How to use the HTML text input recipe import? #5494
-
|
Is the required format / syntax for the HTML text input option documented somewhere? Does this require the OpenAI feature to be configured? I've been looking but haven't found any guidance. I tried writing a very simple recipe in HTML and pasting it in just using a super simple format, but when I tap "Create Recipe" nothing happens and I don't get any kind of error message (currently trying on a mobile browser). |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
The HTML import option is identical to the regular scraper, except instead of fetching the HTML with a network request we already have it. If you don't have OpenAI integration enabled, then the HTML must have JSON-LD data in valid schema.org recipe format. If you're not copy-pasting existing HTML from a real website, you'll have an easier time just using the JSON import instead of the HTML import (which has the same restrictions, but you don't have to worry about fitting the JSON-LD into HTML). Alternatively, if OpenAI is enabled, we first attempt to parse it normally (using the schema.org definition), then just throw the whole thing into an OpenAI request and ask it to format it into a particular JSON format. TL;DR only use HTML for directly importing real websites. Use the JSON import for JSON that adheres to the schema.org recipe spec, or enable OpenAI and use any format you want (either HTML or JSON, whichever is easier). |
Beta Was this translation helpful? Give feedback.
-
|
Awesome, thanks, that clears things up. I just got OpenAI working and used an image successfully, I will see what it can do with text versions now 🙂 |
Beta Was this translation helpful? Give feedback.
The HTML import option is identical to the regular scraper, except instead of fetching the HTML with a network request we already have it. If you don't have OpenAI integration enabled, then the HTML must have JSON-LD data in valid schema.org recipe format. If you're not copy-pasting existing HTML from a real website, you'll have an easier time just using the JSON import instead of the HTML import (which has the same restrictions, but you don't have to worry about fitting the JSON-LD into HTML).
Alternatively, if OpenAI is enabled, we first attempt to parse it normally (using the schema.org definition), then just throw the whole thing into an OpenAI request and ask it to format it into a p…