Update cxjs_substring() to match exp_fn_substring()

[Lightning11wins]

I caused a bug in the gift_associations.app because the client-side version of substring() functioned differently from the server side version, as documented in #121. This PR will close #121 by solving this inconsistency issue.

[greptile-apps]

Greptile Summary

This PR updates cxjs_substring() in ht_render.js to align its behavior with the server-side exp_fn_substring() in exp_functions.c, fixing a client/server inconsistency reported in #121.

• Adds explicit type validation (typeof s != 'string', Number.isInteger(p/l)) mirroring the server's DATA_T_STRING/DATA_T_INTEGER checks, and returns null for invalid types instead of coercing via String(s).
• Replaces deprecated substr() with slice() and adds proper clamping of the start position and length to match the server's i < 0 → 0 and i > n → n logic, which also fixes incorrect behavior for zero or negative position arguments.

Confidence Score: 5/5

Safe to merge — the change is narrowly scoped to one JS helper function and correctly mirrors the C reference implementation.

The new logic precisely replicates what exp_fn_substring() does in C: 1-based position adjusted to 0-based, both start and length clamped to valid range, and explicit type checks returning null for invalid input types. Edge cases (zero/negative position, length larger than remaining string, null third argument) were verified against the C implementation and all match. Replacing the deprecated substr() with slice() is a clean improvement with no behavioral difference for valid inputs.

No files require special attention — only one function in one JS file changed.

Important Files Changed

Filename	Overview
centrallix-os/sys/js/ht_render.js	cxjs_substring() updated to match server-side behavior: adds type guards, fixes negative/zero position clamping, replaces deprecated substr() with slice(), and clamps length correctly.

Sequence Diagram

sequenceDiagram
    participant Client as Browser cxjs_substring
    participant Engine as Server exp_fn_substring

    Note over Client,Engine: Before this PR

    Client->>Client: "String(s).substr(p-1) — no clamping"
    Engine->>Engine: "validate types, clamp start and length"

    Note over Client,Engine: After this PR — aligned behavior

    Client->>Client: "typeof s != string → return null"
    Engine->>Engine: "DataType != DATA_T_STRING → return error"
    Client->>Client: "i = p-1, clamp to 0 or s.length"
    Engine->>Engine: "i = i1-1, clamp to 0 or n"
    Client->>Client: "len = l ?? rest.length, clamp to rest.length"
    Engine->>Engine: "i = i2->Integer or strlen(ptr), clamp"
    Client->>Client: "return rest.slice(0, len)"
    Engine->>Engine: "memcpy into tree->String"

Loading

_{Reviews (1): Last reviewed commit: "Update cxjs_substring() to be more simil..." | Re-trigger Greptile}

[Lightning11wins]

This PR is ready for human review.

[gbeeley]

Change needed for string object compatibility.

[gbeeley]

Note that for String objects, typeof returns 'object', not 'string'. Both string objects and string primitives need to work in these functions (for compatibility reasons).

[Lightning11wins]

Oh, I see. Fixed.