first iteration of parsing metadata and chapter names from ncc.html file

2026-03-01 13:39:41 +00:00 · 2026-02-07 16:45:40 +01:00 · 2026-02-07 16:45:40 +01:00 · 6c9bf8c2bd
commit 6c9bf8c2bd
parent fe13456a2b
10 changed files with 394 additions and 6 deletions
--- a/test/server/utils/parsers/parseDaisyMetadata.test.js
+++ b/test/server/utils/parsers/parseDaisyMetadata.test.js
@ -0,0 +1,73 @@
+const chai = require('chai')
+const expect = chai.expect
+const { parseDaisyMetadata } = require('../../../../server/utils/parsers/parseDaisyMetadata')
+
+describe('parseDaisyMetadata', () => {
+  it('returns null if htmlText is empty', () => {
+    const result = parseDaisyMetadata('')
+    expect(result).to.be.null
+  })
+
+  it('parses common metadata values from DAISY ncc.html', () => {
+    const nccHtml = `
+      <html>
+        <head>
+          <title>Fallback Title</title>
+          <meta name="dc:title" content="The DAISY Book">
+          <meta name="dc:creator" content="Jane Doe & Richard Roe">
+          <meta name="ncc:narrator" content="Reader One; Reader Two">
+          <meta name="dc:publisher" content="Talking Books Inc">
+          <meta name="dc:date" content="2021-06-04">
+          <meta name="dc:language" content="en">
+          <meta name="dc:subject" content="Fiction, Mystery">
+          <meta name="ncc:keywords" content="audio; daisy">
+          <meta name="dc:identifier" content="ISBN 978-1-4028-9462-6">
+          <meta name="dc:identifier:asin" content="ASIN: B012345678">
+        </head>
+      </html>
+    `
+
+    const result = parseDaisyMetadata(nccHtml)
+    expect(result.title).to.equal('The DAISY Book')
+    expect(result.authors).to.deep.equal(['Jane Doe', 'Richard Roe'])
+    expect(result.narrators).to.deep.equal(['Reader One', 'Reader Two'])
+    expect(result.publisher).to.equal('Talking Books Inc')
+    expect(result.publishedYear).to.equal('2021')
+    expect(result.language).to.equal('en')
+    expect(result.genres).to.deep.equal(['Fiction', 'Mystery'])
+    expect(result.tags).to.deep.equal(['audio', 'daisy'])
+    expect(result.isbn).to.equal('978-1-4028-9462-6')
+    expect(result.asin).to.equal('B012345678')
+  })
+
+  it('falls back to title tag when dc:title is not set', () => {
+    const nccHtml = `
+      <html>
+        <head>
+          <title>Title From Head</title>
+        </head>
+      </html>
+    `
+    const result = parseDaisyMetadata(nccHtml)
+    expect(result.title).to.equal('Title From Head')
+  })
+
+  it('parses chapter names from heading entries in ncc.html', () => {
+    const nccHtml = `
+      <html>
+        <body>
+          <h1><a href="book.smil#id1">Chapter 1</a></h1>
+          <h2><a href="book.smil#id2">Chapter 2: The Road</a></h2>
+          <h3>Part 1</h3>
+        </body>
+      </html>
+    `
+
+    const result = parseDaisyMetadata(nccHtml)
+    expect(result.chapters).to.deep.equal([
+      { title: 'Chapter 1' },
+      { title: 'Chapter 2: The Road' },
+      { title: 'Part 1' }
+    ])
+  })
+})
--- a/test/server/utils/scandir.test.js
+++ b/test/server/utils/scandir.test.js
@ -49,4 +49,22 @@ describe('scanUtils', async () => {
      'Author/Series2/Book5/deeply/nested': ['cd 01/audiofile.mp3', 'cd 02/audiofile.mp3']
    })
  })
+
+  it('should include DAISY ncc.html changes when includeNonMediaFiles is enabled', async () => {
+    const filePath = 'Author/Book3/ncc.html'
+    const dirname = Path.dirname(filePath)
+    const fileItems = [
+      {
+        name: Path.basename(filePath),
+        reldirpath: dirname === '.' ? '' : dirname,
+        extension: Path.extname(filePath),
+        deep: filePath.split('/').length - 1
+      }
+    ]
+
+    const libraryItemGrouping = scanUtils.groupFileItemsIntoLibraryItemDirs('book', fileItems, false, true)
+    expect(libraryItemGrouping).to.deep.equal({
+      'Author/Book3': ['ncc.html']
+    })
+  })
 })