ManYue OPDS-PSE templates: a three-page count maps to zero through two

An OPDS cover does not verify a page template’s first and last pages. For an owned three-page sample explicitly offering PSE, check count, zero-based index and actual images. This extends connection and link-role tasks without claiming page streaming for all OPDS or EPUB entries, or requiring an in-app server-template editor.

Confirm that the entry actually offers page streaming

Record library alias, entry identifier, edition, time and account role. Inspect permitted feed evidence for a PSE stream link and count, not just a cover or archive filename. Ordinary OPDS acquisition can describe a CBZ/PDF archive without making that acquisition directly readable/downloadable through this comic integration. Follow published scope and actual controls. Without PSE evidence, retain directory observations rather than inventing image URLs against public servers. Use administrator-supplied or existing-tool evidence, redact domains/credentials when sharing, and do not claim access to unseen internal app requests.

Retain the template and total count exactly

{pageNumber} is a template variable; count denotes total pages, not the value to substitute as the last page. The example has count three with first/middle/last marks; a separate cover link does not automatically add a fourth page. Retain stream relation, image type, count and template, marking missing fields unprovided rather than inferring count from a chapter title. For other variables, consult the specification and server convention instead of deleting braces to obtain a plausible URL. This task uses an explicitly zero-based simple template, not a guarantee that every extension field is consumed by the client.

Separate visible ordinal positions from protocol indices

For a zero-based three-page stream, protocol indices are zero, one, two while reader ordinals are first, second, third. Map first to zero, middle to one and last to two: final index is count minus one, not three. This difference is not a missing cover, and index zero must not automatically be excluded. Keep protocol numbering separate from app display. A displayed first-page 1 does not prove the request used index one. Fill request fields only from actual accessible evidence; otherwise leave them unknown.

Check legal first, middle and last without guessing out-of-range behaviour

Use normal available controls to read first, middle, last in the permitted sample. If the administrator supplies corresponding request logs, check substitutions zero, one, two. Index three is outside this declared sample; repeated boundary requests are unnecessary, and no particular error code is promised. A placeholder or repeated ending from an implementation would not justify changing count to four merely because an image appeared. Separate declared range, observed legal content and untested out-of-range cases. A specification is not a server test report, and requests not performed must not be recorded as verified.

Distinguish cover artwork from the first streamed page

Catalog artwork may match the first streamed page or be a separate thumbnail; similar appearance does not merge different link roles. Retain sources for artwork and index-zero image, checking dimensions, marks and neighbours. A loaded cover with a failed first page calls for template/access checks, not a claim of complete downloading. A readable first page likewise does not certify the cover link. Keep archive/sample links separate, never constructing a stream by appending numbers to the artwork URL or validating a full first page with a thumbnail.

A count control cannot create missing content

If an administrator makes an owned test-feed control changing count from three to four while retaining three images, declaration may no longer match available content. Do not duplicate the last image as a fourth to make completion appear successful. No production edit is required; a paper mapping shows that a larger declaration creates no image bytes. Count two can exclude the original ending from the declared range, requiring edition-specific records rather than being called an original repair. State separate changes to count, template and files; switching works simultaneously would not be a single-variable control.

For one-page shifts, check indexing conventions first

For a middle image at the start or a repeated ending, check count, start convention, template and entry identity before investigating an old edition or another server object. Do not equate displayed page one with request index one or erase evidence through bulk cache deletion first. Provide a redacted template, count and mark mapping when seeking help, excluding private authentication. Without request visibility, do not invent the URL the app supposedly sent. Keep legal-page observations and insufficient index evidence separate; isolating server versus client behaviour requires more than one wrong image.

Exit and return to complete count-to-image mappings

Exit the reader and return to the same entry, revisiting first, middle, last and artwork. Check that the last confirmed image still corresponds to the third reader position. Save six fields: entry identity, stream role, count, index convention, actual images and evidence scope, leaving unseen requests unknown. Completion means traceable three-page/zero-through-two mapping, not identical out-of-range responses everywhere or support for every OPDS extension. Recheck endings after count changes; a past sample does not certify an entire library indefinitely.

Related guides and sources

ManYue OPDS connections: why a catalogue may not support page streaming; ManYue: Distinguish OPDS artwork, sample and acquisition links.

Checked on November 11, 2026: Apple public information lists ManYue 2.0.5, requiring iOS / iPadOS 15.0 or later. See the official App Store for published capabilities; verify specific controls in the installed version.

Method reference: Anansi OPDS-PSE1.2: Page Templates and Counts. Samples, passages and controls are original exercises. Sources establish definitions and checking methods rather than new automatic analysis in the app. Manual records may be kept externally, and unknown fields should remain explicit. This article does not claim step-by-step testing on a physical device, private server or every output environment.