lee/hpr-knowledge-base
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Initial commit: HPR Knowledge Base MCP Server

Browse Source 
- MCP server with stdio transport for local use
- Search episodes, transcripts, hosts, and series
- 4,511 episodes with metadata and transcripts
- Data loader with in-memory JSON storage

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Lee Hanken
2025-10-26 10:54:13 +00:00
commit 7c8efd2228
4494 changed files with 1705541 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

196
hpr_transcripts/hpr4407.txt Normal file
View File

@@ -0,0 +1,196 @@
Episode: 4407
Title: HPR4407: A 're-response' Bash script
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr4407/hpr4407.mp3
Transcribed: 2025-10-26 00:20:31
---
This is Hacker Public Radio Episode 4407 for Tuesday the 24th of June 2025.
Today's show is entitled, A Rear Response Bash Script.
It is hosted by Dave Morris and is about 13 minutes long.
It carries an explicit flight.
The summary is, my take on Ken's response to Kevin's show 4398.
Hello, this is Dave Morris, doing a show today entitled, A Rear Response Bash Script.
So what happened was today, although I actually wrote this few days ago, today Ken's show
came out 4404 and it was responding to Kevin's show 4398 and today is the 19th of June.
Kevin's show came out on the 11th of June.
Now, Kevin was using a bash pipeline in his show to find the latest episode in an RSS feed
and then to download it, it did a bit more, but this was the point really.
And he used GREP to parse the XML of the feed.
Ken's response to this was to suggest the use of the program XML style it to parse the XML
because such a complex structured format has XML.
You can't really parse with anything other than a program that understands, if you like,
the intricacies of the format structure.
And these sorts of programs that understand HTML, XML, YAML, JSON, etc. exist.
So using them is the wisest thing to do.
And his show Ken presented a bash script which dealt with this problem.
And it also dealt with the ordering of the episodes in the feed, which he wasn't quite clear
whether Kevin's example would do to what he was expecting it to do.
Kevin was expecting it to do an ordering stuff by date or by number or whatever.
Anyway, Ken attempted to achieve that as well in a rather clever way.
I did enjoy Ken's show a lot.
Actually, it was a nice way of taking a different look at a problem which I think was really warranted
because I made a reaction to listening to Kevin's show was just thinking,
hmm, don't think I'd do it quite that way.
Ken asked for any responses to his show, anybody had to find out how anybody else would write such a script.
So I thought that was a call to action on my part.
So I've put together an alternative script I'm talking about here.
My script is a remodeling of Ken's.
It's not really a completely different solution.
It just contains a few alternative ways of doing what Ken did.
And I'll reordering the parts of his original.
It doesn't do things in the same order as his scripted.
And I'll come onto that in a minute.
So I've presented the actual script in the notes.
It's not very long.
Ken's number lines, but it's not much longer than Ken's.
But I've changed it a fair bit and what I've done is I've put comments in it.
You can run it with the comments in and check.
They're numbered comments, so I'm just going to refer to each of them
to explain what I did.
So we start with comment number one.
The format of the pipeline in the script is different.
It starts by defining a while loop.
But the data which the read command in the while loop receives
comes from a process substitution.
That's that thing where you have a less than sign, or greater than sign,
but a less than sign in this case, with a parenthesized list of statements.
And I did a show on this subject, which included this subject a long time ago,
and I've made a reference to it in the notes.
I've arranged the pipeline this way because it's bad practice to place a while loop in a pipeline.
You can have it at the start of the pipeline, but don't put it in it,
because it creates a separate sub-process,
which can't communicate with the rest of the pipeline very well.
And again, I've written a show explaining why it's the bad idea, so I've referenced that.
I also added a hyphen R to the read, because I use shell check with VIM,
and it always nags me that I'll get unexpected results of a dog.
On to comment number two.
So the lines coming from the process substitution are from running curl, as candid,
and then, in this case, they're feeding the curl output, which is the XML feed,
into XML style it.
And that's then picking out the pub date field of the item,
so that's the date relating to the item in the feed, and the URL of the enclosure.
XML style it, thing does is to put those two fields together, separated by a semicolon.
So I just did the same thing as Ken's did.
And each of the lines coming from this process substitution fed into the read command in the while loop,
and then we pick apart the elements of it.
And this is done to get the first one, which is pub date, with the expression pub date equals,
and then in quotes, dollar, open, curly bracket, item, colon, semicolon, star,
close, curly bracket, close quotes.
What that does is it's a type of parameter manipulation, which again I did a show on,
show 1648 a long time ago, and there's a particular mode of operation,
which is titled remove matching suffix pattern for this from a parameter.
So what it's actually doing is it's returning the part of the string that's in item before,
not including the semicolon.
So we've gone to comment three, then pub date's got a date in it,
and Ken made the point that this is the very weird date format,
that the feed, XML feed, RSS feed specifies, which is a weird, weird format.
And he uses the date command to convert it into an ISO 8601 date in time.
I did this as well, but I changed the output format to a shorter form,
which is plus percent capital F, capital T, percent T.
So percent F means the date caution in ISO 8601, and percent T is the time format,
and the T in the middle is what you need as part of the full date time format.
That's just me showing it can be done, which I would do because I'm lazy.
Command 4 refers to the bit of code that gets the URL out of the item string,
and that uses another one of these parameter manipulation expressions,
which is, double quote.
So it's a good idea to quote these things,
so you probably don't have to, but it's wise to do so.
Double quote, double quote, dollar, open code bracket, item hash, hash sign,
number sign, asterisk semicolon, close bracket, close quote.
What that is doing is removing the matching prefix pattern.
Okay, so what it's doing is it says, give me everything,
it's removing everything up to the semicolon.
So it returns the URL, which follows the semicolon.
Then there's an echo, which just can kind of mix these two together,
and returns them as the output of the while loop,
which is then going to be fed into a pipe.
And what this one does is, rather than using a tab character between the two elements,
I said, just use the semicolon.
I don't think the tab character had any significance against original,
so I didn't use it.
And that's then fed to sort the sort command,
where it's sorted numerically, based on the time,
the date stamp, really, the time and date stamp.
So it just doesn't numeric sort.
It doesn't care that there are two parts to the line.
It just sorts by what it sees at the beginning.
If you had two absolutely identical time stamps,
then it would then sort on the URL.
But that's pretty unlikely, incredibly unlikely.
So it's fairly safe thing to do.
If you want to be really fussy about it,
you can use sorts field capability, but I didn't do that.
And sort does the sorting and reverses it in time order.
So the one at the top of the list, the newest one.
Now, I followed on with what Ken had done,
and used head iPhone 1 to get just the first one from that list.
So that's the sort, and we throw everything away.
But the first one, then I used cut rather than orc,
which can use, which is a thing where you can chop things up by field.
And the field delimiter is a semicolon,
and I just wanted to field number two.
In other words, the URL.
And then I piped that to WGET.
Now, I used a slightly different form of WGET from anybody else,
because it's at the end of the pipe.
And I told it that what WGET is actually getting the URL
on its standard in, in order to make it use that
as the input file to be searched.
It will collect it and download it, I should say.
So point number six is using this WGET,
which uses the option, iPhone, iPhone input,
iPhone file, equals.
And then an iPhone in that context means the standard in.
So that's what it does.
So it's actually pretty simple.
My changes are not massive.
It's just they've cleaned things up a little bit,
rather than calling several programs to manipulate bits of data.
I use BASH's inbuilt manipulation tools,
which I prefer, and they're likely to be a bit more efficient.
Not significantly, this is not going to cause
tremendous difference in the time things,
but just because I like to do things more tidally,
having come from a mainframe background
where every CPU cycle counted.
I want to show my solutions better in a significant way.
As I said, I like to use the BASH functionality.
The other thing I thought, as I was assembling these notes,
was that instead of using head and cut,
I would have attempted, if I thought more about it,
at the time, to use said.
And I've shown how I could use said instead of the head and cut.
So it's a said expression, which is,
you know, pretty much always in single quotes,
one, then open curly bracket.
S slash carriage, let's see, up arrow, we think,
dot backslash plus semicolon slash slash semicolon,
Q, close curly brackets.
What that means is, operate only on line one,
so that's the line we care about.
And then when you've on line one, replace the characters that begin
from the start of the line, that's what a carrot means,
dot backslash plus means any other characters that happen to be,
and we expect there to be at least one,
so we'll use the plus, then, and the semicolon,
and replace them by nothing.
And then once you've done that, quit from said.
So that does the equivalent of the head and the cut,
and then that feeds into double-getness before.
So yeah, I was going to make reference to my said series
from many years ago, I thought, wow, these people will be.
So I've fed up looking back at Dave's old shows that I won't bother.
But I've referenced all the various things that I've mentioned
and linked to in the notes, as a link section at the end.
Okay, that's it, and thanks for watching listening.
Okay, bye.
You have been listening to Hacker Public Radio
at Hacker Public Radio, does work.
Today's show was contributed by a HBR listener like yourself.
If you ever thought of recording podcasts,
you click on our contribute link to find out how easy it really is.
Hosting for HBR has been kindly provided by
an honesthost.com, the internet archive, and our sings.net.
On the Sadois status, today's show is released
under Creative Commons, Attribution 4.0 International License.