Validate xml before parsing.

Author: johnspurlock-skymethod

validator-worker/common/deps_xml.ts

@@ -1,2 +1,2 @@
-export { parseXml, findChildElements, findElementRecursive, computeAttributeMap, qnameEq, qnamesInclude } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/xml_parser.ts';
-export type { ExtendedXmlNode, Qname } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/xml_parser.ts';
+export { parseXml, validateXml, findChildElements, findElementRecursive, computeAttributeMap, qnameEq, qnamesInclude } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/xml_parser.ts';
+export type { ExtendedXmlNode, Qname } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/xml_parser.ts';

validator-worker/common/validation_job_vm.ts

@@ -2,7 +2,7 @@ import { checkMatches, checkEqual } from './check.ts';
 import { Qnames } from './qnames.ts';
 import { isReadonlyArray } from './util.ts';
 import { RuleReference, MessageOptions, ValidationCallbacks, validateFeedXml, podcastIndexReference } from './validator.ts';
-import { computeAttributeMap, ExtendedXmlNode, parseXml } from './deps_xml.ts';
+import { computeAttributeMap, ExtendedXmlNode, parseXml, validateXml } from './deps_xml.ts';
 import { setIntersect } from './sets.ts';
 import { InMemoryCache, Callbacks, Comment, makeRateLimitedFetcher, makeThreadcap, Threadcap, updateThreadcap, Fetcher as ThreadcapFetcher } from './deps_comments.ts';
 
@@ -189,6 +189,11 @@ export class ValidationJobVM {
 
                     let xml: ExtendedXmlNode | undefined;
                     try {
+                        const result = validateXml(text);
+                        if (result !== true) {
+                            const { code, col, line, msg } = result;
+                            throw new Error(`${code} (at line ${line}, col ${col}): ${msg}`);
+                        }
                         xml = parseXml(text);
                         console.log(xml);
                     } catch (e) {

validator-worker/deps_cli.ts

@@ -1,7 +1,7 @@
 export { basename, dirname, join, fromFileUrl, resolve, toFileUrl, extname, relative, isAbsolute, normalize, parse as parsePath, globToRegExp } from 'https://deno.land/std@0.140.0/path/mod.ts';
 export { parse as parseFlags } from 'https://deno.land/std@0.140.0/flags/mod.ts';
-export { ModuleWatcher } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/cli/module_watcher.ts';
-export { bundle } from 'https://raw.githubusercontent.com/skymethod/denoflare/e5a9ba66237f9e3a5d555442252539ec267ae989/cli/bundle.ts';
-export { fileExists } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/cli/fs_util.ts';
-export { Bytes } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/bytes.ts';
-export { parseJsonc } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/cli/jsonc.ts';
+export { ModuleWatcher } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/cli/module_watcher.ts';
+export { bundle } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/cli/bundle.ts';
+export { fileExists } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/cli/fs_util.ts';
+export { Bytes } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/bytes.ts';
+export { parseJsonc } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/cli/jsonc.ts';

validator-worker/deps_worker.ts

@@ -1,4 +1,4 @@
-export { Bytes } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/bytes.ts';
-export { importText } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/import_text.ts';
-export type { IncomingRequestCf, ModuleWorkerContext, DurableObjectNamespace, DurableObjectState } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/common/cloudflare_workers_types.d.ts';
-export { ColoFromTrace } from 'https://raw.githubusercontent.com/skymethod/denoflare/v0.5.1/examples/tiered-websockets-worker/colo_from_trace.ts';
+export { Bytes } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/bytes.ts';
+export { importText } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/import_text.ts';
+export type { IncomingRequestCf, ModuleWorkerContext, DurableObjectNamespace, DurableObjectState } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/common/cloudflare_workers_types.d.ts';
+export { ColoFromTrace } from 'https://raw.githubusercontent.com/skymethod/denoflare/eeacf004d8ec15913f94d5644d10f8f04d5340d2/examples/tiered-websockets-worker/colo_from_trace.ts';

validator-worker/static/app.js

@@ -2389,11 +2389,13 @@ var getTraversalObj$1 = parser.getTraversalObj;
 parser.j2xParser;
 parser.parse;
 parser.parseToNimn;
-parser.validate;
-function decodeXml(encoded) {
+var validate$1 = parser.validate;
+function decodeXml(encoded, additionalEntities = {}) {
     return encoded.replaceAll(/&(#(\d+)|[a-z]+);/g, (str, entity, decimal)=>{
         if (typeof decimal === 'string') return String.fromCharCode(parseInt(decimal));
         if (typeof entity === 'string') {
+            const additional = additionalEntities[entity];
+            if (additional) return additional;
             const rt = ENTITIES_TO_UNENCODED_CHARS[entity];
             if (rt) return rt;
         }
@@ -2407,12 +2409,18 @@ const ENTITIES_TO_UNENCODED_CHARS = {
     'apos': `'`,
     'quot': '"'
 };
-function parseXml(xml) {
+function validateXml(xml) {
+    const rt = validate$1(xml);
+    return rt === true ? rt : rt.err;
+}
+function parseXml(xml, opts = {}) {
+    const { additionalEntities  } = opts;
+    const tagValueProcessor = (tagName)=>decodeXml(tagName, additionalEntities);
     const rt = getTraversalObj$1(xml, {
         ignoreAttributes: false,
         parseAttributeValue: false,
         parseNodeValue: false,
-        tagValueProcessor: decodeXml
+        tagValueProcessor
     });
     const namespaces = new XmlNamespaces();
     applyQnames(rt, namespaces);
@@ -4215,6 +4223,11 @@ class ValidationJobVM {
                     start = Date.now();
                     let xml;
                     try {
+                        const result = validateXml(text);
+                        if (result !== true) {
+                            const { code , col , line , msg  } = result;
+                            throw new Error(`${code} (at line ${line}, col ${col}): ${msg}`);
+                        }
                         xml = parseXml(text);
                         console.log(xml);
                     } catch (e) {

validator-worker/static/app.js.sha1

@@ -1 +1 @@
-343c7c4c43753e414235693f577f48d9f9382172
+d053ae71a909e6e2ac4854bff6e2559e57f525a7