prefix-safe-json 0.0.1-alpha.3 → 0.1.0

package/dist/parser.js

@@ -1,949 +0,0 @@
-// ---------------------------------------------------------------------------
-// IncrementalJsonParser — main parser implementation
-// ---------------------------------------------------------------------------
-//
-// Integrates: UTF-8 decoder → Lexer/Scanner → Grammar Stack → Event Builder
-//
-// Public API: push(), snapshot(), drainEvents(), finish()
-// ---------------------------------------------------------------------------
-import { DiagnosticCode } from "./diagnostics/codes.js";
-import { DEFAULT_LIMITS, sanitizeLimit } from "./limits.js";
-import { Utf8Decoder, stringToUtf8 } from "./utf8/decoder.js";
-import { Scanner } from "./lexer/scanner.js";
-import { ScannerState } from "./lexer/states.js";
-import { TokenType } from "./lexer/tokens.js";
-import { GrammarStack } from "./grammar/stack.js";
-import { EventBuilder } from "./semantic/builder.js";
-import { SnapshotBuilder } from "./semantic/snapshot.js";
-import { createDiagnostic } from "./diagnostics/factory.js";
-/**
- * Create a new IncrementalJsonParser.
- */
-export function createParser(options) {
-    return new Parser(options);
-}
-class Parser {
-    limits;
-    utf8;
-    scanner;
-    grammar;
-    events;
-    snapshot_;
-    allDiagnostics = [];
-    allRepairs = [];
-    // Sticky flags mirroring the *existence* of certain diagnostics/repairs,
-    // set unconditionally in addDiagnostic()/addRepair() regardless of
-    // whether the corresponding array is at its maxQueuedEvents cap. Several
-    // checks (isExecutable, determineOutcome) need an exact, monotonic answer
-    // to "did X ever happen for this stream" — re-deriving that by scanning
-    // allDiagnostics/allRepairs is unsound once those arrays can be truncated,
-    // since a disqualifying entry could be the one silently dropped. This is
-    // the same reasoning that already made grammar.hasDuplicate a dedicated
-    // boolean instead of an array scan.
-    everHadFatalDiagnostic = false;
-    everHadNonRecoverableError = false;
-    everHadStructuralOrLossyRepair = false;
-    everHadCloseContainerRepair = false;
-    phase = "collecting";
-    receivedBytes = 0;
-    consumedBytes = 0;
-    rootComplete = false;
-    terminal = false;
-    hasSeenValue = false;
-    syntax_ = "empty";
-    trailingDataSeen = false;
-    trailingDataBytes = 0;
-    rootScalarValue = undefined;
-    repairs;
-    // Track if we're in a duplicate-key value (to skip it)
-    skipValueDepth = 0;
-    isSkippingValue = false;
-    // For finish
-    finishMeta;
-    constructor(options) {
-        const userLimits = options?.limits ?? {};
-        this.limits = {
-            maxInputBytes: sanitizeLimit(userLimits.maxInputBytes, DEFAULT_LIMITS.maxInputBytes),
-            maxDepth: sanitizeLimit(userLimits.maxDepth, DEFAULT_LIMITS.maxDepth),
-            maxStringBytes: sanitizeLimit(userLimits.maxStringBytes, DEFAULT_LIMITS.maxStringBytes),
-            maxQueuedEvents: sanitizeLimit(userLimits.maxQueuedEvents, DEFAULT_LIMITS.maxQueuedEvents),
-            maxTrailingDataBytes: sanitizeLimit(userLimits.maxTrailingDataBytes, DEFAULT_LIMITS.maxTrailingDataBytes),
-        };
-        this.repairs = {
-            rawControlCharacters: options?.repairs?.rawControlCharacters ?? "reject",
-            trailingData: options?.repairs?.trailingData ?? "isolate",
-            closeContainersAtFinish: options?.repairs?.closeContainersAtFinish ?? "disabled",
-        };
-        this.utf8 = new Utf8Decoder();
-        this.scanner = new Scanner(options);
-        this.grammar = new GrammarStack(this.limits.maxDepth);
-        this.events = new EventBuilder(this.limits.maxQueuedEvents);
-        this.snapshot_ = new SnapshotBuilder();
-    }
-    push(chunk) {
-        if (this.phase === "finished") {
-            throw new Error("E_PUSH_AFTER_FINISH: push() called after finish()");
-        }
-        if (this.terminal) {
-            return {
-                acceptedBytes: 0,
-                emittedEvents: 0,
-                syntax: this.syntax_,
-                terminal: true,
-            };
-        }
-        // Convert string to bytes
-        const bytes = typeof chunk === "string" ? stringToUtf8(chunk) : chunk;
-        // Check input limit
-        if (this.receivedBytes + bytes.length > this.limits.maxInputBytes) {
-            const diag = createDiagnostic(DiagnosticCode.E_LIMIT_INPUT_BYTES, "fatal", this.receivedBytes, `Input exceeds maximum of ${this.limits.maxInputBytes} bytes`, false);
-            this.addDiagnostic(diag);
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return {
-                acceptedBytes: 0,
-                emittedEvents: 0,
-                syntax: this.syntax_,
-                terminal: true,
-            };
-        }
-        this.receivedBytes += bytes.length;
-        this.events.resetPushCount();
-        // Decode UTF-8
-        const decoded = this.utf8.decode(bytes);
-        if (decoded.strippedBom) {
-            const repair = {
-                code: "R_STRIP_UTF8_BOM",
-                byteRange: [0, 3],
-                impact: "representation_preserving",
-                description: "Stripped UTF-8 BOM at index 0",
-            };
-            this.addRepair(repair);
-        }
-        // Handle UTF-8 errors
-        for (const err of decoded.errors) {
-            const code = err.kind === "overlong"
-                ? DiagnosticCode.E_INVALID_UTF8
-                : err.kind === "out_of_range"
-                    ? DiagnosticCode.E_INVALID_UTF8
-                    : err.kind === "invalid_start_byte"
-                        ? DiagnosticCode.E_INVALID_UTF8
-                        : DiagnosticCode.E_INVALID_UTF8;
-            const diag = createDiagnostic(code, "error", err.byteOffset, `Invalid UTF-8: ${err.kind} at byte ${err.byteOffset}`, false);
-            this.addDiagnostic(diag);
-            this.syntax_ = "invalid";
-            this.terminal = true;
-            return {
-                acceptedBytes: decoded.consumed,
-                emittedEvents: this.events.emittedDuringPush,
-                syntax: this.syntax_,
-                terminal: true,
-            };
-        }
-        // Feed decoded text to scanner character by character
-        const text = decoded.text;
-        // Track byte positions via UTF-8 encoding
-        let bytePos = this.consumedBytes;
-        for (const ch of text) {
-            if (this.terminal)
-                break;
-            const charBytes = getUtf8ByteLength(ch);
-            // If root is already complete, handle trailing content at parser level
-            if (this.rootComplete) {
-                const isWs = ch === " " || ch === "\t" || ch === "\n" || ch === "\r";
-                if (!isWs) {
-                    if (!this.trailingDataSeen) {
-                        this.trailingDataSeen = true;
-                        if (this.repairs.trailingData === "reject") {
-                            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_TRAILING_DATA, "error", bytePos, `Trailing data rejected`, false));
-                            this.syntax_ = "invalid";
-                            this.terminal = true;
-                            return {
-                                acceptedBytes: decoded.consumed,
-                                emittedEvents: this.events.emittedDuringPush,
-                                syntax: this.syntax_,
-                                terminal: true,
-                            };
-                        }
-                        this.addDiagnostic(createDiagnostic(DiagnosticCode.E_TRAILING_DATA, "warning", bytePos, `Unexpected data after root JSON value`, true));
-                        const repair = {
-                            code: "R_ISOLATE_TRAILING_DATA",
-                            byteRange: [bytePos, bytePos + charBytes],
-                            impact: "root_preserving",
-                            description: "Isolated trailing data after complete JSON root",
-                        };
-                        this.addRepair(repair);
-                    }
-                    this.trailingDataBytes += charBytes;
-                    if (this.trailingDataBytes > this.limits.maxTrailingDataBytes) {
-                        this.addDiagnostic(createDiagnostic(DiagnosticCode.E_LIMIT_INPUT_BYTES, "error", bytePos, `Maximum trailing data bytes exceeded`, false));
-                        this.syntax_ = "invalid";
-                        this.terminal = true;
-                    }
-                }
-                bytePos += charBytes;
-                continue;
-            }
-            // Set up scanner for object key context
-            const frame = this.grammar.current;
-            if (frame &&
-                frame.containerType === "object" &&
-                (frame.objectExpectation === "first_key_or_end" || frame.objectExpectation === "key_after_comma") &&
-                this.scanner.currentState === ScannerState.Structural) {
-                this.scanner.setNextStringIsKey(true);
-            }
-            this.scanner.feedChar(ch, bytePos, charBytes);
-            // Process any emitted tokens
-            const tokens = this.scanner.takeTokens();
-            for (const token of tokens) {
-                if (this.terminal)
-                    break;
-                this.processToken(token);
-            }
-            // Process any scanner diagnostics and repairs
-            const scanDiags = this.scanner.takeDiagnostics();
-            const scanRepairs = this.scanner.takeRepairs();
-            for (const repair of scanRepairs) {
-                this.addRepair(repair);
-            }
-            for (const diag of scanDiags) {
-                this.addDiagnostic(diag);
-                if (diag.severity === "fatal" ||
-                    (diag.severity === "error" && !diag.recoverable)) {
-                    this.syntax_ = "invalid";
-                    this.terminal = true;
-                }
-            }
-            // Process grammar diagnostics
-            const grammarDiags = this.grammar.takeDiagnostics();
-            for (const diag of grammarDiags) {
-                this.addDiagnostic(diag);
-                if (diag.severity === "fatal") {
-                    this.terminal = true;
-                    this.syntax_ = "invalid";
-                }
-            }
-            if (this.events.isTerminal) {
-                this.terminal = true;
-                this.syntax_ = "invalid";
-                break;
-            }
-            bytePos += charBytes;
-        }
-        this.consumedBytes = bytePos;
-        return {
-            acceptedBytes: decoded.consumed,
-            emittedEvents: this.events.emittedDuringPush,
-            syntax: this.syntax_,
-            terminal: this.terminal,
-        };
-    }
-    snapshot() {
-        const pending = this.buildPendingTokens();
-        return {
-            phase: this.phase,
-            syntax: this.syntax_,
-            stableValue: this.buildStableValue(),
-            rootComplete: this.rootComplete,
-            executable: this.isExecutable(),
-            pending,
-            repairs: Object.freeze([...this.allRepairs]),
-            diagnostics: Object.freeze([...this.allDiagnostics]),
-            receivedBytes: this.receivedBytes,
-            consumedBytes: this.consumedBytes,
-        };
-    }
-    drainEvents() {
-        return this.events.drain();
-    }
-    finish(meta) {
-        if (this.phase === "finished") {
-            throw new Error("E_PUSH_AFTER_FINISH: finish() called more than once");
-        }
-        this.phase = "finished";
-        this.finishMeta = meta;
-        const reason = meta?.reason ?? "unknown";
-        // Try to finalize any pending number or literal at stream end
-        if (!this.terminal) {
-            // Check for pending number — at stream end, numbers can be finalized
-            // only if the stream completed normally
-            const pendingInfo = this.scanner.getPendingInfo();
-            if (pendingInfo.type === "number") {
-                if (reason === "complete") {
-                    // Stream ended normally — the number terminator is implicit EOF
-                    const finalized = this.scanner.finalizeNumber();
-                    if (finalized === "finalized") {
-                        const tokens = this.scanner.takeTokens();
-                        for (const token of tokens) {
-                            this.processToken(token);
-                        }
-                    }
-                    else if (finalized === "invalid") {
-                        this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", pendingInfo.byteStart, `Invalid number at stream end: ${pendingInfo.buffer}`, false));
-                        // Unlike every other error branch in this method, this one
-                        // never routed through commitScalar()/the scanner's own
-                        // Invalid-state transition, so nothing else here sets
-                        // terminal/syntax_. Without this, a definitively malformed
-                        // number (leading zero, trailing "."/exponent - not fixable by
-                        // more input, unlike a genuine truncation) fell through
-                        // determineOutcome()'s generic `!rootComplete` case and was
-                        // misreported as outcome "truncated" instead of "invalid" for
-                        // a bare top-level root (e.g. plain "01" or "-01" as the whole
-                        // document) - the same malformation wrapped in a container
-                        // (e.g. `{"a":01}`) already reported "invalid" correctly via
-                        // the scanner's ScannerState.Invalid path.
-                        this.terminal = true;
-                        this.syntax_ = "invalid";
-                    }
-                    // "not_ready": scanner wasn't in a finalizable number state at all
-                    // (e.g. buffer is just "-", or ends right at a bare "e"/"E") —
-                    // preserved exactly as before: no diagnostic added here.
-                }
-                else {
-                    // Truncated — number is incomplete
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_INCOMPLETE_NUMBER, "error", pendingInfo.byteStart, `Incomplete number at stream end: ${pendingInfo.buffer}`, false));
-                }
-            }
-            else if (pendingInfo.type === "literal") {
-                if (reason === "complete" &&
-                    pendingInfo.buffer.length ===
-                        (pendingInfo.buffer === "tru" || pendingInfo.buffer === "fal"
-                            ? 0
-                            : pendingInfo.buffer.length)) {
-                    const finalized = this.scanner.finalizeLiteral();
-                    if (finalized) {
-                        const tokens = this.scanner.takeTokens();
-                        for (const token of tokens) {
-                            this.processToken(token);
-                        }
-                    }
-                    else {
-                        this.addDiagnostic(createDiagnostic(DiagnosticCode.E_INCOMPLETE_LITERAL, "error", pendingInfo.byteStart, `Incomplete literal at stream end: ${pendingInfo.buffer}`, false));
-                    }
-                }
-                else {
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_INCOMPLETE_LITERAL, "error", pendingInfo.byteStart, `Incomplete literal at stream end: ${pendingInfo.buffer}`, false));
-                }
-            }
-            else if (pendingInfo.type === "string" ||
-                pendingInfo.type === "object_key") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNTERMINATED_STRING, "error", pendingInfo.byteStart, "Unterminated string at stream end", false));
-            }
-            else if (pendingInfo.type === "unicode_escape") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_INCOMPLETE_UNICODE_ESCAPE, "error", pendingInfo.byteStart, "Incomplete unicode escape at stream end", false));
-            }
-            else if (pendingInfo.type === "surrogate") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNPAIRED_SURROGATE, "error", pendingInfo.byteStart, "Unpaired surrogate at stream end", false));
-            }
-            // Check for incomplete UTF-8
-            if (this.utf8.hasIncomplete()) {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_INCOMPLETE_UTF8, "error", this.utf8.incompleteByteOffset(), "Incomplete UTF-8 sequence at stream end", false));
-            }
-        }
-        // Check for truncation
-        if (!this.rootComplete &&
-            reason !== "complete") {
-            // Attempt structural salvage
-            let salvaged = false;
-            if (this.repairs.closeContainersAtFinish === "safe-only" &&
-                this.scanner.currentState === ScannerState.Structural &&
-                this.scanner.getPendingInfo().type === "none" &&
-                this.grammar.canSafelyCloseAll() &&
-                !this.hasFatalDiagnostic() &&
-                (this.syntax_ !== "invalid")) {
-                salvaged = true;
-                const frames = this.grammar.getFrames();
-                // We know exactly what to close, just close them all
-                for (let i = frames.length - 1; i >= 0; i--) {
-                    const frame = frames[i];
-                    if (frame) {
-                        this.events.emitContainerClosed(frame.path, frame.containerType);
-                    }
-                }
-                const repair = {
-                    code: "R_CLOSE_CONTAINER",
-                    byteRange: [this.consumedBytes, this.consumedBytes],
-                    impact: "structural",
-                    description: `Safely closed ${frames.length} containers at stream end`,
-                };
-                this.addRepair(repair);
-                this.rootComplete = true;
-            }
-            if (!salvaged) {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_STREAM_TRUNCATED, "error", this.consumedBytes, `Stream truncated: ${reason}`, false));
-            }
-        }
-        else if (!this.rootComplete && reason === "complete") {
-            // Stream said complete but root isn't closed
-            if (this.hasSeenValue) {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_STREAM_TRUNCATED, "error", this.consumedBytes, "Stream marked complete but JSON root is not closed", false));
-            }
-        }
-        const outcome = this.determineOutcome(reason);
-        const executable = this.isExecutable();
-        // Emit finish events
-        if (this.rootComplete && !this.grammar.hasDuplicate && !this.hasFatalDiagnostic()) {
-            this.events.emitDocumentComplete(executable);
-        }
-        this.events.emitStreamFinished(outcome);
-        return {
-            syntax: this.syntax_,
-            outcome,
-            stableValue: this.buildStableValue(),
-            executable,
-            repairs: Object.freeze([...this.allRepairs]),
-            diagnostics: Object.freeze([...this.allDiagnostics]),
-            receivedBytes: this.receivedBytes,
-            consumedBytes: this.consumedBytes,
-        };
-    }
-    // -----------------------------------------------------------------------
-    // Token processing
-    // -----------------------------------------------------------------------
-    processToken(token) {
-        // Handle value skipping (duplicate key value)
-        if (this.isSkippingValue) {
-            if (token.type === TokenType.ObjectStart ||
-                token.type === TokenType.ArrayStart) {
-                this.skipValueDepth++;
-            }
-            else if (token.type === TokenType.ObjectEnd ||
-                token.type === TokenType.ArrayEnd) {
-                this.skipValueDepth--;
-                if (this.skipValueDepth <= 0) {
-                    this.isSkippingValue = false;
-                    this.skipValueDepth = 0;
-                    // Resume normal flow — need to advance the object expectation
-                    const frame = this.grammar.current;
-                    if (frame && frame.containerType === "object") {
-                        frame.objectExpectation = "comma_or_end";
-                    }
-                }
-            }
-            else if (this.skipValueDepth === 0) {
-                // Scalar value for duplicate key — skip it
-                if (token.type === TokenType.String ||
-                    token.type === TokenType.Number ||
-                    token.type === TokenType.True ||
-                    token.type === TokenType.False ||
-                    token.type === TokenType.Null) {
-                    this.isSkippingValue = false;
-                    const frame = this.grammar.current;
-                    if (frame && frame.containerType === "object") {
-                        frame.objectExpectation = "comma_or_end";
-                    }
-                }
-            }
-            return;
-        }
-        switch (token.type) {
-            case TokenType.ObjectStart:
-                this.handleObjectStart(token);
-                break;
-            case TokenType.ObjectEnd:
-                this.handleObjectEnd(token);
-                break;
-            case TokenType.ArrayStart:
-                this.handleArrayStart(token);
-                break;
-            case TokenType.ArrayEnd:
-                this.handleArrayEnd(token);
-                break;
-            case TokenType.Colon:
-                this.handleColon(token);
-                break;
-            case TokenType.Comma:
-                this.handleComma(token);
-                break;
-            case TokenType.String:
-                this.handleString(token);
-                break;
-            case TokenType.Number:
-                this.handleNumber(token);
-                break;
-            case TokenType.True:
-            case TokenType.False:
-            case TokenType.Null:
-                this.handleLiteral(token);
-                break;
-        }
-    }
-    handleObjectStart(token) {
-        const frame = this.grammar.current;
-        // Validate context
-        if (frame) {
-            if (frame.containerType === "object") {
-                if (frame.objectExpectation !== "value") {
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected '{'", false));
-                    this.terminal = true;
-                    this.syntax_ = "invalid";
-                    return;
-                }
-            }
-            else if (frame.containerType === "array") {
-                if (frame.arrayExpectation !== "first_value_or_end" &&
-                    frame.arrayExpectation !== "value_after_comma") {
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected '{' in array", false));
-                    this.terminal = true;
-                    this.syntax_ = "invalid";
-                    return;
-                }
-            }
-        }
-        const newFrame = this.grammar.pushObject(token.byteStart);
-        if (!newFrame) {
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        this.hasSeenValue = true;
-        if (this.syntax_ === "empty") {
-            this.syntax_ = "incomplete";
-        }
-        // Init container in snapshot builder
-        if (!this.grammar.isEmpty()) {
-            if (this.grammar.depth === 1) {
-                this.snapshot_.initRootObject(newFrame);
-            }
-            else {
-                this.snapshot_.initContainerForFrame(newFrame, "object");
-            }
-        }
-    }
-    handleObjectEnd(token) {
-        const frame = this.grammar.current;
-        if (!frame || frame.containerType !== "object") {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected '}' — no matching open object", false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        if (frame.objectExpectation !== "first_key_or_end" &&
-            frame.objectExpectation !== "comma_or_end") {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, `Unexpected '}' while expecting ${frame.objectExpectation}`, false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        this.grammar.pop();
-        this.events.emitContainerClosed(frame.path, "object");
-        // If this closes the root, emit the full object as committed
-        if (this.grammar.isEmpty()) {
-            this.rootComplete = true;
-            this.syntax_ = "root_complete";
-        }
-        else {
-            // Value committed for parent
-            this.commitContainerToParent(frame.path, token);
-        }
-    }
-    handleArrayStart(token) {
-        const frame = this.grammar.current;
-        // Validate context
-        if (frame) {
-            if (frame.containerType === "object") {
-                if (frame.objectExpectation !== "value") {
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected '['", false));
-                    this.terminal = true;
-                    this.syntax_ = "invalid";
-                    return;
-                }
-            }
-            else if (frame.containerType === "array") {
-                if (frame.arrayExpectation !== "first_value_or_end" &&
-                    frame.arrayExpectation !== "value_after_comma") {
-                    this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected '[' in array", false));
-                    this.terminal = true;
-                    this.syntax_ = "invalid";
-                    return;
-                }
-            }
-        }
-        const newFrame = this.grammar.pushArray(token.byteStart);
-        if (!newFrame) {
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        this.hasSeenValue = true;
-        if (this.syntax_ === "empty") {
-            this.syntax_ = "incomplete";
-        }
-        // Init container in snapshot builder
-        if (this.grammar.depth === 1) {
-            this.snapshot_.initRootArray(newFrame);
-        }
-        else {
-            this.snapshot_.initContainerForFrame(newFrame, "array");
-        }
-    }
-    handleArrayEnd(token) {
-        const frame = this.grammar.current;
-        if (!frame || frame.containerType !== "array") {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ']' — no matching open array", false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        if (frame.arrayExpectation !== "first_value_or_end" &&
-            frame.arrayExpectation !== "comma_or_end") {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ']'", false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        this.grammar.pop();
-        this.events.emitContainerClosed(frame.path, "array");
-        if (this.grammar.isEmpty()) {
-            this.rootComplete = true;
-            this.syntax_ = "root_complete";
-        }
-        else {
-            this.commitContainerToParent(frame.path, token);
-        }
-    }
-    handleColon(token) {
-        const frame = this.grammar.current;
-        if (!frame || frame.containerType !== "object" || frame.objectExpectation !== "colon") {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ':'", false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        frame.objectExpectation = "value";
-    }
-    handleComma(token) {
-        const frame = this.grammar.current;
-        if (!frame) {
-            this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ','", false));
-            this.terminal = true;
-            this.syntax_ = "invalid";
-            return;
-        }
-        if (frame.containerType === "object") {
-            if (frame.objectExpectation !== "comma_or_end") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ',' in object", false));
-                this.terminal = true;
-                this.syntax_ = "invalid";
-                return;
-            }
-            frame.objectExpectation = "key_after_comma";
-        }
-        else {
-            if (frame.arrayExpectation !== "comma_or_end") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected ',' in array", false));
-                this.terminal = true;
-                this.syntax_ = "invalid";
-                return;
-            }
-            frame.arrayExpectation = "value_after_comma";
-        }
-    }
-    handleString(token) {
-        const frame = this.grammar.current;
-        if (!frame) {
-            // Root-level string value
-            this.hasSeenValue = true;
-            this.rootComplete = true;
-            this.syntax_ = "root_complete";
-            this.events.emitValueCommitted("", token.value, [
-                token.byteStart,
-                token.byteEnd,
-            ]);
-            this.rootScalarValue = token.value;
-            return;
-        }
-        if (frame.containerType === "object") {
-            if (frame.objectExpectation === "first_key_or_end" ||
-                frame.objectExpectation === "key_after_comma") {
-                // This is an object key
-                const isNew = this.grammar.registerObjectKey(token.value, token.byteStart);
-                if (!isNew) {
-                    // Duplicate key — skip the upcoming value
-                    // The grammar registered the diagnostic already
-                    const dupDiags = this.grammar.takeDiagnostics();
-                    for (const d of dupDiags) {
-                        this.addDiagnostic(d);
-                    }
-                    frame.objectExpectation = "colon";
-                    this.isSkippingValue = true;
-                    this.skipValueDepth = 0;
-                    return;
-                }
-                frame.objectExpectation = "colon";
-            }
-            else if (frame.objectExpectation === "value") {
-                // This is a string value for an object field
-                const path = this.grammar.currentValuePath();
-                this.events.emitValueCommitted(path, token.value, [
-                    token.byteStart,
-                    token.byteEnd,
-                ]);
-                this.snapshot_.processEvent({
-                    type: "value_committed",
-                    sequence: 0,
-                    path,
-                    operation: "add",
-                    value: token.value,
-                    byteRange: [token.byteStart, token.byteEnd],
-                });
-                frame.objectExpectation = "comma_or_end";
-            }
-        }
-        else if (frame.containerType === "array") {
-            if (frame.arrayExpectation !== "first_value_or_end" &&
-                frame.arrayExpectation !== "value_after_comma") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected string in array", false));
-                this.terminal = true;
-                this.syntax_ = "invalid";
-                return;
-            }
-            // Array element
-            const path = this.grammar.currentValuePath();
-            this.events.emitValueCommitted(path, token.value, [
-                token.byteStart,
-                token.byteEnd,
-            ]);
-            this.snapshot_.processEvent({
-                type: "value_committed",
-                sequence: 0,
-                path,
-                operation: "add",
-                value: token.value,
-                byteRange: [token.byteStart, token.byteEnd],
-            });
-            this.grammar.advanceArrayIndex();
-            frame.arrayExpectation = "comma_or_end";
-        }
-    }
-    handleNumber(token) {
-        const value = Number(token.value);
-        this.commitScalar(value, token);
-    }
-    handleLiteral(token) {
-        let value;
-        if (token.type === TokenType.True)
-            value = true;
-        else if (token.type === TokenType.False)
-            value = false;
-        else
-            value = null;
-        this.commitScalar(value, token);
-    }
-    commitScalar(value, token) {
-        const frame = this.grammar.current;
-        if (!frame) {
-            // Root-level scalar
-            this.hasSeenValue = true;
-            this.rootComplete = true;
-            this.syntax_ = "root_complete";
-            this.events.emitValueCommitted("", value, [
-                token.byteStart,
-                token.byteEnd,
-            ]);
-            // Store root scalar in snapshot builder
-            // For root scalars, we directly set the root value
-            this.rootScalarValue = value;
-            return;
-        }
-        if (frame.containerType === "object") {
-            if (frame.objectExpectation === "value") {
-                const path = this.grammar.currentValuePath();
-                this.events.emitValueCommitted(path, value, [
-                    token.byteStart,
-                    token.byteEnd,
-                ]);
-                this.snapshot_.processEvent({
-                    type: "value_committed",
-                    sequence: 0,
-                    path,
-                    operation: "add",
-                    value,
-                    byteRange: [token.byteStart, token.byteEnd],
-                });
-                frame.objectExpectation = "comma_or_end";
-            }
-        }
-        else if (frame.containerType === "array") {
-            if (frame.arrayExpectation !== "first_value_or_end" &&
-                frame.arrayExpectation !== "value_after_comma") {
-                this.addDiagnostic(createDiagnostic(DiagnosticCode.E_UNEXPECTED_TOKEN, "error", token.byteStart, "Unexpected scalar in array", false));
-                this.terminal = true;
-                this.syntax_ = "invalid";
-                return;
-            }
-            const path = this.grammar.currentValuePath();
-            this.events.emitValueCommitted(path, value, [
-                token.byteStart,
-                token.byteEnd,
-            ]);
-            this.snapshot_.processEvent({
-                type: "value_committed",
-                sequence: 0,
-                path,
-                operation: "add",
-                value,
-                byteRange: [token.byteStart, token.byteEnd],
-            });
-            this.grammar.advanceArrayIndex();
-            frame.arrayExpectation = "comma_or_end";
-        }
-    }
-    commitContainerToParent(_containerPath, _token) {
-        const parentFrame = this.grammar.current;
-        if (!parentFrame)
-            return;
-        if (parentFrame.containerType === "object") {
-            // The closed container was a value in the parent object
-            // The value_committed for the object field is implicitly done
-            // when the container closes
-            parentFrame.objectExpectation = "comma_or_end";
-        }
-        else if (parentFrame.containerType === "array") {
-            this.grammar.advanceArrayIndex();
-            parentFrame.arrayExpectation = "comma_or_end";
-        }
-    }
-    // -----------------------------------------------------------------------
-    // Helpers
-    // -----------------------------------------------------------------------
-    addDiagnostic(diag) {
-        // Freeze before storing: snapshot()/finish() freeze the ARRAY they
-        // return (Object.freeze([...this.allDiagnostics])), but that's shallow
-        // - it doesn't protect the Diagnostic objects themselves, which are the
-        // *same* references stored here. Without this, a consumer mutating a
-        // field on a diagnostic object obtained from one snapshot()/finish()
-        // call corrupts this parser's internal history, silently visible in
-        // every later snapshot()/finish() call.
-        Object.freeze(diag);
-        // Cap accumulation independently of the EventBuilder's own queue limit:
-        // a consumer who drains events after every push() keeps that queue
-        // short, which would otherwise let this array grow without bound for
-        // the lifetime of the parser (see limits.maxQueuedEvents).
-        if (diag.severity === "fatal")
-            this.everHadFatalDiagnostic = true;
-        if ((diag.severity === "error" || diag.severity === "fatal") &&
-            !diag.recoverable) {
-            this.everHadNonRecoverableError = true;
-        }
-        if (this.allDiagnostics.length < this.limits.maxQueuedEvents) {
-            this.allDiagnostics.push(diag);
-        }
-        this.events.emitDiagnostic(diag);
-    }
-    addRepair(repair) {
-        // See addDiagnostic()'s matching comment: freeze before storing so
-        // external mutation of a repair object obtained from a prior
-        // snapshot()/finish() call can't corrupt this parser's internal
-        // history. byteRange is a nested array - Object.freeze on the repair
-        // itself doesn't reach into it, so it needs its own freeze.
-        Object.freeze(repair.byteRange);
-        Object.freeze(repair);
-        if (repair.impact === "structural" || repair.impact === "lossy") {
-            this.everHadStructuralOrLossyRepair = true;
-        }
-        if (repair.code === "R_CLOSE_CONTAINER") {
-            this.everHadCloseContainerRepair = true;
-        }
-        if (this.allRepairs.length < this.limits.maxQueuedEvents) {
-            this.allRepairs.push(repair);
-        }
-        this.events.emitRepairApplied(repair);
-    }
-    isExecutable() {
-        if (this.phase !== "finished")
-            return false;
-        if (!this.rootComplete)
-            return false;
-        // this.terminal/syntax_ is the same authoritative "something went
-        // wrong" signal determineOutcome() checks first (syntax_ === "invalid").
-        // Some fatal conditions — e.g. EventBuilder's own event-queue-capacity
-        // cutoff (see semantic/builder.ts's enqueue()) — set these directly
-        // without going through addDiagnostic(), so they don't set
-        // everHadFatalDiagnostic below. Checking them explicitly here closes
-        // that gap instead of only catching diagnostics that happen to have
-        // been routed through addDiagnostic().
-        if (this.terminal || this.syntax_ === "invalid")
-            return false;
-        const reason = this.finishMeta?.reason ?? "unknown";
-        if (reason === "length" ||
-            reason === "network_error" ||
-            reason === "provider_error" ||
-            reason === "cancelled" ||
-            reason === "unknown") {
-            return false;
-        }
-        if (this.grammar.hasDuplicate)
-            return false;
-        if (this.hasFatalDiagnostic())
-            return false;
-        if (this.hasNonRecoverableError())
-            return false;
-        if (this.everHadStructuralOrLossyRepair)
-            return false;
-        const pendingInfo = this.scanner.getPendingInfo();
-        if (pendingInfo.type !== "none")
-            return false;
-        if (this.trailingDataSeen)
-            return false;
-        return true;
-    }
-    hasFatalDiagnostic() {
-        return this.everHadFatalDiagnostic;
-    }
-    hasNonRecoverableError() {
-        return this.everHadNonRecoverableError;
-    }
-    determineOutcome(reason) {
-        if (this.syntax_ === "invalid")
-            return "invalid";
-        if (this.hasFatalDiagnostic())
-            return "invalid";
-        if (this.grammar.hasDuplicate)
-            return "invalid";
-        // Check if salvaged
-        if (this.everHadCloseContainerRepair)
-            return "salvaged";
-        if (!this.rootComplete)
-            return "truncated";
-        if (reason !== "complete" && reason !== "unknown") {
-            // Reason indicates stream was cut off, but root is magically complete?
-            // Since it's complete, the parsed JSON is fully intact.
-            // We consider it valid, but potentially non-executable depending on policy.
-            return "valid";
-        }
-        return "valid";
-    }
-    buildStableValue() {
-        if (this.rootScalarValue !== undefined)
-            return this.rootScalarValue;
-        return this.snapshot_.getStableValue();
-    }
-    buildPendingTokens() {
-        const pending = [];
-        const info = this.scanner.getPendingInfo();
-        if (info.type !== "none") {
-            pending.push({
-                type: info.type === "object_key" ? "object_key" : info.type,
-                path: this.grammar.currentValuePath(),
-                buffered: info.buffer,
-                byteOffset: info.byteStart,
-            });
-        }
-        return pending;
-    }
-}
-/**
- * Get the UTF-8 byte length of a character.
- */
-function getUtf8ByteLength(ch) {
-    const code = ch.codePointAt(0) ?? 0;
-    if (code <= 0x7f)
-        return 1;
-    if (code <= 0x7ff)
-        return 2;
-    if (code <= 0xffff)
-        return 3;
-    return 4;
-}
-//# sourceMappingURL=parser.js.map