From c936c09db93df97412a24e550a2cecdc1aad0f3b Mon Sep 17 00:00:00 2001
From: Sanders Kleinfeld <sanders@oreilly.com>
Date: Tue, 26 Aug 2014 09:47:15 -0400
Subject: [PATCH] More robust trim-url function that strips initial "www."
 (without preceding "http://") and trailing forward slashes.

---
 htmlbook-xsl/xrefgen.xsl         |  43 ++++++++---
 htmlbook-xsl/xspec/xrefgen.xspec | 127 ++++++++++++++++++++++++++++++-
 2 files changed, 156 insertions(+), 14 deletions(-)

diff --git a/htmlbook-xsl/xrefgen.xsl b/htmlbook-xsl/xrefgen.xsl
index f0d2b9b6..ffb72779 100755
--- a/htmlbook-xsl/xrefgen.xsl
+++ b/htmlbook-xsl/xrefgen.xsl
@@ -619,17 +619,38 @@
   <!-- Template to trim http:// and http://www from URLs -->
   <xsl:template name="trim-url">
     <xsl:param name="url-to-trim"/>
-    <xsl:choose>
-      <xsl:when test="contains($url-to-trim, 'http://www.')">
-	<xsl:value-of select="substring-after($url-to-trim, 'http://www.')"/>
-      </xsl:when>
-      <xsl:when test="contains($url-to-trim, 'http://')">
-	<xsl:value-of select="substring-after($url-to-trim, 'http://')"/>
-      </xsl:when>
-      <xsl:otherwise>
-	<xsl:value-of select="$url-to-trim"/>
-      </xsl:otherwise>
-    </xsl:choose>
+
+    <!-- First, trim http://www., http://, or www. prefixes -->
+    <xsl:variable name="prefix-trimmed">
+      <xsl:choose>
+	<xsl:when test="contains($url-to-trim, 'http://www.') and substring-before($url-to-trim, 'http://www.') = ''">
+	  <xsl:value-of select="substring-after($url-to-trim, 'http://www.')"/>
+	</xsl:when>
+	<xsl:when test="contains($url-to-trim, 'http://') and substring-before($url-to-trim, 'http://') = ''">
+	  <xsl:value-of select="substring-after($url-to-trim, 'http://')"/>
+	</xsl:when>
+	<xsl:when test="contains($url-to-trim, 'www.') and substring-before($url-to-trim, 'www.') = ''">
+	  <xsl:value-of select="substring-after($url-to-trim, 'www.')"/>
+	</xsl:when>
+	<xsl:otherwise>
+	  <xsl:value-of select="$url-to-trim"/>
+	</xsl:otherwise>
+      </xsl:choose>
+    </xsl:variable>
+
+    <!-- Then trim trailing forward slashes -->
+    <xsl:variable name="suffix-trimmed">
+      <xsl:choose>
+	<xsl:when test="substring($prefix-trimmed, string-length($prefix-trimmed), 1) = '/'">
+	  <xsl:value-of select="substring($prefix-trimmed, 1, string-length($prefix-trimmed) - 1)"/>
+	</xsl:when>
+	<xsl:otherwise>
+	  <xsl:value-of select="$prefix-trimmed"/>
+	</xsl:otherwise>
+      </xsl:choose>
+    </xsl:variable>
+    
+    <xsl:value-of select="$suffix-trimmed"/>
   </xsl:template>
 
 </xsl:stylesheet> 
diff --git a/htmlbook-xsl/xspec/xrefgen.xspec b/htmlbook-xsl/xspec/xrefgen.xspec
index 698b1619..944523a7 100755
--- a/htmlbook-xsl/xspec/xrefgen.xspec
+++ b/htmlbook-xsl/xspec/xrefgen.xspec
@@ -708,6 +708,11 @@
         <p><a id="hyperlink_same_text_node" href="http://oreilly.com">http://oreilly.com</a></p>
         <p><a id="hyperlink_same_text_node_2" href="http://oreilly.com">oreilly.com</a></p>
         <p><a id="hyperlink_same_text_node_3" href="http://www.oreilly.com">oreilly.com</a></p>
+	<p><a id="hyperlink_same_text_node_4" href="http://www.oreilly.com">www.oreilly.com</a></p>
+        <p><a id="hyperlink_same_text_node_5" href="http://oreilly.com/">http://oreilly.com</a></p>
+        <p><a id="hyperlink_same_text_node_6" href="http://oreilly.com/">oreilly.com</a></p>
+        <p><a id="hyperlink_same_text_node_7" href="http://www.oreilly.com/">oreilly.com</a></p>
+	<p><a id="hyperlink_same_text_node_8" href="http://www.oreilly.com/">www.oreilly.com</a></p>
         <p><a id="link" data-type="link" href="#some_id">some awesome location in the text</a></p>
 	<p><a href="#another_id" id="stealth_xref">XREF masquerading as an external hyperlink (no proper data-type)</a></p>
         <p><a id="email_me" href="mailto:tools@oreilly.com">tools@oreilly.com</a></p>
@@ -745,6 +750,46 @@
       
       <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
     </x:scenario> 
+
+    <x:scenario label="with url.in.parens disabled (hyperlink same text node as href 4)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_4']">
+        <x:param name="url.in.parens" select="0"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
+
+    <x:scenario label="with url.in.parens disabled (hyperlink same text node as href 5)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_5']">
+        <x:param name="url.in.parens" select="0"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
+
+    <x:scenario label="with url.in.parens disabled (hyperlink same text node as href 6)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_6']">
+        <x:param name="url.in.parens" select="0"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
+
+    <x:scenario label="with url.in.parens disabled (hyperlink same text node as href 7)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_7']">
+        <x:param name="url.in.parens" select="0"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
+
+    <x:scenario label="with url.in.parens disabled (hyperlink same text node as href 8)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_8']">
+        <x:param name="url.in.parens" select="0"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
     
     <x:scenario label="with url.in.parens disabled (internal link)">
       <x:context select="//h:a[@id='link']">
@@ -801,6 +846,46 @@
         <x:param name="url.in.parens" select="1"/>
       </x:context>
       
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario>
+
+    <x:scenario label="with url.in.parens enabled (hyperlink same text node as href 4)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_4']">
+        <x:param name="url.in.parens" select="1"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario> 
+
+    <x:scenario label="with url.in.parens enabled (hyperlink same text node as href 5)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_5']">
+        <x:param name="url.in.parens" select="1"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario> 
+
+    <x:scenario label="with url.in.parens enabled (hyperlink same text node as href 6)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_6']">
+        <x:param name="url.in.parens" select="1"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario> 
+
+    <x:scenario label="with url.in.parens enabled (hyperlink same text node as href 7)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_7']">
+        <x:param name="url.in.parens" select="1"/>
+      </x:context>
+      
+      <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
+    </x:scenario> 
+
+    <x:scenario label="with url.in.parens enabled (hyperlink same text node as href 8)">
+      <x:context select="//h:a[@id='hyperlink_same_text_node_8']">
+        <x:param name="url.in.parens" select="1"/>
+      </x:context>
+      
       <x:expect label="No URL should be generated in parentheses after the 'a'" test="not(exists(//h:span[@class='print_url_in_parens']))"/>
     </x:scenario> 
     
@@ -1033,18 +1118,54 @@ sidebar
     </x:call>
     <x:expect label="Trimmed URL should be returned without http://">oreilly.com</x:expect>
   </x:scenario>
+
+  <x:scenario label="If trim-url is called on URL text starting with www.">
+    <x:call template="trim-url">
+      <x:param name="url-to-trim">www.oreilly.com</x:param>
+    </x:call>
+    <x:expect label="Trimmed URL should be returned without http://">oreilly.com</x:expect>
+  </x:scenario>
   
-  <x:scenario label="If trim-url is called on URL text starting with http://www">
+  <x:scenario label="If trim-url is called on URL text starting with http://www.">
     <x:call template="trim-url">
       <x:param name="url-to-trim">http://www.oreilly.com</x:param>
     </x:call>
-    <x:expect label="Trimmed URL should be returned without http://">oreilly.com</x:expect>
+    <x:expect label="Trimmed URL should be returned without http://www.">oreilly.com</x:expect>
+  </x:scenario>
+
+  <x:scenario label="If trim-url is called on URL text that ends in a forward slash">
+    <x:call template="trim-url">
+      <x:param name="url-to-trim">oreilly.com/</x:param>
+    </x:call>
+    <x:expect label="Trimmed URL should be returned without ending forward slash">oreilly.com</x:expect>
+  </x:scenario>
+
+  <x:scenario label="If trim-url is called on URL text starting with http:// and ending with a forward slash">
+    <x:call template="trim-url">
+      <x:param name="url-to-trim">http://oreilly.com/</x:param>
+    </x:call>
+    <x:expect label="Trimmed URL should be returned without http:// and trailing forward slash">oreilly.com</x:expect>
+  </x:scenario>
+
+  <x:scenario label="If trim-url is called on URL text starting with www. and ending with a forward slash">
+    <x:call template="trim-url">
+      <x:param name="url-to-trim">www.oreilly.com/</x:param>
+    </x:call>
+    <x:expect label="Trimmed URL should be returned without http:// and trailing forward slash">oreilly.com</x:expect>
   </x:scenario>
   
-  <x:scenario label="If trim-url is called on URL text that does not start with http://">
+  <x:scenario label="If trim-url is called on URL text starting with http://www. and ending with a forward slash">
+    <x:call template="trim-url">
+      <x:param name="url-to-trim">http://www.oreilly.com/</x:param>
+    </x:call>
+    <x:expect label="Trimmed URL should be returned without http://www. and trailing forward slash">oreilly.com</x:expect>
+  </x:scenario>
+  
+  <x:scenario label="If trim-url is called on URL text that does not start with http:// or www, or end with a forward slash.">
     <x:call template="trim-url">
       <x:param name="url-to-trim">oreilly.com</x:param>
     </x:call>
     <x:expect label="Trimmed URL should be returned as is">oreilly.com</x:expect>
   </x:scenario>
+
 </x:description>