tlsfuzzer/extract.py: Add support for {r,s} tuples in sigs file

[GeorgePantelakis]

Checklist

• I have read the CONTRIBUTING.md document and my PR follows change requirements therein
• all new and existing tests pass (see CI results)

---

This change is 

[tomato42]

so, that's generally called raw encoding, see https://ecdsa.readthedocs.io/en/latest/basics.html#signature-formats

I wonder if instead of adding a "r_and_s", we shouldn't add a generic input_format parameter and allow it to have either "DER" (the default), or "raw", for the new option.

Reviewed 6 of 6 files at r1, all commit messages.
Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @GeorgePantelakis)

---

tlsfuzzer/extract.py line 82 at r1 (raw file):

    print(" --raw-sigs FILE Read the signatures from an external file.")
    print("                The file must be in binary format.")
    print(" --r-and-s      Specifies that the signatures in the binary file")

again, probably should use raw format

---

tlsfuzzer/extract.py line 920 at r1 (raw file):

        """Iterator. Read the signatures from file provided"""
        with open(filename if filename else self.sigs, "rb") as sigs_fp:
            if self.r_or_s_size:

instead of increasing indentation level, why not create two new private methods, and call them from here based on the expected signature format?

---

tlsfuzzer/extract.py line 932 at r1 (raw file):

                sig = sigs_fp.read(1)
                while sig:
                    if not ecdsa.der.is_sequence(sig):  # pragma: no cover

no, that's a branch that we ought to have test coverage for

---

tlsfuzzer/extract.py line 939 at r1 (raw file):

                    try:
                        sig_length = ecdsa.der.read_length(length_bytes)[0]
                    except ecdsa.UnexpectedDER:  # pragma: no cover

same here

---

tlsfuzzer/extract.py line 986 at r1 (raw file):

        if self.r_or_s_size:
            r_bytes = sig[:self.r_or_s_size]
            s_bytes = sig[self.r_or_s_size:]

why not use ecdsa.util.sigdecode_string: https://ecdsa.readthedocs.io/en/latest/ecdsa.util.html#ecdsa.util.sigdecode_string ?

[tomato42]

thinking a bit more: the command line option can also be a --sig-format [DER|RAW] instead of a flag, that will be more future proof

[GeorgePantelakis]

Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @tomato42)

---

tlsfuzzer/extract.py line 932 at r1 (raw file):

Previously, tomato42 (Hubert Kario) wrote…

no, that's a branch that we ought to have test coverage for

Can you elaborate a bit more on why this (and next) path needs to have coverage? I mean it is not tlsfuzzer code and in general, it is a very simple misformating error, perhaps I am missing something.

[tomato42]

Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @GeorgePantelakis)

---

tlsfuzzer/extract.py line 932 at r1 (raw file):

Previously, GeorgePantelakis (George Pantelakis) wrote…

Can you elaborate a bit more on why this (and next) path needs to have coverage? I mean it is not tlsfuzzer code and in general, it is a very simple misformating error, perhaps I am missing something.

This error will be raised if the user passes in wrong signature file (i.e. in raw format when the system is configured for DER), or the signature file is damaged.

i.e. it's a user facing check and error, therefore we should have test coverage that this kind of mistake is detected and reported appropriately

[tomato42]

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @GeorgePantelakis)

[tomato42]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @GeorgePantelakis)

[tomato42]

looks good, thanks!