trying to fix

var/elfeed/db/data/a8/a8272751619566f2f5cdb353b226bae7231689bc

@@ -0,0 +1,85 @@
+
+         
+         <p>Earlier today I wanted to convert a bunch of Org meta data keywords to
+lower case notation, with the help of <code>isearch-forward-regexp</code> and
+<code>query-replace-regexp</code>.  This included everything from special comments,
+like <code>#+TITLE:</code> to property drawers in the form of <code>:PROPERTIES:</code>.
+There were 340 such cases in my file and I was reluctant to do so
+manually.  Thankfully, Emacs makes such a task fairly simple once you
+get past the essentials, because it lets you evaluate arbitrary Elisp
+forms for the replacement text.  This includes the possibility of
+running some function on a matched regexp group.  In this case that
+function was <code>(downcase OBJECT)</code>.</p>
+
+<p>First I needed a regular expression that would capture all targets.
+<code>isearch-forward-regexp</code> and its <code>query-replace-regexp</code> counterpart
+expect a single backslash for the escape character, so I ended up with
+this pattern:</p>
+
+<pre><code>^\(#\|:\)[^ ].*?:
+</code></pre>
+
+<p>If I were to test this in the buffer, I could use <code>M-x re-builder</code>,
+which however requires double backslashes (same for when you write Elisp
+code):</p>
+
+<pre><code>^\\(#\\|:\\)[^ ].*?:
+</code></pre>
+
+<ul>
+  <li>
+    <p>Once I was sure of the regular expression I had to use, I went to the
+top of the buffer and invoked <code>isearch-forward-regexp</code> (bound to
+<code>C-M-s</code> by default).</p>
+  </li>
+  <li>
+    <p>Entered the pattern <code>^\(#\|:\)[^ ].*?:</code>, got live feedback of the
+matching items, and switched to <code>query-replace</code> with <code>C-%</code>
+(<code>isearch-query-replace</code>).  When the Isearch is regexp-aware so is the
+corresponding <code>query-replace</code>.</p>
+  </li>
+  <li>
+    <p>For the replacement text I instructed the command to evaluate the
+<code>downcase</code> function.  This is done by escaping the comma operator
+(<code>,</code>) and then supplying the function with a regexp group.  Because I
+wished to match everything, the group should be <code>\0</code>.  Which means
+that the replacement should be expressed thus: <code>\,(downcase \0)</code>.</p>
+  </li>
+</ul>
+
+<p>Try this with any string in a buffer, say, <code>Hello</code>:</p>
+
+<ul>
+  <li><code>M-x query-replace-regexp</code></li>
+  <li><code>Hello</code></li>
+  <li><code>\,(upcase \0)</code></li>
+</ul>
+
+<p>Confirm the operation and you should get <code>HELLO</code>.  This works with
+multiple groups and can read a series of Elisp forms.  An example with
+<code>Hello world</code>:</p>
+
+<ul>
+  <li><code>M-x query-replace-regexp</code></li>
+  <li><code>\(Hello\) \(world\)</code></li>
+  <li><code>\,(downcase \1) \,(capitalize \2)</code></li>
+</ul>
+
+<p>Which should give you <code>hello World</code>.</p>
+
+<p>The elegant minimalism of the <code>query-replace</code> interface grants you the
+power to either replace each match one at a time or hit the exclamation
+mark (<code>!</code>) to answer “yes to all”.  This is what I did to downcase all
+340 matches.  Voila!  All of my Org file’s meta data were converted to
+lower case in one go.</p>
+
+<p>Finally, I discovered <code>query-replace-regexp-eval</code> which saves you from
+adding the escaped comma operator for the replacement’s Elisp form.
+However its doc string reads thus:</p>
+
+<pre><code>Interactive use of this function is deprecated in favor of the
+‘\,’ feature of ‘query-replace-regexp’.  For non-interactive use, a loop
+using ‘search-forward-regexp’ and ‘replace-match’ is preferred.
+</code></pre>
+        
+