init
This commit is contained in:
commit
5e1b2ac657
26 changed files with 393 additions and 0 deletions
12
bash/next/args.sh
Executable file
12
bash/next/args.sh
Executable file
|
|
@ -0,0 +1,12 @@
|
|||
# why can't i use bash if statement, before appending an empty arg string?
|
||||
|
||||
last=`ls | egrep '^\d.*\.md|txt' | tail -n 1`
|
||||
|
||||
if ($* != "")
|
||||
last=last + $*;
|
||||
|
||||
awk '
|
||||
{
|
||||
print $0
|
||||
}
|
||||
END {print NR}' last
|
||||
2
bash/next/emacs_script.sh
Executable file
2
bash/next/emacs_script.sh
Executable file
|
|
@ -0,0 +1,2 @@
|
|||
# emacs --debug-init blah
|
||||
emacs `ls | grep '^\d' | tail -n 1`
|
||||
113
bash/next/next.sh
Executable file
113
bash/next/next.sh
Executable file
|
|
@ -0,0 +1,113 @@
|
|||
# Get the file name in any numbered series of text files and open with emacs
|
||||
|
||||
# Usage:
|
||||
# $ . next [descriptor]
|
||||
|
||||
# Handles an optional string descriptor. Like this:
|
||||
# $ ls
|
||||
# 011.blah.md
|
||||
# $ . next foo
|
||||
# 012.foo.md
|
||||
|
||||
# File format:
|
||||
# in regex terms:
|
||||
# [0-9]+\.[.*\.]*md|txt
|
||||
# in plain terms:
|
||||
# [leadings numbers with consistent width].[optional string descriptor and . separator][md|txt]
|
||||
|
||||
# what this does:
|
||||
# ls | grep [regex for number.*.md] | get last | awk script to increment the digits, preserve the zero prefix, add string descriptor if supplied, and add back extension
|
||||
|
||||
# Note the egrep '^\d.*\.md|txt' (matches whole string) to account for an optional string between the initial number and the file extension. As in '435.blah.md'.
|
||||
|
||||
# old version, relied on python, which didn't handle leading zeros properly, but treated them as binary:
|
||||
# ls | grep '\d.*\.md' | tail -n 1 | sed s/'\..*md'// | sed s/.*/'print & + 1'/ | python - | sed s/.*/'&.md/'
|
||||
|
||||
# BUGS
|
||||
|
||||
# 1. Emacs won't accept piped input as a file name. I'd rather keep this as a separate utility, which only printed the right filename, and then use an alias to pipe the filename to emacs, or invoke it manually.
|
||||
|
||||
# A.
|
||||
# What I want, is this syntax:
|
||||
# next [optional arg]
|
||||
# To both get the file and open emacs with it.
|
||||
|
||||
# B.
|
||||
# But this doesn't work:
|
||||
# next | emacs
|
||||
# Because it sees the stdin the pipe wants to feed it as a substitute for the tty in the new session, rather than as input for startup only, i.e. the filename.
|
||||
|
||||
# C.
|
||||
# Wrapping the whole shell command here works, like this:
|
||||
# emacs `ls | grep ...`
|
||||
|
||||
# But emacs doesn't like being run inside a child level shell, because it can't load .emacs properly. So you must run this file within the current shell using ".", like this:
|
||||
# . next
|
||||
# . next foo
|
||||
|
||||
# D.
|
||||
# This does not work, because .profile must be executed, meaning that the "" expression is peeked by the shell:
|
||||
# alias next="emacs `next.sh $1`"
|
||||
|
||||
# E.
|
||||
# This does not work either, because the arg never gets filled out when evoked:
|
||||
# alias next='emacs `next.sh $1`'
|
||||
|
||||
# F.
|
||||
# This works and is the current solution:
|
||||
# alias next='. next.sh'
|
||||
|
||||
# Which relies on 2 somewhat obscured facts:
|
||||
# 1. This script calls emacs itself, wrapping the real work in ``
|
||||
# 2. The alias is a simple substitution, which fortunately seems to happen before args are processed by the shell, meaning that the shell sees: ". next.sh arg"
|
||||
# 3. NOTE: the symbolic link in the PATH should not be simply "next", because it
|
||||
# may in certain situations conflict with the alias. The link should be "next.sh" -> ~/config/..
|
||||
# 4. NOTE: I use 'e' rather than 'emacs', to use my clever emacs/emacsclient function.
|
||||
|
||||
e `
|
||||
|
||||
ls | egrep '^[[:digit:]].*\.md|txt' | tail -n 1 | awk '
|
||||
|
||||
BEGIN {
|
||||
# set field separator, blank by default
|
||||
FS = "."
|
||||
}
|
||||
|
||||
{
|
||||
newnum = $1 + 1
|
||||
nzeros = length($1) - length(int(newnum))
|
||||
|
||||
# could I do this with a simpler regex to select all leading zeros?
|
||||
leadingzeros = ""
|
||||
if (nzeros > 0)
|
||||
for (i = 1; i <= nzeros; i++)
|
||||
leadingzeros = sprintf("%s%s", leadingzeros, "0")
|
||||
|
||||
if (nzeros > 0)
|
||||
newnum = sprintf("%s%s", leadingzeros, newnum)
|
||||
|
||||
# start by prefixing the number, now with leading zeros
|
||||
finalstr = newnum
|
||||
|
||||
# allow an arg input to fill out descriptor string.
|
||||
# single quote break to allow the shell to fill it out, double quotes so awk will treat it as a string
|
||||
arg = "'$1'"
|
||||
|
||||
if (arg != "")
|
||||
finalstr = sprintf("%s.%s", finalstr, arg)
|
||||
|
||||
# append last field by default
|
||||
finalstr = sprintf("%s.%s", finalstr, $NF)
|
||||
|
||||
# previous loop to reassemble all fields. But I do not want the old middle string here
|
||||
# Keeping it here for reference.
|
||||
# for (i = 2; i <= NF; i++)
|
||||
# finalstr = sprintf("%s.%s", finalstr, $i)
|
||||
# }
|
||||
|
||||
}
|
||||
|
||||
END {
|
||||
print finalstr
|
||||
}'
|
||||
`
|
||||
0
bash/next/test/001.md
Executable file
0
bash/next/test/001.md
Executable file
0
bash/next/test/002.blah.md
Executable file
0
bash/next/test/002.blah.md
Executable file
0
bash/next/test/004.foo.md
Executable file
0
bash/next/test/004.foo.md
Executable file
0
bash/next/test/005.md
Executable file
0
bash/next/test/005.md
Executable file
0
bash/next/test/009.boo.md
Executable file
0
bash/next/test/009.boo.md
Executable file
0
bash/next/test/098.baa.md
Executable file
0
bash/next/test/098.baa.md
Executable file
0
bash/next/test/099.poo.md
Executable file
0
bash/next/test/099.poo.md
Executable file
Loading…
Reference in a new issue