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

348
hpr_transcripts/hpr1952.txt Normal file
View File

@@ -0,0 +1,348 @@
Episode: 1952
Title: HPR1952: Time now Ladies and Gents
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr1952/hpr1952.mp3
Transcribed: 2025-10-18 11:47:28
---
This is HPR Episode 1952 entitled, Time Hour Aids and Gents, and in part of the series,
Bash-Cripting.
It is hosted by Ken Fallon and is about 31 minutes long, the summary is how to get the
total duration of a lot of media files.
This episode of HPR is brought to you by An Honesthost.com.
Get 15% discount on all shared hosting with the offer code HPR15, that's HPR15.
Get your web hosting that's Honest and Fair at An Honesthost.com.
Hi everybody, my name is Ken Fallon and you're listening to another episode of Hacker
Public Radio.
I've been just listening to HPR1943, the HPR audiobook club 11.3, where our very own
David Colons Rivera, aka Lost and Bruns, was interviewed about his audiobook, a very,
very riveting episode and I had to pause a few times to make a few notes myself, but
one of the comments that got me to do this show was the comments that Polky made, was
there a way to get the duration for all the media in a directory, for example, if they
download one file, then to be able to guess the duration of that file is relatively easy,
but to be able to get the duration of all the episodes in an audiobook, especially if
it's serialized.
Well, this is something I tend to do quite a lot, given the nature of my professional
and private life here on HPR.
So there's three ways that springs to mind immediately about that, to do it.
One is our customer tool here on Hacker Public Radio that we use, called FixTags, which
gives you the length in ISO 801 human readable format and I'll make comments about that
as we go along.
So that's basically two digits for the hour, two digits for the minute, two digits for
the second.
That's my personal favorite format for use in the world and also that would be UTC.
One thing Dave added for me was that he also converts that to seconds.
Now I'm going to be rather unfair here on Dave because his application doesn't know
a lot more than just read that out, but it is such a handy tool that I use that quite
a lot myself, so much so that I've written a blog post about that very self-same thing.
So if you look in the show notes, I've done extensive show notes in this, which I did
in Markdown everybody and converted to HTML.
The syntax is relatively simple, you just do Fix underscore tags space and the file I'm
looking at is intro.flac.
So what I'm going to be doing is I'll do some examples with the intro.flac file and
then four other flac files that we keep in the processing directory, the intro, the outro,
the promo for non-host host and the promo for archive.org, so those four just to give
you a comparison.
So the output of filter tags gives you things like Alba Martis, John Lenth and this Lenth
is the one that we're interested in and Dave has a colon to separate the field names
from the values.
And then he has a nice old date, in this case it's 0, 0 colon, 0, 0 colon, 3, 9 space, open
bracket, 3, 9 space, SEC, close bracket.
Now we've got a better problem here because I want to filter out the 39 seconds from the
bit between the bracket and I need to get rid of the sec that string space SEC.
And I'm going to do that is Fix tags, a strict stuff, flac.
And then we're going to pipe it into arc and arc is this really cool arc.
I would love people to do some shows on it, but it's got this really cool feature, two
of which I'm using here and that is to allow two different delimiters and you can use
specified delimitor with the capital F sign.
So if you've got a comma separated file, X dot text, for instance, that you can go
arc space, F single quote, comma single quote, and then print, blah.
In this case, I'm going to use the open and closed brackets as the delimiters and they're
going to print dollar two, but arc also allows you to do filtering.
So I'm using the filter to grip, you know, air quotes here, grip for the world length.
So the entire command is arc dash capital F, single quote, back slash back slash open
parentheses, and that's to escape the parentheses, the pipe to show that there's going to be two
delimiters, back slash back, slash closed parentheses, a closed single quote, space open single
quote.
And this is the bit now where we're going to filter on the length forward slash length,
forward slash space, open curly bracket, print space dollar two, close curly brackets,
close single quote, that's going to print the length and it'll filter just the length
and then that particular line, it'll only print off the thing that is between open and
closed brace.
In this case, it'll return three, nine space SEC.
So I'm just going to pipe that into said space, single quote, S, forward slash space SEC,
back slash back slash G, single quote.
And what that does is it will search for anything that's piped into it.
And if it sees the string space SEC, it'll replace it with basically nothing.
So what gives me if I run the command for those four files is a list, 39 new line, 16
new line, 17 new line and 11.
That's all very well, but it's not really very handy for me.
So what I want to actually use is the BC command.
I could have actually piped those into variables and said let count equals 39 and then add
on the next iteration add 60 to 39 and then add 17 to the sum and add 11 to the sum
of those.
But BC is a command that allows you to do mathematics here.
And I think Charles and J cover this before as well.
So more episodes than that would be also.
So what I want to do is pipe that into a BC and the normal way you can add two numbers,
say one, one plus one, you would type echo space one plus one.
And then the pipe character and BC and BC would return to as a result.
So the first thing I want to do is convert these lines one on top of the other onto one
single line.
And the best way I found to do that was to use command substitution, thank you Dave Morris,
who just replied to an email and he covered that in the show 1903, which is basically
there's two ways to do this back ticks, but I prefer to use the dollar open braces and
close brace way of doing it because you can do those nested for a start and you can also
do anything that goes inside of there, you can put your quotes and stuff.
So if you've got double quotes on the outside, you can continue tag quotes on the inside.
So it's absolutely awesome.
So the entire command I had before, the fixed acts, yadda yadda yadda up to the end of
said, I just enclose all of that in a dollar bracket in front of the echo out.
So that's what it says, whatever is output from this command, you just echo it out.
And that gets rid of the new lines, a really neat way of doing that.
And then I pipe the output of that again into said and they're called s forward slash
space, forward slash plus forward slash g sync close quote.
And that replaces any space with the literal character plus normally when you see a plus
and these sort of things, it means something else.
But in this case, I'm just having that.
And then I wrap all of that again in another dollar at brace command substitution thing.
And I echo that again into the BC command, actually, no, I'm not.
I'm just piping the remainder of that into the BC command, so there's no more.
So the number I get is one, two, seven.
Now that's the number of seconds.
And what I actually want to do that is make that human readable because it's nice to have
it in seconds for calculation, but I actually want to be able to, you know, Poki wants
to be able to look at this and see how long the show is.
And it's a lot easier to say it's two hours, 43 minutes, eight seconds rather than give
a big number of seconds.
So there's several ways of doing this.
And the best way I found, I actually wrote a program myself, but I found that the data
command that comes with bash is more than capable of doing the majority of calculations
of this type.
You can use the dash D argument, which displays time, string, defined by the string.
So you push dash D and you post last Tuesday or dash D and now you're so it's just going
to four days.
It's very intelligent.
Instead of what it normally does is it, you know, if you type date, it'll give you
whatever it is now.
But if you do a dash D, followed by the at sign, then it will convert it into the number
of seconds since the epoch, which everybody knows was the first of the first 1970 in UTC.
I don't like to convert that to a date.
So if you do that conversion, don't forget to use the dash U, which is the universal
coordinated time because it's actually going to add the numbers.
If you don't put the dash U in there, it'll not add the, it'll add your offset.
So I'm plus one at the moment and it will add an extra hour to my calculation, which
is something that I want.
And with date, you have a wide array of options that you can specify with a percent command.
So in this case, I'm going to be using percent capital T, which formats to a ISO 801 time format,
which is the equivalent of percent H for hours, colon, percent M for minutes, colon,
percent S for seconds, all of those are in capital.
And if you do a mandate, get it.
If you do a mandate, you can get all of this is very short, man page and perfectly readable.
So in the following command, I do put a forward slash in front of the day command.
And that's because I've a list, earliest they command days to be specific to my system.
I have it, when I do days, I have four masses the way I like it, oh, excuse me, I have
a forward format it, the way that I like it.
So the entire command is forwards slash date to on the aliases and use the default date,
space dash U for universal time, D for to use date from a argument, rather than use the
system date as in now, the at sign to tell you that there's going to be a number provided,
which is going to be the number of seconds since 1970, first of January 1970.
And of course in there, I'm just putting in copying and pasting that entire string, the
echo dollar thingy fixed brackets all the way over to BC, where I get that number, I
close that again, space plus and percent T, this will be in the show notes.
And basically what that does is it converts one, two, seven seconds into two minutes and
seven seconds, no surprise there.
So as a little bit of a check, what I did was I ran the time command, which I don't
work day in time, I know, I know, I know, I know, but the time command is actually something
that you can put in front of any other command and it will give you the real, the user and
the system time, which just gives you an idea of what's going on.
You got to be careful using it because if you're using the comparison for accessing
files, for instance, the first time you run the file will be read into memory and kept
in memory.
We've got a lot of cache memory, Linux will do this by default.
So that the second time you run it'll look a lot faster and if you're comparing it against
others, so you just keep that in mind and you might need to offload that.
So the exact same command again that I did before, I just put days at time in front of
the slash date part.
Instead of fix tags, asterix flag, I have fix tags, asterix mp3 and asterix.org.
And that returned three hours, nine minutes and forty nine seconds of files that are
waiting in the queue to be read.
And that actually took two point nine seconds real to basically process, about three seconds
to run.
Now, so that's option number one, let's move on to option number two.
And this is a command that I use, two of these commands I use every single day, hundreds
of times a day, probably the order of mice, the commands that I use will be grep, then
media info and then XML starless or sorry XML starless and then quite often I use media
info as well.
It is a utility that's available for basically all operation systems and the even
supply it as a library.
And you can install it by, it's in all the repels.
And if you do that on the media info flag, what it does is give you a general section and
a audio section that is based on the, on the fact that the mp3 file is a mpeg standard
and it's got different, different streams inside.
I knew I'm not going to go into this, that's covered before, maybe I'll cover it again.
So they, so just as a comparison, it gives you the complete file name, the format is
flak, it's the format info is free loss, audio, codec, the file size, it gives the duration,
which is 39s, space 50ms, so 39.39 seconds, 50 milliseconds, and it's a variable bit
rate of 701 and the album, the soccer, proper, radio track name, genre genre.
And another thing it does is it gives you the audio track, so this is the general track
and then the audio track, and you might think that's a bit odd.
But what you can have in some, in some files, especially in a mpeg file, in a video file,
you might have the audio track a shorter, then the overall length of the full track.
So there's an introduction and you don't have an audio track, so the audio track only
starts two minutes into the movie when all the studio flitzing things are on and gone.
So that's normal, you wouldn't normally see that, but it is something that is possible
to happen.
So we've got a few issues here with this standard duration and the first one is the fact
that it's human readable in so far as it says 39s and 50ms, and that can go up to hours
and I have no idea if I could go today or whatever, so we have no way of actually knowing
what's going to happen there and then it's going to be a nightmare to sort of deal with that.
It's a lot easier to do calculations, you know, if you're just dealing with seconds.
The other issue with that is that we've got the same feel name described in the general
and audio section, audio sections, so for a given file there will be two entries where
there's a duration.
So one way to fix the first issue, to get rid of the human readable and give a second
at least, well actually it gives you milliseconds, you can use the dash dash full option, so
media info, space dash dash full, space intro.flack and it gives you so much information there,
I just grept in the example and the show notes on the duration and it gives something
10 or so different durations, all go space as a colon as a as a separator, and then our
the first option is 39050, then the second is 390s, space 50s, that's repeated three times,
and then we finally got the ISO, it's 601 format, 0, 0, 0, colon 39.050 milliseconds,
and then that's repeated twice because obviously you have the general section and you've got the
audio section, so that's actually a bit of a pain. So although there are a variety of all,
it's pretty not, ISO it's 601 in this case, it's useful for displaying it to humans as
a less disambiguous, it's not very nice for us because we want to do calculations as I explained
before, and the duplicate things are a little bit of a concern as well because what you could do,
what you might be tempted to do there is, okay, it's okay, for one file you sort it and you do a
unique, so either sort space dash u or sort pipe unique, and that will get rid of the duplicates
of unit 1, you got 1, and then you can eliminate the s and the ms and the colon, and then you just left
with the single one that has got seconds on it, that's fine, but what happens if you've got,
if you're dealing with all files, like I don't know, files that are deliberately supposed to be
exactly, and now we're on the button, you know, you've got two files exactly the same size,
deliberately padded to be that size, then when you do a sort and unique them, you've suddenly
an hour short because you haven't taken those ones into account. So to get around ZS,
MediaInfo has got the dash dash output equals XML, where the O in output is capital,
capitalized, and the XML is also in capitals, and what that does is it gives you
it as an element tree, so as an XML file, ideally what I would have liked there was that the
duration would also have an element say, duration, milliseconds, duration, humor, you know,
format one, duration, format two, format equals one, format equals two, or something like that,
or that there was at least an idea, not that you could count on, but that is sadly not the case.
So now we have an XML file, and what I'm doing with that is I pipe it into my old trusty friend
XML starlet, which is something, yes, I need to do some shows on that because it's a brilliant,
brilliant tool for working with XML files, and what it does is I'm going to use it
basically as a way of gripping XML files, and you should never grip XML files. Every time you
grip an XML file, a cute furry kitten dies, so bear that in mind. A colleague of mine told me
that one day I cracked up, but it's quite true because XML is completely independent. You could have
you know, 15 carriage returns and spaces, and you get a squash shop or mixed round or the
elements and the attributes could be all over the shop. So it's not very wise to, you basically
can trust gripping an XML file, unless you want to see does it occur in that file, which I do from
time to time. But XML starlet is quite a nice tool. I'm just going to go briefly into what the
options are. XML starlet space select that turns it into selection mode, so I'm looking for a value,
I'm looking, I'm I'm going to ask it for a value based on the next path expression, next path
expressions. So if you're similar to this, you know, CD dot CD dot dot CD backslash, that sort of
thing. It allows you to do that with an XML file. So the first one is a dash t, which from my memory
means use text space dash t lowercase. So the first was an uppercase lowercase. It says I'm going
to build a template. And the template you always have a template and you need to find me this
and give me that. So the find me this part is the dash m match. And then basically it's medium for
slash file slash track square bracket attribute and the outside type equals audio, close square bracket
forward slash duration, open square bracket, the number one close square bracket. Single quote.
So what that does is if you look at the the the the dumb the object model of that file,
you'll see that it's a it starts with XML medium for. So it tells you what it is. Then there's a file,
you know, this particular file happens to be this. And then we're looking at a particular track.
And this in the square brackets, we're asking for the track of type audio as opposed to the track
of type general, which was up earlier. And then of that, I want the duration and just give me the
first duration. So it treats it like an array. And then we do a space dash fee and the dot,
which just shows me, yes, whatever that value is short, the dash in shows new line and the dash
would pipe it to standard out with her. I don't actually need the dash in there because,
but I put it in a force habit. That's wonderful except for the fact that medium for gives me
things a milliseconds. The reason that does that is because it's a it's a professional level tool,
I guess. And it there are 25 frames in a second. So we need to be able to break it down to that.
There can even be higher. But like the 25 24 frames per second. So you need that that's a amount
of granularity on your frame rate. So to get rid of that, a cool trick I just found on
Dr. Go recently was to put it to cut off the last three characters of the file of that output,
which in this case, I know will always be the millisecond component because I'm not interested in
that. And the command for that is said single quote s. So search for forward slash dot. So any
character and then keep going until and you can need to read this back to front. So it's escaped
curly bracket. So slash curly bracket three slash curly bracket and then the dollar. So that
says come back. Find me anything. Go to the end and then come back three characters and slash slash
means replace it with whatever's between the slashes, which is nothing. And then it replaces that.
And I of course get 39 60 17 and 11. And I do exactly the same thing again, replacing the
spaces with the plus and I get two minutes and seven seconds again. And when I run that against the
the list of podcasts that I have, I get three hours, nine seconds and 49, 49 seconds as a total,
which is exactly the same as the one we got with fixed tax earlier. The difference with this,
of course, is that it is done in point six of a second as opposed to the two three seconds
that fixed tags took Dave, of course, is now yelling as the microphone. And again,
media info and XML starlet are designed for this sort of stuff. So there may be faster for that
reason. I'm not 100% sure. I might be just doing this incorrectly. I want to give you one more
option because fixed tags is definitely not going to be installed in your system unless you have
happened to be working for HPR. And it's quite quite a few difficult to install libraries, I think,
would be a safe way of putting it. And I think Dave be in agreement with me on that one.
FFM, sorry, media info and XML starlet are probably in your repositories, but you may not have
them installed on your system. Quite a lot of systems do come though with FFM peg installed.
And one of the tools as part of that is a tool called FF probe, which as far as I can tell it's
it's the same actual binary that's run. And you can run that. And if you run FF probe space
dashi to say this is the input file intro.flacking in our case, it'll print off a nice details list.
A lot of just information about the program itself and then at the bottom it'll give you some
information about the track, not a whole lot, but a little bit. It happens to send that to standard
error. FFM peg and FF probe have weird ways of, or have non-standard ways of dealing with standard
errors and also the way they, they initiate processes. You need to be careful to use an F loop
instead of a if loop instead of a while loop, because they will run at once in a while loop and
then stop due to the way that the open sessions. It's, I've got caught with that a few times,
just something to keep in mind. But anyway, I send a quick way to send a standard error to standard
input is to use a, after the input to flag FF dashi intro.flack space to greater than
a percent one. So that was anything going to two, which a standard error will be redirected to
standard output. And then I can grip for the word duration. And that gives me aduration colon
space, 0, 0 colon, 0, 0 colon, 3, 9 dot 0, 5, comma start colon, lots and lots and bit raise 7,
blah, blah, blah. So I get some information there. So what we could do in this case is
well, the problem is here we've got a human readable day format, which I said earlier,
it's a little bit more difficult to to process it because, I mean, once it goes to 13,
once it goes to 59, then suddenly a digit starts appearing in one one as opposed to, and the,
you know, the minute section goes to 0, 0 and the hour section has been incremental by one.
That's a little bit more difficult to deal with. So luckily, though, the, you can deal with it
using the day command, which was something that I rather enjoyed. And again, I'm using the
forward slash date to on alias. You won't need to do that probably. The dash U to be UTC and the dash D.
And in this case, I'm putting in, so the number that I got was 0, 0 colon, 0, 0 colon, 3, 9,
so this time I'm putting in a date dash UD space 1970 dash 0 1 dash 0 1 T, 0, 0 colon, 0, 0 colon,
3, 9, 0, 5 space plus percent S. And the percent S will form it as a percent floor case S will
format it to seconds. So what that does is it counts the number of seconds since epoch. And
we're convert, we're not using now, we're using that particular date, which is why I've had to put
in 1970, 0, 1, 0, 1 T at the beginning was to give me the number of seconds that would have passed
since the first of January, since 1970, which is when the epoch started. Okay, so that gives us
a number. Now, let's see if we can do that for this intro to a flag, which of course we can.
We put the date space that's U 1970, 0, 1 T. And then we use our command substitution percent
open-square brackets f of probe dash I flag redirect standard output term to square F4D.
Then we print the second field, which gives us a nice clean one except there's a comma
tag down to the end and we use said S, F4C, F4C, F4C, G to get rid of that. We close the command
substitution and we do plus percent S and we get a 39. Problem is, unlike the other ones,
you can't just replace intro to a flag with asterix dot MP3 or asterix dot flag. The only way
you can get around that is to put it into a loop and as I said, a while loop won't work
without doing some weird directions. So I do a 4i in asterix flag do and then I replace
intro dot flag with dollar i and at the end of it do it done. So that gives me a new line,
39, 60, 17, 11 exactly what we had before. And I won't bore you with the details of how we escaped
all that again and we presented an ISO at 601 format and we get 0, 0, 0, 0, 0, 0, 0, 0, 7.
Now this time when we run it takes 4 seconds, 4.2 seconds basically to run as opposed to
fixed tags which was 3 seconds and medium for which was 0.6 seconds. But again, the conclusions
about this is really all kind of loosely based on what you haven't installed. So the best tool
is the tool that you have. And some people might argue that okay because we're short of shows
again because Ahuka and John Kulp are busy with life as they have not that employment for
that far from it. They have as we pointed out later during the year last year they have been
carrying quite a lot of the burden of HPR. So in the last few weeks we've been getting quite a
few times where our seven day warning has come up. That's something I was so happy not to see
and unfortunately now this turns me into what they call here in the Netherlands's Threskip
which is a stressed out chicken. So if you could all kindly do some shows that would be fantastic.
So that's it. Links are in the show notes. I did it using John Kulp's suggestion of
retextmarkdown which is actually very good. I find the focusing your mind you don't get too involved
in the formatting and then you can go back and tidy it all up again. Well with that I'll
call it a halt and as you can see even the tiniest little thing can trigger geeks and one of my
nerds supposed to do along and possibly monumentally boring show. But there you go. If you're ever
wanting to convert dates this is date command is absolutely very useful. I'm very fast actually
for doing quite a lot of this stuff. All right, tune in tomorrow for another exciting episode of
Hacker Public Radio.
You've been listening to Hacker Public Radio at Hacker Public Radio. We are a community podcast
network that releases shows every weekday Monday through Friday. Today's show, like all our shows,
was contributed by an HBR listener like yourself. If you ever thought of recording a podcast
then click on our contribute link to find out how easy it really is. Hacker Public Radio was
founded by the Digital Dove Pound and the Infonomicon Computer Club and is part of the
binary revolution at binrev.com. If you have comments on today's show, please email the host directly,
leave a comment on the website or record a follow-up episode yourself. Unless otherwise status,
today's show is released on the Creative Commons, Attribution, ShareLite, 3.0 license.