I'm writing man pages for my Tcl library, one file per command. Command names contain ::, for example mypackage::mycommand.
I'd like to call proper man page by following command: man mypackage::mycommand. For this purpose corresponding man file is called mypackage::mycommand.n (I'm using "n" section of manual).
It works in Linux, but it does not in Windows, because some applications cannot properly work with files containing ":" in names. So I'd like to rename man files, say to mypackage_mycommand.n.
Question: is it possible to call my manual by man mypackage::mycommand, if corresponding man file has different name?
Thanks.
Related
I have a list of 400 websites from where I'm downloading PDF's. 100 of these websites share the same pdf name: templates.pdf
When running wget, it either ignores the pdfs that have name "templates" or overwrites them. I was searching for a command that would make a new templates2.pdf for 2 hours, but I couldn't find anything.
The default behavior of wget is to use the .1, .2 prefixes when a file is downloaded multiple times into the same target directory. This appears to be what you are asking for. (The poorly named -nc option causes subsequent downloads of files with the same name to be ignored, which you don't want).
If the default behavior is not what you want, the -O option looks promising, as it allows you to _choose) the output file name. Here is a brief article explaining its use.
Of course, if you go the -O route, you'd need to ensure the output file does not exist, and do the suffix incrementing on your own.
I would like to write a Doxygen comment that names the file in which the comment occurs. Rather than write the filename explicitly, I would like Doxygen to supply me with it. Thus, if I change the name of the file, or move some of the content into a different file, I don't need to change hard-coded instances of the name.
For a concrete example, let's say I'm adding comments to functions in array.hpp, and I want the comment for certain functions to say "This function should only be used within array.hpp." I want to be able to write
/**
* This function should only be used within #thisfile.
*/
where #thisfile is a Doxygen expression that translates into array.hpp within the file array.hpp.
I've looked at the Doxygen documentation, including "Automatic link generation/Links to files" and the entire "Special Commands" section, but I haven't found what I'm looking for. Does such functionality exist?
Note that essentially the same question was asked on the Doxygen mailing list a few weeks ago. It has not received any replies.
General
As far as I know such functionality does not exist out-of-the-box. But you can add it by configuring an INPUT_FILTER in your Doxyfile. The path to the file is passed as an argument to the filter by doxygen. This can be used by the filter to replace your keyword (for example #thisfile) with the path to the file.
Below I give an example how to implement this with bash. A solution for other shells or Windows should be quite similar.
Example for bash
Write a short bash script infiltrate_filename.sh:
#!/bin/bash
pathToScript=`pwd`"/"
sed -e "s:#thisfile:${1/$pathToScript/}:g" $1
This script truncates the path to the file by the working directory. The resulting string is used to replace the keyword of your choice (here: #thisfile).
Make your script executable: chmod +x infiltrate_filename.sh
Set the INPUT_FILTER in your Doxyfile to INPUT_FILTER = ./infiltrate_filename.sh
That's it! 🎉 Now you can use #thisfile in your documentation blocks and it will be replaced by the path to the file. As the paths are relative to Doxygen's working directory they will automatically be linked to the file.
Notes
This solution assumes that the filter script is located in the working directory of doxygen (for example ~/my_project) and that the INPUT files are in subdirectories of the working directory (for example ~/my_project/src/foo/bar).
I have tested this example on a minimum working example. I am not a bash or sed expert. This solution may be improvable.
In vim, I loaded a series of web pages (one at a time) into a vim buffer (using the vim netrw plugin) and then parsed the html (using the vim elinks plugin). All good. I then wrote a series of vim scripts using regexes with a final result of a few thousand lines where each line was formatted correctly (csv) for uploading into a database.
In order to do that I had to use vim's marking functionality so that I could loop over specific points of the document and reassemble it back together into one csv line. Now, I am considering automating this by using Perl's "Mechanize" library of classes (UserAgent, etc).
Questions:
Can vim's ability to "mark" sections of a document (in order to
perform substitutions on) be accomplished in Perl?
It was suggested to use "elinks" directly - which I take to mean to
load the page into a headless browser using ellinks and perform Perl
scripts on the content from there(?)
If that's correct, would there become a deployment problem with
elinks when I migrate the site from my localhost LAMP stack setup to
a hosting company like Bluehost?
Thanks
Edit 1:
TYRING TO MIGRATE KNOWLEDGE FROM VIM TO PERL:
If #flesk (below) is right, then how would I go about performing this routine (written in vim) that "marks" lines in a text file ("i" and "j") and then uses that as a range ('i,'j) to perform the last two substitutions?
:g/^\s*\h/d|let#"=substitute(#"[:-2],'\s\+and\s\+',',','')|ki|/\n\s*\h\|\%$/kj|
\ 'i,'js/^\s*\(\d\+\)\s\+-\s\+The/\=#".','.submatch(1).','/|'i,'js/\s\+//g
I am not seeing this capability in the perldoc perlre manual. Am I missing either a module or some basic Perl understanding of m/ or qr/ ??
I'm sure all you need is some kind of HTML parser. For example I'm using HTML::TreeBuilder::XPath.
I've looked through several similar questions, but either I didn't understand their answer or my question is different than theirs. So, I have a project contains many subdirecties and different type of files. I would like to search a function name among those .C files only.
Some information on the web suggest to use "Esc x dired-do-query-replace-regexp". However, this will search not just C files, but also other file like .elf which isn't helpfule in my case. Other people sugget to use TAG function, but it will require me to type "etags *.c" for every subdirectory which is also impossible.
How should I do this while working on those large scale software project?
Thanks
Lee
Use ack-grep on linux
ack-grep "keyword" -G *.c
My favorite: igrep-find, found in the package igrep.el. Usage is:
M-x igrep-find some_thing RET *.C
There's the built in grep-find, docs here, but I find it awkward to use.
For a more general answer, see this similar question: Using Emacs For Big Big Projects.
if you're on linux, you can use grep to find files with a certain text in them. you would then do this outside of emacs, in your shell/command prompt. here's a nice syntax:
grep --color=auto --include=*.c -iRnH 'string to search for' /dir/to/search/
the directory to search can be specified relative, so if you're in the directory you want to use as the root directory for your recursive search, you can just skip the whole directory address and specify a single dot.
grep --color=auto --include=*.c -iRnH 'string to search for' .
the part --color=auto makes some text highlighted. --include=*.c is the part that specifies what files to search. in this case, only files with the c-extension. the flag i makes stuff case insensitive, the flag R makes the search recursive, the flag n adds the line number to the report, and the flag H adds the file path to the report.
To breed find and grep there is find-grep function, there you can change the invocation string to find . -name *.c etc. Make it a function, if You like. Then You use eg. C-x` et al. to navigate the results.
To search among the files in one directory i use lgrep, it prompts you in which files to search.
You can use cscope and xcscope.el : http://www.emacswiki.org/emacs/CScopeAndEmacs
Try with dired: place the cursor on the directory name to search, type A and in the minibuffer the text to find.
I am new to UNIX and when I was reading a book about UNIX, I came across following two problems that I didn't understand. I would really appreciate your help.
1) Look up the man page for the file command, and then use it on all files in the /dev directory. Can you group these files into two categories?
2) Run the tty command, and note the device name of your terminal. Now use this device name(/dev/pst/6) in the command cp /etc/passwd /dev/pts/6. what do you observe?
Fair question really... it's so easy for us to take so much for granted.
To read the manual page for the command called file, just type...
man file
...which will present a lot of information that will probably be quite confusing, but you'll get used to this stuff pretty quick if you keep at it. Crucially, file is a program that tries to categorise the files you ask it to. If you type...
file /dev/*
...that will do what the question asked, and invoke file with a list of the files in the /dev/ subdirectory. The list is actually prepared by the "shell" program that you're typing into, which then executes the file program and passes it the list. file then outputs some description of the files. On my computer, and where [SHELL-PROMPT] will be different on your computer, I typed file /dev/* and part of the output looked like:
[SHELL-PROMPT] file /dev/*
...lots of stuff...
/dev/cevt: character special (255/176)
/dev/console: character special (5/1)
/dev/core: symbolic link to `/proc/kcore'
/dev/cpqci: character special (10/209)
/dev/cpqhealth: directory
/dev/crom: character special (255/180)
...lots of stuff...
/dev/md8: block special (9/8)
/dev/md9: block special (9/9)
/dev/mem: character special (1/1)
/dev/mice: character special (13/63)
/dev/mouse0: character special (13/32)
/dev/mptctl: character special (10/220)
/dev/net: directory
/dev/nflog: character special (36/5)
/dev/null: character special (1/3)
/dev/parport0: character special (99/0)
...lots of stuff...
There's a filesystem entry for each directory/file combination (known as a path) in the left column, and file is describing the content in the right. Those descriptions may not make a lot of sense, but you can see that some patterns: some entries are "block special", others "character special", some are directory which implies you may find more files underneath (i.e. ls /dev/net/*). The numbers after "special" files are just operating system identifiers to differentiate the files mentioned. The import of this is that input and output from some devices connected to the computer is being made possible as if the device was a file in the filesystem. That "file" abstraction is being used as a general model for input and output. So, /dev/tty for example is tty - or terminal - device. Any data you try to read from there will actually be taken from the keyboard you're using to type into the shell (in the simple case), and anything you write there will become visible in the same terminal you're typing into. /dev/null is another interesting one: you can read and write from it, but it's an imaginary thing that never actually provides data (just indicates and End-of-File condition, and throws away any data written into it). You can keep reading from /dev/random and it will produce random values each time... good if you need random numbers or file content for encryption or some kind of statistical work.
2) Run the tty command, and note the
device name of your terminal. Now use
this device name(/dev/pst/6) in the
command cp /etc/passwd /dev/pts/6.
what do you observe?
By typing "tty" you can ask for the device representing your terminal...
[SHELL-PROMPT] tty
/dev/pts/11
But, I just said /dev/tty is another name for the same thing, so there's normally no need to use the "tty" program to find this more specific name. Still, if you create a couple terminal windows to your host, and type tty in each, you will see that each shell is connected to a different pseudo-terminal device. Still, each shell - and program run from the shell - can by default also refer to its own terminal input and output device as /dev/tty... it's a convenient context-sensitive name. The command...
cp /etc/passwd /dev/pts/6
...where you replace 6 with whatever your tty program really reported (e.g. 11 in my case), does the same thing as...
cp /etc/passwd /dev/tty
...it just reads the contents of the file /etc/passwd and writes them out on your screen. Now, the problem is that /etc/password looks like a lot of unintelligible junk to the average person - it's no wonder you couldn't make sense of what was happening. Try this instead...
echo "i said hello" > /tmp/hello.file
cp /tmp/hello.file /dev/tty
...and you'll see how to direct some specific, recognisable content into a new file (in this case putting it in the tmp "temporary" directory (the file will disappear when you reboot your PC), then copying that file content back to your screen.
(If you have logged on in two terminal windows, you can even go into one shell and copy the file to the /dev/pts/NN device reported by the other shell, effectively sending a message to the other window. You can even bypass the file and echo 'boo' > /dev/tty/NN. You'll only have permissions to do this if the same userid is logged into both windows.)