From 140a38fd39213c76d0d803de3635a3732ad94a67 Mon Sep 17 00:00:00 2001
From: Kevin Pham <kevin@deoxy.net>
Date: Thu, 30 Nov 2023 22:47:15 -0600
Subject: [PATCH] improve /api/content metadata and tag handling

---
 cmd/main.go                                   |  2 +-
 handlers/api_content.go                       | 44 +++++++++++++++++++
 .../responsemodifers/api/outline_api.go       | 41 ++++++++++++++++-
 3 files changed, 85 insertions(+), 2 deletions(-)
 create mode 100644 handlers/api_content.go

diff --git a/cmd/main.go b/cmd/main.go
index a7840f0..e6aaf1f 100644
--- a/cmd/main.go
+++ b/cmd/main.go
@@ -183,7 +183,7 @@ func main() {
 		RulesetPath: *ruleset,
 	}
 
-	app.Get("api/content/*", handlers.NewAPIOutlineHandler("api/outline/*", proxyOpts))
+	app.Get("api/content/*", handlers.NewAPIContentHandler("api/outline/*", proxyOpts))
 
 	app.Get("outline/*", handlers.NewOutlineHandler("outline/*", proxyOpts))
 
diff --git a/handlers/api_content.go b/handlers/api_content.go
new file mode 100644
index 0000000..71608e5
--- /dev/null
+++ b/handlers/api_content.go
@@ -0,0 +1,44 @@
+package handlers
+
+import (
+	"ladder/proxychain"
+	rx "ladder/proxychain/requestmodifers"
+	tx "ladder/proxychain/responsemodifers"
+
+	"github.com/gofiber/fiber/v2"
+)
+
+func NewAPIContentHandler(path string, opts *ProxyOptions) fiber.Handler {
+	// TODO: implement ruleset logic
+	/*
+		var rs ruleset.RuleSet
+		if opts.RulesetPath != "" {
+			r, err := ruleset.NewRuleset(opts.RulesetPath)
+			if err != nil {
+				panic(err)
+			}
+			rs = r
+		}
+	*/
+
+	return func(c *fiber.Ctx) error {
+		proxychain := proxychain.
+			NewProxyChain().
+			WithAPIPath(path).
+			SetDebugLogging(opts.Verbose).
+			SetRequestModifications(
+				rx.MasqueradeAsGoogleBot(),
+				rx.ForwardRequestHeaders(),
+				rx.SpoofReferrerFromGoogleSearch(),
+			).
+			AddResponseModifications(
+				tx.DeleteIncomingCookies(),
+				tx.RewriteHTMLResourceURLs(),
+				tx.APIContent(),
+			).
+			SetFiberCtx(c).
+			Execute()
+
+		return proxychain
+	}
+}
diff --git a/proxychain/responsemodifers/api/outline_api.go b/proxychain/responsemodifers/api/outline_api.go
index d96b05f..f17f42e 100644
--- a/proxychain/responsemodifers/api/outline_api.go
+++ b/proxychain/responsemodifers/api/outline_api.go
@@ -27,6 +27,15 @@ type TextContent struct {
 	Data string `json:"data"`
 }
 
+type ListContent struct {
+	Type      string            `json:"type"`
+	ListItems []ListItemContent `json:"listItems"`
+}
+
+type ListItemContent struct {
+	Data string `json:"data"`
+}
+
 type JSONDocument struct {
 	Success  bool         `json:"success"`
 	Error    ErrorDetails `json:"error"`
@@ -35,6 +44,7 @@ type JSONDocument struct {
 		Author      string   `json:"author"`
 		URL         string   `json:"url"`
 		Hostname    string   `json:"hostname"`
+		Image       string   `json:"image"`
 		Description string   `json:"description"`
 		Sitename    string   `json:"sitename"`
 		Date        string   `json:"date"`
@@ -58,11 +68,13 @@ func ExtractResultToAPIResponse(extract *trafilatura.ExtractResult) *JSONDocumen
 	jsonDoc.Metadata.URL = extract.Metadata.URL
 	jsonDoc.Metadata.Hostname = extract.Metadata.Hostname
 	jsonDoc.Metadata.Description = extract.Metadata.Description
+	jsonDoc.Metadata.Image = extract.Metadata.Image
 	jsonDoc.Metadata.Sitename = extract.Metadata.Sitename
 	jsonDoc.Metadata.Date = extract.Metadata.Date.Format("2006-01-02")
 	jsonDoc.Metadata.Categories = extract.Metadata.Categories
 	jsonDoc.Metadata.Tags = extract.Metadata.Tags
 	jsonDoc.Metadata.License = extract.Metadata.License
+	jsonDoc.Metadata.Hostname = extract.Metadata.Hostname
 
 	// Populate content
 	if extract.ContentNode != nil {
@@ -120,7 +132,34 @@ func parseContent(node *html.Node) []interface{} {
 			}
 			content = append(content, text)
 
-		// continue with other tags
+		case "h4":
+			text := TextContent{
+				Type: "h4",
+				Data: dom.InnerText(child),
+			}
+			content = append(content, text)
+
+		case "h5":
+			text := TextContent{
+				Type: "h5",
+				Data: dom.InnerText(child),
+			}
+			content = append(content, text)
+
+		case "ul", "ol":
+			list := ListContent{
+				Type:      child.Data,
+				ListItems: []ListItemContent{},
+			}
+			for listItem := child.FirstChild; listItem != nil; listItem = listItem.NextSibling {
+				if listItem.Data == "li" {
+					listItemContent := ListItemContent{
+						Data: dom.InnerText(listItem),
+					}
+					list.ListItems = append(list.ListItems, listItemContent)
+				}
+			}
+			content = append(content, list)
 
 		default:
 			text := TextContent{