mdld-parse 0.7.3 → 0.7.5

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "mdld-parse",
-	"version": "0.7.3",
+	"version": "0.7.5",
 	"description": "A standards-compliant parser for **MD-LD (Markdown-Linked Data)** — a human-friendly RDF authoring format that extends Markdown with semantic annotations.",
 	"type": "module",
 	"main": "index.js",
@@ -38,8 +38,5 @@
 	"homepage": "https://mdld.js.org",
 	"bugs": {
 		"url": "https://github.com/davay42/mdld-parse/issues"
-	},
-	"dependencies": {
-		"rdfa-parse": "^1.0.1"
 	}
 }

package/src/constants.js

@@ -0,0 +1,30 @@
+/**
+ * Shared utilities for MD-LD Parser and Renderer
+ * Ensures DRY code and consistent CommonMark processing
+ */
+
+
+
+export const DEFAULT_CONTEXT = {
+    '@vocab': "http://www.w3.org/2000/01/rdf-schema#",
+    rdf: 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
+    rdfs: 'http://www.w3.org/2000/01/rdf-schema#',
+    xsd: 'http://www.w3.org/2001/XMLSchema#',
+    sh: "http://www.w3.org/ns/shacl#",
+    prov: 'http://www.w3.org/ns/prov#'
+};
+
+// CommonMark patterns - shared between parser and renderer
+export const URL_REGEX = /^(https?|ftp|mailto|tag|nih|urn|uuid|did|web|ipfs|ipns|data|file|urn:uuid):/;
+export const FENCE_REGEX = /^(`{3,}|~{3,})(.*)/;
+export const PREFIX_REGEX = /^\[([^\]]+)\]\s*<([^>]+)>/;
+export const HEADING_REGEX = /^(#{1,6})\s+(.+?)(?:\s*(\{[^}]+\}))?$/;
+export const UNORDERED_LIST_REGEX = /^(\s*)([-*+]|\d+\.)\s+(.+?)(?:\s*(\{[^}]+\}))?\s*$/;
+export const BLOCKQUOTE_REGEX = /^>\s+(.+?)(?:\s*(\{[^}]+\}))?$/;
+export const STANDALONE_SUBJECT_REGEX = /^\s*\{=(.*?)\}\s*$/;
+
+// Pre-compiled carrier patterns for performance
+export const CARRIER_PATTERN_ARRAY = [
+    ['EMPHASIS', /[*__`]+(.+?)[*__`]+\s*\{([^}]+)\}/y],
+    ['CODE_SPAN', /``(.+?)``\s*\{([^}]+)\}/y]
+];

package/src/generate.js

@@ -1,21 +1,26 @@
 import { shortenIRI, expandIRI, DataFactory } from './utils.js';
-import { DEFAULT_CONTEXT } from './shared.js';
-
-// Helper functions for cleaner term type checking
-function isLiteral(term) {
-    return term?.termType === 'Literal';
-}
-
-function isNamedNode(term) {
-    return term?.termType === 'NamedNode';
-}
-
-function isRdfType(term) {
-    return term?.value === 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type';
-}
-
+import { DEFAULT_CONTEXT } from './constants.js';
+import {
+    isLiteral,
+    collectUsedPrefixes,
+    sortQuadsByPredicate,
+    generatePrefixDeclaration,
+    generateLiteralText,
+    generateObjectText,
+    filterQuadsByType
+} from './shared.js';
+
+export function extractLocalName(iri, ctx = {}) {
+    if (!iri) return iri;
+
+    // Check for exact prefix matches first
+    for (const [prefix, namespace] of Object.entries(ctx)) {
+        if (iri.startsWith(namespace) || iri.startsWith(namespace.slice(0, -1))) {
+            return iri.substring(namespace.length);
+        }
+    }
 
-function extractLocalName(iri) {
+    // Fallback to original logic for local names
     const separators = ['#', '/', ':'];
     for (const sep of separators) {
         const lastSep = iri.lastIndexOf(sep);
@@ -83,14 +88,14 @@ function groupQuadsBySubject(quads) {
 
 function buildDeterministicMDLD(subjectGroups, context) {
     let text = '';
+    const usedPrefixes = collectUsedPrefixes(subjectGroups, context);
 
     // Add prefixes first (deterministic order), but exclude default context prefixes
     const sortedPrefixes = Object.entries(context).sort(([a], [b]) => a.localeCompare(b));
     for (const [prefix, namespace] of sortedPrefixes) {
         // Skip default context prefixes - they're implicit in MDLD
-        if (prefix !== '@vocab' && !prefix.startsWith('@') && !DEFAULT_CONTEXT[prefix]) {
-            const prefixDecl = `[${prefix}] <${namespace}>\n`;
-            text += prefixDecl;
+        if (prefix !== '@vocab' && !prefix.startsWith('@') && !DEFAULT_CONTEXT[prefix] && usedPrefixes.has(prefix)) {
+            text += generatePrefixDeclaration(prefix, namespace);
         }
     }
 
@@ -105,47 +110,25 @@ function buildDeterministicMDLD(subjectGroups, context) {
         const subjectQuads = subjectGroups.get(subjectIRI);
         const shortSubject = shortenIRI(subjectIRI, context);
 
-        // Separate types, literals, and objects using helper functions
-        const types = subjectQuads.filter(q => isRdfType(q.predicate));
-        const literals = subjectQuads.filter(q => isLiteral(q.object) && !isRdfType(q.predicate));
-        const objects = subjectQuads.filter(q => isNamedNode(q.object) && !isRdfType(q.predicate));
+        // Separate types, literals, and objects using shared utility
+        const { types, literals, objects } = filterQuadsByType(subjectQuads);
 
         // Generate heading
-        const localSubjectName = extractLocalName(subjectIRI);
+        const localSubjectName = extractLocalName(subjectIRI, context);
         const typeAnnotations = types.length > 0
-            ? ' ' + types.map(t => '.' + extractLocalName(t.object.value)).sort().join(' ')
+            ? ' ' + types.map(t => '.' + shortenIRI(t.object.value, context)).sort().join(' ')
             : '';
 
-        const headingText = `# ${localSubjectName} {=${shortSubject}${typeAnnotations}}\n\n`;
-
-        text += headingText;
-
-        // Add literals (deterministic order)
-        const sortedLiterals = literals.sort((a, b) => a.predicate.value.localeCompare(b.predicate.value));
-        for (const quad of sortedLiterals) {
-            const predShort = shortenIRI(quad.predicate.value, context);
-            let annotation = predShort;
-
-            // Use DataFactory XSD constants for datatype comparison
-            const xsdString = 'http://www.w3.org/2001/XMLSchema#string';
-            if (quad.object.language) {
-                annotation += ` @${quad.object.language}`;
-            } else if (quad.object.datatype.value !== xsdString) {
-                annotation += ` ^^${shortenIRI(quad.object.datatype.value, context)}`;
-            }
+        text += `# ${localSubjectName} {=${shortSubject}${typeAnnotations}}\n`;
 
-            const literalText = `[${quad.object.value}] {${annotation}}\n`;
-            text += literalText;
-        }
+        // Add literals and objects using shared utilities
+        sortQuadsByPredicate(literals).forEach(quad => {
+            text += generateLiteralText(quad, context);
+        });
 
-        // Add objects (deterministic order)
-        const sortedObjects = objects.sort((a, b) => a.predicate.value.localeCompare(b.predicate.value));
-        for (const quad of sortedObjects) {
-            const objShort = shortenIRI(quad.object.value, context);
-            const predShort = shortenIRI(quad.predicate.value, context);
-            const objectText = `[${objShort}] {+${objShort} ?${predShort}}\n`;
-            text += objectText;
-        }
+        sortQuadsByPredicate(objects).forEach(quad => {
+            text += generateObjectText(quad, context);
+        });
 
         text += '\n';
     }

package/src/index.js

@@ -3,7 +3,7 @@ export { merge } from './merge.js';
 export { generate } from './generate.js';
 export { locate } from './locate.js';
 export { render } from './render.js';
-export { DEFAULT_CONTEXT } from './shared.js';
+export { DEFAULT_CONTEXT } from './constants.js';
 export {
     DataFactory,
     hash,

package/src/locate.js

@@ -18,21 +18,6 @@ export function locate(quad, origin) {
         return null;
     }
 
-    // Find the origin entry in quadIndex
-    const entry = origin.quadIndex.get(quadKey);
-    if (!entry) {
-        return null;
-    }
-
-    // Return the lean origin entry structure
-    return {
-        blockId: entry.blockId,
-        range: entry.range,
-        carrierType: entry.carrierType,
-        subject: entry.subject,
-        predicate: entry.predicate,
-        context: entry.context,
-        value: entry.value,
-        polarity: entry.polarity
-    };
+    // Return the origin entry directly - no need to create new object
+    return origin.quadIndex.get(quadKey) || null;
 }

package/src/merge.js

@@ -1,15 +1,14 @@
 import { parse } from './parse.js';
-import { DEFAULT_CONTEXT } from './shared.js';
+import { quadToKeyForOrigin } from './utils.js';
+import { DEFAULT_CONTEXT } from './constants.js';
 
 /**
- * Creates a unique key for quad identity matching
+ * Creates a unique key for quad identity matching - using shared utility
  * @param {Quad} quad 
  * @returns {string}
  */
 function quadKey(quad) {
-    const datatype = quad.object.datatype?.value || '';
-    const language = quad.object.language || '';
-    return `${quad.subject.value}|${quad.predicate.value}|${quad.object.value}|${datatype}|${language}`;
+    return quadToKeyForOrigin(quad);
 }
 
 /**
@@ -42,6 +41,7 @@ export function merge(docs, options = {}) {
     const allDocuments = [];
     const quadIndex = new Map();
     const allStatements = []; // Collect statements from all documents
+    const accumulatedContext = new Map(); // Track all unique prefixes across documents
 
     // Process each document in order
     for (let i = 0; i < docs.length; i++) {
@@ -53,6 +53,16 @@ export function merge(docs, options = {}) {
         // Normalize input to ParseResult
         const doc = normalizeInput(input, options, docContext);
 
+        // Accumulate context from this document
+        if (doc.context) {
+            for (const [prefix, namespace] of Object.entries(doc.context)) {
+                // Don't override default context entries unless explicitly provided in options
+                if (!accumulatedContext.has(prefix) && !DEFAULT_CONTEXT[prefix]) {
+                    accumulatedContext.set(prefix, namespace);
+                }
+            }
+        }
+
         // Create document origin
         const documentOrigin = {
             index: i,
@@ -74,14 +84,12 @@ export function merge(docs, options = {}) {
             sessionBuffer.set(key, quad);
 
             // Create quad origin with document index and polarity
-            const existingOrigin = doc.origin.quadIndex.get(quadKey(quad));
-            if (existingOrigin) {
-                quadIndex.set(quadKey(quad), {
-                    ...existingOrigin,
-                    documentIndex: i,
-                    polarity: '+'
-                });
-            }
+            const existingOrigin = doc.origin.quadIndex.get(key);
+            quadIndex.set(key, {
+                ...(existingOrigin || {}),
+                documentIndex: i,
+                polarity: '+'
+            });
         }
 
         // Fold retractions
@@ -97,14 +105,12 @@ export function merge(docs, options = {}) {
             }
 
             // Create quad origin for remove quads
-            const existingOrigin = doc.origin.quadIndex.get(quadKey(quad));
-            if (existingOrigin) {
-                quadIndex.set(quadKey(quad), {
-                    ...existingOrigin,
-                    documentIndex: i,
-                    polarity: '-'
-                });
-            }
+            const existingOrigin = doc.origin.quadIndex.get(key);
+            quadIndex.set(key, {
+                ...(existingOrigin || {}),
+                documentIndex: i,
+                polarity: '-'
+            });
         }
     }
 
@@ -119,7 +125,11 @@ export function merge(docs, options = {}) {
     };
 
     // Build final context (union of all contexts)
-    const finalContext = { ...DEFAULT_CONTEXT, ...options.context };
+    const finalContext = {
+        ...DEFAULT_CONTEXT,
+        ...options.context,
+        ...Object.fromEntries(accumulatedContext)
+    };
 
     // Enforce hard invariant
     const quadKeys = new Set(finalQuads.map(quadKey));

package/src/parse.js

@@ -1,93 +1,128 @@
 import {
     DataFactory,
     expandIRI,
-    parseSemanticBlock,
     quadIndexKey,
     createLiteral,
     hash
 } from './utils.js';
-import { DEFAULT_CONTEXT, URL_REGEX, FENCE_REGEX, PREFIX_REGEX, HEADING_REGEX, UNORDERED_LIST_REGEX, BLOCKQUOTE_REGEX, STANDALONE_SUBJECT_REGEX, INLINE_CARRIER_PATTERNS } from './shared.js';
+import {
+    DEFAULT_CONTEXT,
+    URL_REGEX,
+    FENCE_REGEX,
+    PREFIX_REGEX,
+    HEADING_REGEX,
+    UNORDERED_LIST_REGEX,
+    BLOCKQUOTE_REGEX,
+    STANDALONE_SUBJECT_REGEX,
+    CARRIER_PATTERN_ARRAY,
+
+} from './constants.js';
+import {
 
-// Cache for fence regex patterns to avoid recreation
-const FENCE_CLOSE_PATTERNS = new Map();
+    getFenceClosePattern,
+    calcRangeInfo,
+    calcAttrsRange,
+    createToken,
+    createCarrier,
+    createListToken,
+    parseSemCached,
+    parseLangAndAttrs,
+    findMatchingBracket,
+    extractUrlFromBrackets,
+    extractAttributesFromText,
+    determineCarrierType,
+    calcCarrierRanges,
+    extractCleanText,
+    RDF_TYPE,
+    RDF_STATEMENT,
+    RDF_SUBJECT,
+    RDF_PREDICATE,
+    RDF_OBJECT,
+    createLeanOriginEntry,
+    resolveSubject,
+    resolveObject,
+    processTokenWithBlockTracking
+} from './shared.js';
 
-function getFenceClosePattern(fenceChar) {
-    if (!FENCE_CLOSE_PATTERNS.has(fenceChar)) {
-        FENCE_CLOSE_PATTERNS.set(fenceChar, new RegExp(`^(${fenceChar}{3,})`));
-    }
-    return FENCE_CLOSE_PATTERNS.get(fenceChar);
-}
 
-function parseLangAndAttrs(langAndAttrs) {
-    const spaceIndex = langAndAttrs.indexOf(' ');
-    const braceIndex = langAndAttrs.indexOf('{');
-    const langEnd = Math.min(
-        spaceIndex > -1 ? spaceIndex : Infinity,
-        braceIndex > -1 ? braceIndex : Infinity
-    );
-    return {
-        lang: langAndAttrs.substring(0, langEnd),
-        attrsText: langAndAttrs.substring(langEnd).match(/\{[^{}]*\}/)?.[0] || null
+export function parse(text, options = {}) {
+    const state = {
+        ctx: { ...DEFAULT_CONTEXT, ...(options.context || {}) },
+        df: options.dataFactory || DataFactory,
+        quads: [],
+        quadBuffer: new Map(),
+        removeSet: new Set(),
+        origin: {
+            quadIndex: new Map(),
+            blocks: new Map(),
+            documentStructure: []
+        },
+        currentSubject: null,
+        tokens: null,
+        currentTokenIndex: -1,
+        statements: [],
+        statementCandidates: new Map(),
+        currentBlock: null,
+        blockStack: []
     };
-}
 
-const semCache = {};
-const EMPTY_SEM = Object.freeze({ predicates: [], types: [], subject: null });
+    state.tokens = scanTokens(text);
+
+    // Single-pass processing: resolve prefixes AND process tokens together
+    for (let i = 0; i < state.tokens.length; i++) {
+        const token = state.tokens[i];
+        state.currentTokenIndex = i;
+
+        // Handle prefix tokens immediately during main pass
+        if (token.type === 'prefix') {
+            let resolvedIri = token.iri;
+            if (token.iri.includes(':')) {
+                const colonIndex = token.iri.indexOf(':');
+                const potentialPrefix = token.iri.substring(0, colonIndex);
+                const reference = token.iri.substring(colonIndex + 1);
+                if (state.ctx[potentialPrefix] && potentialPrefix !== '@vocab') {
+                    resolvedIri = state.ctx[potentialPrefix] + reference;
+                }
+            }
+            state.ctx[token.prefix] = resolvedIri;
+            continue; // Skip token processor for prefixes
+        }
 
-function parseSemCached(attrs) {
-    if (!attrs) return EMPTY_SEM;
-    let sem = semCache[attrs];
-    if (!sem) {
-        sem = Object.freeze(parseSemanticBlock(attrs));
-        semCache[attrs] = sem;
+        // Process all other tokens
+        TOKEN_PROCESSORS[token.type]?.(token, state);
     }
-    return sem;
-}
 
-function calcRangeInfo(line, attrs, lineStart, prefixLength, valueLength) {
-    const wsLength = prefixLength < line.length && line[prefixLength] === ' ' ? 1 :
-        line.slice(prefixLength).match(/^\s+/)?.[0]?.length || 0;
-    const valueStartInLine = prefixLength + wsLength;
-    return {
-        valueRange: [lineStart + valueStartInLine, lineStart + valueStartInLine + valueLength],
-        attrsRange: calcAttrsRange(line, attrs, lineStart)
-    };
-}
+    // Optimized quad filtering - use Set.has() instead of array.includes()
+    const quadKeys = new Set();
+    for (const quad of state.quads) {
+        quadKeys.add(quadIndexKey(quad.subject, quad.predicate, quad.object));
+    }
 
-function calcAttrsRange(line, attrs, lineStart) {
-    if (!attrs) return null;
-    const attrsStartInLine = line.lastIndexOf(attrs);
-    return attrsStartInLine >= 0 ? [lineStart + attrsStartInLine, lineStart + attrsStartInLine + attrs.length] : null;
-}
+    // Direct Set iteration - more efficient than filter()
+    const filteredRemove = [];
+    for (const quad of state.removeSet) {
+        const key = quadIndexKey(quad.subject, quad.predicate, quad.object);
+        if (!quadKeys.has(key)) {
+            filteredRemove.push(quad);
+        }
+    }
 
-function createToken(type, range, text, attrs = null, attrsRange = null, valueRange = null, extra = {}) {
-    const token = { type, range, text, attrs, attrsRange, valueRange, ...extra };
-    Object.defineProperty(token, '_carriers', {
-        enumerable: false, writable: true, value: null
-    });
-    return token;
+    return { quads: state.quads, remove: filteredRemove, statements: state.statements, origin: state.origin, context: state.ctx };
 }
 
+
+// Cache for fence regex patterns - using shared utility
+
 function getCarriers(token) {
     if (token.type === 'code') return [];
     return token._carriers || (token._carriers = extractInlineCarriers(token.text, token.range[0]));
 }
 
-const createListToken = (type, line, lineStart, pos, match) => {
-    const attrs = match[4] || null;
-    const prefix = match[1].length + (match[2] ? match[2].length : 0);
-    const rangeInfo = calcRangeInfo(line, attrs, lineStart, prefix, match[3].length);
-    return createToken(type, [lineStart, pos - 1], match[3].trim(), attrs,
-        rangeInfo.attrsRange, rangeInfo.valueRange, { indent: match[1].length });
-};
-
 function scanTokens(text) {
     const tokens = [];
     const lines = text.split('\n');
     let pos = 0;
     let codeBlock = null;
-
-    // Direct lookup instead of linear search
     const PROCESSORS = [
         { type: 'fence', test: line => FENCE_REGEX.test(line.trim()), process: handleFence },
         { type: 'content', test: () => codeBlock, process: line => codeBlock.content.push(line) },
@@ -194,16 +229,6 @@ function scanTokens(text) {
     return tokens;
 }
 
-function createCarrier(type, text, attrs, attrsRange, valueRange, range, pos, extra = {}) {
-    return { type, text, attrs, attrsRange, valueRange, range, pos, ...extra };
-}
-
-// Pre-compiled carrier patterns for better performance
-const CARRIER_PATTERN_ARRAY = [
-    ['EMPHASIS', /[*__`]+(.+?)[*__`]+\s*\{([^}]+)\}/y],
-    ['CODE_SPAN', /``(.+?)``\s*\{([^}]+)\}/y]
-];
-
 function extractInlineCarriers(text, baseOffset = 0) {
     const carriers = [];
     let pos = 0;
@@ -268,74 +293,6 @@ function extractInlineCarriers(text, baseOffset = 0) {
     return carriers;
 }
 
-function calcCarrierRanges(match, baseOffset, matchStart) {
-    const valueStart = baseOffset + matchStart + match[0].indexOf(match[1]);
-    const valueEnd = valueStart + match[1].length;
-    const attrsStart = baseOffset + matchStart + match[0].indexOf('{');
-    const attrsEnd = attrsStart + match[2].length + 2; // +2 for { and }
-    return {
-        valueRange: [valueStart, valueEnd],
-        attrsRange: [attrsStart + 1, attrsEnd - 1], // Exclude braces
-        range: [baseOffset + matchStart, attrsEnd],
-        pos: matchStart + match[0].length // pos should be relative to current text, not document
-    };
-}
-
-function findMatchingBracket(text, bracketStart) {
-    let bracketDepth = 1;
-    let bracketEnd = bracketStart + 1;
-
-    while (bracketEnd < text.length && bracketDepth > 0) {
-        if (text[bracketEnd] === '[') bracketDepth++;
-        else if (text[bracketEnd] === ']') bracketDepth--;
-        bracketEnd++;
-    }
-
-    return bracketDepth > 0 ? null : bracketEnd;
-}
-
-function extractUrlFromBrackets(text, bracketEnd) {
-    let url = null;
-    let spanEnd = bracketEnd;
-
-    if (text[spanEnd] === '(') {
-        const parenEnd = text.indexOf(')', spanEnd);
-        if (parenEnd !== -1) {
-            url = text.substring(spanEnd + 1, parenEnd);
-            spanEnd = parenEnd + 1;
-        }
-    }
-
-    return { url, spanEnd };
-}
-
-function extractAttributesFromText(text, spanEnd, baseOffset) {
-    let attrs = null;
-    let attrsRange = null;
-    const remaining = text.substring(spanEnd);
-
-    const wsMatch = remaining.match(/^\s+/);
-    const attrsStart = wsMatch ? wsMatch[0].length : 0;
-
-    if (remaining[attrsStart] === '{') {
-        const braceEnd = remaining.indexOf('}', attrsStart);
-        if (braceEnd !== -1) {
-            attrs = remaining.substring(attrsStart, braceEnd + 1);
-            const absStart = baseOffset + spanEnd + attrsStart;
-            attrsRange = [absStart, absStart + attrs.length];
-            spanEnd += braceEnd + 1;
-        }
-    }
-
-    return { attrs, attrsRange, finalSpanEnd: spanEnd };
-}
-
-function determineCarrierType(url) {
-    if (url && !url.startsWith('=')) {
-        return { carrierType: 'link', resourceIRI: url };
-    }
-    return { carrierType: 'span', resourceIRI: null };
-}
 
 function createBlockEntry(token, state) {
     const blockId = token._blockId || hash(`${token.type}:${token.range?.[0]}:${token.range?.[1]}`);
@@ -364,31 +321,6 @@ function createBlockEntry(token, state) {
     return blockEntry;
 }
 
-function extractCleanText(token) {
-    if (!token.text) return '';
-
-    let text = token.text;
-
-    // Remove semantic annotations
-    if (token.attrsRange) {
-        const beforeAttrs = text.substring(0, token.attrsRange[0] - (token.range?.[0] || 0));
-        const afterAttrs = text.substring(token.attrsRange[1] - (token.range?.[0] || 0));
-        text = beforeAttrs + afterAttrs;
-    }
-
-    // Clean based on token type
-    switch (token.type) {
-        case 'heading':
-            return text.replace(/^#+\s*/, '').trim();
-        case 'list':
-            return text.replace(/^[-*+]\s*/, '').trim();
-        case 'blockquote':
-            return text.replace(/^>\s*/, '').trim();
-        default:
-            return text.trim();
-    }
-}
-
 function enrichBlockFromAnnotation(blockEntry, sem, carrier, state) {
     // Update subject if available
     if (sem.subject && sem.subject !== 'RESET') {
@@ -539,17 +471,8 @@ function emitQuad(quads, quadBuffer, removeSet, quadIndex, block, subject, predi
         // Detect rdf:Statement pattern during single-pass parsing
         detectStatementPatternSinglePass(quad, dataFactory, meta, statements, statementCandidates);
 
-        // Create lean origin entry - avoid spread operator for better performance
-        const originEntry = {
-            blockId: block.id,
-            range: block.range,
-            carrierType: block.carrierType,
-            subject: subject.value,
-            predicate: predicate.value,
-            context: block.context, // Direct reference instead of spread
-            polarity: meta?.remove ? '-' : '+',
-            value: block.text || ''
-        };
+        // Create lean origin entry using shared utility
+        const originEntry = createLeanOriginEntry(block, subject, predicate, meta);
 
         quadIndex.set(quadKey, originEntry);
 
@@ -563,13 +486,6 @@ function emitQuad(quads, quadBuffer, removeSet, quadIndex, block, subject, predi
     }
 }
 
-// Extract RDF constants once at module level for efficiency
-const RDF_TYPE = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type';
-const RDF_STATEMENT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#Statement';
-const RDF_SUBJECT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#subject';
-const RDF_PREDICATE = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate';
-const RDF_OBJECT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#object';
-
 function detectStatementPatternSinglePass(quad, dataFactory, meta, statements = null, statementCandidates = null) {
     // Skip if not called from parse context (for testing compatibility)
     if (!statements || !statementCandidates) return;
@@ -619,30 +535,6 @@ function detectStatementPatternSinglePass(quad, dataFactory, meta, statements =
     }
 }
 
-const resolveFragment = (fragment, state) => {
-    if (!state.currentSubject) return null;
-    const subjectValue = state.currentSubject.value;
-    const hashIndex = subjectValue.indexOf('#');
-    const baseIRI = hashIndex > -1 ? subjectValue.slice(0, hashIndex) : subjectValue;
-    return state.df.namedNode(baseIRI + '#' + fragment);
-};
-
-function resolveSubject(sem, state) {
-    if (!sem.subject) return null;
-    if (sem.subject === 'RESET') {
-        state.currentSubject = null;
-        return null;
-    }
-    if (sem.subject.startsWith('=#')) return resolveFragment(sem.subject.substring(2), state);
-    return state.df.namedNode(expandIRI(sem.subject, state.ctx));
-}
-
-function resolveObject(sem, state) {
-    if (!sem.object) return null;
-    if (sem.object.startsWith('#')) return resolveFragment(sem.object.substring(1), state);
-    return state.df.namedNode(expandIRI(sem.object, state.ctx));
-}
-
 const createTypeQuad = (typeIRI, subject, state, block, entryIndex = null) => {
     const expandedType = expandIRI(typeIRI, state.ctx);
     const typeInfo = typeof entryIndex === 'object' ? entryIndex : { entryIndex, remove: false };
@@ -748,123 +640,9 @@ function processStandaloneSubject(token, state) {
 }
 
 const TOKEN_PROCESSORS = {
-    heading: (token, state) => {
-        const blockEntry = createBlockEntry(token, state);
-        state.currentBlock = blockEntry;
-        state.blockStack.push(blockEntry.id);
-
-        processTokenAnnotations(token, state, token.type);
-
-        state.blockStack.pop();
-        state.currentBlock = state.blockStack.length > 0 ?
-            state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
-    },
-    code: (token, state) => {
-        const blockEntry = createBlockEntry(token, state);
-        state.currentBlock = blockEntry;
-        state.blockStack.push(blockEntry.id);
-
-        processTokenAnnotations(token, state, token.type);
-
-        state.blockStack.pop();
-        state.currentBlock = state.blockStack.length > 0 ?
-            state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
-    },
-    blockquote: (token, state) => {
-        const blockEntry = createBlockEntry(token, state);
-        state.currentBlock = blockEntry;
-        state.blockStack.push(blockEntry.id);
-
-        processTokenAnnotations(token, state, token.type);
-
-        state.blockStack.pop();
-        state.currentBlock = state.blockStack.length > 0 ?
-            state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
-    },
-    para: (token, state) => {
-        const blockEntry = createBlockEntry(token, state);
-        state.currentBlock = blockEntry;
-        state.blockStack.push(blockEntry.id);
-
-        processStandaloneSubject(token, state);
-        processTokenAnnotations(token, state, token.type);
-
-        state.blockStack.pop();
-        state.currentBlock = state.blockStack.length > 0 ?
-            state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
-    },
-    list: (token, state) => {
-        const blockEntry = createBlockEntry(token, state);
-        state.currentBlock = blockEntry;
-        state.blockStack.push(blockEntry.id);
-
-        processTokenAnnotations(token, state, token.type);
-
-        state.blockStack.pop();
-        state.currentBlock = state.blockStack.length > 0 ?
-            state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
-    },
+    heading: (token, state) => processTokenWithBlockTracking(token, state, processTokenAnnotations, createBlockEntry),
+    code: (token, state) => processTokenWithBlockTracking(token, state, processTokenAnnotations, createBlockEntry),
+    blockquote: (token, state) => processTokenWithBlockTracking(token, state, processTokenAnnotations, createBlockEntry),
+    para: (token, state) => processTokenWithBlockTracking(token, state, processTokenAnnotations, createBlockEntry, [processStandaloneSubject]),
+    list: (token, state) => processTokenWithBlockTracking(token, state, processTokenAnnotations, createBlockEntry),
 };
-
-export function parse(text, options = {}) {
-    const state = {
-        ctx: { ...DEFAULT_CONTEXT, ...(options.context || {}) },
-        df: options.dataFactory || DataFactory,
-        quads: [],
-        quadBuffer: new Map(),
-        removeSet: new Set(),
-        origin: {
-            quadIndex: new Map(),
-            blocks: new Map(),
-            documentStructure: []
-        },
-        currentSubject: null,
-        tokens: null,
-        currentTokenIndex: -1,
-        statements: [],
-        statementCandidates: new Map(), // Track incomplete rdf:Statement patterns
-        currentBlock: null,
-        blockStack: []
-    };
-
-    state.tokens = scanTokens(text);
-
-    // Single loop instead of filter+forEach for better performance
-    for (const token of state.tokens) {
-        if (token.type === 'prefix') {
-            let resolvedIri = token.iri;
-            if (token.iri.includes(':')) {
-                const colonIndex = token.iri.indexOf(':');
-                const potentialPrefix = token.iri.substring(0, colonIndex);
-                const reference = token.iri.substring(colonIndex + 1);
-                if (state.ctx[potentialPrefix] && potentialPrefix !== '@vocab') {
-                    resolvedIri = state.ctx[potentialPrefix] + reference;
-                }
-            }
-            state.ctx[token.prefix] = resolvedIri;
-        }
-    }
-
-    for (let i = 0; i < state.tokens.length; i++) {
-        const token = state.tokens[i];
-        state.currentTokenIndex = i;
-        TOKEN_PROCESSORS[token.type]?.(token, state);
-    }
-
-    // Optimize array operations - avoid Array.from() and filter()
-    const quadKeys = new Set();
-    for (const quad of state.quads) {
-        quadKeys.add(quadIndexKey(quad.subject, quad.predicate, quad.object));
-    }
-
-    // Direct iteration instead of Array.from() + filter()
-    const filteredRemove = [];
-    for (const quad of state.removeSet) {
-        const key = quadIndexKey(quad.subject, quad.predicate, quad.object);
-        if (!quadKeys.has(key)) {
-            filteredRemove.push(quad);
-        }
-    }
-
-    return { quads: state.quads, remove: filteredRemove, statements: state.statements, origin: state.origin, context: state.ctx };
-}

package/src/render.js

@@ -6,7 +6,12 @@ import {
     parseSemanticBlock,
     hash
 } from './utils.js';
-import { DEFAULT_CONTEXT } from './shared.js';
+import {
+    escapeHtml,
+    getIndentLevel,
+    processPredicates
+} from './shared.js';
+import { DEFAULT_CONTEXT } from './constants.js';
 
 /**
  * Render MD-LD to HTML+RDFa
@@ -302,17 +307,6 @@ function parseMarkdownList(markdownList, blocks, state) {
     return html;
 }
 
-/**
- * Get indent level from source text
- */
-function getIndentLevel(block, sourceText) {
-    if (!block.range || !sourceText) return 0;
-
-    const text = sourceText.substring(block.range.start, block.range.end);
-    const indentMatch = text.match(/^(\s*)/);
-    return indentMatch ? indentMatch[1].length : 0;
-}
-
 /**
  * Render a single block
  */
@@ -408,26 +402,9 @@ function buildRDFaAttrsFromBlock(block, ctx) {
         attrs.push(`typeof="${escapeHtml(types)}"`);
     }
 
-    // Predicates
+    // Predicates using shared utility
     if (block.predicates && block.predicates.length > 0) {
-        const literalProps = [];
-        const objectProps = [];
-        const reverseProps = [];
-
-        block.predicates.forEach(pred => {
-            const iri = typeof pred === 'string' ? pred : pred.iri;
-            const expanded = expandIRI(iri, ctx);
-            const shortened = shortenIRI(expanded, ctx);
-            const form = typeof pred === 'string' ? '' : (pred.form || '');
-
-            if (form === '!') {
-                reverseProps.push(shortened);
-            } else if (form === '?') {
-                objectProps.push(shortened);
-            } else {
-                literalProps.push(shortened);
-            }
-        });
+        const { literalProps, objectProps, reverseProps } = processPredicates(block.predicates, ctx);
 
         if (literalProps.length > 0) {
             attrs.push(`property="${escapeHtml(literalProps.join(' '))}"`);
@@ -474,17 +451,3 @@ function wrapWithRDFaContext(html, ctx) {
 
     return `<div${prefixDecl}${vocabDecl}>${html}</div>`;
 }
-
-/**
- * Escape HTML special characters
- */
-function escapeHtml(text) {
-    const map = {
-        '&': '&amp;',
-        '<': '&lt;',
-        '>': '&gt;',
-        '"': '&quot;',
-        "'": '&#39;'
-    };
-    return String(text || '').replace(/[&<>"']/g, m => map[m]);
-}

package/src/shared.js

@@ -1,37 +1,5 @@
-/**
- * Shared utilities for MD-LD Parser and Renderer
- * Ensures DRY code and consistent CommonMark processing
- */
-
-export const DEFAULT_CONTEXT = {
-    '@vocab': "http://www.w3.org/2000/01/rdf-schema#",
-    rdf: 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
-    rdfs: 'http://www.w3.org/2000/01/rdf-schema#',
-    xsd: 'http://www.w3.org/2001/XMLSchema#',
-    sh: "http://www.w3.org/ns/shacl#",
-    prov: 'http://www.w3.org/ns/prov#'
-};
-
-// CommonMark patterns - shared between parser and renderer
-export const URL_REGEX = /^(https?|ftp|mailto|tag|nih|urn|uuid|did|web|ipfs|ipns|data|file|urn:uuid):/;
-export const FENCE_REGEX = /^(`{3,}|~{3,})(.*)/;
-export const PREFIX_REGEX = /^\[([^\]]+)\]\s*<([^>]+)>/;
-export const HEADING_REGEX = /^(#{1,6})\s+(.+?)(?:\s*(\{[^}]+\}))?$/;
-export const UNORDERED_LIST_REGEX = /^(\s*)([-*+]|\d+\.)\s+(.+?)(?:\s*(\{[^}]+\}))?\s*$/;
-export const BLOCKQUOTE_REGEX = /^>\s+(.+?)(?:\s*(\{[^}]+\}))?$/;
-export const STANDALONE_SUBJECT_REGEX = /^\s*\{=(.*?)\}\s*$/;
-
-// Inline carrier patterns - shared extraction logic
-export const INLINE_CARRIER_PATTERNS = {
-    EMPHASIS: /[*__`]+(.+?)[*__`]+\s*\{([^}]+)\}/y,
-    CODE_SPAN: /``(.+?)``\s*\{([^}]+)\}/y
-};
-
-// Pre-compiled carrier patterns for performance
-export const CARRIER_PATTERN_ARRAY = [
-    ['EMPHASIS', /[*__`]+(.+?)[*__`]+\s*\{([^}]+)\}/y],
-    ['CODE_SPAN', /``(.+?)``\s*\{([^}]+)\}/y]
-];
+import { DEFAULT_CONTEXT, STANDALONE_SUBJECT_REGEX, FENCE_REGEX, PREFIX_REGEX, HEADING_REGEX, UNORDERED_LIST_REGEX, BLOCKQUOTE_REGEX } from './constants.js';
+import { parseSemanticBlock, expandIRI, shortenIRI } from './utils.js';
 
 // Cache for fence regex patterns
 export const FENCE_CLOSE_PATTERNS = new Map();
@@ -156,6 +124,172 @@ export const PROCESSORS = [
     { test: line => true, process: null } // Default: paragraph
 ];
 
+// Token scanning processors - shared between parser and renderer
+export const TOKEN_PROCESSORS = [
+    { type: 'fence', test: line => FENCE_REGEX.test(line.trim()), process: null }, // Will be overridden in parse.js
+    { type: 'content', test: line => false, process: null }, // Will be overridden in parse.js  
+    { type: 'prefix', test: line => PREFIX_REGEX.test(line), process: null }, // Will be overridden in parse.js
+    { type: 'heading', test: line => HEADING_REGEX.test(line), process: null }, // Will be overridden in parse.js
+    { type: 'list', test: line => UNORDERED_LIST_REGEX.test(line), process: null }, // Will be overridden in parse.js
+    { type: 'blockquote', test: line => BLOCKQUOTE_REGEX.test(line), process: null }, // Will be overridden in parse.js
+    { type: 'para', test: line => line.trim(), process: null } // Will be overridden in parse.js
+];
+
+// Language and attributes parsing
+export function parseLangAndAttrs(langAndAttrs) {
+    const spaceIndex = langAndAttrs.indexOf(' ');
+    const braceIndex = langAndAttrs.indexOf('{');
+    const langEnd = Math.min(
+        spaceIndex > -1 ? spaceIndex : Infinity,
+        braceIndex > -1 ? braceIndex : Infinity
+    );
+    return {
+        lang: langAndAttrs.substring(0, langEnd),
+        attrsText: langAndAttrs.substring(langEnd).match(/\{[^{}]*\}/)?.[0] || null
+    };
+}
+
+// Carrier extraction utilities
+export function findMatchingBracket(text, bracketStart) {
+    let bracketDepth = 1;
+    let bracketEnd = bracketStart + 1;
+
+    while (bracketEnd < text.length && bracketDepth > 0) {
+        if (text[bracketEnd] === '[') bracketDepth++;
+        else if (text[bracketEnd] === ']') bracketDepth--;
+        bracketEnd++;
+    }
+
+    return bracketDepth > 0 ? null : bracketEnd;
+}
+
+export function extractUrlFromBrackets(text, bracketEnd) {
+    let url = null;
+    let spanEnd = bracketEnd;
+
+    if (text[spanEnd] === '(') {
+        const parenEnd = text.indexOf(')', spanEnd);
+        if (parenEnd !== -1) {
+            url = text.substring(spanEnd + 1, parenEnd);
+            spanEnd = parenEnd + 1;
+        }
+    }
+
+    return { url, spanEnd };
+}
+
+export function extractAttributesFromText(text, spanEnd, baseOffset) {
+    let attrs = null;
+    let attrsRange = null;
+    const remaining = text.substring(spanEnd);
+
+    const wsMatch = remaining.match(/^\s+/);
+    const attrsStart = wsMatch ? wsMatch[0].length : 0;
+
+    if (remaining[attrsStart] === '{') {
+        const braceEnd = remaining.indexOf('}', attrsStart);
+        if (braceEnd !== -1) {
+            attrs = remaining.substring(attrsStart, braceEnd + 1);
+            const absStart = baseOffset + spanEnd + attrsStart;
+            attrsRange = [absStart, absStart + attrs.length];
+            spanEnd += braceEnd + 1;
+        }
+    }
+
+    return { attrs, attrsRange, finalSpanEnd: spanEnd };
+}
+
+export function determineCarrierType(url) {
+    if (url && !url.startsWith('=')) {
+        return { carrierType: 'link', resourceIRI: url };
+    }
+    return { carrierType: 'span', resourceIRI: null };
+}
+
+export function calcCarrierRanges(match, baseOffset, matchStart) {
+    const valueStart = baseOffset + matchStart + match[0].indexOf(match[1]);
+    const valueEnd = valueStart + match[1].length;
+    const attrsStart = baseOffset + matchStart + match[0].indexOf('{');
+    const attrsEnd = attrsStart + match[2].length + 2; // +2 for { and }
+    return {
+        valueRange: [valueStart, valueEnd],
+        attrsRange: [attrsStart + 1, attrsEnd - 1], // Exclude braces
+        range: [baseOffset + matchStart, attrsEnd],
+        pos: matchStart + match[0].length // pos should be relative to current text, not document
+    };
+}
+
+// Clean text extraction utilities
+export function extractCleanText(token) {
+    if (!token.text) return '';
+
+    let text = token.text;
+
+    // Remove semantic annotations
+    if (token.attrsRange) {
+        const beforeAttrs = text.substring(0, token.attrsRange[0] - (token.range?.[0] || 0));
+        const afterAttrs = text.substring(token.attrsRange[1] - (token.range?.[0] || 0));
+        text = beforeAttrs + afterAttrs;
+    }
+
+    // Clean based on token type
+    switch (token.type) {
+        case 'heading':
+            return text.replace(/^#+\s*/, '').trim();
+        case 'list':
+            return text.replace(/^[-*+]\s*/, '').trim();
+        case 'blockquote':
+            return text.replace(/^>\s*/, '').trim();
+        default:
+            return text.trim();
+    }
+}
+
+// Quad emission utilities
+export const RDF_TYPE = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type';
+export const RDF_STATEMENT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#Statement';
+export const RDF_SUBJECT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#subject';
+export const RDF_PREDICATE = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate';
+export const RDF_OBJECT = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#object';
+
+export function createLeanOriginEntry(block, subject, predicate, meta = null) {
+    return {
+        blockId: block.id,
+        range: block.range,
+        carrierType: block.carrierType,
+        subject: subject.value,
+        predicate: predicate.value,
+        context: block.context, // Direct reference instead of spread
+        polarity: meta?.remove ? '-' : '+',
+        value: block.text || ''
+    };
+}
+
+// Fragment resolution utilities
+export function resolveFragment(fragment, currentSubject, dataFactory) {
+    if (!currentSubject) return null;
+    const subjectValue = currentSubject.value;
+    const hashIndex = subjectValue.indexOf('#');
+    const baseIRI = hashIndex > -1 ? subjectValue.slice(0, hashIndex) : subjectValue;
+    return dataFactory.namedNode(baseIRI + '#' + fragment);
+}
+
+export function resolveSubject(sem, state) {
+    if (!sem.subject) return null;
+    if (sem.subject === 'RESET') {
+        state.currentSubject = null;
+        return null;
+    }
+    if (sem.subject.startsWith('=#')) return resolveFragment(sem.subject.substring(2), state.currentSubject, state.df);
+    return state.df.namedNode(expandIRI(sem.subject, state.ctx));
+}
+
+export function resolveObject(sem, state) {
+    if (!sem.object) return null;
+    if (sem.object.startsWith('#')) return resolveFragment(sem.object.substring(1), state.currentSubject, state.df);
+    return state.df.namedNode(expandIRI(sem.object, state.ctx));
+}
+
 // HTML escaping - shared utility
 export function escapeHtml(text) {
     if (!text) return '';
@@ -167,6 +301,77 @@ export function escapeHtml(text) {
         .replace(/'/g, '&#x27;');
 }
 
+// RDF term type checking utilities - shared across modules
+export function isLiteral(term) {
+    return term?.termType === 'Literal';
+}
+
+export function isNamedNode(term) {
+    return term?.termType === 'NamedNode';
+}
+
+export function isRdfType(term) {
+    return term?.value === 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type';
+}
+
+// IRI prefix extraction utility
+export function getPrefixFromIRI(iri, context) {
+    if (!iri) return null;
+    const shortened = shortenIRI(iri, context);
+    if (shortened.includes(':')) {
+        return shortened.split(':')[0];
+    }
+    return null;
+}
+
+// Prefix collection utility - used by generate.js
+export function collectUsedPrefixes(subjectGroups, context) {
+    const usedPrefixes = new Set();
+
+    for (const subjectQuads of subjectGroups.values()) {
+        for (const quad of subjectQuads) {
+            // Check subject prefix
+            const subjectPrefix = getPrefixFromIRI(quad.subject.value, context);
+            if (subjectPrefix) usedPrefixes.add(subjectPrefix);
+
+            // Check predicate prefix
+            const predicatePrefix = getPrefixFromIRI(quad.predicate.value, context);
+            if (predicatePrefix) usedPrefixes.add(predicatePrefix);
+
+            // Check object prefix if it's a named node
+            if (isNamedNode(quad.object)) {
+                const objectPrefix = getPrefixFromIRI(quad.object.value, context);
+                if (objectPrefix) usedPrefixes.add(objectPrefix);
+            }
+
+            // Check datatype prefix if present
+            if (quad.object.datatype && quad.object.datatype.value) {
+                const datatypePrefix = getPrefixFromIRI(quad.object.datatype.value, context);
+                if (datatypePrefix) usedPrefixes.add(datatypePrefix);
+            }
+        }
+    }
+
+    return usedPrefixes;
+}
+
+// Token processing utility - eliminates duplication in TOKEN_PROCESSORS
+export function processTokenWithBlockTracking(token, state, processAnnotations, createBlockEntry, additionalProcessors = []) {
+    const blockEntry = createBlockEntry(token, state);
+    state.currentBlock = blockEntry;
+    state.blockStack.push(blockEntry.id);
+
+    // Run any additional processors first
+    additionalProcessors.forEach(processor => processor(token, state));
+
+    // Process annotations
+    processAnnotations(token, state, token.type);
+
+    state.blockStack.pop();
+    state.currentBlock = state.blockStack.length > 0 ?
+        state.origin.blocks.get(state.blockStack[state.blockStack.length - 1]) : null;
+}
+
 // Quad key generation - shared between parser and renderer
 export function quadIndexKey(subject, predicate, object) {
     const datatype = object.datatype?.value || '';
@@ -199,14 +404,126 @@ export function resolveSubjectType(subjectDecl) {
     return 'full-iri';
 }
 
-// Fragment resolution - shared logic
-export function resolveFragment(fragment, currentSubject) {
-    if (!currentSubject) {
-        throw new Error('Fragment requires current subject');
+// Constants - shared across modules (bundle-size optimized)
+export const XSD_STRING = 'http://www.w3.org/2001/XMLSchema#string';
+
+// Optimized sorting utilities - inline for better minification
+export function sortQuadsByPredicate(quads) {
+    return quads.sort((a, b) => a.predicate.value.localeCompare(b.predicate.value));
+}
+
+// Optimized text generation - template literals for smaller bundle
+export const generatePrefixDeclaration = (prefix, namespace) => `[${prefix}] <${namespace}>\n`;
+
+export function generateLiteralText(quad, context) {
+    const predShort = shortenIRI(quad.predicate.value, context);
+    let annotation = predShort;
+
+    if (quad.object.language) {
+        annotation += ` @${quad.object.language}`;
+    } else if (quad.object.datatype.value !== XSD_STRING) {
+        annotation += ` ^^${shortenIRI(quad.object.datatype.value, context)}`;
+    }
+
+    return `[${quad.object.value}] {${annotation}}\n`;
+}
+
+export const generateObjectText = (quad, context) => {
+    const objShort = shortenIRI(quad.object.value, context);
+    const predShort = shortenIRI(quad.predicate.value, context);
+    return `[${objShort}] {+${objShort} ?${predShort}}\n`;
+};
+
+// Optimized quad filtering - destructuring for smaller minified output
+export function filterQuadsByType(subjectQuads) {
+    const types = [], literals = [], objects = [];
+    for (const q of subjectQuads) {
+        if (isRdfType(q.predicate)) {
+            types.push(q);
+        } else if (isLiteral(q.object)) {
+            literals.push(q);
+        } else if (isNamedNode(q.object)) {
+            objects.push(q);
+        }
     }
-    const fragmentName = fragment.substring(2); // Remove =#
-    const baseIRI = currentSubject.value;
-    const hashIndex = baseIRI.indexOf('#');
-    const base = hashIndex > -1 ? baseIRI.slice(0, hashIndex) : baseIRI;
-    return base + '#' + fragmentName;
+    return { types, literals, objects };
+}
+
+// Predicate processing utilities - common RDFa patterns
+export function processPredicates(predicates, ctx) {
+    const literalProps = [];
+    const objectProps = [];
+    const reverseProps = [];
+
+    predicates.forEach(pred => {
+        const iri = typeof pred === 'string' ? pred : pred.iri;
+        const expanded = expandIRI(iri, ctx);
+        const shortened = shortenIRI(expanded, ctx);
+        const form = typeof pred === 'string' ? '' : (pred.form || '');
+
+        if (form === '!') {
+            reverseProps.push(shortened);
+        } else if (form === '?') {
+            objectProps.push(shortened);
+        } else {
+            literalProps.push(shortened);
+        }
+    });
+
+    return { literalProps, objectProps, reverseProps };
+}
+
+// Deterministic sorting utilities - ensure consistent output
+export function sortDeterministic(array, keyFn) {
+    return array.sort((a, b) => {
+        const keyA = keyFn(a);
+        const keyB = keyFn(b);
+        return keyA.localeCompare(keyB);
+    });
+}
+
+export function sortQuadsDeterministically(quads) {
+    return quads.sort((a, b) => {
+        // Deterministic sorting: subject -> predicate -> object
+        const sComp = a.subject.value.localeCompare(b.subject.value);
+        if (sComp !== 0) return sComp;
+        const pComp = a.predicate.value.localeCompare(b.predicate.value);
+        if (pComp !== 0) return pComp;
+        const oA = isLiteral(a.object) ? a.object.value : a.object.value;
+        const oB = isLiteral(b.object) ? b.object.value : b.object.value;
+        return oA.localeCompare(oB);
+    });
+}
+
+// Optimized deterministic prefix generation
+export function generateDeterministicPrefixes(context, usedPrefixes) {
+    const sortedEntries = Object.entries(context).sort(([a], [b]) => a.localeCompare(b));
+    let text = '';
+
+    for (const [prefix, namespace] of sortedEntries) {
+        if (prefix !== '@vocab' && !prefix.startsWith('@') && !DEFAULT_CONTEXT[prefix] && usedPrefixes.has(prefix)) {
+            text += generatePrefixDeclaration(prefix, namespace);
+        }
+    }
+
+    return text;
+}
+
+// Memory-efficient block creation
+export function createOptimizedBlockEntry(token, state) {
+    const id = hash(`${token.range[0]}-${token.range[1]}-${token.text.slice(0, 50)}`);
+    const block = {
+        id,
+        type: token.type,
+        carrierType: token.type,
+        range: token.range,
+        text: token.text,
+        carriers: [],
+        predicates: [],
+        subject: state.currentSubject,
+        context: { ...state.ctx }
+    };
+
+    state.origin.blocks.set(id, block);
+    return block;
 }

package/src/utils.js

@@ -1,4 +1,4 @@
-import { URL_REGEX, DEFAULT_CONTEXT } from './shared.js';
+import { URL_REGEX, DEFAULT_CONTEXT } from './constants.js';
 
 // Base Term class for RDF/JS compatibility
 export class Term {