New posts

content/blog/blogroll-from-subscriptions.md

@@ -0,0 +1,58 @@
+---
+title: "Blogroll From Subscriptions"
+date: 2022-12-18T13:05:49-05:00
+draft: false
+tags: []
+math: false
+---
+
+While I was browsing around personal websites, I found a fun little piece of code from Jake Bauer's [links page](https://www.paritybit.ca/links). 
+
+```bash
+grep "xmlUrl" static/subscriptions.opml |\
+sed 's/.*text=\"\(.*\)\" xmlUrl=\"\(https\?:\/\/[^\/]*\/\)\(.*\)\" .*/<li><a href=\"\2\">\1<\/a> (<a href=\"\2\3\">feed<\/a>)<\/li>/g'
+```
+
+This takes the subscriptions exported from [yarr](https://github.com/nkanaev/yarr) and generates a HTML list which you can then include in a blogroll page.
+
+Running this script on my export from [Feedbin](https://feedbin.com/) yielded some extra metadata being shown in the HTML.  For example: `Joke Bauer type="rss" (Feed)`. So let's edit the code snippet above so that it works for my subscription export.
+
+From my `subscriptions.xml` here's an example entry:
+
+```xml
+<outline text="Jake Bauer" title="Jake Bauer" type="rss" xmlUrl="https://www.paritybit.ca/feed.xml" htmlUrl="https://www.paritybit.ca/"/>
+```
+
+It looks like I need to extract the `title`, `xmlUrl`, and `htmlUrl` attributes in that specific order. I'll use the same technique from a previous post on [capturing quoated strings](/capturing-quoted-string-sed).
+
+```bash
+grep "xmlUrl" subscriptions.xml |\
+sed 's/.*title=\"\([^\"]*\)\".*xmlUrl=\"\([^\"]*\)\".*htmlUrl=\"\([^\"]*\)\".*/<li><a href=\"\3\">\1<\/a> (<a href=\"\2\">feed<\/a>)<\/li>/g'
+```
+
+We can then clean this up into a `create_blogroll` script saved within [`~/.local/bin`](/blog/customexec/).
+
+```bash
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+show_usage() {
+    echo "Usage: create_blogroll [subscriptions.xml]"
+    exit 1
+}
+
+# Check argument count
+if [ "$#" -ne 1 ]; then
+    show_usage
+fi
+
+QUOTED_STR="\"\([^\"]*\)\""
+XML_EXPR=".*title=$QUOTED_STR.*xmlUrl=$QUOTED_STR.*htmlUrl=$QUOTED_STR.*"
+HTML_EXPR="<li><a href=\"\3\">\1<\/a> (<a href=\"\2\">feed<\/a>)<\/li>"
+REPLACE_EXPR="s/$XML_EXPR/$HTML_EXPR/g"
+
+grep "xmlUrl" "$1" | sed "$REPLACE_EXPR"
+```
+

content/blog/capturing-quoted-string-sed.md

@@ -0,0 +1,65 @@
+---
+title: "Capturing Quoted Strings in Sed"
+date: 2022-12-18T12:55:32-05:00
+draft: false
+tags: []
+math: false
+---
+
+*Disclaimer: This posts assumes some knowledge about regular expressions.*
+
+Recently I was trying to capture an HTML attribute in `sed`. For example, let's say I want to extract the `href` attribute in the following example:
+
+```
+<a href="https://brandonrozek.com" rel="me"></a>
+```
+
+Advice you commonly see on the Internet is to use a capture group for anything between the quotes of the href.
+
+In regular expression land, we can represent anything as `.*` and define a capture group of some regular expression `X` as `\(X\)`.
+
+```bash
+sed "s/.*href=\"\(.*\)\".*/\1/g"
+```
+
+What does this look like for our input?
+
+```bash
+echo \<a href=\"https://brandonrozek.com\" rel=\"me\"\>\</a\> |\
+sed "s/.*href=\"\(.*\)\".*/\1/g"
+```
+
+```
+https://brandonrozek.com" rel="me
+```
+
+It matches all the way until the second `"`! What we want, is to not match *any* character within the quotations, but match any character that is not the quotation itself `[^\"]*`
+
+```bash
+sed "s/.*href=\"\([^\"]*\)\".*/\1/g"
+```
+
+This then works for our example:
+
+```bash
+echo \<a href=\"https://brandonrozek.com\" rel=\"me\"\>\</a\> |\
+sed "s/.*href=\"\([^\"]*\)\".*/\1/g"
+```
+
+```
+https://brandonrozek.com
+```
+
+Within a bash script, we can make this a little more readable by using multiple variables.
+
+```bash
+QUOTED_STR="\"\([^\"]*\)\""
+BEFORE_TEXT=".*href=$QUOTED_STR.*"
+AFTER_TEXT="\1"
+REPLACE_EXPR="s/$BEFORE_TEXT/$AFTER_TEXT/g"
+
+INPUT="\<a href=\"https://brandonrozek.com\" rel=\"me\"\>\</a\>"
+
+echo "$INPUT" | sed "$REPLACE_EXPR"
+```
+