Examples: use the converters the API spec documents - #125

Three examples call converters that the production API spec (https://v2.convertapi.com/info/openapi, 332 paths on 2026-11-09) does not list. They still convert today, but developers copy these examples and the how-to guides on convertapi.com reuse them, so they should call documented converters. This is the Node part of what ConvertAPI/convertapi-library-dotnet#86 does for .NET.

https://www.convertapi.com/web-to-pdf is not in the site's sitemap; https://www.convertapi.com/html-to-pdf is. The other examples already call documented converters. CI runs the examples against staging, which served the same spec as production on 2026-11-09 (4,729,541 bytes).

Verified against production (2026-11-09)

Node 24.20.0, npm ci && npm run build, then node examples/<name>.js from the repo root, once on main (a402bd7) and once on this branch, with the token in API_TOKEN. The [trace] lines come from a --require preload that prints each request's path; they are not part of the examples' output. <tmp> is os.tmpdir().

conversion_chaining.js on this branch:

Converting PDF to JPG and compressing result files with ZIP

[trace] POST v2.convertapi.com/upload -> HTTP 200
[trace] POST v2.convertapi.com/convert/pdf/to/jpg -> HTTP 200
[trace] POST v2.convertapi.com/convert/file/to/zip -> HTTP 200
Files saved to
<tmp>/test.zip

On main the output is the same except the third request, POST v2.convertapi.com/convert/any/to/zip. Both runs saved a 775,249-byte test.zip with test.jpg to test-6.jpg; the JPGs are byte-identical before and after.

convert_url_to_pdf.js on this branch (on main the request was /convert/web/to/pdf, the rest the same):

[trace] POST v2.convertapi.com/convert/html/to/pdf -> HTTP 200
The web page PDF saved to
<tmp>/web-example.pdf

get_result_file_stream.js on this branch (on main: /convert/web/to/pdf, 91,412 bytes):

[trace] POST v2.convertapi.com/convert/html/to/pdf -> HTTP 200
Result url: https://v2.convertapi.com/d/<file-id>/web-example.pdf
Downloaded data size: 91424

The PDF text: on main and in both runs on this branch, Wikimedia answered ConvertAPI's renderer with its "Too many requests" page (1 page, 62 words), the same page the old web -> pdf got. This does not depend on the converter: in the same hour the same html -> pdf call rendered the article (3 pages, 1,925 words) from the Go, Python, PHP and Ruby examples. The demo URL is being looked at separately.

Seen, not changed: the library itself still picks undocumented converters when no source format is given: any for every zip target and url when the parameters have a Url (src/utils.js, detectFormat). The examples now pass the source format, so they do not depend on it.

CI

The test (Node 20/22/24) jobs ran every example against staging and passed on the first attempt; there get_result_file_stream.js downloaded 286,113 bytes, the full article. The build job fails at "Audit the runtime dependencies": npm audit reports follow-redirects 1.16.0 - 1.16.1 (GHSA-mrwp-f3q8-3g9p), the same failure main has had since 2026-11-07 (https://github.com/ConvertAPI/convertapi-library-node/actions/runs/42359841164). It is not caused by this change.

Same change in the other SDKs

• Go: Examples: use the converters the API spec documents convertapi-library-go#22
• Java: Examples: use the converters the API spec documents convertapi-library-java#58
• PHP: Examples: use the converters the API spec documents convertapi-library-php#73
• Python: Examples: use the converters the API spec documents convertapi-library-python#62
• Ruby: Examples: use the converters the API spec documents convertapi-library-ruby#43
• .NET: Examples: use documented converters, read the production catalog, save SplitPDF to temp, hide account secrets convertapi-library-dotnet#86
• JS (browser): no change needed; its README calls only documented converters.

🤖 Generated with Claude Code