From 6caf274300093b978efed56b95c75b4e6891e7bd Mon Sep 17 00:00:00 2001
From: Matt Baer <matt@write.as>
Date: Wed, 21 Feb 2024 14:20:22 -0500
Subject: [PATCH] Add text.ExtractImages func

---
 text/extract.go | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 text/extract.go

diff --git a/text/extract.go b/text/extract.go
new file mode 100644
index 0000000..3c2e0e0
--- /dev/null
+++ b/text/extract.go
@@ -0,0 +1,33 @@
+package text
+
+import (
+	"github.com/kylemcc/twitter-text-go/extract"
+	"net/url"
+	"regexp"
+)
+
+var imageURLRegex = regexp.MustCompile(`(?i)[^ ]+\.(gif|png|jpg|jpeg|image)$`)
+
+func ExtractImages(content string) []string {
+	matches := extract.ExtractUrls(content)
+	urls := map[string]bool{}
+	for i := range matches {
+		uRaw := matches[i].Text
+		// Parse the extracted text so we can examine the path
+		u, err := url.Parse(uRaw)
+		if err != nil {
+			continue
+		}
+		// Ensure the path looks like it leads to an image file
+		if !imageURLRegex.MatchString(u.Path) {
+			continue
+		}
+		urls[uRaw] = true
+	}
+
+	resURLs := make([]string, 0)
+	for k := range urls {
+		resURLs = append(resURLs, k)
+	}
+	return resURLs
+}