gnu grep on R-Pi

I’m sure I’m doing something incredibly stupid here, but I can’t work out what… I installed !grep from PackMan on an R-pi, but cannot for the life of me work out how to use wildcards for the filename. If I run ‘grep pattern *’ I get no output. If I run, say, ‘grep pattern *.htm’ I get a match in a single file, when I know that the pattern is in multiple files.

Do I need to escape the ‘*’ character in the filename somehow?

Sorry, I know I should be able to work this one out, but I’ve been trying now for over an hour to get a simple grep to work across a bunch of html files, and I just can’t get it to work the way the documentation suggests it should.

Thanks,

Im not sure. It is to long ago. You can download my grep frontend based on the frontend module. A !Help with a wildcard-section is inside.

Thanks for that, but I still can’t get grep to return the results I’d expect. Has anyone else got a Pi that they could test with? I still reckon I’m doing something stupid, but I’m just going round in circles now. How should it work? Like unix? Or am I missing some RISC OS specific convention? All I’ve done is install grep from !PackMan and attempted to use it as per the helptext, with “grep pattern *” or “grep pattern */htm” – the current directory is correctly set.

OK, here’s an odd thing. I execute the same command twice, and it returns different results. This is a straight cut&paste from a command line:

*grep TZR *.htm
*grep TZR *.htm
He mentions that a TZR250 is a good way to start, and this seems good
bike? TZR250, as it's (relatively) cheap, won't scare the crap out of
BMCRC, as they're close by, and run a large TZR class. So, now I knew
will be running the TZR class for a while yet, and it's a good class
Ian Bassett who has a ready to race TZR up for grabs. So far
Drive over to Thamesmead to meet Ian and his amazing performing TZR. I
the rest of the TZRs there. Some of them look frighteningly serious,
but at least there's one guy there who's got a TZR with a worse colour
entered the three TZR races on the day, as a) I'm not sure how

I genuinely don’t understand this now!

I tend to just use ‘grep pattern .’ to search all files. Potentially ‘grep pattern @’ on RISC OS if grep is using RISC OS format filenames instead of unix (@ is RISC OS’s equivalent of unix’s .)

I’m not sure what style filenames the version of grep in packman expects, but I’d expect at least one of the above to produce what you’re after. I’ve never really used wildcarded filenames with grep on RISC OS, so I’m not sure whether they usually work or not.

Nope, no joy:

*grep TZR .
/SDFS::RISCOSpi.$/Apps/Utilities/!grep/grep: .
*grep TZR @
/SDFS::RISCOSpi.$/Apps/Utilities/!grep/grep: @

That string should be in multiple files. I’ve got a copy of the same directory on a linux machine:

neil@raspberrypi /mnt/usbdisk/race diary $ grep -c TZR *.htm
d_199707.htm:9
d_199708.htm:4
d_199709.htm:2
d_199710.htm:10
d_199711.htm:2
d_199801.htm:5
d_199803.htm:1
d_199804.htm:2
d_199805.htm:3
d_199806.htm:4
d_199807.htm:4
d_199808.htm:2
d_199809.htm:4
d_199810.htm:6
d_199811.htm:1
d_199901.htm:9
d_199903.htm:4
d_199904.htm:7
d_199905.htm:8
d_199906.htm:2
d_199907.htm:7
d_199908.htm:0
d_199909.htm:4
d_199910.htm:6
d_199911.htm:7
d_200001.htm:6
d_200003.htm:2
d_200004.htm:4
d_200005.htm:1
d_200006.htm:1
d_200007.htm:1
d_200008.htm:0
d_200009.htm:0
d_200010.htm:0
d_200011.htm:0
d_200101.htm:0
d_200103.htm:0
d_200104.htm:0
d_200105.htm:0
d_200106.htm:0
d_200203.htm:0
d_200204.htm:1
d_200205.htm:0
d_200207.htm:0
d_200211.htm:0
d_200212.htm:2
d_200301.htm:1
d_200302.htm:7
d_200303.htm:0
d_200304.htm:5
d_200306.htm:1
d_200308.htm:5

If I execute exactly the same command on RISC OS, it stops after the first file:

*grep -c TZR *.htm
9
*

*grep -c TZR *.htm

What is “TZR”?

Have you tried “*grep -c TZR */htm” ?

it stops after the first file

…and doesn’t appear to output the filename either.

The UNIX shell (e.g. bash) is expanding the wildcard before running grep. The normal RISC OS shell doesn’t do that and will pass ‘*.htm’ directly to grep. So, unless the copy of grep you are using was extended to simulate this wildcard expansion it isn’t going do what you want.

What I suspect is happening is that ‘*.htm’ is getting interpreted a the name of a single file. IIRC this would cause the ‘first’ matching file to be opened.

I have not a RISC OS Computer here so I’m not sure but I mean there is a filename wildcard enviroment to set. I remember a little bit it is desribe in my helpfile. It is long time ago as I need this.
Will you search only in html-files? Why? You can search in the complet directory.
Sorry if I misunderstand any thinks.

Try adding the ‘-r’ switch (as @ is a directory). With that added it works fine here.

Yes, I tried ‘grep pattern */htm’ and received an error that /htm wasn’t found. Which kind of makes sense (to me, at least), as I’m assuming that as a Unix utility it’s going to use Unix filenaming conventions?

So, if ‘*.htm’ matches only one file, what should I use to search all files in a directory? ‘grep pattern *’ returns no matches. It doesn’t matter if I put the pattern to search for in quotes or not. And I don’t want to search recursively, as I’m already in the directory that I want to perform the grep. And the reason for searching only html files is that there are also many jpegs and other, non-text files in that directory that do not need searching.

Is there a ‘native’ RISC OS grep I can use? I mean, somebody must have wanted to locate all incidences of a string in a set of files before now? I’m sure I must just be missing something staggeringly obvious.

Textseek?

Ah-ha! Looks promising. I’ll take a look. Thanks!

Yes, I tried ‘grep pattern */htm’ and received an error that /htm wasn’t found. Which kind of makes sense (to me, at least), as I’m assuming that as a Unix utility it’s going to use Unix filenaming conventions?

Which is why ports that bring in filename conventions from other filing systems are horrible and should be expunged from reality…with prejudice. [hint: there’s rather more to porting than just (re)compiling it on a different system]

You might feel that is a strong opinion, but really, we can’t have this utility doing something with filenames, and that utility doing something totally different.
In RISC OS, ‘$’ is root, ‘.’ is a directory separator, and we use ‘/’ to separate extensions from names (for filesystems that need such a thing). Anything else is wrong.
Just.
Plain.
Wrong.

[And if you don’t believe me, imagine the stink that would be created if you made a nifty utility for Linux, but insisted upon RISC OS style paths!]

‘grep pattern *’ returns no matches.

That’s bizarre as it is pretty universal that a star is a multicharacter wildcard.

I guess I ought to find grep and drop it in the emulator for a play.

I can use? I mean, somebody must have wanted to locate all incidences of a string in a set of files before now?

Yes – I wrote a little program to do it over two decades ago.
Simplistic but slow (chunk-load file into a buffer and byte-look for character 1, then look for character 2, and if those match, do a string comparison… noteable bug, would be blind to strings spanning the chunk boundary… but even then it manages to be better than Windows’ “Find” which seems to miss more than it finds). I would give it to you if I had it, but I think it is on tape and my streamer gave up the ghost back circa Y2K…

I’m off to download Textseek. ;-)

hint: there’s rather more to porting than just (re)compiling it on a different system

Yup, I completely agree on this point. You can see the confusion that it causes to a beginner! Thanks to everyone for their help here.

(Now to work out how to do quotes here…)

Has anybody tried Hvor (Danish for where) which you can download from http://www.wra1th.plus.com/zip/hvor.zip ? I wrote it as an exercise back in 2007. It uses Lua pattern syntax rather than Perl. I also wrote Gap which searches for two words separated by less than a given number of words. See http://www.wra1th.plus.com/throw.html .

Has anybody tried Hvor

I’m getting the following on my BB-xM (5.19 16.05.2012)

Internal error: abort on data transfer at &000217C8

Postmortem requested
217c0 in anonymous function
21918 in anonymous function
103d8 in anonymous function
f5e0 in anonymous function
Arg5: 0×00000018 24 → [0xe59ff2b8 0xe24ee004 0xe28fd0d8 0xe28fa00c]
Arg4: 0×000000f0 240 → [00000000 00000000 00000000 00000000]
Arg3: 0×00029338 168760 → [0×00029364 0×0003b768 00000000 0×00000020]
Arg2: 0×0001036c 66412 → [0xe1a0c00d 0xe92dd8f0 0xe24cb004 0xe15d000a]
Arg1: 0×0002b988 178568 → [0×0003138c 0×00006208 0×000360c4 0×000360ac]
102f8 in anonymous function
104c0 in anonymous function
cf00 in anonymous function
b86c in anonymous function
15650 in anonymous function
fda0 in anonymous function
Arg2: 0×00000002 2
Arg1: 0×0002b988 178568 → [0×0003138c 0×00006208 0×000360c4 0×000360ac]
1261c in anonymous function
ffb4 in anonymous function
f5e0 in anonymous function
Arg5: 0×00000018 24 → [0xe59ff2b8 0xe24ee004 0xe28fd0d8 0xe28fa00c]
Arg4: 0×00000020 32 → [0xe28fd0d8 0xe28fa00c 0xe92d400f 0xe1a0e00f]
Arg3: 0×00029500 169216 → [0×0002b770 0xffffffff 0×0002b988 0×00000002]
Arg2: 0×0000cd34 52532 → [0xe5912004 0xe5911000 0xea000c75 0xe1a0c00d]
Arg1: 0×0002b988 178568 → [0×0003138c 0×00006208 0×000360c4 0×000360ac]
102f8 in anonymous function
cdc0 in anonymous function
84a4 in anonymous function
8bec in anonymous function
8f88 in anonymous function
fda0 in anonymous function
ffa4 in anonymous function
f5e0 in anonymous function
Arg5: 00000000 0
Arg4: 0×00000008 8 → [0xe59ff2b8 0xe59ff2b8 0xe59ff2b8 0xe59ff2b8]
Arg3: 0×00029690 169616 → [0×00008e6c 0×000296a8 0×000296cc 0×000296a8]
Arg2: 0×0000cde8 52712 → [0xe1a0c00d 0xe92dd830 0xe24cb004 0xe15d000a]
Arg1: 0×0002b988 178568 → [0×0003138c 0×00006208 0×000360c4 0×000360ac]
102f8 in anonymous function
ceac in anonymous function
9034 in anonymous function
Arg2: 0×00008fdc 36828 → [0xe1a0c00d 0xe92dd870 0xe24cb004 0xe15d000a]
Arg1: 0×0002973c 169788 → [0×736e693a 0×6c6c6174 0×2e346468 0×42212e24]
fc13c628 in shared library function
2632c in anonymous function

Now to work out how to do quotes here…

Ah you will be wanting to type “b” and a “q” and a dot (as “bq.”) with a space after and then the quoted text. Then a space line after the quoted text.

BTW. feel free to say how naff textile is, there won’t be any arguments.

The forum used to block quote with a simple email style chevron, but that function went a while ago, unfortunately.

BTW. feel free to say how naff textile is, there won’t be any arguments.

Well… as a newcomer I thought it best to keep my head below the parapet for a while first :) But I’ll admit to a certain amount of frustration trying to use the forum and wiki here from NetSurf. I just assumed it was some kind of initiation test…

Intwernal Error: abort on data transfer …

Sorry. If I had tested it first in the RPi I would have remembered it needs recompiling. 2007 is a long time ago.

I have now scrapped Hvor. Try !L—cate instead, from http://www.wra1th.plus.com/zip/l—cate.zip . I have tested it on the Raspberry Pi. The interface is unsophisticated, but it lets you search for a pattern in files or directories, recursively or not, allowing files with given filetypes whose leafnames match a given pattern.

Just a quick note to say that I grabbed !Textseek, and it works beautifully. Thank you! I’ll also give !L-cate a whizz this afternoon when I’ve got a few minutes. Again, thanks all for the help.

Having used Textseek for a bit, can I ask what the protocol would be for getting it included as a package that can be installed from PackMan? (assuming that the original author would permit this). I just think that it’s a very useful thing to have, which would have saved me a bit of grief had I found it initially. I think that other newcomers to RISC OS could also benefit from having it easily available.

I’m happy to help out here if I can – I just don’t want to tread on anyones toes while I’m doing it. Ta.

Having used Textseek for a bit, can I ask what the protocol would be for getting it included as a package that can be installed from PackMan?

It already is. I found this to be a very useful program and the author allowed me to package it a while ago. I’ve just double checked and it it in the package lists shipped with the RPi.

So it is! I looked through the list, got as far as THHeap and THSound, and assumed that it wasn’t. Of course, I hadn’t banked on H coming before e in the !PackMan sort order. Mea culpa.