From bed96aee397b036ccbd0b8686540d0de6ecb5d4c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Lu=C3=ADs=20Simas?= <luissimas@protonmail.com>
Date: Wed, 26 Jun 2024 06:55:53 -0300
Subject: [PATCH] feat: exclude http links from metrics

---
 internal/collector/note.go      | 11 ++++++++++-
 internal/collector/note_test.go | 10 +++++++++-
 2 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/internal/collector/note.go b/internal/collector/note.go
index 722c959..c83eb2a 100644
--- a/internal/collector/note.go
+++ b/internal/collector/note.go
@@ -2,6 +2,7 @@ package collector
 
 import (
 	"log/slog"
+	"net/url"
 	"slices"
 
 	"github.com/luissimas/zettelkasten-exporter/internal/metrics"
@@ -46,7 +47,10 @@ func collectLinks(content []byte) map[string]uint {
 				return ast.WalkContinue, nil
 			}
 
-			// TODO: check if target is not a http link
+			if isUrl(target) {
+				return ast.WalkContinue, nil
+			}
+
 			v, ok := links[target]
 			if !ok {
 				links[target] = 0
@@ -61,3 +65,8 @@ func collectLinks(content []byte) map[string]uint {
 	slog.Debug("Collected links", slog.Any("links", links))
 	return links
 }
+
+func isUrl(s string) bool {
+	u, err := url.Parse(s)
+	return err == nil && u.Scheme != "" && u.Host != ""
+}
diff --git a/internal/collector/note_test.go b/internal/collector/note_test.go
index 83f4834..abc6821 100644
--- a/internal/collector/note_test.go
+++ b/internal/collector/note_test.go
@@ -55,12 +55,20 @@ func TestCollectNoteMetrics(t *testing.T) {
 		},
 		{
 			name:    "ignore embeddedlinks",
-			content: "![[target.png]]\n!()[another.jpeg]\n[[link]]",
+			content: "![[target.png]]\n![](another.jpeg)\n[[link]]",
 			expected: metrics.NoteMetrics{
 				Links:     map[string]uint{"link": 1},
 				LinkCount: 1,
 			},
 		},
+		{
+			name:    "ignore http links",
+			content: "[[one]][this is an http link](https://go.dev/)[[not/an/http/link]]",
+			expected: metrics.NoteMetrics{
+				Links:     map[string]uint{"one": 1, "not/an/http/link": 1},
+				LinkCount: 1,
+			},
+		},
 	}
 
 	for _, d := range data {