summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--src/HTML5/Parser/Scanner.php49
-rw-r--r--test/HTML5/ScannerTest.php63
2 files changed, 108 insertions, 4 deletions
diff --git a/src/HTML5/Parser/Scanner.php b/src/HTML5/Parser/Scanner.php
index 0260391..2500dab 100644
--- a/src/HTML5/Parser/Scanner.php
+++ b/src/HTML5/Parser/Scanner.php
@@ -85,16 +85,57 @@ class Scanner {
}
}
+ /**
+ * Get the next group of that is a hex value.
+ *
+ * Note, along with getting the characters the pointer in the data will be
+ * moved as well.
+ *
+ * @todo There is a potential for many false positives with this method. Make it more accurate.
+ *
+ * @return string
+ * The next group that is a hex value.
+ */
public function getHex() {
- $this->charsWhile(self::CHARS_HEX);
+ return $this->is->charsWhile(self::CHARS_HEX);
}
+
+ /**
+ * Get the next group of characters that are ASCII Alpha characters.
+ *
+ * Note, along with getting the characters the pointer in the data will be
+ * moved as well.
+ *
+ * @return string
+ * The next group of ASCII alpha characters.
+ */
public function getAsciiAlpha() {
- $this->charsWhile(self::CHARS_ALPHA);
+ return $this->is->charsWhile(self::CHARS_ALPHA);
}
+
+ /**
+ * Get the next group of characters that are ASCII Alpha characters and numbers.
+ *
+ * Note, along with getting the characters the pointer in the data will be
+ * moved as well.
+ *
+ * @return string
+ * The next group of ASCII alpha characters and numbers.
+ */
public function getAsciiAlphaNum() {
- $this->charsWhile(self::CHARS_ALNUM);
+ return $this->is->charsWhile(self::CHARS_ALNUM);
}
+
+ /**
+ * Get the next group of numbers.
+ *
+ * Note, along with getting the characters the pointer in the data will be
+ * moved as well.
+ *
+ * @return string
+ * The next group of numbers.
+ */
public function getNumeric() {
- $this->charsWhile('0123456789');
+ return $this->is->charsWhile('0123456789');
}
}
diff --git a/test/HTML5/ScannerTest.php b/test/HTML5/ScannerTest.php
index 919f8fc..251dcfa 100644
--- a/test/HTML5/ScannerTest.php
+++ b/test/HTML5/ScannerTest.php
@@ -62,4 +62,67 @@ class ScannerTest extends TestCase {
$c = $s->next();
$this->assertEquals($c, $s->current());
}
+
+ public function testUnconsume() {
+ $s = new Scanner(new InputStream("abcdefghijklmnopqrst"));
+
+ // Get initial position.
+ $s->next();
+ $start = $s->position();
+
+ // Move forward a bunch of positions.
+ $amount = 7;
+ for($i = 0; $i < $amount; $i++) {
+ $s->next();
+ }
+
+ // Roll back the amount we moved forward.
+ $s->unconsume($amount);
+
+ $this->assertEquals($start, $s->position());
+ }
+
+ // public function testGetHex() {
+ // $s = new Scanner(new InputStream("abcdef%mnop*"));
+
+ // $s->next();
+
+ // $this->assertEquals('bcdef', $s->getHex());
+
+ // echo $s->next(); echo $s->next(); echo $s->position(); echo $s->getHex();
+
+ // //$this->assertEquals('mnop', $s->getHex());
+ // }
+
+ public function testGetAsciiAlpha() {
+ $s = new Scanner(new InputStream("abcdef1%mnop*"));
+
+ $this->assertEquals('abcdef', $s->getAsciiAlpha());
+
+ // Move past the 1% to scan the next group of text.
+ $s->next();
+ $s->next();
+ $this->assertEquals('mnop', $s->getAsciiAlpha());
+ }
+
+ public function testGetAsciiAlphaNum() {
+ $s = new Scanner(new InputStream("abcdef1ghpo#mn94op"));
+
+ $this->assertEquals('abcdef1ghpo', $s->getAsciiAlphaNum());
+
+ // Move past the # to scan the next group of text.
+ $s->next();
+ $this->assertEquals('mn94op', $s->getAsciiAlphaNum());
+ }
+
+ public function testGetNumeric() {
+ $s = new Scanner(new InputStream("1784a 45 9867 #"));
+
+ $this->assertEquals('1784', $s->getNumeric());
+
+ // Move past the 'a ' to scan the next group of text.
+ $s->next();
+ $s->next();
+ $this->assertEquals('45', $s->getNumeric());
+ }
} \ No newline at end of file