tudurom/usam

blending sam with the shell experiment

usam

micro-sam

An experiment in blending sam with the shell.

After the experiment is done, the plan is to build a shell around structural regular expressions.

Mode of operation

Each sam command is implemented as a stand-alone command line tool.

The commands are chained together via pipes.

Many commands accept a new dot as an argument. The argument dot is evaluated relative to the current dot. Works for loops (x commands) too.

Tools

e <filename>

Opens the file for editing. Dot is set to 0. It is the start of the pipe chain.

Example:

e file.txt | further editing

po and pc

If you want to manipulate text from a pipe, you can use po (pipe open) and pc (pipe close).

pc writes the resulting text to stdout.

po reads all the pipe's content before continuing the command chain.

Example:

dmesg | po | further editing | pc > new_dmesg.txt

el <new_dot>

Sets the dot to a new address. When used in a loop, it ends the loop and sets the dot relative to the last dot of the loop.

Example:

e file.txt | el 2,3 | p

p [dot]

Prints the content of the dot.

Example:

e file.txt | p 2,3 # does the exact thing as the above example

c <text> [dot]

Changes the dot's value with text. Sets dot to the changed text.

Example:

<<EOF > file.txt
Happy 2018!
EOF

e file.txt | c 2019 '/2018/' | p ,
# Prints: Happy 2019!

i <text> [dot]

Like c, but inserts text right before the dot.

a <text> [dot]

Like c, but inserts text right after the dot.

d [dot]

Deletes the dot's content.

Example:

<<EOF > file.txt
We live in a different society.
EOF

e file.txt | d '/ different/' | p ,
# Prints: We live in a society.

s <regexp> <text> [n|g] [dot]

Substitute text for the first match to the regexp in the dot. Set dot to the modified range. In text, $ signs are interpreted as in Go's Expand. If you want to change dot and substitute the first match, you must call it like so:

s <regexp> <text> 1 <dot>

Example:

<<EOF > file.txt
y y a x y y
EOF

e file.txt | s '(a) (x)' '$2 $1' 1 1 | p ,
# Prints: y y x a y y

x [regexp]

For each regexp match in the dot, sets dot to that and execute the next command on that dot.

Because x doesn't accept a dot argument, you must use el first. Ironically, el comes from "end loop".

You can also compose x commands.

Example:

e vim.1 | el , | x 'vim' | p +-
# Prints all lines that contain the word 'vim'.
# If a line has 'vim' in it more than once, the line will pe printed each time.