do not create char class entries for atBOL matching

The scanning engine is testing for newline on raw input characters without
translation through the char-class map, so there is no need for a separate
class in that case.

javatests/jflex/testcase/bol/BUILD

@@ -7,11 +7,19 @@ jflex(
     outputs = ["BolScanner.java"],
 )
 
+jflex(
+    name = "gen_bol2_scanner",
+    srcs = ["bol2.flex"],
+    jflex_bin = "//jflex:jflex_bin",
+    outputs = ["Bol2Scanner.java"],
+)
+
 java_test(
     name = "BolTest",
     srcs = [
         "BolTest.java",
         "State.java",
+        ":gen_bol2_scanner",
         ":gen_bol_scanner",
     ],
     deps = [

javatests/jflex/testcase/bol/BolTest.java

@@ -10,16 +10,25 @@
 
 /** Tests BOL and EOL operators. */
 public class BolTest {
+  // standard BOL and EOL tests
   BolScanner scanner;
+  // test BOL in absence of char class for NL characters
+  Bol2Scanner scanner2;
 
   @Before
   public void testMustInitializeScanner() {
     scanner = null;
+    scanner2 = null;
   }
 
   @After
   public void end() throws Exception {
-    assertThat(scanner.yylex()).isEqualTo(State.END_OF_FILE);
+    if (scanner != null) {
+      assertThat(scanner.yylex()).isEqualTo(State.END_OF_FILE);
+    }
+    if (scanner2 != null) {
+      assertThat(scanner2.yylex()).isEqualTo(State.END_OF_FILE);
+    }
   }
 
   @Test
@@ -35,6 +44,30 @@ public void bolAndEol() throws Exception {
     assertThat(scanner.yylex()).isEqualTo(State.LINE_FEED);
   }
 
+  @Test
+  public void bolAndEol2() throws Exception {
+    scanner = createScanner("hello\r\n");
+    assertThat(scanner.yylex()).isEqualTo(State.HELLO_AT_BOL_AND_EOL);
+    assertThat(scanner.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner.yylex()).isEqualTo(State.LINE_FEED);
+  }
+
+  @Test
+  public void bolAndEol3() throws Exception {
+    scanner = createScanner("\nhello\r\n");
+    assertThat(scanner.yylex()).isEqualTo(State.LINE_FEED);
+    assertThat(scanner.yylex()).isEqualTo(State.HELLO_AT_BOL_AND_EOL);
+    assertThat(scanner.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner.yylex()).isEqualTo(State.LINE_FEED);
+  }
+
+  @Test
+  public void bolAndEof() throws Exception {
+    // EOF does not count as EOL. Should it?
+    scanner = createScanner("hello");
+    assertThat(scanner.yylex()).isEqualTo(State.HELLO_AT_BOL);
+  }
+
   @Test
   public void eol() throws Exception {
     scanner = createScanner("  hello\n");
@@ -79,7 +112,35 @@ public void other() throws Exception {
     assertThat(scanner.yylex()).isEqualTo(State.LINE_FEED);
   }
 
+  @Test
+  public void bolNoNLCCL() throws Exception {
+    scanner2 = createScanner2("hello");
+    assertThat(scanner2.yylex()).isEqualTo(State.HELLO_AT_BOL);
+  }
+
+  @Test
+  public void bolNoNLCCL2() throws Exception {
+    scanner2 = createScanner2("hello\n");
+    assertThat(scanner2.yylex()).isEqualTo(State.HELLO_AT_BOL);
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+  }
+
+  @Test
+  public void noBolNoNLCCL() throws Exception {
+    scanner2 = createScanner2("\n  hello  ");
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner2.yylex()).isEqualTo(State.HELLO_SIMPLY);
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+    assertThat(scanner2.yylex()).isEqualTo(State.OTHER);
+  }
+
   private static BolScanner createScanner(final String content) throws IOException {
     return new BolScanner(CharSource.wrap(content).openStream());
   }
+
+  private static Bol2Scanner createScanner2(final String content) throws IOException {
+    return new Bol2Scanner(CharSource.wrap(content).openStream());
+  }
 }

javatests/jflex/testcase/bol/State.java

@@ -6,7 +6,6 @@ public enum State {
   HELLO_AT_BOL,
   HELLO_AT_EOL,
   HELLO_SIMPLY,
-  CARRIAGE_RETURN,
   LINE_FEED,
   SPACE,
   OTHER,

javatests/jflex/testcase/bol/bol.flex

@@ -24,10 +24,9 @@ package jflex.testcase.bol;
 "hello"$   { return State.HELLO_AT_EOL; }
 "hello"    { return State.HELLO_SIMPLY; }
 
-\r         { return State.CARRIAGE_RETURN; }
 \n         { return State.LINE_FEED; }
 
 " "        { return State.SPACE; }
-.          { return State.OTHER; }
+[^]        { return State.OTHER; }
 
 <<EOF>>    { return State.END_OF_FILE; }

javatests/jflex/testcase/bol/bol2.flex

@@ -0,0 +1,19 @@
+package jflex.testcase.bol;
+
+%%
+
+%public
+%class Bol2Scanner
+%type State
+%debug
+
+%unicode
+
+%%
+
+^"hello"   { return State.HELLO_AT_BOL; }
+"hello"    { return State.HELLO_SIMPLY; }
+
+[^]        { return State.OTHER; }
+
+<<EOF>>    { return State.END_OF_FILE; }

jflex/src/main/cup/LexParse.cup

@@ -596,11 +596,7 @@ states        ::=  IDENT:id COMMA states:list
                 ;
 
 hatOPT        ::=  HAT
-                   {:
-                      // assumption: newline chars have no uppercase variant
-                      charClasses.makeClass("\n\r\u000B\u000C\u0085\u2028\u2029", false);
-                      RESULT = true;
-                   :}
+                   {: RESULT = true; :}
                 |  /* empty */
                    {: RESULT = false; :}
                 ;

testsuite/testcases/src/test/cases/bol/.gitignore

@@ -1 +1,2 @@
 Bol.java
+Bol2.java

testsuite/testcases/src/test/cases/bol/bol-0.input

@@ -6,3 +6,4 @@ hello
 hello 
 
 sdfa
+hello

testsuite/testcases/src/test/cases/bol/bol-0.output

@@ -2,87 +2,90 @@ line: 1 col: 1 char: 0 match: --hello--
 action [19] { System.out.println("hello at BOL and EOL"); }
 hello at BOL and EOL
 line: 1 col: 6 char: 5 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 2 col: 1 char: 6 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 2 col: 2 char: 7 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 2 col: 3 char: 8 match: --hello--
 action [21] { System.out.println("hello at EOL"); }
 hello at EOL
 line: 2 col: 8 char: 13 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 3 col: 1 char: 14 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 3 col: 2 char: 15 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 3 col: 3 char: 16 match: --hello--
 action [22] { System.out.println("just hello"); }
 just hello
 line: 3 col: 8 char: 21 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 3 col: 9 char: 22 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 3 col: 10 char: 23 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 4 col: 1 char: 24 match: --hello--
 action [20] { System.out.println("hello at BOL"); }
 hello at BOL
 line: 4 col: 6 char: 29 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 4 col: 7 char: 30 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 5 col: 1 char: 31 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 5 col: 2 char: 32 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 5 col: 3 char: 33 match: --hello--
 action [21] { System.out.println("hello at EOL"); }
 hello at EOL
 line: 5 col: 8 char: 38 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 6 col: 1 char: 39 match: --hello--
 action [20] { System.out.println("hello at BOL"); }
 hello at BOL
 line: 6 col: 6 char: 44 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 6 col: 7 char: 45 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 7 col: 1 char: 46 match: -- --
-action [27] { System.out.println( "\" \"" ); }
+action [26] { System.out.println( "\" \"" ); }
 " "
 line: 7 col: 2 char: 47 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
 line: 8 col: 1 char: 48 match: --s--
-action [28] { System.out.println( yytext() ); }
+action [27] { System.out.println( yytext() ); }
 s
 line: 8 col: 2 char: 49 match: --d--
-action [28] { System.out.println( yytext() ); }
+action [27] { System.out.println( yytext() ); }
 d
 line: 8 col: 3 char: 50 match: --f--
-action [28] { System.out.println( yytext() ); }
+action [27] { System.out.println( yytext() ); }
 f
 line: 8 col: 4 char: 51 match: --a--
-action [28] { System.out.println( yytext() ); }
+action [27] { System.out.println( yytext() ); }
 a
 line: 8 col: 5 char: 52 match: --\u000A--
-action [25] { System.out.println("\\n"); }
+action [24] { System.out.println("\\n"); }
 \n
+line: 9 col: 1 char: 53 match: --hello--
+action [20] { System.out.println("hello at BOL"); }
+hello at BOL
 -1