Serializer: reserve worst-case JSON escape space so values are not dropped

[ManoharPaturi]

Problem

The JSON serializer reserves escape space assuming at most 2x expansion, but ndpi_json_string_escape() can emit 6 bytes per input byte (\u00XX for control characters, 3-byte U+FFFD for invalid bytes):

• src/lib/ndpi_serializer.c (ndpi_serialize_binary_raw): needed += 16 + klen + vlen; — no escape headroom at all
• (ndpi_serialize_uint32_binary): needed += 24 + slen*2 /* account escape (x2) */;

When a value's escaped form exceeds the reserved space, the escape hits its internal bound and returns -1, which the consumers add into u_int32_t size_used — moving the write position backwards, dropping the value, and still returning success. The exported JSON is left truncated/unquoted.

Verified on a dev build: ndpi_serialize_string_string(&s, "key", <2047 '\x01's>) returns rc=0 and produces {"key"} — a record every downstream JSON consumer (ntopng flow export) fails to parse. Smaller control-heavy values vanish the same way once the buffer has grown organically.

Fix

Reserve the worst case (6x) at both sites. With the escape never hitting its bound, the unchecked += can no longer receive -1.

Verification

make unit serializer tests pass; the 2047-control-char case now serializes the full \u0001... sequence into valid JSON (verified in a scratch build).

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[IvanNardi]

@ManoharPaturi, could you rebase, please?

[ManoharPaturi]

@IvanNardi Rebased onto the current dev branch and force-pushed. The diff is unchanged only the two reservation-size adjustments in ndpi_serializer.c. The failing checks were against the four-day-old head; the full matrix is re-running now, and I will follow up if anything still fails.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[ManoharPaturi]

For clarity on the checks: the failing "Tarball/Package/Symbols checks (ubuntu-latest)" job also fails on the current dev branch itself (run 37127699665), so it appears to be pre-existing and unrelated to this change. The remaining matrix jobs are still running on the rebased head; I will keep an eye on them.

[IvanNardi]

For clarity on the checks: the failing "Tarball/Package/Symbols checks (ubuntu-latest)" job also fails on the current dev branch itself (run 37127699665), so it appears to be pre-existing and unrelated to this change. The remaining matrix jobs are still running on the rebased head; I will keep an eye on them.

don't worry about that; the rebase was needed to fix the gcc-4.9 job (completely unrelated to your change)

[IvanNardi]

LGTM, thank you!