feat: implement comprehensive ANSI/NIST support (NIEM XML & Traditional) by Drawcris · Pull Request #145 · BiometricsUBB/Forensic-Biometrics-Studio

Drawcris · 2026-05-05T17:43:42Z

Key Features Added

1. Modern XML (NIEM) Support

Import/Export: Full support for ANSI/NIST-ITL 1-2025 XML records.
NIEM Compatibility: Seamlessly parses and generates NIEM-compliant biometric data packages.

2. Legacy Traditional Binary Support (.an2 / .eft)

Robust Binary Parser: Implemented a native TypeScript parser based on Record Length headers to prevent data corruption from internal delimiter collisions.
Legacy Records: Added support for Record Types 1 (Header), 4 (High-Res Grayscale), 9 (Minutiae), 13 (Latent Image), and 14 (Tenprint Image).
Type-4 Support: Specialized handling for legacy fixed-binary header records.

3. Image Decoding & Biometric Overlay

WSQ Support: Integrated WSQ image decoding via @li0ard/wsq.
Raw Grayscale Conversion: Added on-the-fly conversion of raw grayscale pixel data to PNG for browser-native rendering.
Minutiae Extraction: Automatic parsing of Type-9 minutiae with coordinate and angle translation, overlaid directly onto the biometric image.

4. UI/UX Improvements

Integrated Menus: Added dedicated "Load ANSI/NIST (XML)" and "Load Traditional ANSI/NIST" options to the canvas header.
Localization: Full translation support (EN/PL) for all new features and tooltips.

Technical Changes

Added loadAnsiNistWithDialog.ts and loadTraditionalAnsiNistWithDialog.ts for specialized loading logic.
Implemented a binary-safe record extraction utility to handle non-textual image data safely.
Updated canvas-header.tsx and translation files.
Ensured strict TypeScript compliance and linting (SonarJS/Security rules).

Verification

Validated against the BioCTS conformance dataset (both pass/fail scenarios).
Tested with real-world .an2 files containing compressed WSQ and raw grayscale images.

…al Binary)

Drawcris

All code review feedback has been addressed. Added i18n, normalized minutiae angles, optimized XML queries, and pinned the WSQ library version.

DGrygierczyk

Few comments from my side. Feel free to disagree or resolve if not applicable as im not really deep into this repository

DGrygierczyk · 2026-06-18T10:34:15Z

+        const { source } = sprite.texture.baseTexture.resource;
+        if (source) {
+            ctx.drawImage(source, 0, 0);
+            return canvas.toDataURL("image/jpeg").split(",")[1] ?? null;


image is exported as lossy JPEG - this bakes compression artifacts into evidence-grade fingerprint data. Use lossless PNG (the load path already decodes PNG):

return canvas.toDataURL("image/png").split(",")[1] ?? null;

DGrygierczyk · 2026-06-18T10:34:15Z

+            angleDeg = ((Math.round(angleDeg) % 360) + 360) % 360;
+
+            let categoryCode = "UNK";
+            if (m.typeId === "e6cbde52-5a18-4236-8287-7a1daf941ba9") {


These two UUIDs are duplicated from autoMarkWithSourceafis.ts, where they already exist as TYPE_ID_RIDGE_ENDING / TYPE_ID_BIFURCATION (just not exported). Export them at the source and import here so the ids can't drift:

// autoMarkWithSourceafis.ts export const TYPE_ID_RIDGE_ENDING = "e6cbde52-5a18-4236-8287-7a1daf941ba9"; export const TYPE_ID_BIFURCATION = "f47c4b97-2d62-4959-aa21-edebfa7a756a";

DGrygierczyk · 2026-06-18T10:34:15Z

+
+    while (pos < buffer.length) {
+        const { len, rType, isTagValue } = getRecordInfo(buffer, pos);
+        if (len <= 0 || pos + len > buffer.length) break;


If the length field is malformed, parseInt returns NaN. NaN <= 0 is false, so this guard passes, then pos += len makes pos NaN and the loop exits - silently dropping every remaining record. Add a finite check:

if (!Number.isFinite(len) || len <= 0 || pos + len > buffer.length) break;

DGrygierczyk · 2026-06-18T10:34:15Z

+    }
+
+    if (minutiaeNodes.length > 0) {
+        const importedMarkings = minutiaeNodes.map((node, index) => {


This minutiae→RayMarking mapping (coord parsing, the (angleDeg - 90) * π/180 conversion, type resolution + fallback id) is duplicated in loadTraditionalAnsiNistWithDialog.ts. Extract a shared helper so the angle/type logic lives in one place and can't diverge.

DGrygierczyk · 2026-06-18T10:34:15Z

-    const defaultMarkingsFilePath = `${filePath}.json`;
-    if (await exists(defaultMarkingsFilePath)) {
-        await loadMarkingsData(defaultMarkingsFilePath, canvasId);
+    if (typeof filePathOrData === "string") {


Both branches run the same reset trio when no markings file applies. Collapse to one path:

const markingsPath = typeof filePathOrData === "string" ? `${filePathOrData}.json` : null; if (markingsPath && (await exists(markingsPath))) { await loadMarkingsData(markingsPath, canvasId); } else { MarkingsStore(canvasId).actions.markings.reset(); MarkingsStore(canvasId).actions.labelGenerator.reset(); MarkingsStore(getOppositeCanvasId(canvasId)).actions.labelGenerator.reset(); }

DGrygierczyk · 2026-06-18T10:34:15Z

+            const typeId =
+                resolveSourceafisTypeId(typeStr) ||
+                MarkingTypesStore.state.types[0]?.id ||
+                "unknown-type-id";


Fallback id is "unknown-type-id" here but "unknown" in the traditional loader - both are invalid ids that yield markings the UI can't categorize. Pick one real default (e.g. first configured marking type) and use it consistently, or skip the minutia.

DGrygierczyk · 2026-06-18T10:42:03Z

+/**
+ * Parses fields within a record buffer.
+ */


do we need all of those comments for those lines and all of the below ones? the method names and return values should be descriptive enough for all devs to quickly get a grasp what the method do (which for me they are). Not sure what is the convention in repo so @Drawcris and @TheMultii feel free to disagree on that one. Some of them such as the one in line 117 etc. are fine as they are describing not descriptive things but rest of them for me is just a bloat

I've got a fairly large branch, just waiting for the semester to be over. It removes all those unnecessary comments (emojis included - along with some other stuff). Adding new ones at this point is pointless for me, especially when they add zero value and the functions are self-explanatory anyway.

Drawcris · 2026-06-18T14:51:26Z

All feedback from the code review has been addressed. I switched the image export to lossless PNG, deduplicated the UUID constants, extracted a shared minutiae mapping helper, added a safety check for malformed record lengths to prevent infinite loops, and consolidated the reset logic.

TheMultii · 2026-06-19T13:54:02Z

@Drawcris resolve conflicts

Drawcris · 2026-06-19T18:26:52Z

@TheMultii Conflicts have been resolved.

feat: implement comprehensive ANSI/NIST support (NIEM XML & Tradition…

f3b814e

…al Binary)

TheMultii requested changes May 22, 2026

View reviewed changes

refactor: address code review comments for ANSI/NIST support

6f3b356

Drawcris commented Jun 5, 2026

View reviewed changes

DGrygierczyk reviewed Jun 18, 2026

View reviewed changes

refactor: address code review comments for ANSI/NIST support

2d19960

Drawcris requested a review from TheMultii June 18, 2026 14:51

chore: resolve conflicts with upstream/master

34ce3a0

TheMultii merged commit 380756d into BiometricsUBB:master Jun 20, 2026

Conversation

Drawcris commented May 5, 2026

Key Features Added

1. Modern XML (NIEM) Support

2. Legacy Traditional Binary Support (.an2 / .eft)

3. Image Decoding & Biometric Overlay

4. UI/UX Improvements

Technical Changes

Verification

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Drawcris left a comment

Choose a reason for hiding this comment

Uh oh!

DGrygierczyk left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Drawcris commented Jun 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

TheMultii commented Jun 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Drawcris commented Jun 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Drawcris commented Jun 18, 2026 •

edited

Loading

TheMultii commented Jun 19, 2026 •

edited

Loading

Drawcris commented Jun 19, 2026 •

edited

Loading