@segnavia/html 0.27.0 → 0.28.0

package/dist/index.d.ts

@@ -3,11 +3,17 @@ import { REFUSED_ELEMENTS, SanitizeOptions, findActiveContent, sanitize } from "
 import { BINDINGS, FillOptions, Filled, bindingNames, bindingRoot, fill, hasBindings } from "./bindings.js";
 import { BookDocument, BookDocumentOptions, BookHeading, BookPrintPart, BookSectionInfo, bookPrintPart } from "./book/types.js";
 import { bookDocument } from "./book/book-document.js";
+import { bookView, bookViewFromHtml } from "./book/book-view.js";
+import { PaginateBlock, PaginateBookInput, PaginateBookPage, PaginateBookRefusal, PaginateBookResult, PaginateFlow, PaginateFlowParents, PaginateLine, PaginateMark, PaginateNote, PaginateNoteRef, PaginateOpens, PaginatePageMarks, PaginatePageNote, PaginatePageRole, PaginateSequence, PaginateWindow } from "./book/paginate-types.js";
+import { BookFurnitureValues, embedFunction, fillBookFurniture, furnitureScript } from "./book/furniture.js";
+import { PAGINATE_FUNCTIONS, paginateBook, paginateScript } from "./book/paginate-book.js";
+import { PaginateTarget, PaginateTargetPage, heldPageNumber, openerPageIndex, pageIndexForTarget, windowPageIndex } from "./book/paginate-target.js";
+import { bookStylesheet, sanitizeBookCss } from "./book-css.js";
 import { FontCssRefusedError, fontCss, fontFaceCss } from "./font-css.js";
 import { fontFormat } from "./font-format.js";
 import { fontReferenceRefusals } from "./font-references.js";
 import { ExpandMotifsOptions, ExpandedMotifs, MOTIF_DEPTH_LIMIT, MotifRefusedError, PlacedMotif, expandMotifs, hasMotifPlacement, libraryMotifPlacementRefusal, motifPlacementRefusals, renderMotifPlacements } from "./motifs.js";
-import { BookPageFacts, BookPrintedPagePlan, BookPrintedPageRole, PrintedPagePlan, PrintedPageSide, planBookPages, planPrintedPages, printedPageAt } from "./printed-pages.js";
+import { BookPageFacts, BookPrintedPagePlan, BookPrintedPageRole, BookSectionParents, PrintedPagePlan, PrintedPageSide, planBookPages, planPrintedPages, printedPageAt, selectBookParent } from "./printed-pages.js";
 import { SlotFillEntry, SlotFillRefusal, SlotFillsConflictError, SlotFillsRefusedError, SlotPrintOptions, assertSingleSlotCopy, assertSlotPrintOptions, collectSlotPrintCopies, isSlotCopyRefusal, slotFillCopyId, slotFillsConflict } from "./slot-print.js";
 import { RequiredSlotsRefusedError, UnmeasuredSlotsRefusedError, mergeSlotFitResults, requiredSlotRefusal, unmeasuredSlotRefusal } from "./slot-policy.js";
 import { MAX_SLOT_DECLARATIONS, PreparedSlots, SlotFillRefusedError, hasSlotWork, prepareSlotCollection, prepareSlots, slotFillRefusal } from "./slots.js";
@@ -17,4 +23,4 @@ export type HtmlKind = 'plain' | 'live-data' | 'scripted' | 'refused';
 export declare function classify(html: string, prefix: string): HtmlKind;
 export declare function searchText(html: string): string;
 //#endregion
-export { BINDINGS, type BookDocument, type BookDocumentOptions, type BookHeading, BookPageFacts, type BookPrintPart, BookPrintedPagePlan, BookPrintedPageRole, type BookSectionInfo, type ExpandMotifsOptions, type ExpandedMotifs, type FillOptions, type Filled, FontCssRefusedError, MAX_SLOT_DECLARATIONS, MOTIF_DEPTH_LIMIT, MotifRefusedError, type PlacedMotif, type PreparedSlots, PrintedPagePlan, PrintedPageSide, REFUSED_ELEMENTS, RequiredSlotsRefusedError, type SanitizeOptions, type SlotFillEntry, type SlotFillRefusal, SlotFillRefusedError, SlotFillsConflictError, SlotFillsRefusedError, type SlotPrintOptions, UNICODE_RANGE_CLEARANCE, type UnicodeRangeParse, UnmeasuredSlotsRefusedError, assertSingleSlotCopy, assertSlotPrintOptions, base64ToBytes, bindingNames, bindingRoot, bookDocument, bookPrintPart, bytesToBase64, collectSlotPrintCopies, expandMotifs, fill, findActiveContent, fontCss, fontFaceCss, fontFormat, fontReferenceRefusals, hasBindings, hasMotifPlacement, hasSlotWork, isSlotCopyRefusal, libraryMotifPlacementRefusal, mergeSlotFitResults, motifPlacementRefusals, parseUnicodeRange, planBookPages, planPrintedPages, prepareSlotCollection, prepareSlots, printedPageAt, renderMotifPlacements, requiredSlotRefusal, sanitize, slotFillCopyId, slotFillRefusal, slotFillsConflict, unmeasuredSlotRefusal };
+export { BINDINGS, type BookDocument, type BookDocumentOptions, type BookFurnitureValues, type BookHeading, BookPageFacts, type BookPrintPart, BookPrintedPagePlan, BookPrintedPageRole, type BookSectionInfo, type BookSectionParents, type ExpandMotifsOptions, type ExpandedMotifs, type FillOptions, type Filled, FontCssRefusedError, MAX_SLOT_DECLARATIONS, MOTIF_DEPTH_LIMIT, MotifRefusedError, PAGINATE_FUNCTIONS, type PaginateBlock, type PaginateBookInput, type PaginateBookPage, type PaginateBookRefusal, type PaginateBookResult, type PaginateFlow, type PaginateFlowParents, type PaginateLine, type PaginateMark, type PaginateNote, type PaginateNoteRef, type PaginateOpens, type PaginatePageMarks, type PaginatePageNote, type PaginatePageRole, type PaginateSequence, type PaginateTarget, type PaginateTargetPage, type PaginateWindow, type PlacedMotif, type PreparedSlots, PrintedPagePlan, PrintedPageSide, REFUSED_ELEMENTS, RequiredSlotsRefusedError, type SanitizeOptions, type SlotFillEntry, type SlotFillRefusal, SlotFillRefusedError, SlotFillsConflictError, SlotFillsRefusedError, type SlotPrintOptions, UNICODE_RANGE_CLEARANCE, type UnicodeRangeParse, UnmeasuredSlotsRefusedError, assertSingleSlotCopy, assertSlotPrintOptions, base64ToBytes, bindingNames, bindingRoot, bookDocument, bookPrintPart, bookStylesheet, bookView, bookViewFromHtml, bytesToBase64, collectSlotPrintCopies, embedFunction, expandMotifs, fill, fillBookFurniture, findActiveContent, fontCss, fontFaceCss, fontFormat, fontReferenceRefusals, furnitureScript, hasBindings, hasMotifPlacement, hasSlotWork, heldPageNumber, isSlotCopyRefusal, libraryMotifPlacementRefusal, mergeSlotFitResults, motifPlacementRefusals, openerPageIndex, pageIndexForTarget, paginateBook, paginateScript, parseUnicodeRange, planBookPages, planPrintedPages, prepareSlotCollection, prepareSlots, printedPageAt, renderMotifPlacements, requiredSlotRefusal, sanitize, sanitizeBookCss, selectBookParent, slotFillCopyId, slotFillRefusal, slotFillsConflict, unmeasuredSlotRefusal, windowPageIndex };

package/dist/index.js

@@ -3,12 +3,17 @@ import { BINDINGS, bindingNames, bindingRoot, fill, hasBindings } from "./bindin
 import { base64ToBytes, bytesToBase64 } from "./base64.js";
 import { bookPrintPart } from "./book/types.js";
 import { bookDocument } from "./book/book-document.js";
+import { bookView, bookViewFromHtml } from "./book/book-view.js";
+import { embedFunction, fillBookFurniture, furnitureScript } from "./book/furniture.js";
+import { PAGINATE_FUNCTIONS, paginateBook, paginateScript } from "./book/paginate-book.js";
+import { heldPageNumber, openerPageIndex, pageIndexForTarget, windowPageIndex } from "./book/paginate-target.js";
+import { bookStylesheet, sanitizeBookCss } from "./book-css.js";
 import { UNICODE_RANGE_CLEARANCE, parseUnicodeRange } from "./unicode-range.js";
 import { FontCssRefusedError, fontCss, fontFaceCss } from "./font-css.js";
 import { fontFormat } from "./font-format.js";
 import { fontReferenceRefusals } from "./font-references.js";
 import { MOTIF_DEPTH_LIMIT, MotifRefusedError, expandMotifs, hasMotifPlacement, libraryMotifPlacementRefusal, motifPlacementRefusals, renderMotifPlacements } from "./motifs.js";
-import { planBookPages, planPrintedPages, printedPageAt } from "./printed-pages.js";
+import { planBookPages, planPrintedPages, printedPageAt, selectBookParent } from "./printed-pages.js";
 import { RequiredSlotsRefusedError, UnmeasuredSlotsRefusedError, mergeSlotFitResults, requiredSlotRefusal, unmeasuredSlotRefusal } from "./slot-policy.js";
 import { MAX_SLOT_DECLARATIONS, SlotFillRefusedError, hasSlotWork, prepareSlotCollection, prepareSlots, slotFillRefusal } from "./slots.js";
 import { SlotFillsConflictError, SlotFillsRefusedError, assertSingleSlotCopy, assertSlotPrintOptions, collectSlotPrintCopies, isSlotCopyRefusal, slotFillCopyId, slotFillsConflict } from "./slot-print.js";
@@ -24,4 +29,4 @@ function searchText(html) {
 	return html.replace(/<(script|style)\b[^>]*>[\s\S]*?<\/\1\s*>/gi, " ").replace(/<[^>]+>/g, " ").replace(/&nbsp;/g, " ").replace(/\s+/g, " ").trim();
 }
 //#endregion
-export { BINDINGS, FontCssRefusedError, MAX_SLOT_DECLARATIONS, MOTIF_DEPTH_LIMIT, MotifRefusedError, REFUSED_ELEMENTS, RequiredSlotsRefusedError, SlotFillRefusedError, SlotFillsConflictError, SlotFillsRefusedError, UNICODE_RANGE_CLEARANCE, UnmeasuredSlotsRefusedError, assertSingleSlotCopy, assertSlotPrintOptions, base64ToBytes, bindingNames, bindingRoot, bookDocument, bookPrintPart, bytesToBase64, classify, collectSlotPrintCopies, expandMotifs, fill, findActiveContent, fontCss, fontFaceCss, fontFormat, fontReferenceRefusals, hasBindings, hasMotifPlacement, hasSlotWork, isSlotCopyRefusal, libraryMotifPlacementRefusal, mergeSlotFitResults, motifPlacementRefusals, parseUnicodeRange, planBookPages, planPrintedPages, prepareSlotCollection, prepareSlots, printedPageAt, renderMotifPlacements, requiredSlotRefusal, sanitize, searchText, slotFillCopyId, slotFillRefusal, slotFillsConflict, unmeasuredSlotRefusal };
+export { BINDINGS, FontCssRefusedError, MAX_SLOT_DECLARATIONS, MOTIF_DEPTH_LIMIT, MotifRefusedError, PAGINATE_FUNCTIONS, REFUSED_ELEMENTS, RequiredSlotsRefusedError, SlotFillRefusedError, SlotFillsConflictError, SlotFillsRefusedError, UNICODE_RANGE_CLEARANCE, UnmeasuredSlotsRefusedError, assertSingleSlotCopy, assertSlotPrintOptions, base64ToBytes, bindingNames, bindingRoot, bookDocument, bookPrintPart, bookStylesheet, bookView, bookViewFromHtml, bytesToBase64, classify, collectSlotPrintCopies, embedFunction, expandMotifs, fill, fillBookFurniture, findActiveContent, fontCss, fontFaceCss, fontFormat, fontReferenceRefusals, furnitureScript, hasBindings, hasMotifPlacement, hasSlotWork, heldPageNumber, isSlotCopyRefusal, libraryMotifPlacementRefusal, mergeSlotFitResults, motifPlacementRefusals, openerPageIndex, pageIndexForTarget, paginateBook, paginateScript, parseUnicodeRange, planBookPages, planPrintedPages, prepareSlotCollection, prepareSlots, printedPageAt, renderMotifPlacements, requiredSlotRefusal, sanitize, sanitizeBookCss, searchText, selectBookParent, slotFillCopyId, slotFillRefusal, slotFillsConflict, unmeasuredSlotRefusal, windowPageIndex };

package/dist/motifs.d.ts

@@ -8,6 +8,7 @@ export interface ExpandMotifsOptions {
   scripts: Scripts;
   prefix: string;
   linkOrigins?: Iterable<string>;
+  dataFonts?: boolean;
   signal: AbortSignal;
   deferPaths?: Iterable<string>;
 }

package/dist/motifs.js

@@ -102,6 +102,7 @@ async function fillVersion(html, placement, element, state) {
 		prefix: state.options.prefix,
 		data: data ?? {},
 		linkOrigins: state.options.linkOrigins,
+		dataFonts: state.options.dataFonts,
 		registerHrefOrigins: true,
 		deferPaths: state.options.deferPaths
 	});
@@ -181,7 +182,10 @@ async function expandPlacement(element, inheritedLang, stack, state) {
 		lang,
 		fields
 	};
-	let versionHtml = sanitize(choice.version.html, { linkOrigins: state.options.linkOrigins });
+	let versionHtml = sanitize(choice.version.html, {
+		linkOrigins: state.options.linkOrigins,
+		dataFonts: state.options.dataFonts
+	});
 	if (bound && !state.validateOnly) {
 		const filled = await fillVersion(versionHtml, placement, element, state);
 		if (filled === void 0) return;

package/dist/printed-pages.d.ts

@@ -1,4 +1,4 @@
-import { NumberSystem, PrintedPageRole, PrintedPageRules } from "@segnavia/format";
+import { BookSectionParents as BookSectionParents$1, NumberSystem, PrintedPageRole, PrintedPageRules } from "@segnavia/format";
 //#region src/printed-pages.d.ts
 export type PrintedPageSide = 'left' | 'right';
 export interface PrintedPagePlan {
@@ -9,13 +9,15 @@ export interface PrintedPagePlan {
   number: string;
   total: string;
 }
+export type BookPrintedPageRole = 'recto' | 'verso' | 'opener' | 'blank' | 'fixed';
 export interface BookPageFacts {
   blank: boolean;
   sectionOpening: boolean;
   folio?: string;
   runningHead?: string;
+  role?: BookPrintedPageRole | 'body';
+  parents?: BookSectionParents$1;
 }
-export type BookPrintedPageRole = 'recto' | 'verso' | 'opener' | 'blank';
 export interface BookPrintedPagePlan extends Omit<BookPageFacts, 'blank' | 'sectionOpening'> {
   index: number;
   role: BookPrintedPageRole;
@@ -24,5 +26,17 @@ export interface BookPrintedPagePlan extends Omit<BookPageFacts, 'blank' | 'sect
 }
 export declare function printedPageAt(rules: PrintedPageRules, index: number, count: number, direction: 'ltr' | 'rtl'): Pick<PrintedPagePlan, 'role' | 'parent' | 'side'>;
 export declare function planPrintedPages(rules: PrintedPageRules, count: number, direction: 'ltr' | 'rtl', numbering?: NumberSystem): PrintedPagePlan[];
-export declare function planBookPages(facts: BookPageFacts[], direction: 'ltr' | 'rtl'): BookPrintedPagePlan[];
-//#endregion
+export declare function selectBookParent(fact: BookPageFacts, index: number, count: number, direction: 'ltr' | 'rtl', names?: {
+  readonly recto: "segnavia:recto";
+  readonly verso: "segnavia:verso";
+  readonly opener: "segnavia:opener";
+  readonly blank: "segnavia:blank";
+}): Pick<BookPrintedPagePlan, 'role' | 'parent' | 'side'>;
+export declare function planBookPages(facts: BookPageFacts[], direction: 'ltr' | 'rtl', names?: {
+  readonly recto: "segnavia:recto";
+  readonly verso: "segnavia:verso";
+  readonly opener: "segnavia:opener";
+  readonly blank: "segnavia:blank";
+}): BookPrintedPagePlan[];
+//#endregion
+export type { BookSectionParents$1 as BookSectionParents };

package/dist/printed-pages.js

@@ -39,37 +39,59 @@ function planPrintedPages(rules, count, direction, numbering = "arabic") {
 		};
 	});
 }
-const BOOK_RULES = {
-	every: BOOK_PARENT_NAMES.recto,
-	recto: BOOK_PARENT_NAMES.recto,
-	verso: BOOK_PARENT_NAMES.verso
-};
-function planBookPages(facts, direction) {
+function selectBookParent(fact, index, count, direction, names = BOOK_PARENT_NAMES) {
+	const physical = printedPageAt({
+		every: names.recto,
+		recto: names.recto,
+		verso: names.verso
+	}, index, count, direction);
+	const named = fact.parents;
+	if (fact.blank) return {
+		role: "blank",
+		parent: named?.blank ?? names.blank,
+		side: physical.side
+	};
+	if (fact.role === "fixed") return {
+		role: "fixed",
+		parent: named?.fixed ?? names.blank,
+		side: physical.side
+	};
+	if (fact.sectionOpening && fact.role !== "body") return {
+		role: "opener",
+		parent: named?.opener ?? names.opener,
+		side: physical.side
+	};
+	if (physical.role === "verso") return {
+		role: "verso",
+		parent: named?.verso ?? names.verso,
+		side: physical.side
+	};
+	return {
+		role: "recto",
+		parent: named?.recto ?? names.recto,
+		side: physical.side
+	};
+}
+function planBookPages(facts, direction, names = BOOK_PARENT_NAMES) {
 	return facts.map((fact, offset) => {
 		const index = offset + 1;
-		const selected = printedPageAt(BOOK_RULES, index, facts.length, direction);
-		if (fact.blank) return {
+		const selected = selectBookParent(fact, index, facts.length, direction, names);
+		if (selected.role === "blank") return {
 			index,
-			role: "blank",
-			parent: BOOK_PARENT_NAMES.blank,
-			side: selected.side
+			...selected
 		};
-		if (fact.sectionOpening) return {
+		if (selected.role === "opener") return {
 			index,
-			role: "opener",
-			parent: BOOK_PARENT_NAMES.opener,
-			side: selected.side,
+			...selected,
 			...fact.folio === void 0 ? {} : { folio: fact.folio }
 		};
 		return {
 			index,
-			role: selected.role === "verso" ? "verso" : "recto",
-			parent: selected.parent,
-			side: selected.side,
+			...selected,
 			...fact.folio === void 0 ? {} : { folio: fact.folio },
 			...fact.runningHead === void 0 ? {} : { runningHead: fact.runningHead }
 		};
 	});
 }
 //#endregion
-export { planBookPages, planPrintedPages, printedPageAt };
+export { planBookPages, planPrintedPages, printedPageAt, selectBookParent };

package/dist/sanitize-css.d.ts

@@ -0,0 +1,6 @@
+//#region src/sanitize-css.d.ts
+export interface CssSanitizeResult {
+  css: string;
+  removed: string[];
+}
+//#endregion

package/dist/sanitize-css.js

@@ -0,0 +1,308 @@
+import { cssEscape, cssTokens } from "./css-tokens.js";
+//#region src/sanitize-css.ts
+const DATA_IMAGE = /^data:image\//i;
+const DATA_FONT = /^data:(?:font\/[^;,]+|application\/(?:x-)?font-[^;,]+)(?:[;,])/i;
+const DECLARATION_AT_RULES = /* @__PURE__ */ new Set([
+	"counter-style",
+	"font-face",
+	"font-palette-values",
+	"page",
+	"property"
+]);
+const FETCH_FUNCTIONS = [
+	"url",
+	"src",
+	"image-set",
+	"-webkit-image-set"
+];
+const FETCH_AT_KEYWORDS = ["import"];
+const FETCH_CONSTRUCT = new RegExp(`@(?:${FETCH_AT_KEYWORDS.join("|")})|(?:${FETCH_FUNCTIONS.join("|")})\\(`, "gi");
+const FETCH_FUNCTION_SET = new Set(FETCH_FUNCTIONS);
+const FETCH_AT_KEYWORD_SET = new Set(FETCH_AT_KEYWORDS);
+function normalizedUrl(value, decodeEscapes = true) {
+	let normalized = "";
+	for (let at = 0; at < value.length;) if (value.startsWith("/*", at)) {
+		const close = value.indexOf("*/", at + 2);
+		at = close < 0 ? value.length : close + 2;
+	} else if (decodeEscapes && value[at] === "\\") {
+		const escaped = cssEscape(value, at);
+		normalized += escaped[0];
+		at = escaped[1];
+	} else normalized += value[at++] ?? "";
+	normalized = normalized.trim();
+	if ((normalized[0] === "\"" || normalized[0] === "'") && normalized.at(-1) === normalized[0]) normalized = normalized.slice(1, -1);
+	return [...normalized].filter((character) => character.charCodeAt(0) > 32 && character.charCodeAt(0) !== 127).join("").toLowerCase();
+}
+function decodedCss(source) {
+	let decoded = "";
+	for (let at = 0; at < source.length;) if (source[at] === "\\") {
+		const escaped = cssEscape(source, at);
+		decoded += escaped[0];
+		at = escaped[1];
+	} else decoded += source[at++] ?? "";
+	return decoded.replace(/\/\*[\s\S]*?(?:\*\/|$)/g, "").replace(/\s+/g, " ").toLowerCase();
+}
+function functionArgument(value) {
+	let depth = 1;
+	let quote = "";
+	for (let at = 0; at < value.length; at++) {
+		const character = value[at] ?? "";
+		if (quote) {
+			if (character === quote) quote = "";
+			continue;
+		}
+		if (character === "\"" || character === "'") quote = character;
+		else if (character === "(") depth++;
+		else if (character === ")" && --depth === 0) return value.slice(0, at);
+	}
+	return value;
+}
+function imageSetItems(value) {
+	const items = [];
+	let start = 0;
+	let depth = 0;
+	let quote = "";
+	for (let at = 0; at < value.length; at++) {
+		const character = value[at] ?? "";
+		if (quote) {
+			if (character === quote) quote = "";
+			continue;
+		}
+		if (character === "\"" || character === "'") quote = character;
+		else if (character === "(") depth++;
+		else if (character === ")") depth--;
+		else if (character === "," && depth === 0) {
+			items.push(value.slice(start, at));
+			start = at + 1;
+		}
+	}
+	items.push(value.slice(start));
+	return items;
+}
+function allowedAuditData(value, dataFonts) {
+	let argument = value.trimStart();
+	if (/^url\(/i.test(argument)) argument = functionArgument(argument.slice(argument.indexOf("(") + 1)).trimStart();
+	argument = argument.replace(/^["']/, "");
+	return DATA_IMAGE.test(argument) || dataFonts && DATA_FONT.test(argument);
+}
+function lexicalFindings(source, dataFonts) {
+	const text = decodedCss(source);
+	const found = [];
+	for (const match of text.matchAll(FETCH_CONSTRUCT)) {
+		const construct = match[0].toLowerCase();
+		const after = text.slice((match.index ?? 0) + construct.length);
+		if (construct.startsWith("@")) {
+			if (!allowedAuditData(after, dataFonts)) found.push(construct);
+			continue;
+		}
+		const name = construct.slice(0, -1);
+		const argument = functionArgument(after);
+		if (!(name === "image-set" || name === "-webkit-image-set" ? imageSetItems(argument).every((item) => allowedAuditData(item, dataFonts)) : allowedAuditData(argument, dataFonts))) found.push(`${name === "-webkit-image-set" ? "image-set" : name}()`);
+	}
+	return [...new Set(found)];
+}
+function matchingClose(list, open, limit = list.length) {
+	let depth = 1;
+	for (let at = open + 1; at < limit; at++) {
+		if (list[at]?.kind === "open" || list[at]?.kind === "function") depth++;
+		if (list[at]?.kind === "close" && --depth === 0) return at;
+	}
+	return limit;
+}
+function unsafeImageSet(list, start, close) {
+	let depth = 0;
+	for (let at = start + 1; at < close; at++) {
+		const token = list[at];
+		if (token?.kind === "function" || token?.kind === "open") depth++;
+		else if (token?.kind === "close") depth--;
+		else if (token?.kind === "string" && depth === 0 && !DATA_IMAGE.test(normalizedUrl(token.value, false))) return true;
+	}
+	return false;
+}
+function urlFinding(value, dataFonts, fontFace, finding = "url()") {
+	if (DATA_IMAGE.test(value)) return void 0;
+	if (fontFace && DATA_FONT.test(value)) return dataFonts ? void 0 : "data font";
+	return finding;
+}
+function functionFinding(source, list, at, end, dataFonts, fontFace) {
+	const token = list[at];
+	if (token?.kind === "bad-url") return "url()";
+	if (token?.kind === "url") return urlFinding(token.value, dataFonts, fontFace);
+	if (token?.kind !== "function") return void 0;
+	const close = matchingClose(list, at, end);
+	if (!FETCH_FUNCTION_SET.has(token.value)) return void 0;
+	if (token.value === "image-set" || token.value === "-webkit-image-set") return unsafeImageSet(list, at, close) ? "image-set()" : void 0;
+	return urlFinding(normalizedUrl(source.slice(token.end, close < end ? list[close]?.start : source.length)), dataFonts, fontFace, `${token.value}()`);
+}
+function findings(source, list, start, end, dataFonts, fontFace) {
+	const found = [];
+	for (let at = start; at < end; at++) {
+		const token = list[at];
+		if (token?.kind === "at" && FETCH_AT_KEYWORD_SET.has(token.value)) found.push(`@${token.value}`);
+		const finding = functionFinding(source, list, at, end, dataFonts, fontFace);
+		if (finding) found.push(finding);
+	}
+	return [...new Set(found)];
+}
+function boundary(list, start, end) {
+	let depth = 0;
+	for (let at = start; at < end; at++) {
+		const token = list[at];
+		if (!token) continue;
+		if (depth === 0 && (token.kind === "semicolon" || token.kind === "open" && token.value === "{")) return at;
+		if (token.kind === "function" || token.kind === "open") depth++;
+		else if (token.kind === "close") depth--;
+	}
+	return end;
+}
+function cleanStatement(source, list, start, edge, dataFonts, removed) {
+	const first = list[start];
+	const delimiter = list[edge];
+	if (!first || !delimiter) return {
+		character: source.length,
+		next: edge,
+		text: ""
+	};
+	const unsafe = first.kind === "at" && first.value === "import" ? ["@import"] : findings(source, list, start, edge + 1, dataFonts, false);
+	if (unsafe.length > 0) removed.push(...unsafe);
+	return {
+		character: delimiter.end,
+		next: edge + 1,
+		text: unsafe.length === 0 ? source.slice(first.start, delimiter.end) : ""
+	};
+}
+function cleanBlockRule(source, list, start, edge, end, dataFonts, removed) {
+	const first = list[start];
+	const delimiter = list[edge];
+	if (!first || !delimiter) return {
+		character: source.length,
+		next: end,
+		text: ""
+	};
+	const close = matchingClose(list, edge, end);
+	const ruleName = first.kind === "at" ? first.value : "";
+	const fontFace = ruleName === "font-face";
+	const unsafe = findings(source, list, start, edge, dataFonts, false);
+	if (fontFace) unsafe.push(...findings(source, list, edge + 1, close, dataFonts, true));
+	if (unsafe.length > 0) removed.push(...unsafe);
+	let text = "";
+	if (unsafe.length === 0) {
+		const body = first.kind !== "at" || DECLARATION_AT_RULES.has(ruleName) ? cleanDeclarations(source, list, edge + 1, close, dataFonts, removed, fontFace) : cleanRules(source, list, edge + 1, close, dataFonts, removed);
+		text = source.slice(first.start, delimiter.end) + body;
+		if (close < end) text += source.slice(list[close]?.start, list[close]?.end);
+	}
+	return {
+		character: close < end ? list[close]?.end ?? delimiter.end : source.length,
+		next: close < end ? close + 1 : end,
+		text
+	};
+}
+function cleanRuleUnit(source, list, start, end, dataFonts, removed) {
+	const first = list[start];
+	if (!first) return {
+		character: source.length,
+		next: end,
+		text: ""
+	};
+	const edge = boundary(list, start, end);
+	const delimiter = list[edge];
+	if (edge >= end && delimiter?.kind === "close") {
+		const unsafe = findings(source, list, start, end, dataFonts, false);
+		if (unsafe.length > 0) removed.push(...unsafe);
+		return {
+			character: delimiter.start,
+			next: end,
+			text: unsafe.length === 0 ? source.slice(first.start, delimiter.start) : ""
+		};
+	}
+	if (!delimiter) {
+		const unsafe = findings(source, list, start, end, dataFonts, false);
+		if (unsafe.length > 0) removed.push(...unsafe);
+		return {
+			character: source.length,
+			next: end,
+			text: unsafe.length === 0 ? source.slice(first.start) : ""
+		};
+	}
+	return delimiter.kind === "semicolon" ? cleanStatement(source, list, start, edge, dataFonts, removed) : cleanBlockRule(source, list, start, edge, end, dataFonts, removed);
+}
+function cleanRules(source, list, start, end, dataFonts, removed) {
+	let output = "";
+	let character = start === 0 ? 0 : list[start - 1]?.end ?? source.length;
+	let at = start;
+	while (at < end) {
+		const first = list[at];
+		if (!first) break;
+		output += source.slice(character, first.start);
+		const unit = cleanRuleUnit(source, list, at, end, dataFonts, removed);
+		output += unit.text;
+		character = unit.character;
+		at = unit.next;
+	}
+	const rangeEnd = end < list.length ? list[end]?.start ?? source.length : source.length;
+	return output + source.slice(character, rangeEnd);
+}
+function cleanDeclarationSegment(source, list, start, end, character, dataFonts, fontFace, removed) {
+	const delimiter = list[end];
+	const hasSemicolon = delimiter?.kind === "semicolon";
+	const segmentEnd = hasSemicolon ? end + 1 : end;
+	const finish = hasSemicolon ? delimiter.end : end < list.length ? delimiter?.start ?? source.length : source.length;
+	const unsafe = findings(source, list, start, segmentEnd, dataFonts, fontFace);
+	if (unsafe.length === 0) return [source.slice(character, finish), finish];
+	removed.push(...unsafe);
+	return [source.slice(character, list[start]?.start ?? finish), finish];
+}
+function declarationEnd(list, start, end) {
+	let depth = 0;
+	for (let at = start; at < end; at++) {
+		const token = list[at];
+		if (depth === 0 && token?.kind === "semicolon") return at;
+		if (token?.kind === "function" || token?.kind === "open") depth++;
+		else if (token?.kind === "close") depth--;
+	}
+	return end;
+}
+function isDeclaration(list, start) {
+	return list[start]?.value.startsWith("--") === true && list[start + 1]?.value === ":";
+}
+function cleanDeclarations(source, list, start, end, dataFonts, removed, fontFace = false) {
+	let output = "";
+	let character = start === 0 ? 0 : list[start - 1]?.end ?? source.length;
+	let at = start;
+	while (at < end) {
+		const first = list[at];
+		if (!first) break;
+		output += source.slice(character, first.start);
+		const edge = boundary(list, at, end);
+		if (list[edge]?.kind === "open" && !isDeclaration(list, at)) {
+			const unit = cleanBlockRule(source, list, at, edge, end, dataFonts, removed);
+			output += unit.text;
+			character = unit.character;
+			at = unit.next;
+		} else {
+			const finish = declarationEnd(list, at, end);
+			const cleaned = cleanDeclarationSegment(source, list, at, finish, first.start, dataFonts, fontFace, removed);
+			output += cleaned[0];
+			character = cleaned[1];
+			at = finish < end ? finish + 1 : end;
+		}
+	}
+	const rangeEnd = end < list.length ? list[end]?.start ?? source.length : source.length;
+	return output + source.slice(character, rangeEnd);
+}
+function sanitizeCss(source, declarations, dataFonts) {
+	const list = cssTokens(source);
+	const removed = [];
+	const css = declarations ? cleanDeclarations(source, list, 0, list.length, dataFonts, removed) : cleanRules(source, list, 0, list.length, dataFonts, removed);
+	const auditRemoved = lexicalFindings(css, dataFonts);
+	if (auditRemoved.length > 0) return {
+		css: "",
+		removed: [.../* @__PURE__ */ new Set([...removed, ...auditRemoved])]
+	};
+	return {
+		css,
+		removed: [...new Set(removed)]
+	};
+}
+//#endregion
+export { lexicalFindings, sanitizeCss };

package/dist/sanitize.d.ts

@@ -4,6 +4,7 @@ export declare const REFUSED_ELEMENTS: Set<string>;
 export interface SanitizeOptions {
   linkOrigins?: Iterable<string>;
   keepScripts?: boolean;
+  dataFonts?: boolean;
   refuseElements?: Iterable<string>;
 }
 export declare function findActiveContent(html: string, options?: SanitizeOptions): string[];

package/dist/sanitize.js

@@ -1,3 +1,4 @@
+import { sanitizeCss } from "./sanitize-css.js";
 import { isTag } from "domhandler";
 import render from "dom-serializer";
 import { parseDocument } from "htmlparser2";
@@ -44,7 +45,9 @@ const URL_ATTRIBUTES = /* @__PURE__ */ new Set([
 	"longdesc",
 	"ping"
 ]);
-const DATA_IMAGE = /^data:image\/(?:png|jpeg|gif|webp|svg\+xml)(?:[;,])/i;
+const DATA_IMAGE_TYPES = String.raw`(?:png|jpeg|gif|webp|svg\+xml)`;
+const DATA_IMAGE = new RegExp(`^data:image/${DATA_IMAGE_TYPES}(?:[;,])`, "i");
+const DATA_IMAGE_SRCSET = new RegExp(`^data:image/${DATA_IMAGE_TYPES}(?:;[^,]*)?,[^,]*$`, "i");
 function parse(html) {
 	return parseDocument(html, {
 		lowerCaseAttributeNames: true,
@@ -94,24 +97,13 @@ function safeUrl(attribute, value, allowed) {
 			return false;
 		}
 	}
-	if (attribute === "srcset") return /^data:image\/(?:png|jpeg|gif|webp|svg\+xml)(?:;[^,]*)?,[^,]*$/i.test(url);
+	if (attribute === "srcset") return DATA_IMAGE_SRCSET.test(url);
 	if (attribute === "src" || attribute === "poster") return DATA_IMAGE.test(url);
 	return false;
 }
-function safeStyle(value) {
-	const matches = [...value.matchAll(/url\(\s*(["']?)(.*?)\1\s*\)/gi)];
-	if ((value.match(/url\(/gi)?.length ?? 0) !== matches.length) return false;
-	return matches.every((match) => normalizeUrl(match[2] ?? "").startsWith("data:image/"));
-}
-function refusedStyleElement(element) {
-	if (element.name.toLowerCase() !== "style") return false;
-	const text = element.children.map((child) => "data" in child ? child.data : "").join("");
-	return /@import/i.test(text) || !safeStyle(text);
-}
 function refusedElement(element, refused, keepScripts) {
 	const name = element.name.toLowerCase();
 	if (name === "script") return !keepScripts;
-	if (refusedStyleElement(element)) return true;
 	if (refused.has(name)) return true;
 	if (insideSvg(element) && REFUSED_SVG_ELEMENTS.has(name)) return true;
 	return name === "meta" && Object.keys(element.attribs).some((attribute) => attribute.toLowerCase() === "http-equiv");
@@ -119,11 +111,23 @@ function refusedElement(element, refused, keepScripts) {
 function refusedAttribute(name, value, allowed, keepScripts) {
 	const attribute = name.toLowerCase();
 	if (attribute.startsWith("on")) return !keepScripts;
-	return REFUSED_ATTRIBUTES.has(attribute) || URL_ATTRIBUTES.has(attribute) && !safeUrl(attribute, value, allowed) || attribute === "style" && !safeStyle(value);
+	return REFUSED_ATTRIBUTES.has(attribute) || URL_ATTRIBUTES.has(attribute) && !safeUrl(attribute, value, allowed);
 }
 function refusedSet(options) {
 	return new Set([...REFUSED_ELEMENTS, ...options.refuseElements ?? []].map((name) => name.toLowerCase()));
 }
+function styleText(element) {
+	return element.children.map((child) => "data" in child ? child.data : "").join("");
+}
+function styleFindings(element, dataFonts) {
+	const found = [];
+	if (element.name.toLowerCase() === "style") for (const construct of sanitizeCss(styleText(element), false, dataFonts).removed) found.push(`<style> ${construct}`);
+	for (const [name, value] of Object.entries(element.attribs)) {
+		if (name.toLowerCase() !== "style") continue;
+		for (const construct of sanitizeCss(value, true, dataFonts).removed) found.push(`${element.name}[${name}] ${construct}`);
+	}
+	return found;
+}
 function findActiveContent(html, options = {}) {
 	const allowed = origins(options.linkOrigins);
 	const refused = refusedSet(options);
@@ -131,10 +135,27 @@ function findActiveContent(html, options = {}) {
 	const found = [];
 	for (const element of elements(parse(html).children)) {
 		if (refusedElement(element, refused, keepScripts)) found.push(`<${element.name}>`);
+		found.push(...styleFindings(element, options.dataFonts ?? false));
 		for (const [name, value] of Object.entries(element.attribs)) if (refusedAttribute(name, value, allowed, keepScripts)) found.push(`${element.name}[${name}]`);
 	}
 	return found;
 }
+function cleanStyleElement(element, dataFonts) {
+	if (element.name.toLowerCase() !== "style") return;
+	const children = element.children.filter((candidate) => "data" in candidate);
+	const child = children[0];
+	if (!child || !("data" in child)) return;
+	child.data = sanitizeCss(styleText(element), false, dataFonts).css;
+	for (const extra of children.slice(1)) if ("data" in extra) extra.data = "";
+}
+function cleanAttributes(element, allowed, keepScripts, dataFonts) {
+	for (const [name, value] of Object.entries(element.attribs)) if (refusedAttribute(name, value, allowed, keepScripts)) delete element.attribs[name];
+	else if (name.toLowerCase() === "style") {
+		const css = sanitizeCss(value, true, dataFonts).css.trim();
+		if (css) element.attribs[name] = css;
+		else delete element.attribs[name];
+	}
+}
 function sanitize(html, options = {}) {
 	const allowed = origins(options.linkOrigins);
 	const refused = refusedSet(options);
@@ -143,7 +164,8 @@ function sanitize(html, options = {}) {
 	const prune = (nodes) => nodes.filter((node) => {
 		if (!isTag(node)) return true;
 		if (refusedElement(node, refused, keepScripts)) return false;
-		for (const [name, value] of Object.entries(node.attribs)) if (refusedAttribute(name, value, allowed, keepScripts)) delete node.attribs[name];
+		cleanStyleElement(node, options.dataFonts ?? false);
+		cleanAttributes(node, allowed, keepScripts, options.dataFonts ?? false);
 		node.children = prune(node.children);
 		return true;
 	});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@segnavia/html",
-  "version": "0.27.0",
+  "version": "0.28.0",
   "type": "module",
   "license": "MIT",
   "repository": {
@@ -26,7 +26,7 @@
     }
   },
   "dependencies": {
-    "@segnavia/format": "0.27.0",
+    "@segnavia/format": "0.28.0",
     "dom-serializer": "^2.0.0",
     "domhandler": "^5.0.3",
     "htmlparser2": "^10.0.0",