Merge branch 'main' into beta

# Conflicts:
#	bun.lock
#	package.json

Author: macieklamberski

bun.lock

@@ -54,13 +54,13 @@
     "docs:build": "vitepress build docs"
   },
   "dependencies": {
-    "entities": "^8.0.0",
-    "fast-xml-parser": "~5.4.2"
+    "entities": "^7.0.1",
+    "fast-xml-parser": "~5.5.12"
   },
   "devDependencies": {
-    "@types/bun": "^1.3.11",
+    "@types/bun": "^1.3.12",
     "kvalita": "^1.12.4",
-    "tsdown": "^0.21.7",
+    "tsdown": "^0.21.8",
     "vitepress": "^2.0.0-alpha.17"
   }
 }

package.json

@@ -107,6 +107,7 @@ export const parserConfig: X2jOptions = {
   ignorePiTags: true,
   ignoreDeclaration: true,
   attributeNamePrefix: '@',
+  jPath: false,
   transformTagName: (name) => name.toLowerCase(),
   transformAttributeName: (name) => name.toLowerCase(),
 }

src/common/config.ts

@@ -887,3 +887,46 @@ export const parseJsonObject = (value: unknown): unknown => {
     return JSON.parse(value)
   } catch {}
 }
+
+// TODO: Remove expandStopNodes and namespaceContainers after fast-xml-parser
+// resolves stop node performance regression. Revert feed configs from
+// ...expandStopNodes(namespaceStopNodes, [...]) back to ...namespaceStopNodes.
+// https://github.com/NaturalIntelligence/fast-xml-parser/issues/816
+
+// Namespace elements that act as containers for other namespace elements.
+const namespaceContainers = [
+  'media:group',
+  'media:content',
+  'media:group.media:content',
+  'media:embed',
+  'podcast:liveitem',
+  'itunes:owner',
+]
+
+// Expands wildcard stop nodes (*.prefix:element) into explicit paths for
+// each feed container, including paths through namespace containers (e.g.
+// media:group, podcast:liveitem). Only crosses containers with leaves from
+// the same namespace prefix.
+export const expandStopNodes = (
+  stopNodes: Array<string>,
+  feedContainerPaths: Array<string>,
+): Array<string> => {
+  const set = new Set<string>()
+
+  for (const node of stopNodes) {
+    const suffix = node.slice(2)
+    const prefix = suffix.split(':')[0]
+
+    for (const feedPath of feedContainerPaths) {
+      set.add(`${feedPath}.${suffix}`)
+
+      for (const nsContainer of namespaceContainers) {
+        if (nsContainer.split(':')[0] === prefix) {
+          set.add(`${feedPath}.${nsContainer}.${suffix}`)
+        }
+      }
+    }
+  }
+
+  return [...set]
+}

src/common/utils.ts

@@ -1,14 +1,15 @@
 import { XMLParser } from 'fast-xml-parser'
+import { Expression } from 'path-expression-matcher'
 import {
   namespacePrefixes,
   namespaceStopNodes,
   namespaceUris,
   parserConfig,
 } from '../../../common/config.js'
-import { createNamespaceNormalizator } from '../../../common/utils.js'
+import { createNamespaceNormalizator, expandStopNodes } from '../../../common/utils.js'
 
 export const stopNodes = [
-  ...namespaceStopNodes,
+  ...expandStopNodes(namespaceStopNodes, ['feed', 'feed.entry', 'feed.entry.source']),
   'feed.author.name',
   'feed.author.uri',
   'feed.author.url', // Atom 0.3.
@@ -72,7 +73,7 @@ export const stopNodes = [
 
 export const parser = new XMLParser({
   ...parserConfig,
-  stopNodes,
+  stopNodes: stopNodes.map((node) => new Expression(node)),
 })
 
 export const normalizeNamespaces = createNamespaceNormalizator(namespaceUris, namespacePrefixes, [

src/feeds/atom/parse/config.ts

@@ -1,14 +1,20 @@
 import { XMLParser } from 'fast-xml-parser'
+import { Expression } from 'path-expression-matcher'
 import {
   namespacePrefixes,
   namespaceStopNodes,
   namespaceUris,
   parserConfig,
 } from '../../../common/config.js'
-import { createNamespaceNormalizator } from '../../../common/utils.js'
+import { createNamespaceNormalizator, expandStopNodes } from '../../../common/utils.js'
 
 export const stopNodes = [
-  ...namespaceStopNodes,
+  ...expandStopNodes(namespaceStopNodes, [
+    'rdf:rdf.channel',
+    'rdf:rdf.item',
+    'rdf:rdf.image',
+    'rdf:rdf.textinput',
+  ]),
   'rdf:rdf.channel.title',
   'rdf:rdf.channel.link',
   'rdf:rdf.channel.description',
@@ -26,7 +32,7 @@ export const stopNodes = [
 
 export const parser = new XMLParser({
   ...parserConfig,
-  stopNodes,
+  stopNodes: stopNodes.map((node) => new Expression(node)),
 })
 
 export const normalizeNamespaces = createNamespaceNormalizator(namespaceUris, namespacePrefixes, [

src/feeds/rdf/parse/config.ts

@@ -1,11 +1,12 @@
 import { XMLParser } from 'fast-xml-parser'
+import { Expression } from 'path-expression-matcher'
 import {
   namespacePrefixes,
   namespaceStopNodes,
   namespaceUris,
   parserConfig,
 } from '../../../common/config.js'
-import { createNamespaceNormalizator } from '../../../common/utils.js'
+import { createNamespaceNormalizator, expandStopNodes } from '../../../common/utils.js'
 
 // These elements can appear both inside <channel> and as direct children of
 // <rss> in malformed feeds, so stop nodes are generated for both paths.
@@ -35,7 +36,7 @@ const sharedStopNodes = [
 ]
 
 export const stopNodes = [
-  ...namespaceStopNodes,
+  ...expandStopNodes(namespaceStopNodes, ['rss.channel', 'rss.channel.item', 'rss', 'rss.item']),
   'rss.channel.title',
   'rss.channel.link',
   'rss.channel.description',
@@ -59,7 +60,7 @@ export const stopNodes = [
 
 export const parser = new XMLParser({
   ...parserConfig,
-  stopNodes,
+  stopNodes: stopNodes.map((node) => new Expression(node)),
 })
 
 export const normalizeNamespaces = createNamespaceNormalizator(namespaceUris, namespacePrefixes)

src/feeds/rss/parse/config.ts

@@ -1,20 +1,21 @@
 import { XMLParser } from 'fast-xml-parser'
+import { Expression } from 'path-expression-matcher'
 import { parserConfig } from '../../common/config.js'
 
 export const stopNodes = [
   'opml.head.title',
-  'opml.head.dateCreated',
-  'opml.head.dateModified',
-  'opml.head.ownerName',
-  'opml.head.ownerEmail',
-  'opml.head.ownerId',
+  'opml.head.datecreated',
+  'opml.head.datemodified',
+  'opml.head.ownername',
+  'opml.head.owneremail',
+  'opml.head.ownerid',
   'opml.head.docs',
-  'opml.head.expansionState',
-  'opml.head.vertScrollState',
-  'opml.head.windowTop',
-  'opml.head.windowLeft',
-  'opml.head.windowBottom',
-  'opml.head.windowRight',
+  'opml.head.expansionstate',
+  'opml.head.vertscrollstate',
+  'opml.head.windowtop',
+  'opml.head.windowleft',
+  'opml.head.windowbottom',
+  'opml.head.windowright',
   // Not a stop node because it supports recursive nesting that requires parser traversal.
   // '*.outline',
   // Not a stop node because *.X.Y wildcard patterns don't work in fast-xml-parser
@@ -24,5 +25,5 @@ export const stopNodes = [
 
 export const parser = new XMLParser({
   ...parserConfig,
-  stopNodes,
+  stopNodes: stopNodes.map((node) => new Expression(node)),
 })