fix(author): a source's own fields no longer hide its content, for a chat with no skill text

design/author.md

@@ -15,7 +15,7 @@ already holds text to the content check's rules, from the pack an untrusted chec
 `apps/skills` plays every example untrusted too. Measured on a five-question quiz, at four
 characters a token: `content.json` was about 3,470 tokens to write by hand, after reading 3,900 to
 9,500 of a type's semantics; the source is 1,000 to 1,800 tokens, written from a type file of 200
-to 3,000. Eighteen things the code does not say by itself:
+to 3,000. Nineteen things the code does not say by itself:
 
 - **One generator, three ways to run it.** The third, `saved`, checks content as H5P stores it
   (*Saved content is checked as the build checks what it writes*, below). `ContentGenerator` was the sweep's
@@ -396,6 +396,47 @@ to 3,000. Eighteen things the code does not say by itself:
   Interactive Video, Image Hotspots, Image Choice, Game Map): of 12 first replies 2 were right at
   once with the sentence and 2 without, the Accordions. The skill gives the type files' address
   on GitHub too, for a chat that has its text and can open a link.
+- **A source's own fields do not hide its content, and its outer shape is taken as it is meant**
+  (2026-11-10, #174). Gemini in Chrome, given a link to the skill's folder, which shows a chat the
+  file names and none of the text, wrote a book its own way: `contentType` for `type`, and the
+  title, licence and authors in a `metadata` object, as H5P keeps them. The build checked a
+  source's own fields, then the type's name, then the content, and stopped at each: seven
+  problems of the outer shape, then thirty of the content, and a package at the third try.
+  Measured on Haiku in that position, the request as it was worded and no skill text, six content
+  types, two runs each, the build page's copy pasted back: with 0.7.0 none was right at once, 3
+  by the second round, 8 of 12 by the fourth, and four never. All twelve first sources had the
+  content's fields beside `type` with no `params`, nine named the type as it is spoken of, and six
+  of the nine second rounds were one problem, `the library pack has no content type
+  interactive-book`, with no name offered: the next guesses were `book` and `flashcards`.
+
+  So a build reads a source with the pack's names (`sourceOf`), and takes what can only be meant:
+  the content type under `contentType` or `library`, or named loosely as `type`, the same letters
+  and digits in any case with or without `H5P.` (`names.mjs`); the source's own fields in
+  `metadata`, and in no other object, since Question Set's `introPage` has a title too; the
+  content's fields beside `type`. Each is a problem, said once with what to write, and the content
+  is generated and checked in the same build wherever the type is known, with a stand-in for what
+  is missing. A name that is no type has the nearest offered: those its letters are part of, and
+  those they end with, not one they begin with, since the "image" of "image choice" is no lead.
+  A title is not tried, "Multiple Choice" or "Image Choice": the pack's index has none, and
+  reading them means fetching every part, 147 of them and some 15 MB, to name one type. A number
+  with no point is no version: "Question Set 2" is no Question Set. Sub-content that names its
+  library under `type` or `contentType` is taken the same way, its parameters in `params` or
+  beside the name, where one of the two names a library the field takes, at the major written
+  where the field takes it, and the loader loads that library, so what is written for it is
+  checked against its fields and not against none. No other key: in the 328 sources the tests
+  stored, every name found under another one was a text that reads like one, a card's answer
+  "table" six times, `main: "Question"` three times, `title` and `libraryTitle` "Text". A title beside the
+  parameters that is no field of the library is the sub-content's own, its `metadata.title`, in a
+  library field's place or in its group's, which had been two problems a chapter.
+
+  With it 10 of the 12 were right by the second round and all 12 by the third, and every one
+  plays. Of the 263 sources the earlier tests stored, written with the skill's text, 259 have the
+  same problems, the other four being books whose chapters had a title beside them, 13 problems
+  now 7, and none builds with one build and not the other; the 36 sources of the type
+  files and the quiz format build the same package. `readSource` keeps its contract, a source's
+  own problems thrown, without the pack's names. What the build page must do with it is its own:
+  hand the type's file back for a type that is named loosely, which its rule, the source's `type`
+  as written, does not.
 
 Not built: editing an existing package, which a tool beside it does with `checkContent` as its
 check; layout for Drag and Drop and Crossword; and any judgement of whether content is right, which

packages/author/README.md

@@ -97,6 +97,20 @@ h5p-author: The source has 3 problems:
   params.slides[0].elements[0].params.text: has <h1>, which this field does not take; it takes <a> <br> …
 ```
 
+A source's own fields are read the same way, for a chat that wrote one with no skill beside it.
+What its outer shape plainly means is taken and said once, and the content is checked in the same
+round: the content type under `contentType` or `library`, or named as it is spoken of
+(`"interactive-book"`, `"Question Set"`), the title, language, licence and authors in `metadata`,
+the content's fields beside `type`.
+
+```
+h5p-author: The source has 4 problems:
+  contentType: is not a field of a source: the content type is its type: write "type": "H5P.InteractiveBook"
+  metadata: is not a field of a source: title, license, authors are written beside type, not in an object: write "title": …, "license": …, "authors": … at the top
+  language: expected a BCP 47 language tag, such as "en" or "de-AT"
+  params.chapters[0]: is the params of sub-content, without the sub-content around them: write params.chapters[0] as { "library": "H5P.Column", "metadata": { "title": … }, "params": { "content": … } }
+```
+
 The quiz format the h5p-author skill published first (`{ "kind": "question-set", "questions": [ … ] }`)
 is still taken, for a Multiple Choice question or a Question Set of them.
 

packages/author/lib/browser.mjs

@@ -1,9 +1,10 @@
 import { BlobReader, BlobWriter, ZipWriter } from '@zip.js/zip.js'
 import { crc32 } from '@missing-elements/h5p-normalize/crc32'
 import { packLibrariesIn } from '@missing-elements/h5p-offline-player/plan'
-import { composePackage, readSource, writePackage } from './build.mjs'
+import { composePackage, sourceOf, typeNames, writePackage } from './build.mjs'
 import { MediaStore } from './media.mjs'
 import { Pack } from './pack.mjs'
+import { SourceError } from './source-error.mjs'
 
 export * from './core.mjs'
 
@@ -154,11 +155,13 @@ export async function buildPackage(input, options) {
   secure()
   const { libraries, files = {}, fetch: fetchImpl = pageFetch } = options ?? {}
   if (!libraries) throw new TypeError('Name the library pack: a pack openPack opened, the pack export of @missing-elements/h5p-libraries, or the address of a copy of its parts/ folder')
-  const source = readSource(input)
   const pack = libraries instanceof Pack ? libraries : await openPack(libraries)
   try {
+    // With the pack's names, as the command's build reads a source (#174).
+    const read = sourceOf(input, typeNames(pack))
+    if (!read.typed) throw new SourceError(read.problems)
     const media = new MediaStore(pageMedia(files), { fetch: fetchImpl })
-    const built = await composePackage(source, pack, media)
+    const built = await composePackage(read.source, pack, media, read.problems)
     const writer = new ZipWriter(new BlobWriter('application/zip'), { useWebWorkers: false })
     await writePackage(writer, built, pack, async () => {
       for (const item of media.files) {