Merge branch 'v1-webscraper' of https://github.com/mendableai/firecrawl into v1-webscraper

This commit is contained in:
Nicolas 2024-08-26 18:17:17 -03:00
commit 8c37ea6d96

View File

@ -27,7 +27,10 @@ const url = z.preprocess(
.string() .string()
.url() .url()
.regex(/^https?:\/\//, "URL uses unsupported protocol") .regex(/^https?:\/\//, "URL uses unsupported protocol")
.regex(/\.[a-z]{2,}$/i, "URL must have a valid top-level domain") .refine(
(x) => /\.[a-z]{2,}(\/|$)/i.test(x),
"URL must have a valid top-level domain or be a valid path"
)
.refine( .refine(
(x) => !isUrlBlocked(x), (x) => !isUrlBlocked(x),
"Firecrawl currently does not support social media scraping due to policy restrictions. We're actively working on building support for it." "Firecrawl currently does not support social media scraping due to policy restrictions. We're actively working on building support for it."