Finally fix Shift_JIS seeker

lib/Encoding/ShiftJIS.php

@@ -120,19 +120,19 @@ class ShiftJIS extends AbstractEncoding implements StatelessEncoding {
             }
             // go back one byte
             $b1 = ord(@$this->string[--$this->posByte]);
-            if ($b1 < 0x40 || $b1 > 0xFC || $b1 == 0x7F || $this->posByte === 0 || $this->posByte === $this->errMark) { // these bytes never appear in sequences, and the first byte is necessarily the start of a sequence
+            if ($b1 < 0x40 || $b1 > 0xFC || $b1 === 0x7F || $this->posByte === 0 || $this->posByte === $this->errMark) { // these bytes never appear in sequences, and the first byte is necessarily the start of a sequence
                 // the byte is a character
                 continue;
             }
             // go back a second byte
             $b2 = ord(@$this->string[--$this->posByte]);
-            if ($this->posByte === $this->errMark || $this->posByte === 0) { // position is unambiguously the start of a character
-                // the two bytes form a character
-                continue;
-            } elseif ($b2 < 0x81 || $b2 > 0xFC || ($b2 >= 0xA0 && $b2 <= 0xDF)) { // these bytes never appear in the lead of a sequence
+            if ($b2 < 0x81 || $b2 > 0xFC || ($b2 >= 0xA0 && $b2 <= 0xDF)) { // these bytes never appear in the lead of a sequence
                 // the first byte was a character
                 $this->posByte += 1;
                 continue;
+            } elseif ($this->posByte === $this->errMark || $this->posByte === 0) { // position is unambiguously the start of a character
+                // the two bytes form a character
+                continue;
             } else { // the second byte is part of a two-byte sequence, but it's unclear if it's the lead or trail byte
                 $start = $this->posByte + 2;
                 $pos = $this->posByte;

tests/cases/Encoding/TestGB18030.php

@@ -144,6 +144,14 @@ class TestGB18030 extends \MensBeam\Intl\Test\CoderDecoderTest {
         return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
     }
 
+
+    /**
+     * @covers MensBeam\Intl\Encoding\GB18030::seekBack
+     */
+    public function testSeekBackOverRandomData() {
+        return parent::testSeekBackOverRandomData();
+    }
+
     public function provideCodePoints() {
         // bytes confirmed using Firefox
         $series_gb18030 = [
@@ -200,14 +208,6 @@ class TestGB18030 extends \MensBeam\Intl\Test\CoderDecoderTest {
         }
     }
 
-
-    /**
-     * @covers MensBeam\Intl\Encoding\GB18030::seekBack
-     */
-    public function testSeekBackOverRandomData() {
-        return parent::testSeekBackOverRandomData();
-    }
-
     public function provideStrings() {
         return [
             'empty string' => ["", []],

tests/cases/Encoding/TestShiftJIS.php

@@ -12,16 +12,6 @@ use MensBeam\Intl\Encoding\EncoderException;
 
 class TestShiftJIS extends \MensBeam\Intl\Test\CoderDecoderTest {
     protected $testedClass = ShiftJIS::class;
-    /*
-        Char 0 U+007A   (1 byte)  Offset 0
-        Char 1 U+86CC   (2 bytes) Offset 1
-        Char 2 U+6C34   (2 bytes) Offset 3
-        Char 3 U+00CA   (0 bytes) Offset 5
-        Char 4 U+0304   (2 bytes) Offset 5
-        Char 5 U+00EA   (0 bytes) Offset 7
-        Char 6 U+030C   (2 bytes) Offset 7
-        End of string at char 7, offset 9
-    */
     protected $seekString = "";
     protected $seekCodes = [];
     protected $seekOffsets = [];
@@ -151,6 +141,12 @@ class TestShiftJIS extends \MensBeam\Intl\Test\CoderDecoderTest {
 
     public function provideStrings() {
         return [
+            'empty string' => ["", []],
+            'sanity check' => ["40", [64]],
+            'former ASCII deviations' => ["5C 7E", [92, 126]],
+            'JIS X 0201 range' => ["A1 DF", [65377, 65439]],
+            'EUDC range' => ["F040 F9FC", [57344, 59223]],
+            'JIS X 0208 assigned range' => ["8140 9F7E 8180 9FFC", [12288, 27631, 247, 28364]],
         ];
     }
 

tests/lib/DecoderTest.php

@@ -54,7 +54,7 @@ abstract class DecoderTest extends \PHPUnit\Framework\TestCase {
         }
         $this->assertSame(sizeof($exp), $pos);
         while ($s->posChar()) {
-            $this->assertSame(0, $s->seek(-1));
+            $this->assertSame(0, $s->seek(-1), "Error stepping back to position ".($pos - 1));
             $this->assertSame(--$pos, $s->posChar());
             $act[] = $s->nextCode();
             $s->seek(-1);

tools/test-shiftjis.html

@@ -0,0 +1,20 @@
+<!DOCTYPE html>
+<meta charset=shift_jis>
+<!-- Chromium does NOT produce correct results as of this writing; use Firefox to generate test data -->
+<script>
+var sampleStrings = {
+    'empty string': "",
+    // sanity checks
+    'sanity check': "40",
+    'former ASCII deviations': "5C 7E",
+    'JIS X 0201 range': "A1 DF",
+    'EUDC range': "F040 F9FC",
+    // JIS X 0208
+    'JIS X 0208 assigned range': "8140 9F7E 8180 9FFC",
+};
+var sampleCharacters = {
+};
+var seekCodePoints = [
+];
+</script>
+<script src="test.js"></script>