Merge pull request #45 from glensc/set-sig-regex

add getSignatureRegex/setSignatureRegex
willdurand · Nov 28, 2017 · a1f6a22 · a1f6a22
2 parents d3839b5 + 7c64d7d
commit a1f6a22
Show file tree

Hide file tree

Showing 3 changed files with 84 additions and 4 deletions.
diff --git a/src/EmailReplyParser/Parser/EmailParser.php b/src/EmailReplyParser/Parser/EmailParser.php
@@ -18,12 +18,14 @@
  */
 class EmailParser
 {
+    const QUOTE_REGEX = '/>+$/s';
+
     /**
      * Regex to match signatures
+     *
+     * @var string
      */
-    const SIG_REGEX   = '/(?:^\s*--|^\s*__|^-\w|^-- $)|(?:^Sent from my (?:\s*\w+){1,4}$)|(?:^={30,}$)$/s';
-
-    const QUOTE_REGEX = '/>+$/s';
+    private $signatureRegex = '/(?:^\s*--|^\s*__|^-\w|^-- $)|(?:^Sent from my (?:\s*\w+){1,4}$)|(?:^={30,}$)$/s';
 
     /**
      * @var string[]
@@ -141,6 +143,28 @@ public function setQuoteHeadersRegex(array $quoteHeadersRegex)
         return $this;
     }
 
+    /**
+     * @return string
+     * @since 2.7.0
+     */
+    public function getSignatureRegex()
+    {
+        return $this->signatureRegex;
+    }
+
+    /**
+     * @param string $signatureRegex
+     *
+     * @return EmailParser
+     * @since 2.7.0
+     */
+    public function setSignatureRegex($signatureRegex)
+    {
+        $this->signatureRegex = $signatureRegex;
+
+        return $this;
+    }
+
     /**
      * @param FragmentDTO[] $fragmentDTOs
      *
@@ -174,11 +198,12 @@ private function isQuoteHeader($line)
 
     private function isSignature($line)
     {
-        return preg_match(static::SIG_REGEX, strrev($line)) ? true : false;
+        return preg_match($this->signatureRegex, strrev($line)) ? true : false;
     }
 
     /**
      * @param string $line
+     * @return bool
      */
     private function isQuote($line)
     {
@@ -191,8 +216,10 @@ private function isEmpty(FragmentDTO $fragment)
     }
 
     /**
+     * @param FragmentDTO $fragment
      * @param string  $line
      * @param boolean $isQuoted
+     * @return bool
      */
     private function isFragmentLine(FragmentDTO $fragment, $line, $isQuoted)
     {

diff --git a/tests/EmailReplyParser/Tests/Parser/EmailParserTest.php b/tests/EmailReplyParser/Tests/Parser/EmailParserTest.php
@@ -410,6 +410,24 @@ public function testEmailWithFairAmountOfContent()
         $this->assertRegexp('/^On Thursday/', (string) $fragments[0]);
     }
 
+    /**
+     * override regexp, not to match too greedy signature.
+     *
+     * See: https://github.com/willdurand/EmailReplyParser/pull/42
+     */
+    public function testCustomSignatureRegex()
+    {
+        $signatureRegex = '/(?:^\s*--|^\s*__|^-- $)|(?:^Sent from my (?:\s*\w+){1,3})$/s';
+        $this->parser->setSignatureRegex($signatureRegex);
+        $email = $this->parser->parse($this->getFixtures('email_ls-l.txt'));
+        $fragments = $email->getFragments();
+
+        // this should match two blocks, body and a signature
+        $this->assertCount(2, $fragments);
+        $this->assertFalse($fragments[0]->isSignature());
+        $this->assertTrue($fragments[1]->isSignature());
+    }
+
     /**
      * @dataProvider getDateFormats
      */

diff --git a/tests/Fixtures/email_ls-l.txt b/tests/Fixtures/email_ls-l.txt
@@ -0,0 +1,35 @@
+MIME-Version: 1.0
+Content-Type: text/plain; charset=utf-8; format=flowed
+Content-Transfer-Encoding: 7bit
+
+here's some funny one
+
+
+$ LC_ALL=C ls -l /tmp|grep sess|head
+-rw------- 1 http   http       62 Feb 15 12:45 
+sess_07ncrlhq50obbd5kp1vp02lp97
+-rw------- 1 http   http        0 Feb 15 10:18 
+sess_0g01akj9ccmq6r001p2klb55s0
+-rw------- 1 http   http        0 Feb 14 23:26 
+sess_0gcjo35c35f330p4qm31c1ovv1
+-rw------- 1 http   http     4410 Feb 15 12:32 
+sess_0i16be4lk5derhdfeas2uomnf4
+-rw------- 1 http   http      172 Feb 15 11:59 
+sess_0jv5f5i6eu7qfp41mc6hkfjpq3
+-rw------- 1 http   http        0 Feb 15 10:12 
+sess_0njep2fkt6v5j45t5r0hcfup77
+-rw------- 1 http   http        0 Feb 15 09:57 
+sess_1j44bltbjpkej984sfor5461u3
+-rw------- 1 http   http        0 Feb 14 23:37 
+sess_1r9r8a6kaqscq46psrcf1ssm24
+-rw------- 1 http   http        0 Feb 15 10:17 
+sess_25cvldfhk0nann15asctkrg3b2
+-rw------- 1 http   http       59 Feb 14 23:43 
+sess_25tni1suqgasqk8osnmk098sc6
+
+
+
+-- 
+glen
+
+