More fixes
This commit is contained in:
@@ -113,9 +113,10 @@ export async function POST(request: Request) {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// Normalize content for deduplication (strip HTML entities, URLs, whitespace)
|
// Normalize content for deduplication (strip HTML entities, URLs, whitespace, category suffixes)
|
||||||
const normalizeForDedup = (content: string): string => {
|
const normalizeForDedup = (content: string): string => {
|
||||||
return content
|
return content
|
||||||
|
.replace(/Posted into [\w\s-]+/gi, '') // Remove "Posted into [Category]" patterns
|
||||||
.replace(/&[a-z]+;/gi, '') // Remove HTML entities like ‘
|
.replace(/&[a-z]+;/gi, '') // Remove HTML entities like ‘
|
||||||
.replace(/&#\d+;/g, '') // Remove numeric entities
|
.replace(/&#\d+;/g, '') // Remove numeric entities
|
||||||
.replace(/https?:\/\/[^\s]+/gi, '') // Remove URLs
|
.replace(/https?:\/\/[^\s]+/gi, '') // Remove URLs
|
||||||
@@ -123,7 +124,7 @@ const normalizeForDedup = (content: string): string => {
|
|||||||
.replace(/\s+/g, ' ') // Normalize whitespace
|
.replace(/\s+/g, ' ') // Normalize whitespace
|
||||||
.toLowerCase()
|
.toLowerCase()
|
||||||
.trim()
|
.trim()
|
||||||
.slice(0, 100); // Compare first 100 chars
|
.slice(0, 50); // Compare first 50 chars (article title)
|
||||||
};
|
};
|
||||||
|
|
||||||
// Helper to transform cached remote posts to match local post format
|
// Helper to transform cached remote posts to match local post format
|
||||||
|
|||||||
@@ -71,9 +71,10 @@ const stripFirstUrl = (text: string, url: string) => {
|
|||||||
return `${before} ${after}`.trim();
|
return `${before} ${after}`.trim();
|
||||||
};
|
};
|
||||||
|
|
||||||
// Normalize content for deduplication (strip HTML entities, URLs, whitespace)
|
// Normalize content for deduplication (strip HTML entities, URLs, whitespace, category suffixes)
|
||||||
const normalizeForDedup = (content: string): string => {
|
const normalizeForDedup = (content: string): string => {
|
||||||
return content
|
return content
|
||||||
|
.replace(/Posted into [\w\s-]+/gi, '') // Remove "Posted into [Category]" patterns
|
||||||
.replace(/&[a-z]+;/gi, '') // Remove HTML entities like ‘
|
.replace(/&[a-z]+;/gi, '') // Remove HTML entities like ‘
|
||||||
.replace(/&#\d+;/g, '') // Remove numeric entities
|
.replace(/&#\d+;/g, '') // Remove numeric entities
|
||||||
.replace(/https?:\/\/[^\s]+/gi, '') // Remove URLs
|
.replace(/https?:\/\/[^\s]+/gi, '') // Remove URLs
|
||||||
@@ -81,7 +82,7 @@ const normalizeForDedup = (content: string): string => {
|
|||||||
.replace(/\s+/g, ' ') // Normalize whitespace
|
.replace(/\s+/g, ' ') // Normalize whitespace
|
||||||
.toLowerCase()
|
.toLowerCase()
|
||||||
.trim()
|
.trim()
|
||||||
.slice(0, 100); // Compare first 100 chars
|
.slice(0, 50); // Compare first 50 chars (article title)
|
||||||
};
|
};
|
||||||
|
|
||||||
const parseRemoteHandle = (handle: string) => {
|
const parseRemoteHandle = (handle: string) => {
|
||||||
|
|||||||
Reference in New Issue
Block a user