atdoc-core 0.1.0

package/dist/Lexer.d.ts

@@ -0,0 +1,23 @@
+export type TokenType = 'NODE' | 'PAREN' | 'STYLES' | 'SLOT_OPEN' | 'SLOT_CLOSE' | 'RAW' | 'TEXT';
+export interface Token {
+    type: TokenType;
+    value: string;
+    /** Source character offsets — lets consumers (e.g. editor diagnostics) map a token back to a range. */
+    start: number;
+    end: number;
+    /** RAW tokens only — false if the scan ran out of input before finding the closing "]" (mid-typing, or a genuinely unterminated block). Lets consumers (e.g. the editor's completion provider) tell "caret still inside raw content" apart from "raw content already closed". */
+    closed?: boolean;
+    /** raw-escaped RAW tokens only — each local escape sequence the scan consumed ("@]", "@[", "@@]", "@@["), with its source range. Lets the Parser point an editor marker at the exact escape when one looks like it swallowed the intended closing bracket. */
+    escapes?: RawEscape[];
+}
+/** One consumed local escape inside @raw's content — see scanDepthRaw. */
+export interface RawEscape {
+    start: number;
+    end: number;
+    /** The literal source characters, e.g. "@]" or "@@[". */
+    seq: string;
+    /** True when the next source character after the escape is a newline — the signature of an escape that consumed a "]" the author meant as the node's end-of-line closer (see Parser.diagnoseRawEscapes). */
+    atLineEnd: boolean;
+}
+export declare function tokenize(source: string): Token[];
+//# sourceMappingURL=Lexer.d.ts.map

package/dist/Lexer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Lexer.d.ts","sourceRoot":"","sources":["../src/Lexer.ts"],"names":[],"mappings":"AAcA,MAAM,MAAM,SAAS,GAAG,MAAM,GAAG,OAAO,GAAG,QAAQ,GAAG,WAAW,GAAG,YAAY,GAAG,KAAK,GAAG,MAAM,CAAC;AAElG,MAAM,WAAW,KAAK;IACpB,IAAI,EAAE,SAAS,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;IACd,uGAAuG;IACvG,KAAK,EAAE,MAAM,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,iRAAiR;IACjR,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,8PAA8P;IAC9P,OAAO,CAAC,EAAE,SAAS,EAAE,CAAC;CACvB;AAED,0EAA0E;AAC1E,MAAM,WAAW,SAAS;IACxB,KAAK,EAAE,MAAM,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,yDAAyD;IACzD,GAAG,EAAE,MAAM,CAAC;IACZ,4MAA4M;IAC5M,SAAS,EAAE,OAAO,CAAC;CACpB;AAsJD,wBAAgB,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,KAAK,EAAE,CA6IhD"}

package/dist/Lexer.js

@@ -0,0 +1,292 @@
+// Lexer — implements Inline Syntax Specification §2 (Lexer behavior definition) precisely:
+//
+//   1. "@@"                       → literal "@" (checked before any registry lookup)
+//   2. "@" + known command name   → NODE token
+//   3. "@" + anything else        → literal text
+//
+// It is registry-aware (see registry.ts) rather than a context-free tokenizer,
+// because raw-content nodes (@code, @mermaid, @raw, @kbd, @fn) require the
+// Lexer to switch into an opaque scan mode for their bracket content — see
+// Inline Syntax Specification §9 (@raw Opaque Domain) and Special-Nodes.md §6.
+import { getNodeDef } from './registry';
+const IDENT_CHAR = /[a-zA-Z0-9_-]/;
+/**
+ * Scans a strong-quoted slot, starting just after the opening "{[".
+ *
+ * The content runs verbatim to the first "]}" — no bracket depth, no escapes,
+ * nothing to get wrong. That is the whole point: the ordinary "[...]" form
+ * terminates on depth, so content whose brackets don't balance cannot be
+ * written at all in @code/@mermaid (which define no escapes) and needs
+ * hand-escaping in @raw. Measured over 1,632 code fences in real READMEs, 2
+ * had unbalanced brackets and 4 contained "]}", with no overlap — so the pair
+ * that can't be written one way can always be written the other.
+ *
+ * A fixed terminator can still be defeated by content that contains "]}", the
+ * same way any fixed delimiter can; that content keeps the "[...]" form.
+ */
+function scanStrongRaw(source, start) {
+    const end = source.indexOf(']}', start);
+    if (end === -1)
+        return { text: source.slice(start), endPos: source.length, closed: false };
+    return { text: source.slice(start, end), endPos: end + 2, closed: true };
+}
+function isRawFamily(mode) {
+    return mode === 'raw' || mode === 'raw-escaped' || mode === 'key' || mode === 'integer';
+}
+/**
+ * Scans raw, unparsed content starting right after the opening "[".
+ * Tracks nested "[" / "]" depth so literal brackets inside code/diagram
+ * content don't prematurely terminate the slot (see Text-Formatting.md §4 Raw
+ * and Widget-Blocks.md §4 Mermaid for why this matters in practice).
+ *
+ * `localEscape` enables @raw's four local exceptions (Inline Spec §9):
+ *   "@]"  → literal "]"
+ *   "@@]" → literal "@]"
+ *   "@["  → literal "["
+ *   "@@[" → literal "@["
+ * These do NOT apply to @code/@mermaid, which define no escape mechanism at all.
+ */
+function scanDepthRaw(source, start, localEscape) {
+    let depth = 1;
+    let buf = '';
+    let i = start;
+    const n = source.length;
+    const escapes = [];
+    while (i < n) {
+        if (localEscape && source[i] === '@' && source[i + 1] === '@' && source[i + 2] === ']') {
+            escapes.push({ start: i, end: i + 3, seq: '@@]', atLineEnd: source[i + 3] === '\n' });
+            buf += '@]';
+            i += 3;
+            continue;
+        }
+        if (localEscape && source[i] === '@' && source[i + 1] === '@' && source[i + 2] === '[') {
+            escapes.push({ start: i, end: i + 3, seq: '@@[', atLineEnd: source[i + 3] === '\n' });
+            buf += '@[';
+            i += 3;
+            continue;
+        }
+        if (localEscape && source[i] === '@' && source[i + 1] === ']') {
+            escapes.push({ start: i, end: i + 2, seq: '@]', atLineEnd: source[i + 2] === '\n' });
+            buf += ']';
+            i += 2;
+            continue;
+        }
+        if (localEscape && source[i] === '@' && source[i + 1] === '[') {
+            escapes.push({ start: i, end: i + 2, seq: '@[', atLineEnd: source[i + 2] === '\n' });
+            buf += '[';
+            i += 2;
+            continue;
+        }
+        const ch = source[i];
+        if (ch === '[') {
+            depth++;
+            buf += ch;
+            i++;
+            continue;
+        }
+        if (ch === ']') {
+            depth--;
+            i++;
+            if (depth === 0)
+                return { text: buf, endPos: i, closed: true, escapes };
+            buf += ch;
+            continue;
+        }
+        buf += ch;
+        i++;
+    }
+    // Ran out of input before depth reached 0 — genuinely unterminated (or, for
+    // the editor's incremental re-tokenize of `textBeforeCaret`, just not typed
+    // that far yet). See Token.closed.
+    return { text: buf, endPos: i, closed: false, escapes };
+}
+/** Flat scan for @kbd's `key` and @fn's `integer` — no nesting, no escapes. */
+function scanFlatRaw(source, start) {
+    let i = start;
+    const n = source.length;
+    let buf = '';
+    while (i < n && source[i] !== ']') {
+        buf += source[i];
+        i++;
+    }
+    if (i >= n)
+        return { text: buf, endPos: i, closed: false }; // ran out before finding "]"
+    return { text: buf, endPos: i + 1, closed: true };
+}
+function scanRawContent(source, start, mode) {
+    if (mode === 'key' || mode === 'integer')
+        return scanFlatRaw(source, start);
+    return scanDepthRaw(source, start, mode === 'raw-escaped');
+}
+/**
+ * Finds where a "{styles}" run ends, scanning from just after the "{".
+ * Returns the index of the terminator and whether it was a real "}".
+ *
+ * Deliberately stricter than the EBNF, which defines the inner set as
+ * `text-char - "}"` and leaves `text-char = any-unicode-char` — i.e. a
+ * {styles} run could technically span lines and swallow anything. Taken
+ * literally, a half-typed "{" (`@h(4){` with the "}" not there yet) eats the
+ * rest of the document up to whatever "}" appears next — a brace inside an
+ * unrelated `@code` block, say — silently deleting every node in between
+ * from the AST. Nothing legitimate needs that reach: a styles run is a short
+ * comma-separated token list, no spec example spans lines, and the editor's
+ * Monarch rule (/\{[^}]*\}/, matched per line) never supported it either.
+ * So the scan stops at:
+ *   "}"            — the real terminator (closed: true)
+ *   end of line    — unterminated; the author is mid-typing
+ *   "["            — unterminated; the content slot has started, and no
+ *                    styles value ever contains "["
+ * Both unterminated cases still produce a STYLES token so the Parser can
+ * flag the slot itself (e.g. "@heading has no {styles} slot"), just one that
+ * stops before it can damage the rest of the document.
+ */
+function scanStylesEnd(source, start) {
+    const n = source.length;
+    let i = start;
+    while (i < n) {
+        const ch = source[i];
+        if (ch === '}')
+            return { end: i, closed: true };
+        if (ch === '\n' || ch === '[')
+            return { end: i, closed: false };
+        i++;
+    }
+    return { end: n, closed: false };
+}
+export function tokenize(source) {
+    const tokens = [];
+    const n = source.length;
+    let i = 0;
+    let textBuf = '';
+    let textStart = -1;
+    const appendText = (s, pos) => {
+        if (textBuf === '')
+            textStart = pos;
+        textBuf += s;
+    };
+    const flushText = () => {
+        if (textBuf) {
+            tokens.push({ type: 'TEXT', value: textBuf, start: textStart, end: textStart + textBuf.length });
+            textBuf = '';
+            textStart = -1;
+        }
+    };
+    while (i < n) {
+        const ch = source[i];
+        if (ch === '@') {
+            // Step 1 (Inline Spec §2): "@@" is checked first, purely by pattern —
+            // before any Command Registry lookup. See Special-Nodes.md §5.
+            if (source[i + 1] === '@') {
+                appendText('@', i);
+                i += 2;
+                continue;
+            }
+            const nodeStart = i;
+            // Step 2: maximal-munch identifier, then registry lookup.
+            let j = i + 1;
+            while (j < n && IDENT_CHAR.test(source[j]))
+                j++;
+            const ident = source.slice(i + 1, j);
+            const nodeDef = ident ? getNodeDef(ident) : undefined;
+            if (!nodeDef) {
+                // Step 3: unknown command → literal text (Inline Spec §6).
+                appendText(source.slice(i, j), i);
+                i = j;
+                continue;
+            }
+            flushText();
+            tokens.push({ type: 'NODE', value: ident, start: nodeStart, end: j });
+            i = j;
+            // Optional "(...)" and "{...}" — modifier/level/title/language/uri/id/
+            // options, and styles respectively. Canonical order is (paren) before
+            // {styles} (Block Syntax Specification §6/§7), but a swapped `{styles}
+            // (title)` must still tokenize as STYLES then PAREN rather than let the
+            // out-of-place "(" decay into literal TEXT — otherwise the Parser never
+            // sees a PAREN token to flag with a specific "wrong order" error, and
+            // authors just get a misleading "expects a content slot" instead.
+            // Grammar excludes ")"/"}" from the inner text-char sets, so a naive
+            // indexOf is faithful for "(...)" (no nesting is syntactically valid,
+            // and Block Spec §5's @img example genuinely spans lines). "{...}" gets
+            // the stricter scanStylesEnd() instead — see there for why.
+            let sawParen = false;
+            let sawStyles = false;
+            while (!sawParen || !sawStyles) {
+                if (!sawParen && source[i] === '(') {
+                    const parenStart = i;
+                    const close = source.indexOf(')', i + 1);
+                    const end = close === -1 ? n : close;
+                    const parenEnd = close === -1 ? n : close + 1;
+                    tokens.push({ type: 'PAREN', value: source.slice(i + 1, end), start: parenStart, end: parenEnd });
+                    i = parenEnd;
+                    sawParen = true;
+                }
+                else if (!sawStyles && source[i] === '{' && !(isRawFamily(nodeDef.content) && source[i + 1] === '[')) {
+                    // "{[" is the strong-quote content opener, not a styles slot — see
+                    // the content handling below. There's no ambiguity to resolve:
+                    // scanStylesEnd() already treats "[" as a terminator because no
+                    // styles value ever contains one, so "{[" could never have opened a
+                    // valid styles slot anyway.
+                    const stylesStart = i;
+                    const { end, closed } = scanStylesEnd(source, i + 1);
+                    const stylesEnd = closed ? end + 1 : end;
+                    tokens.push({ type: 'STYLES', value: source.slice(i + 1, end), start: stylesStart, end: stylesEnd });
+                    i = stylesEnd;
+                    sawStyles = true;
+                }
+                else {
+                    break;
+                }
+            }
+            // Void nodes (@hr, @n) take no slot at all — see registry.ts content: 'none'.
+            if (nodeDef.content === 'none') {
+                continue;
+            }
+            // Strong quote: "{[" ... "]}" runs verbatim to the first "]}", with no
+            // depth counting and no escapes at all. It exists because @code and
+            // @mermaid define no escape mechanism (Inline Spec §9), so a fenced
+            // block whose brackets don't balance — an EBNF grammar quoting "[" and
+            // "]" as terminals, say — simply had no representation and had to be
+            // degraded to an inline @raw, losing both its language tag and its block
+            // rendering. Carrying no escapes, it also emits no escape notices.
+            if (isRawFamily(nodeDef.content) && source[i] === '{' && source[i + 1] === '[') {
+                const slotStart = i;
+                const { text, endPos, closed } = scanStrongRaw(source, i + 2);
+                tokens.push({ type: 'RAW', value: text, start: slotStart, end: endPos, closed });
+                i = endPos;
+            }
+            else if (source[i] === '[') {
+                if (isRawFamily(nodeDef.content)) {
+                    const slotStart = i;
+                    const { text, endPos, closed, escapes } = scanRawContent(source, i + 1, nodeDef.content);
+                    const rawTok = { type: 'RAW', value: text, start: slotStart, end: endPos, closed };
+                    if (escapes && escapes.length > 0)
+                        rawTok.escapes = escapes;
+                    tokens.push(rawTok);
+                    i = endPos;
+                }
+                else {
+                    tokens.push({ type: 'SLOT_OPEN', value: '[', start: i, end: i + 1 });
+                    i++;
+                }
+            }
+            continue;
+        }
+        if (ch === '[') {
+            flushText();
+            tokens.push({ type: 'SLOT_OPEN', value: '[', start: i, end: i + 1 });
+            i++;
+            continue;
+        }
+        if (ch === ']') {
+            flushText();
+            tokens.push({ type: 'SLOT_CLOSE', value: ']', start: i, end: i + 1 });
+            i++;
+            continue;
+        }
+        appendText(ch, i);
+        i++;
+    }
+    flushText();
+    return tokens;
+}
+//# sourceMappingURL=Lexer.js.map

package/dist/Lexer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"Lexer.js","sourceRoot":"","sources":["../src/Lexer.ts"],"names":[],"mappings":"AAAA,2FAA2F;AAC3F,EAAE;AACF,qFAAqF;AACrF,+CAA+C;AAC/C,iDAAiD;AACjD,EAAE;AACF,+EAA+E;AAC/E,2EAA2E;AAC3E,2EAA2E;AAC3E,+EAA+E;AAE/E,OAAO,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AA2BxC,MAAM,UAAU,GAAG,eAAe,CAAC;AAEnC;;;;;;;;;;;;;GAaG;AACH,SAAS,aAAa,CAAC,MAAc,EAAE,KAAa;IAClD,MAAM,GAAG,GAAG,MAAM,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;IACxC,IAAI,GAAG,KAAK,CAAC,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;IAC3F,OAAO,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,MAAM,EAAE,GAAG,GAAG,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;AAC3E,CAAC;AAED,SAAS,WAAW,CAAC,IAAiB;IACpC,OAAO,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,aAAa,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,SAAS,CAAC;AAC1F,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,SAAS,YAAY,CAAC,MAAc,EAAE,KAAa,EAAE,WAAoB;IACvE,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,IAAI,CAAC,GAAG,KAAK,CAAC;IACd,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACxB,MAAM,OAAO,GAAgB,EAAE,CAAC;IAEhC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACb,IAAI,WAAW,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YACvF,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;YACtF,GAAG,IAAI,IAAI,CAAC;YACZ,CAAC,IAAI,CAAC,CAAC;YACP,SAAS;QACX,CAAC;QACD,IAAI,WAAW,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YACvF,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;YACtF,GAAG,IAAI,IAAI,CAAC;YACZ,CAAC,IAAI,CAAC,CAAC;YACP,SAAS;QACX,CAAC;QACD,IAAI,WAAW,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YAC9D,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;YACrF,GAAG,IAAI,GAAG,CAAC;YACX,CAAC,IAAI,CAAC,CAAC;YACP,SAAS;QACX,CAAC;QACD,IAAI,WAAW,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YAC9D,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;YACrF,GAAG,IAAI,GAAG,CAAC;YACX,CAAC,IAAI,CAAC,CAAC;YACP,SAAS;QACX,CAAC;QAED,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QACrB,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,KAAK,EAAE,CAAC;YACR,GAAG,IAAI,EAAE,CAAC;YACV,CAAC,EAAE,CAAC;YACJ,SAAS;QACX,CAAC;QACD,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,KAAK,EAAE,CAAC;YACR,CAAC,EAAE,CAAC;YACJ,IAAI,KAAK,KAAK,CAAC;gBAAE,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;YACxE,GAAG,IAAI,EAAE,CAAC;YACV,SAAS;QACX,CAAC;QACD,GAAG,IAAI,EAAE,CAAC;QACV,CAAC,EAAE,CAAC;IACN,CAAC;IAED,4EAA4E;IAC5E,4EAA4E;IAC5E,mCAAmC;IACnC,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;AAC1D,CAAC;AAED,+EAA+E;AAC/E,SAAS,WAAW,CAAC,MAAc,EAAE,KAAa;IAChD,IAAI,CAAC,GAAG,KAAK,CAAC;IACd,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACxB,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,OAAO,CAAC,GAAG,CAAC,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;QAClC,GAAG,IAAI,MAAM,CAAC,CAAC,CAAC,CAAC;QACjB,CAAC,EAAE,CAAC;IACN,CAAC;IACD,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC,6BAA6B;IACzF,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;AACpD,CAAC;AAED,SAAS,cAAc,CAAC,MAAc,EAAE,KAAa,EAAE,IAAiB;IACtE,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,WAAW,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;IAC5E,OAAO,YAAY,CAAC,MAAM,EAAE,KAAK,EAAE,IAAI,KAAK,aAAa,CAAC,CAAC;AAC7D,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,SAAS,aAAa,CAAC,MAAc,EAAE,KAAa;IAClD,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACxB,IAAI,CAAC,GAAG,KAAK,CAAC;IACd,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACb,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QACrB,IAAI,EAAE,KAAK,GAAG;YAAE,OAAO,EAAE,GAAG,EAAE,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAChD,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,GAAG;YAAE,OAAO,EAAE,GAAG,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;QAChE,CAAC,EAAE,CAAC;IACN,CAAC;IACD,OAAO,EAAE,GAAG,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;AACnC,CAAC;AAED,MAAM,UAAU,QAAQ,CAAC,MAAc;IACrC,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACxB,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,IAAI,OAAO,GAAG,EAAE,CAAC;IACjB,IAAI,SAAS,GAAG,CAAC,CAAC,CAAC;IAEnB,MAAM,UAAU,GAAG,CAAC,CAAS,EAAE,GAAW,EAAE,EAAE;QAC5C,IAAI,OAAO,KAAK,EAAE;YAAE,SAAS,GAAG,GAAG,CAAC;QACpC,OAAO,IAAI,CAAC,CAAC;IACf,CAAC,CAAC;IAEF,MAAM,SAAS,GAAG,GAAG,EAAE;QACrB,IAAI,OAAO,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,GAAG,EAAE,SAAS,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;YACjG,OAAO,GAAG,EAAE,CAAC;YACb,SAAS,GAAG,CAAC,CAAC,CAAC;QACjB,CAAC;IACH,CAAC,CAAC;IAEF,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACb,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QAErB,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,sEAAsE;YACtE,+DAA+D;YAC/D,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBAC1B,UAAU,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC;gBACnB,CAAC,IAAI,CAAC,CAAC;gBACP,SAAS;YACX,CAAC;YAED,MAAM,SAAS,GAAG,CAAC,CAAC;YACpB,0DAA0D;YAC1D,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACd,OAAO,CAAC,GAAG,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;gBAAE,CAAC,EAAE,CAAC;YAChD,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC;YACrC,MAAM,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;YAEtD,IAAI,CAAC,OAAO,EAAE,CAAC;gBACb,2DAA2D;gBAC3D,UAAU,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;gBAClC,CAAC,GAAG,CAAC,CAAC;gBACN,SAAS;YACX,CAAC;YAED,SAAS,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC;YACtE,CAAC,GAAG,CAAC,CAAC;YAEN,uEAAuE;YACvE,sEAAsE;YACtE,uEAAuE;YACvE,wEAAwE;YACxE,wEAAwE;YACxE,sEAAsE;YACtE,kEAAkE;YAClE,qEAAqE;YACrE,sEAAsE;YACtE,wEAAwE;YACxE,4DAA4D;YAC5D,IAAI,QAAQ,GAAG,KAAK,CAAC;YACrB,IAAI,SAAS,GAAG,KAAK,CAAC;YACtB,OAAO,CAAC,QAAQ,IAAI,CAAC,SAAS,EAAE,CAAC;gBAC/B,IAAI,CAAC,QAAQ,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;oBACnC,MAAM,UAAU,GAAG,CAAC,CAAC;oBACrB,MAAM,KAAK,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;oBACzC,MAAM,GAAG,GAAG,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC;oBACrC,MAAM,QAAQ,GAAG,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;oBAC9C,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,EAAE,GAAG,CAAC,EAAE,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,CAAC;oBAClG,CAAC,GAAG,QAAQ,CAAC;oBACb,QAAQ,GAAG,IAAI,CAAC;gBAClB,CAAC;qBAAM,IAAI,CAAC,SAAS,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;oBACvG,mEAAmE;oBACnE,+DAA+D;oBAC/D,gEAAgE;oBAChE,oEAAoE;oBACpE,4BAA4B;oBAC5B,MAAM,WAAW,GAAG,CAAC,CAAC;oBACtB,MAAM,EAAE,GAAG,EAAE,MAAM,EAAE,GAAG,aAAa,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;oBACrD,MAAM,SAAS,GAAG,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;oBACzC,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,EAAE,GAAG,CAAC,EAAE,KAAK,EAAE,WAAW,EAAE,GAAG,EAAE,SAAS,EAAE,CAAC,CAAC;oBACrG,CAAC,GAAG,SAAS,CAAC;oBACd,SAAS,GAAG,IAAI,CAAC;gBACnB,CAAC;qBAAM,CAAC;oBACN,MAAM;gBACR,CAAC;YACH,CAAC;YAED,8EAA8E;YAC9E,IAAI,OAAO,CAAC,OAAO,KAAK,MAAM,EAAE,CAAC;gBAC/B,SAAS;YACX,CAAC;YAED,uEAAuE;YACvE,oEAAoE;YACpE,oEAAoE;YACpE,uEAAuE;YACvE,qEAAqE;YACrE,yEAAyE;YACzE,mEAAmE;YACnE,IAAI,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBAC/E,MAAM,SAAS,GAAG,CAAC,CAAC;gBACpB,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,aAAa,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC9D,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,SAAS,EAAE,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC;gBACjF,CAAC,GAAG,MAAM,CAAC;YACb,CAAC;iBAAM,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBAC7B,IAAI,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;oBACjC,MAAM,SAAS,GAAG,CAAC,CAAC;oBACpB,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,cAAc,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC;oBACzF,MAAM,MAAM,GAAU,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,SAAS,EAAE,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC;oBAC1F,IAAI,OAAO,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC;wBAAE,MAAM,CAAC,OAAO,GAAG,OAAO,CAAC;oBAC5D,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;oBACpB,CAAC,GAAG,MAAM,CAAC;gBACb,CAAC;qBAAM,CAAC;oBACN,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;oBACrE,CAAC,EAAE,CAAC;gBACN,CAAC;YACH,CAAC;YACD,SAAS;QACX,CAAC;QAED,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,SAAS,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACrE,CAAC,EAAE,CAAC;YACJ,SAAS;QACX,CAAC;QACD,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,SAAS,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,KAAK,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACtE,CAAC,EAAE,CAAC;YACJ,SAAS;QACX,CAAC;QAED,UAAU,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;QAClB,CAAC,EAAE,CAAC;IACN,CAAC;IAED,SAAS,EAAE,CAAC;IACZ,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/Parser.d.ts

@@ -0,0 +1,136 @@
+import type { Token } from './Lexer';
+import type { DocASTNode, DocDiagnostic } from './types';
+export declare class DocParser {
+    private tokens;
+    private cursor;
+    /** Non-fatal issues collected during parsing (e.g. an unclosed bracket, a node used in the wrong context). */
+    diagnostics: DocDiagnostic[];
+    constructor(tokens: Token[]);
+    /** Records a non-fatal issue without aborting the parse — see the file-level Editor Mode comment. Severity defaults to 'error' (omitted on the diagnostic itself, preserving the historical shape). */
+    private diagnose;
+    /** Best-effort position for a diagnostic when there's no specific token to blame (e.g. end-of-input). */
+    private cursorPos;
+    parse(): DocASTNode[];
+    private isTopLevelBlock;
+    private parseImplicitParagraph;
+    /**
+     * Parses one node starting at the current NODE token.
+     * `parentType` is the immediate containing node's type (or undefined at
+     * document root) — used to enforce `restrictedTo` (Widget-Blocks.md §3,
+     * Structural-Blocks.md §5 Table: @tab/@cols/@data are only valid inside
+     * their specific parent). A violation is a diagnostic, not a throw — the
+     * node still parses normally, just flagged as misplaced (Editor Mode).
+     */
+    private parseNode;
+    private parseContentByMode;
+    /**
+     * Restructures @list's flat inline content into `list-item` nodes (Block
+     * Syntax Specification §5 List). A line is anything between "\n" boundaries
+     * inside the content's string runs; a DocASTNode segment stays attached to
+     * whichever line it falls on.
+     *
+     * - Every non-blank line is its own item — a leading "- " is optional and
+     *   stripped when present, purely for backward compatibility with the old
+     *   dash-required style; it was never required to make something an item.
+     * - A leading "N. " / "N)" is also optional; when present it's stripped and
+     *   kept as `marker` (only meaningful for @list(ordered), see Adapters.ts,
+     *   letting the numbering jump/resume via <li value>).
+     * - A line that's nothing but a single nested `@list[...]` (plus surrounding
+     *   whitespace) isn't a new item — it's folded into the previous item's
+     *   content as that item's sub-list.
+     * - Blank lines are ignored.
+     */
+    private buildListItems;
+    /** Depth-aware: a literal, unpaired "[" typed as plain text (e.g. "array[0]") stays transparent instead of prematurely closing the slot. */
+    private parseSlotContent;
+    /**
+     * Escape-awareness feedback for @raw (Inline Syntax Specification §9) — the
+     * "advance notice" tier of Editor Mode. Never blocks: the content parses
+     * exactly as the escape rules say either way.
+     *
+     * Two tiers:
+     *
+     *  - warning — the node shows swallow symptoms: it never found its closing
+     *    "]" at all, or an escape sat at the very end of a line the content then
+     *    ran past. The near-certain cause is an escape that consumed the "]" the
+     *    author meant as the node's end (`@raw[@mark[hello@@]]`-style, or a
+     *    trailing "@" fusing with the closer). Scoped to escape-at-line-end
+     *    rather than "any newline in the content" so a deliberate multi-line
+     *    @raw with mid-line escapes (e.g. Markdown import preserving an
+     *    unbalanced code block) stays at the info tier.
+     *
+     *  - info — every other consumed escape gets a quiet heads-up that it *is*
+     *    an escape (a literal bracket that neither ends the node nor counts
+     *    toward depth). This is what catches the cases no heuristic can: an
+     *    accidental `@]` that swallows the rest of its own line still parses
+     *    "successfully", and only the author knows it wasn't meant — the note
+     *    tells them what the Parser did with what they wrote.
+     */
+    private diagnoseRawEscapes;
+    /**
+     * @color/@bordered's {styles} value is semantically validated here purely
+     * for editor feedback (Inline Syntax Specification §7 leaves token *meaning*
+     * to the Renderer — an unrecognized value already falls back gracefully at
+     * render time — but it's almost always a typo, or a missing value, the
+     * author would want flagged, not a silent no-op). An empty `{}` counts as
+     * an invalid value here too, same as a real unrecognized token — it's
+     * flagged right alongside the case where {styles} is missing entirely (see
+     * the call site above).
+     */
+    private diagnoseUnknownColorValue;
+    /**
+     * Like parseSlotContent, but for content modes that only ever hold plain text
+     * (currently just @meta's key=value lines). "@@" already resolves to a literal
+     * "@" at the Lexer level (Inline Spec §2 step 1), so it never reaches here as a
+     * NODE token. Void nodes (@n, content:'none') and raw-family nodes (@raw/@code/
+     * @kbd/..., content:'raw'/'raw-escaped'/'key'/'integer') get a narrow carve-out
+     * since they can't recursively contain more of the very commands this slot
+     * forbids. Every other @command is still invalid here, but instead of throwing
+     * and losing the whole document, it gets fully consumed (to keep the cursor in
+     * sync), silently dropped from the output, and recorded in `diagnostics` — the
+     * editor renders that as a squiggly + hover, the rendered doc simply doesn't
+     * contain it.
+     */
+    private collectRawText;
+    /**
+     * Trims leading/trailing whitespace-only string chunks off a cell's inline
+     * content array — the array equivalent of `str.trim()` for a mixed text/node list.
+     */
+    private trimCellEdges;
+    /**
+     * Parses @cols/@data content as comma-separated cells, each cell holding inline
+     * content (text plus a curated set of formatting nodes — @n, @raw/@code/@kbd/...,
+     * and whatever registry.ts's isCellAllowedNode() lets through, e.g. @bold/@mark/
+     * @link). Anything else is unsupported here: fully consumed (cursor stays in
+     * sync), dropped from the output, and recorded as a diagnostic — same policy as
+     * collectRawText, just producing structured cells instead of a flat string.
+     *
+     * Commas only split cells at this slot's own depth — a comma inside a nested
+     * node's own "[...]" (e.g. `@bold[a,b]`) stays literal, since that TEXT token
+     * is emitted while `depth > 0`.
+     */
+    private parseInlineCellList;
+    private parseDataRows;
+    private skipWhitespaceText;
+    /**
+     * Scans forward tracking SLOT_OPEN/SLOT_CLOSE depth until the bracket that
+     * was already opened by the caller (depth starts at 1, representing that
+     * "[") finds its match — used to abandon a structurally malformed construct
+     * (e.g. `@table` missing `@cols`/`@data`) without losing cursor sync with
+     * the rest of the document. Runs out cleanly at end-of-input.
+     */
+    private skipToMatchingSlotClose;
+    /** Like the old expectSlotOpen, but returns success instead of throwing — a missing "[" is a diagnostic, and the node is left with empty/default content. */
+    private trySlotOpen;
+    /**
+     * Like the old expectSlotClose, but records a diagnostic instead of
+     * throwing when the closing "]" is missing. Every content-collecting loop
+     * that calls this (parseSlotContent, parseInlineCellList, parseDataRows,
+     * the @table/@tabs child loops) only stops without consuming a SLOT_CLOSE
+     * when the token stream itself has run out — so reaching here without one
+     * always means end-of-input, i.e. an unclosed bracket. Editor Mode treats
+     * that as "auto-close at EOF" rather than aborting the whole document.
+     */
+    private closeSlot;
+}
+//# sourceMappingURL=Parser.d.ts.map

package/dist/Parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Parser.d.ts","sourceRoot":"","sources":["../src/Parser.ts"],"names":[],"mappings":"AAYA,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,SAAS,CAAC;AAErC,OAAO,KAAK,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,SAAS,CAAC;AAiBzD,qBAAa,SAAS;IACpB,OAAO,CAAC,MAAM,CAAU;IACxB,OAAO,CAAC,MAAM,CAAK;IAEnB,8GAA8G;IACvG,WAAW,EAAE,aAAa,EAAE,CAAM;IAEzC,YAAY,MAAM,EAAE,KAAK,EAAE,EAE1B;IAED,uMAAuM;IACvM,OAAO,CAAC,QAAQ;IAMhB,yGAAyG;IACzG,OAAO,CAAC,SAAS;IAOV,KAAK,IAAI,UAAU,EAAE,CA+B3B;IAED,OAAO,CAAC,eAAe;IAKvB,OAAO,CAAC,sBAAsB;IAkC9B;;;;;;;OAOG;IACH,OAAO,CAAC,SAAS;IAoGjB,OAAO,CAAC,kBAAkB;IAmI1B;;;;;;;;;;;;;;;;OAgBG;IACH,OAAO,CAAC,cAAc;IAmDtB,4IAA4I;IAC5I,OAAO,CAAC,gBAAgB;IAuCxB;;;;;;;;;;;;;;;;;;;;;;OAsBG;IACH,OAAO,CAAC,kBAAkB;IAwC1B;;;;;;;;;OASG;IACH,OAAO,CAAC,yBAAyB;IAUjC;;;;;;;;;;;;OAYG;IACH,OAAO,CAAC,cAAc;IAqDtB;;;OAGG;IACH,OAAO,CAAC,aAAa;IAiBrB;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,mBAAmB;IA+E3B,OAAO,CAAC,aAAa;IA2BrB,OAAO,CAAC,kBAAkB;IAM1B;;;;;;OAMG;IACH,OAAO,CAAC,uBAAuB;IAa/B,6JAA6J;IAC7J,OAAO,CAAC,WAAW;IAUnB;;;;;;;;OAQG;IACH,OAAO,CAAC,SAAS;CAQlB"}