> Another option is to use Python, which is ubiquitous enough that it can be expected to be installed on virtually every machine
Not on macOS. You can do it easily by just invoking /usr/bin/python3, but you’ll get a (short and non-threatening) dialog to install the Xcode Command-Line Developer Tools. For macOS environments, JavaScript for Automation (i.e. JXA, i.e /usr/bin/osascript -l JavaScript) is a better choice.
However, since Sequoia, jq is now installed by default on macOS.
I feel like the takeaway here is that jq should probably be considered an indispensable part of the modern shell environment. Because without it, you can't sensibly deal with JSON in a shell script, and JSON is everywhere now. It's cool that you can write a parser in awk but the point of this kind of scripting language is to not have to do things like that.
One day I wanted to use a TAP parser for the Test Anything Protocol.
But I didn't want to be bogged down by dependencies.. so I didn't want to go nowhere near Python (and pyenv.. and anaconda.. and then probably having to dockerize that for some reason too..) nor nodeJS nor any of that.
Found a bash shell script to parse TAP written by ESR of all people. That sounds fine, I thought. Most everywhere has bash, and there are no other dependencies.
But it was slow. I mean.. painfully, ridiculously slow. Parsing like 400 lines of TAP took almost a minute.
That's when I did some digging and learned about what awk really is. A scripting language that's baked into the POSIX suite. I had heard vaguely about it beforehand, but never realized it had more power than it's sed/ed/cut/etc brethren in the suite.
So I coded up a TAP parser in awk and it went swimmingly, matched the ESR parser's feature set, and ran millions of lines in less than a second. Score! :D
For the record, "python-without-extra-dependencies" is a thing and a very nice one too. I always prefer it over awk.
Highly recommend to everyone - plenty of "batteries" included, like json parser, basic http client and even XML parser, and no venv/conda required. Very good forward compatibility. Fast (compared to bash).
That does sound compelling, but I frequently enough have to work on embedded systems (often running busybox) where python isn't installed and available R/W storage space is measured in tens or hundreds of Kb.
I find that that is one more environment where awk scripting can get the job done, python/perl/php/etc just can't be introduced, bash can _sometimes_ get the job done if it doesn't have to spawn too many subprocesses, and C/other-compiled-options _might_ be able to help if I had some kind of build environment targeting the platform(s) in question and enough patience.
I'll keep an eye out for python with no extra dependency options on the platforms that can handle that though.
Contemplating this, it’s too bad the Unix scripting ecosystem never evolved a tripartite symbiosis of ‘file‘, ‘lex‘, and ‘yacc‘, or similar tools.
That is, one tool to magically identify a file type, one to tokenize it based on that identification, one to correspondingly parse it. All in a streaming/pipe-friendly mode.
Would fit right in, other than the Unix prejudice (nonsensical from Day 0) for LF-separated text records as the “one true format”.
I guess ... I don't think it really gets you much unless it's 1-pass streaming otherwise we're dealing with entire input buffers and then we're just back to files.
You could argue using the vertical separator | is more syntactically graceful but then it's just a shell argument. There's quite a few radically different shells out there these days like xonsh, murex, and nushell so if simply arranging logic on the screen in a different syntax is what you're looking for then that's probably the way.
What I meant was something like the SAX streaming parse model disaggregated and broken into sniff / lex / parse phases all mediated by a modestly structured stream of the kind a tool like awk could process naturally.
Awk is great and this is a great post. But dang, awk really shoots itself so much with its lack of features that it so desperately needs!
Like: printing all but one column somewhere in the middle. It turns into long, long commands that really pull away from the spirit of fast fabrication unix experimentation.
>awk really shoots itself so much with its lack of features that it so desperately needs!
That's why I use Perl instead (besides some short one liners in awk, which in some cases are even shorter than the Perl version) and do my JSON parsing in Perl.
I've been using perl instead of sed because PCRE is just better and it's the same regex that PHP uses which I've been coding in for nearly 20 years.
I still don't actually know perl, but apparently Gemini does. It wrote a particularly crazy find and replace for me.
Never got around to using or learning awk. Only time I see it come up is when you want to parse some tab delimited output
Things are already like that, friend! We have mawk, gawk and nawk. But it's fun to think about how we could improve our ideal tooling if we had a time machine.
JSON is not a friendly format to the Unix shell — it’s hierarchical, and cannot be reasonably split on any character
Yes, shell is definitely too weak to parse JSON!
(One reason I started https://oils.pub is because I saw that bash completion scripts try to parse bash in bash, which is an even worse idea than trying to parse JSON in bash)
I'd argue that Awk is ALSO too weak to parse JSON
The following code assumes that it will be fed valid JSON. It has some basic validation as a function of the parsing and will most likely throw an error if it encounters something strange, but there are no guarantees beyond that.
Yeah I don't like that! If you don't reject invalid input, you're not really parsing
---
OSH and YSH both have JSON built-in, and they have the hierarchical/recursive data structures you need for the common Python/JS-like API:
osh-0.33$ var d = { date: $(date --iso-8601) }
osh-0.33$ json write (d) | tee tmp.txt
{
"date": "2025-06-28"
}
Parse, then pretty print the data structure you got:
Also, OSH is now FASTER than bash, in both computation and I/O.
This is despite garbage collection, and despite being written in typed Python! I hope to publish a post about these recent improvements
> I think they made some choices that eventually led to the parser being too complex, largely due to the problem of representing what was parsed.
No, the complexity of the parser can be attributed to the incremental parsing. ble.sh implements an incremental parser where one can update only the necessary parts of the previous syntax tree when a part of the command line is modified. I'd probably use the same data structure (but better abstracted using classes) even if I could implement the parser in C or in higher-level languages.
I don't really buy that shell / awk is "too weak" to deal with JSON, the ecosystem of tools is just fairly immature as most of the shells common tools predate JSON by at least a decade. `jq` being a pretty reasonable addition to the standard set of tools included in environments by default.
IMO the real problem is that JSON doesn't work very well at as a because it's core abstraction is objects. It's a pain to deal with in pretty much every statically typed non-object oriented language unless you parse it into native, predefined data structures (think annotated Go structs, Rust, etc.).
I'd say that awk really is too weak. Awk has a grand total of 2 data types: strings, and associative arrays mapping strings to strings. There is no support for arbitrarily nested data structures. You can simulate them with arrays if you really want to, or you could shell out to jq, but it's definitely swimming upstream.
Most languages aren't quite that bad. Even if they can't handle JSON very ergonomically, almost every language has at least some concept of nesting objects inside other objects.
What about shell? Just like awk, bash and zsh have a limited number of data types (the same two as awk plus non-associative arrays). So arguably it has the same problem. On the other hand, as you say, in shell it's perfectly idiomatic to use external tools, and jq is one such tool, available on an increasing number of systems. So you may as well store JSON data in your string variables and use jq to access it as needed. Probably won't be any slower than the calls to sed or awk or cut that fill out most shell scripts.
Now, personally, I've gotten into the habit of writing shell scripts with minimal use of external tools. If you stick to shell builtins, your script will run much faster. And both bash and zsh have a pretty decent suite of string manipulation tools, including some regex support, so you often don't actually need sed or awk or cut. However, this also rules out jq, and neither shell has any remotely comparable builtin.
But you might reasonably object that if I care about speed, I would be better off using a real programming language!
The same author already had made the more thorough jawk. They explicitly said they wanted a cut down version. It's not illegal to want a cut down version of something.
Readit News
Not on macOS. You can do it easily by just invoking /usr/bin/python3, but you’ll get a (short and non-threatening) dialog to install the Xcode Command-Line Developer Tools. For macOS environments, JavaScript for Automation (i.e. JXA, i.e /usr/bin/osascript -l JavaScript) is a better choice.
However, since Sequoia, jq is now installed by default on macOS.
But I didn't want to be bogged down by dependencies.. so I didn't want to go nowhere near Python (and pyenv.. and anaconda.. and then probably having to dockerize that for some reason too..) nor nodeJS nor any of that.
Found a bash shell script to parse TAP written by ESR of all people. That sounds fine, I thought. Most everywhere has bash, and there are no other dependencies.
But it was slow. I mean.. painfully, ridiculously slow. Parsing like 400 lines of TAP took almost a minute.
That's when I did some digging and learned about what awk really is. A scripting language that's baked into the POSIX suite. I had heard vaguely about it beforehand, but never realized it had more power than it's sed/ed/cut/etc brethren in the suite.
So I coded up a TAP parser in awk and it went swimmingly, matched the ESR parser's feature set, and ran millions of lines in less than a second. Score! :D
Highly recommend to everyone - plenty of "batteries" included, like json parser, basic http client and even XML parser, and no venv/conda required. Very good forward compatibility. Fast (compared to bash).
I find that that is one more environment where awk scripting can get the job done, python/perl/php/etc just can't be introduced, bash can _sometimes_ get the job done if it doesn't have to spawn too many subprocesses, and C/other-compiled-options _might_ be able to help if I had some kind of build environment targeting the platform(s) in question and enough patience.
I'll keep an eye out for python with no extra dependency options on the platforms that can handle that though.
That is, one tool to magically identify a file type, one to tokenize it based on that identification, one to correspondingly parse it. All in a streaming/pipe-friendly mode.
Would fit right in, other than the Unix prejudice (nonsensical from Day 0) for LF-separated text records as the “one true format”.
You could argue using the vertical separator | is more syntactically graceful but then it's just a shell argument. There's quite a few radically different shells out there these days like xonsh, murex, and nushell so if simply arranging logic on the screen in a different syntax is what you're looking for then that's probably the way.
Like: printing all but one column somewhere in the middle. It turns into long, long commands that really pull away from the spirit of fast fabrication unix experimentation.
jq and sql both have the same problem :)
Whence perl.
https://github.com/moritz/json/blob/master/lib/JSON/Tiny/Gra...
The following command is handy for grepping the output:
That's why I use Perl instead (besides some short one liners in awk, which in some cases are even shorter than the Perl version) and do my JSON parsing in Perl.
This
diff -rs a/ b/ | ask '/identical/ {print $4}' | xargs rm
is one of my often used awk one liners. Unless some filenames contain e.g. whitespace, then it's Perl again
Yes, shell is definitely too weak to parse JSON!
(One reason I started https://oils.pub is because I saw that bash completion scripts try to parse bash in bash, which is an even worse idea than trying to parse JSON in bash)
I'd argue that Awk is ALSO too weak to parse JSON
The following code assumes that it will be fed valid JSON. It has some basic validation as a function of the parsing and will most likely throw an error if it encounters something strange, but there are no guarantees beyond that.
Yeah I don't like that! If you don't reject invalid input, you're not really parsing
---
OSH and YSH both have JSON built-in, and they have the hierarchical/recursive data structures you need for the common Python/JS-like API:
Parse, then pretty print the data structure you got: Create a JSON syntax error on purpose: (now I see the error message could be better)Another example from wezm yesterday: https://mastodon.decentralised.social/@wezm/1147586026608361...
YSH has JSON natively, but for anyone interested, it would be fun to test out the language by writing a JSON parser in YSH
It's fundamentally more powerful than shell and awk because it has garbage-collected data structures - https://www.oilshell.org/blog/2024/09/gc.html
Also, OSH is now FASTER than bash, in both computation and I/O. This is despite garbage collection, and despite being written in typed Python! I hope to publish a post about these recent improvements
Parsing is a trivial, rejecting invalid input is trivial, the problem is representing the parsed content in a meaningful way.
> bash completion scripts try to parse bash in bash
You're talking about ble.sh, right? I investigated it as well.
I think they made some choices that eventually led to the parser being too complex, largely due to the problem of representing what was parsed.
> Also, OSH is now FASTER than bash, in both computation and I/O.
According to my tests, this is true. Congratulations!
No, the complexity of the parser can be attributed to the incremental parsing. ble.sh implements an incremental parser where one can update only the necessary parts of the previous syntax tree when a part of the command line is modified. I'd probably use the same data structure (but better abstracted using classes) even if I could implement the parser in C or in higher-level languages.
But yes, ble.sh also has a shell parser in shell, although it uses a state machine style that's more principled than bash regex / sed crap.
---
Also, distro build systems like Alpine Linux and others tend to parse shell in shell (or with sed).
They often need package metadata without executing package builds, so they do that by trying to parse shell.
In YSH, you will be able to do that with reflection, basically like Lisp/Python/Ruby, rather than ad hoc parsing.
---
I'm glad to hear you can see the effect of the optimizations ! That took a long time :-)
Some more benchmarks here, which I'll write about: https://oils.pub/release/0.33.0/benchmarks.wwz/osh-runtime/
IMO the real problem is that JSON doesn't work very well at as a because it's core abstraction is objects. It's a pain to deal with in pretty much every statically typed non-object oriented language unless you parse it into native, predefined data structures (think annotated Go structs, Rust, etc.).
Most languages aren't quite that bad. Even if they can't handle JSON very ergonomically, almost every language has at least some concept of nesting objects inside other objects.
What about shell? Just like awk, bash and zsh have a limited number of data types (the same two as awk plus non-associative arrays). So arguably it has the same problem. On the other hand, as you say, in shell it's perfectly idiomatic to use external tools, and jq is one such tool, available on an increasing number of systems. So you may as well store JSON data in your string variables and use jq to access it as needed. Probably won't be any slower than the calls to sed or awk or cut that fill out most shell scripts.
Now, personally, I've gotten into the habit of writing shell scripts with minimal use of external tools. If you stick to shell builtins, your script will run much faster. And both bash and zsh have a pretty decent suite of string manipulation tools, including some regex support, so you often don't actually need sed or awk or cut. However, this also rules out jq, and neither shell has any remotely comparable builtin.
But you might reasonably object that if I care about speed, I would be better off using a real programming language!