider,
+      int maxDeterminizedStates) {
+    this(term, syntax_flags, 0, provider, maxDeterminizedStates);
+  }
+
+  /**
+   * Constructs a query for terms matching <code>term</code>.
+   * 
+   * @param term regular expression.
+   * @param syntax_flags optional RegExp features from {@link RegExp}
+   * @param match_flags boolean 'or' of match behavior options such as case insensitivity
    * @param provider custom AutomatonProvider for named automata
    * @param maxDeterminizedStates maximum number of states that compiling the
    *  automaton for the regexp can result in.  Set higher to allow more complex
    *  queries and lower to prevent memory exhaustion.
    */
-  public RegexpQuery(Term term, int flags, AutomatonProvider provider,
+  public RegexpQuery(Term term, int syntax_flags, int match_flags, AutomatonProvider provider,
       int maxDeterminizedStates) {
     super(term,
-          new RegExp(term.text(), flags).toAutomaton(
+          new RegExp(term.text(), syntax_flags, match_flags).toAutomaton(
                        provider, maxDeterminizedStates), maxDeterminizedStates);
   }