Added built_post_subject for stripping tags and Re's added by mailing lists.

2 files changed, 63 insertions, 0 deletions

diff --git a/src/build_post.php b/src/build_post.php
new file mode 100644
index 0000000..6d95be1
--- /dev/null
+++ b/src/build_post.php
@@ -0,0 +1,29 @@
+<?php
+
+function build_tag_pattern($tag) {
+  return '/' . preg_quote($tag, '/') . '\\s*/';
+}
+
+function build_post_subject($listtag, $forumtag, $subject) {
+  // strip the '[list]' and '[forum]' tags
+  $tagpat = '/(' . preg_quote($listtag, '/') .
+             '|' . preg_quote($forumtag, '/') . ')\\s*/';
+  $subject = preg_replace($tagpat, '', $subject);
+
+  // strip leading sequences of Re-equivalents
+  if (preg_match(
+    '/^((RE|AW|SV|VS)(\\[\\d+\\])?:\\s*)+/i',
+    $subject, $m, PREG_OFFSET_CAPTURE
+  )) {
+    $subject = substr($subject, $m[0][1]);
+  }
+
+  // ensure nonempty subject
+  if (trim($subject) == '') {
+    $subject = '(no subject)';
+  }
+
+  return $subject; 
+}
+
+?>
diff --git a/test/build_post_test.php b/test/build_post_test.php
new file mode 100644
index 0000000..72e28a7
--- /dev/null
+++ b/test/build_post_test.php
@@ -0,0 +1,34 @@
+<?php
+
+require_once('PHPUnit/Framework.php');
+
+require_once(__DIR__ . '/../src/build_post.php');
+
+class build_post_test extends PHPUnit_Framework_TestCase {
+  
+  /* @dataProvider buildPostSubjectProvider */
+  public function testBuildPostSubject($ltag, $ftag, $subject, $expected) {
+    $this->assertEquals(
+      build_post_subject($listtag, $forumtag, $subject),
+      $expected
+    );
+  }
+
+  public function builtPostSubjectProvider() {
+    return array(
+      array('[l]', '[f]', '', '(no subject)'),
+      array('[l]', '[f]', 'Re:', '(no subject)'),
+      array('[l]', '[f]', 'Subject', 'Subject'),
+      array('[l]', '[f]', 'Re: Subject', 'Subject'),
+      array('[l]', '[f]', 'Re: Re: Re: Subject', 'Subject'),
+      array('[l]', '[f]', '[f] Subject', 'Subject'),
+      array('[l]', '[f]', '[f] [f] Subject', 'Subject'),
+      array('[l]', '[f]', '[f] [f] Subject [f]', 'Subject'),
+      array('[l]', '[f]', '[l] [f] Re: Subject', 'Subject'),
+      array('[l]', '[f]', 'Re: [l] [f] Subject', 'Subject'),
+      array('[l]', '[f]', 'Re: Subject [l][f] Subject', 'Subject Subject')
+    );
+  }
+}
+
+?>