Tested toText() support

beberlei · May 9, 2010 · 0938d7f · 0938d7f
1 parent 2050537
commit 0938d7f
Show file tree

Hide file tree

Showing 5 changed files with 80 additions and 5 deletions.
diff --git a/README.markdown b/README.markdown
@@ -23,7 +23,7 @@ the jQuery API.
 
 PHPricot is motivated by Rubys HPricot but only a distantly related port.
 
-## Installation
+## Installation of Required Libraries
 
 This currently a linux only installation guide (please contribute Windows or MacOS if
 you find the time).

diff --git a/lib/PHPricot/Document.php b/lib/PHPricot/Document.php
@@ -35,6 +35,7 @@ public function toText()
         foreach ($this->childNodes AS $node) {
             $txt .= $node->toText();
         }
-        return $txt;
+        $txt = preg_replace('(([\s]{3,}))', '  ', $txt);
+        return rtrim($txt);
     }
 }
diff --git a/lib/PHPricot/Nodes/Element.php b/lib/PHPricot/Nodes/Element.php
@@ -142,6 +142,15 @@ public function val()
     public function toText()
     {
         $out = '';
+
+        if ($this->name == 'h1') {
+            $out .= "\n" . str_repeat('*', 65) . "\n";
+        } else if ($this->name == 'h2') {
+            $out .= "\n" . str_repeat('-', 65) . "\n";
+        } else if ($this->name == 'li') {
+            $out .= " * ";
+        }
+
         foreach ($this->childNodes AS $child) {
             $out .= $child->toText();
         }
@@ -150,10 +159,20 @@ public function toText()
             $out .= "\n";
         } else if ($this->name == 'p') {
             $out .= "\n\n";
-        } else if ($this->name == 'a' && isset($this->attributes['href'])) {
-            $out .= "[#" . $this->attributes['href'] . "]";
+        } else if ($this->name == 'a' && isset($this->attributes['href']) && strpos($this->attributes['href'], "#") !== 0) {
+            $out .= "(" . $this->attributes['href'] . ")";
         } else if ($this->name == 'img' && isset($this->attributes['src'])) {
             $out .= "[image:" . $this->attributes['src'] . "]";
+        } else if ($this->name == 'h1') {
+            $out .= "\n" . str_repeat('*', 65) . "\n\n";
+        } else if (in_array($this->name, array('h2', 'h3', 'h4', 'h5', 'h6'))) {
+            $out .= "\n" . str_repeat('-', 65) . "\n\n";
+        } else if ($this->name == 'td') {
+            $out .= "\t";
+        } else if ($this->name == 'li') {
+            $out .= "\n";
+        } else if ($this->name == 'ul') {
+            $out .= "\n";
         }
 
         return $out;

diff --git a/lib/PHPricot/Nodes/Text.php b/lib/PHPricot/Nodes/Text.php
@@ -28,6 +28,6 @@ public function toHtml()
 
     public function toText()
     {
-        return $this->text;
+        return str_replace("&nbsp;", " ", $this->text);
     }
 }
diff --git a/tests/PHPricot/ToTextTest.php b/tests/PHPricot/ToTextTest.php
@@ -0,0 +1,55 @@
+<?php
+
+class PHPricot_ToTextTest extends PHPUnit_Framework_TestCase
+{
+    public function testHeader1()
+    {
+        $query = new PHPricot_Query('<h1>Foo</h1><p>bar <a href="#foo">baz</a></p>');
+
+        $expected = <<<ETT
+
+*****************************************************************
+Foo
+*****************************************************************
+
+bar baz
+ETT;
+        $this->assertEquals($expected, $query->getDocument()->toText());
+    }
+
+    public function testHeader2()
+    {
+        $query = new PHPricot_Query('<h2>Foo</h2>');
+
+        $expected = <<<ETT
+
+-----------------------------------------------------------------
+Foo
+-----------------------------------------------------------------
+ETT;
+        $this->assertEquals($expected, $query->getDocument()->toText());
+    }
+
+
+    public function testHeader3()
+    {
+        $query = new PHPricot_Query('<h3>Foo</h3>');
+
+        $expected = <<<ETT
+Foo
+-----------------------------------------------------------------
+ETT;
+        $this->assertEquals($expected, $query->getDocument()->toText());
+    }
+
+    public function testList()
+    {
+        $query = new PHPricot_Query('<ul><li>Foo</li><li>Bar</li></ul>');
+
+        $expected = <<<ETT
+ * Foo
+ * Bar
+ETT;
+        $this->assertEquals($expected, $query->getDocument()->toText());
+    }
+}