Add validations for brace quantifiers {m}, {m,} and {m,n}

[LoukasPap]

Fixes issue 289

There are two types of errors in quantifiers:

1. Unmatched \{ for unterminated braces
2. Regular expression too big if m and/or n are outside the [0,255] bounds
3. Invalid content of \{\} for all the others

Unmatched \{

sed always captures the unmatched braces error first. To deal with that, I scan ahead the regex for unclosed { } and if it is violated it throws the error, otherwise we retreat back to the beginning to start the processing.

Regular expression too big

• m and/or n must be between [0, 255]

Invalid content of \{\}

It checks for:

• m <= n
• integers and comma are only allowed (the comma between m and n)
• {,} to be converted to *
• {,n} to be converted to {0,n}

[codecov]

Codecov Report

❌ Patch coverage is 98.08219% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.21%. Comparing base (60243b4) to head (461c1d5).

Files with missing lines	Patch %	Lines
src/sed/delimited_parser.rs	98.42%	5 Missing ⚠️
src/sed/compiler.rs	95.55%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #328      +/-   ##
==========================================
+ Coverage   82.29%   83.21%   +0.92%     
==========================================
  Files          13       13              
  Lines        5534     5876     +342     
  Branches      310      338      +28     
==========================================
+ Hits         4554     4890     +336     
- Misses        977      983       +6     
  Partials        3        3              

Flag	Coverage Δ
macos_latest	83.89% <98.08%> (+0.91%)	⬆️
ubuntu_latest	83.97% <98.08%> (+0.89%)	⬆️
windows_latest	0.00% <0.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[LoukasPap]

Forgot to add unit tests. I will add soon.

[codspeed-hq]

Merging this PR will not alter performance

✅ 11 untouched benchmarks

---

_{Comparing LoukasPap:issue289 (461c1d5) with main (60243b4)}

[LoukasPap]

@e-kwsm Ready for review! 🙂

[e-kwsm]

{,n} to be parsed as literal

nitpick: "{,n}" pattern is not specified in ERE, but

<<< xxx sed -E -n -e '/x{,-1}/s//@/p'

• GNU and BSD: Invalid content of \{\}, exit with one
• uutils: prints nothing and exits with zero

<<< xxx sed -E -n -e '/x{,0}/s//@/p'

• GNU and BSD: print @xxx
• uutils: prints nothing

<<< xxx sed -E -n -e '/x{,1}/s//@/p'

• GNU and BSD: print @xx
• uutils: prints nothing

[e-kwsm]

<<< xxx sed -n -E -e '/.{,x}/p'

• GNU and BSD: Invalid content of \{\}
• uutils: prints nothing

[e-kwsm]

There is also an error when m and/or n are very large numbers but I let that for the regex crate to handle

GNU and BSD support 255-repetition.

https://www.gnu.org/software/sed/manual/html_node/BRE-syntax.html#index-GNU-extensions_002c-to-basic-regular-expressions-2

\{i\}

As *, but matches exactly i sequences (i is a decimal integer; for portability, keep it between 0 and 255 inclusive).

https://man.freebsd.org/cgi/man.cgi?re_format(7)

A bound is `{' followed by an unsigned decimal integer, possibly followed by `,' possibly followed by another unsigned decimal integer, always followed by `}'. The integers must lie between 0 and RE_DUP_MAX (255<**>) inclusive, and if there are two of them, the first may not exceed the second.

[LoukasPap]

Nice catches, I'll check them

[LoukasPap]

I am making some changes and I will re-push.

[LoukasPap]

There were many approaches one could follow to add validation for the quantifiers, each with its own advantages and disadvantages. What I did:

• In parse_regex() I process the quantifier in a single place when I match an open brace - { for ERE and \\{ for BRE
• I separate the validation in two functions to not complicate things a lot:
  • validate_quantifier_structure() where I check for unmatched error and invalid content errors
  • validate_quantifier_numbers() where I basically process the content as it is (because validation happened in previous function) but check only the validity of m and/or n (if they are outside the allowed bounds and if m is greater than n).
• Add tests for various cases (I am missing some unit tests that I'll add if necessary)

Other solutions could be to break the parsing to more places in code by adding more match cases (comma and digit), or to contain the entire validation in one function (that's what I had done initially but I though it was too complicated). Another solution could possibly be to even use regex for validating the regex quantifier.

[LoukasPap]

@sylvestre Ready for review 🙂 (I'll squash before merge)

[LoukasPap]

I will make changes in the next few days and re-push.

[LoukasPap]

Just rebased atop main and squashed all commits into one. I am open to new approaches and suggestions for a better solving of the problem, if it's needed :)