Parser Kmp

Parser KMP Skill

Critical: Do Not Trust Internal Knowledge

Scandit's Kotlin Multiplatform (KMP) SDK is new, shipping in 8.6. Your training data almost certainly predates it and contains zero reliable knowledge of its API — do not pattern-match it against the Android or iOS native SDKs you may know. The KMP API packages are com.kmp.datacapture.* (NOT com.scandit.datacapture.*), and shapes diverge from both native SDKs in specific ways (see below).

Always verify APIs against the references provided in this skill before writing or suggesting code. Do not rely on memorized method signatures, parameters, or property names from any other Scandit platform. If you cannot find an API in the provided references, fetch the relevant documentation page before responding.

KMP-specific gotchas worth flagging:

• Import root is com.kmp.datacapture.parser.* — e.g. com.kmp.datacapture.parser.Parser, com.kmp.datacapture.parser.ParserDataFormat. Never write com.scandit.datacapture.parser.* in KMP code.
• Creation is the companion factory Parser.forFormat(dataCaptureContext, format) — takes the shared DataCaptureContext and a ParserDataFormat value. There is no bare constructor, no Parser.create(format), and no Parser.fromJson(...) on KMP.
• ParserDataFormat has exactly seven values: GS1_AI, HIBC, SWISS_QR, VIN, IATA_BCBP, GS1_DIGITAL_LINK, EPC. There is no AAMVA value — do not invent one. AAMVA driver's license data is handled by ID Capture, not this Parser.
• parser.parseString(data: String) and parser.parseRawData(data: ByteArray) both return a ParsedData and are declared @Throws(Exception::class). On failure they throw ParserException (a plain RuntimeException subclass common to both platforms — iOS has no native parser-exception type, so its NSError is mapped into this same common ParserException under the hood). Catch ParserException first (for code, message, additionalInfo) and fall back to Exception only for anything unexpected.
• ParsedData exposes fields: List<ParsedField>, fieldsByName: Map<String, ParsedField>, fieldsWithIssues: List<ParsedField> (fields that have warnings), and jsonString: String. There is no top-level "issues" list on ParsedData itself — per-field warnings live on ParsedField.warnings.
• ParsedField exposes name: String, parsed: Any?, rawString: String, and warnings: List<ParserIssue>. parsed's actual runtime type (string, number, map, etc.) is field-specific — consult the format's field documentation before casting it.
• ParserIssue exposes only code: ParserIssueCode and message: String — there is no severity or field-name property on the issue itself (the field it belongs to is implied by which ParsedField.warnings list it came from). ParserIssueCode and ParserIssueAdditionalInfoKey are plain Kotlin enums shared across platforms.
• parser.setOptions(options: Map<String, Any>) takes a plain Kotlin Map, not a per-platform options builder. Available keys are format-specific and only meaningful for the formats that support them (e.g. GS1 AI: allowHumanReadableCodes, strictMode; Swiss QR: strictMode, minimalVersion; VIN: strictMode, falsePositiveCompensation; GS1 Digital Link: strictMode, outputHumanReadableString). HIBC, IATA BCBP, and EPC currently take no options.
• Parsing itself needs no camera and no camera permission — a Parser only needs a DataCaptureContext constructed with a license key that has the parser feature enabled. Camera setup, AndroidManifest.xml/Info.plist permission entries, and a scanning mode are only needed if you're also capturing barcodes to feed the parser — for that, route the user to barcode-capture-kmp or sparkscan-kmp.
• The license key placeholder is exactly -- ENTER YOUR SCANDIT LICENSE KEY HERE --.

Intent Routing

Based on the user's request, load the appropriate reference file before responding:

• Creating a Parser, parsing a data string/raw data, reading parsed fields, handling parser issues/exceptions, or combining Parser with a barcode scan result (e.g. "add the Parser to my KMP app", "parse this GS1 barcode", "how do I read a HIBC field", "why did parseString throw", "parse the barcode I just scanned") → read references/integration.md and follow the instructions there.
• Setting up the camera, BarcodeCapture, or SparkScan itself (not the parser) → this skill only covers the Parser half; hand the scanning setup off to barcode-capture-kmp or sparkscan-kmp.
• AAMVA / driver's license barcode field extraction → this is not handled by Parser; route to an id-capture-* skill instead.

API Usage Policy

Only use APIs that are explicitly documented in the Scandit references below. Do not invent or guess method signatures, parameters, or property names — and never carry over a signature from the Android-native or iOS-native SDKs without verifying it also holds for KMP. If unsure whether an API exists or how it is called — or if a compile error occurs — fetch the relevant reference page before responding. Do not tell the user to check the docs themselves. After answering, always include the relevant link so the user can explore further.

Never construct or guess documentation URLs. When you need a specific class or property's API page:

1. First check whether the page you already fetched contains a direct hyperlink to it — topic pages link directly to relevant API symbols. Always request links alongside content in your fetch prompt.
2. If no direct link was found, fetch the API index (see Full API reference in the table below), extract the actual link from it, and follow that.

URL structures can vary and guessing will lead to 404s.

References

Direct users to the right resource based on their question: