From 2e2ed16788b1e88c016a6c3842ce3ea00d325887 Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Thu, 25 Mar 2021 15:02:32 -0400
Subject: [PATCH] Tests for ISO-2022-JP spanning

---
 lib/Encoding/ISO2022JP.php             | 58 +++++++++++++++++---------
 tests/cases/Encoding/TestISO2022JP.php |  6 ++-
 2 files changed, 43 insertions(+), 21 deletions(-)

diff --git a/lib/Encoding/ISO2022JP.php b/lib/Encoding/ISO2022JP.php
index b887fee..a195489 100644
--- a/lib/Encoding/ISO2022JP.php
+++ b/lib/Encoding/ISO2022JP.php
@@ -220,7 +220,7 @@ class ISO2022JP extends AbstractEncoding implements ModalCoder, Decoder {
             }
             $effectiveMask = preg_replace($exc, "", $mask);
             // if the byte after the mode switch is a wanted one, consume it and go back to the start
-            if (strspn(@$this->string[$this->posByte + 3], $effectiveMask, $this->posByte)) {
+            if (strspn(@$this->string[$this->posByte + 3], $effectiveMask)) {
                 $out .= $this->nextChar();
                 if (--$left) {
                     goto Process;
@@ -231,28 +231,48 @@ class ISO2022JP extends AbstractEncoding implements ModalCoder, Decoder {
     }
 
     public function asciiSpanNot(string $mask, int $length = null): string {
-        if ($this->mode === self::ASCII_STATE) {
-            $mask .= "\x0E\x0F\x1B";
-        } elseif ($this->mode === self::ROMAN_STATE) {
-            $mask .= "\x0E\x0F\x1B\x5C\x7E";
-        } else {
-            // in other modes ASCII characters are never returned
-            return "";
-        }
         $mask .= self::HIGH_BYTES;
-        if ($length !== null) {
-            $len = strcspn($this->string, $mask, $this->posByte, $length);
+        $out = "";
+        $left = ($length === null) ? -1 : $length;
+        Process:
+        if ($this->mode === self::KATAKANA_STATE || $this->mode === self::LEAD_BYTE_STATE) {
+            // these modes will always return an empty span
         } else {
-            $len = strcspn($this->string, $mask, $this->posByte);
+            if ($this->mode === self::ASCII_STATE) {
+                $effectiveMask = $mask."\x0E\x0F\x1B";
+            } elseif ($this->mode === self::ROMAN_STATE) {
+                $effectiveMask = $mask."\x0E\x0F\x1B\x5C\x7E";
+            }
+            if ($length !== null) {
+                $len = strcspn($this->string, $effectiveMask, $this->posByte, $left);
+            } else {
+                $len = strcspn($this->string, $effectiveMask, $this->posByte);
+            }
+            if ($len) {
+                $out .= substr($this->string, $this->posByte, $len);
+                $this->posByte += $len;
+                $this->posChar += $len;
+                $left -= $len;
+            }
         }
-        if ($len) {
-            $out = substr($this->string, $this->posByte, $len);
-            $this->posByte += $len;
-            $this->posChar += $len;
-            return $out;
-        } else {
-            return "";
+        // check if the current position has a mode change to ASCII or Roman modes and is followed by a desired character
+        if ($left && @$this->string[$this->posByte] === "\x1B") {
+            if (substr($this->string, $this->posByte + 1, 2) === "\x28\x42") {
+                $effectiveMask = $mask."\x0E\x0F\x1B";
+            } elseif (substr($this->string, $this->posByte + 1, 2) === "\x28\x4A") {
+                $effectiveMask = $mask."\x0E\x0F\x1B\x5C\x7E";
+            } else {
+                return $out;
+            }
+            // if the byte after the mode switch is a wanted one, consume it and go back to the start
+            if (strcspn(@$this->string[$this->posByte + 3], $effectiveMask)) {
+                $out .= $this->nextChar();
+                if (--$left) {
+                    goto Process;
+                }
+            }
         }
+        return $out;
     }
 
     protected function stateSave(): array {
diff --git a/tests/cases/Encoding/TestISO2022JP.php b/tests/cases/Encoding/TestISO2022JP.php
index 8be3d07..89d859b 100644
--- a/tests/cases/Encoding/TestISO2022JP.php
+++ b/tests/cases/Encoding/TestISO2022JP.php
@@ -31,6 +31,8 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
     protected $seekOffsets = [0, 1, 5, 6, 11, 13, 15, 19];
     /* This string contains an invalid character sequence sandwiched between two null characters */
     protected $brokenChar = "00 FF 00";
+    /* This string conatins the ASCII characters "A" and "Z" followed by two arbitrary non-ASCII characters, followed by the two ASCII characters "0" and "9" */
+    protected $spanString = "1B284A 41 5A 1B2849 5C 5F 1B2842 30 39";
 
     public function provideCodePoints() {
         return [
@@ -247,14 +249,14 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
      * @covers MensBeam\Intl\Encoding\ISO2022JP::asciiSpan
      */
     public function testExtractAsciiSpans() {
-        $this->markTestIncomplete();
+        parent::testExtractAsciiSpans();
     }
 
     /**
      * @covers MensBeam\Intl\Encoding\ISO2022JP::asciiSpanNot
      */
     public function testExtractNegativeAsciiSpans() {
-        $this->markTestIncomplete();
+        parent::testExtractNegativeAsciiSpans();
     }
 
     /**