Files
xslt-tools/src/transform/saxonJsDynamicCompiler.ts

317 lines
8.3 KiB
TypeScript

type SefNode = Record<string, unknown> & {
N?: string;
C?: SefNode[];
firstChild?: unknown;
parentNode?: SefNode;
};
type SaxonXdmMap = {
inSituPut: (key: unknown, value: unknown[]) => void;
};
let saxonLoadPromise: Promise<void> | null = null;
export function resolveSaxonScriptUrl(
baseUrl = import.meta.env.BASE_URL,
pageUrl = window.location.href
): string {
const appBase = new URL(baseUrl, pageUrl);
return new URL('vendor/saxon/SaxonJS2.js', appBase).href;
}
type SaxonPrivateRuntime = {
getPlatform?: () => {
resource?: (name: string) => unknown;
parseXmlFromString?: (text: string) => unknown;
};
checkOptions?: (options: Record<string, unknown>) => Record<string, unknown>;
internalTransform?: (
stylesheetInternal: unknown,
source: unknown,
checkedOptions: Record<string, unknown>
) => void;
getResource?: (options: {
file?: string;
location?: string;
text?: string;
type?: 'xml' | 'json' | string;
}) => Promise<unknown>;
XPath?: {
sefToJSON?: (node: unknown, keepDebug: boolean) => unknown;
};
XS?: {
QName?: {
fromParts: (prefix: string, uri: string, local: string) => unknown;
};
};
XdmMap?: new () => SaxonXdmMap;
transform?: (
options: Record<string, unknown>,
execution?: 'sync' | 'async'
) => Promise<{ principalResult?: unknown }>;
};
async function getSaxon(): Promise<SaxonPrivateRuntime> {
await ensureSaxonLoaded();
const saxon = window.SaxonJS as SaxonPrivateRuntime | undefined;
if (!saxon) {
throw new Error('window.SaxonJS is not loaded.');
}
return saxon;
}
async function ensureSaxonLoaded(): Promise<void> {
if (window.SaxonJS) return;
saxonLoadPromise ??= new Promise<void>((resolve, reject) => {
const saxonScriptUrl = resolveSaxonScriptUrl();
const existingScript = document.querySelector<HTMLScriptElement>(
`script[src="${saxonScriptUrl}"]`
);
if (existingScript) {
existingScript.addEventListener('load', () => resolve(), { once: true });
existingScript.addEventListener(
'error',
() => reject(new Error('Failed to load SaxonJS2.js.')),
{ once: true }
);
return;
}
const script = document.createElement('script');
script.src = saxonScriptUrl;
script.async = true;
script.onload = () => resolve();
script.onerror = () => reject(new Error('Failed to load SaxonJS2.js.'));
document.head.appendChild(script);
});
await saxonLoadPromise;
}
function addParentPointers(node: SefNode): void {
node.C?.forEach((child) => {
child.parentNode = node;
addParentPointers(child);
});
}
function checksum(sef: SefNode): void {
function hashString(value: string, seed: number): number {
let current = seed << 8;
for (let index = 0; index < value.length; index += 1) {
current = (current << 1) + value.charCodeAt(index);
}
return current;
}
function hashPair(name: string, uri: string, seed: number): number {
return hashString(name, seed) ^ hashString(uri, seed);
}
let hash = 0;
let counter = 0;
function visit(node: SefNode): void {
hash ^= hashPair(
String(node.N ?? ''),
'http://ns.saxonica.com/xslt/export',
counter++
);
for (const [key, value] of Object.entries(node)) {
if (key !== 'N' && key !== 'C' && key !== String.fromCharCode(931)) {
hash ^= hashPair(key, '', counter);
hash ^= hashString(String(value), counter);
}
}
node.C?.forEach((child) => visit(child));
hash ^= 1;
}
visit(sef);
sef[String.fromCharCode(931)] = (
hash < 0 ? 4294967295 + hash + 1 : hash
).toString(16);
}
function getFirstPrincipalNode(principalResult: unknown): SefNode {
const value = Array.isArray(principalResult)
? principalResult[0]
: principalResult;
if (!value) {
throw new Error('The SaxonJS compiler returned no principal result.');
}
return value as SefNode;
}
function getSyntheticStylesheetBaseUri(): string {
return new URL(
'/__xsl-tools__/in-memory-stylesheet.xsl',
window.location.href
).href;
}
function serializeSaxonResult(value: unknown): string {
if (typeof value === 'string') {
return value;
}
if (value instanceof XMLDocument || value instanceof Document) {
return new XMLSerializer().serializeToString(value);
}
if (value instanceof Element) {
return new XMLSerializer().serializeToString(value);
}
if (value instanceof DocumentFragment) {
const container = document.createElement('div');
container.append(
...Array.from(value.childNodes).map((node) => node.cloneNode(true))
);
return container.innerHTML;
}
return String(value ?? '');
}
async function parseXmlForSaxon(
saxon: SaxonPrivateRuntime,
xmlText: string
): Promise<Record<string, unknown>> {
const baseUri = getSyntheticStylesheetBaseUri();
if (saxon.getResource) {
try {
const parsed = (await saxon.getResource({
text: xmlText,
type: 'xml',
})) as Record<string, unknown>;
parsed._saxonBaseUri = baseUri;
parsed._saxonDocUri = baseUri;
return parsed;
} catch {
// Some browser/runtime combinations only support parsing through the platform adapter.
}
}
const parsed = saxon.getPlatform?.().parseXmlFromString?.(xmlText) as
| Record<string, unknown>
| undefined;
if (!parsed) {
throw new Error(
'Could not parse XML for SaxonJS: neither getResource({ text }) nor platform.parseXmlFromString(...) is available.'
);
}
parsed._saxonBaseUri = baseUri;
parsed._saxonDocUri = baseUri;
return parsed;
}
export async function compileXsltTextToSefJson(
xsltText: string,
compilerOptions: Record<string, unknown> = {}
): Promise<string> {
const saxon = await getSaxon();
const platform = saxon.getPlatform?.();
if (!platform?.resource) {
throw new Error('SaxonJS platform.resource(...) is not available.');
}
if (!saxon.checkOptions || !saxon.internalTransform) {
throw new Error(
'SaxonJS internal compiler APIs are not available: checkOptions/internalTransform missing.'
);
}
if (!saxon.XPath?.sefToJSON) {
throw new Error('SaxonJS.XPath.sefToJSON(...) is not available.');
}
if (!saxon.XdmMap || !saxon.XS?.QName?.fromParts) {
throw new Error('SaxonJS XdmMap/QName APIs are not available.');
}
const compiler = platform.resource('compiler') as SefNode | undefined;
if (!compiler || compiler.N !== 'package') {
throw new Error(
'SaxonJS compiler resource is not available. You are probably loading SaxonJS2.rt.js, which contains only the runtime. Use the full SaxonJS2.js browser file for dynamic XSLT-to-SEF compilation.'
);
}
addParentPointers(compiler);
const stylesheetParams = new saxon.XdmMap();
const staticParameters = new saxon.XdmMap();
stylesheetParams.inSituPut(
saxon.XS.QName.fromParts('', '', 'staticParameters'),
[staticParameters]
);
const source = await parseXmlForSaxon(saxon, xsltText);
const options = {
destination: 'application',
initialMode: 'compile-complete',
isDynamicStylesheet: true,
templateParams: {
'Q{}stylesheet-base-uri': getSyntheticStylesheetBaseUri(),
'Q{}options': {
noXPath: false,
...compilerOptions,
},
},
stylesheetParams,
stylesheetInternal: compiler,
outputProperties: {},
};
if (compiler.relocatable === 'true') {
(options as Record<string, unknown>).isRelocatableStylesheet = true;
}
const checkedOptions = saxon.checkOptions(options);
saxon.internalTransform(compiler, source, checkedOptions);
const sefXml = getFirstPrincipalNode(checkedOptions.principalResult);
const sefJson = saxon.XPath.sefToJSON(
sefXml.firstChild ?? sefXml,
false
) as SefNode;
checksum(sefJson);
return JSON.stringify(sefJson);
}
export async function transformXmlWithDynamicSaxon(
xmlText: string,
xsltText: string
): Promise<string> {
const saxon = await getSaxon();
if (!saxon.transform) {
throw new Error('SaxonJS transform API is not available.');
}
const generatedSef = await compileXsltTextToSefJson(xsltText);
const result = await saxon.transform(
{
stylesheetText: generatedSef,
sourceText: xmlText,
destination: 'serialized',
},
'async'
);
return serializeSaxonResult(result.principalResult);
}