From 865dfbf1cf858f2b999ca3609d24c7309208838e Mon Sep 17 00:00:00 2001
From: al2f <al2f@noreply.codeberg.org>
Date: Wed, 24 Jan 2024 12:50:36 +1030
Subject: [PATCH] Use the rendered html readme from the json in a script tag

---
 pages/repo.go | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)

diff --git a/pages/repo.go b/pages/repo.go
index 6a361ce..dcef25a 100644
--- a/pages/repo.go
+++ b/pages/repo.go
@@ -5,7 +5,10 @@ import (
 	"net/http"
 	"os"
 	"strings"
-
+	"fmt"
+	
+	"github.com/tidwall/gjson"
+	
 	"codeberg.org/gothub/gothub/utils"
 	"github.com/gocolly/colly"
 	"github.com/gofiber/fiber/v2"
@@ -63,7 +66,6 @@ func HandleRepo(c *fiber.Ctx) error {
 			"error": "Repository " + c.Params("user") + "/" + repoUrl + branchExists + " not found",
 		})
 	}
-
 	// Scraping
 	Scrape := Repo{}
 
@@ -92,10 +94,22 @@ func HandleRepo(c *fiber.Ctx) error {
 	sc.OnHTML("div#readme", func(e *colly.HTMLElement) {
 		Scrape.Readme = e.ChildText("a[href='#readme']")
 	})
-	sc.OnHTML("div#readme div.Box-body", func(e *colly.HTMLElement) {
-		Content, _ := e.DOM.Html()
-		readmeOutput = strings.Replace(strings.Replace(strings.Replace(strings.Replace(strings.Replace(string(utils.UGCPolicy().SanitizeBytes([]byte(Content))), "https://github.com", "", -1), "user-content-", "", -1), "https://camo.githubusercontent.com", "/camo", -1), "https://raw.githubusercontent.com", "/raw", -1), "https://user-images.githubusercontent.com", "/user-images", -1)
+	
+	// jq '.props.initialPayload.overview.overviewFiles[]|select(.tabName == "README").richText'
+	sc.OnHTML("div.Layout-main react-partial script", func(e *colly.HTMLElement) {
+		jsonMess := e.Text
+		
+		result := gjson.Get(jsonMess, "props.initialPayload.overview.overviewFiles")
+		result.ForEach(func(i, e gjson.Result) bool {
+			if e.Get("tabName").Str == "README" {
+				unsanitizedOutput := e.Get("richText").Str
+				readmeOutput = strings.Replace(strings.Replace(strings.Replace(strings.Replace(strings.Replace(string(utils.UGCPolicy().SanitizeBytes([]byte(unsanitizedOutput))), "https://github.com", "", -1), "user-content-", "", -1), "https://camo.githubusercontent.com", "/camo", -1), "https://raw.githubusercontent.com", "/raw", -1), "https://user-images.githubusercontent.com", "/user-images", -1)
+				return false
+			}
+			return true // keep iterating
+		})
 	})
+
 	sc.OnHTML("div.BorderGrid-cell ul.list-style-none", func(e *colly.HTMLElement) {
 		e.ForEach("li.d-inline .d-inline-flex", func(i int, el *colly.HTMLElement) {
 			var lang Languages
@@ -129,6 +143,7 @@ func HandleRepo(c *fiber.Ctx) error {
 		})
 	})
 	sc.Visit("https://github.com/" + c.Params("user") + "/" + repoUrl + branchExists)
+	
 	// Add scrape-based info to repoArray
 	repoArray = append(repoArray, Scrape)
 	return c.Render("repo", fiber.Map{