feat: parse link targets

Makefile

@@ -21,4 +21,4 @@ build:
 	go build -o bin/$(BINARY_NAME) ./cmd/zettelkasten-exporter/main.go
 
 run: build
-	ZETTELKASTEN_DIRECTORY=~/Documents/vault ./bin/$(BINARY_NAME) 
+	ZETTELKASTEN_DIRECTORY=./sample ./bin/$(BINARY_NAME)

README.md

@@ -11,10 +11,14 @@
 - [X] Read config
 - [X] Find all files recursivelly
 - [X] Parse markdown links
-- [ ] Exclude links to non existing files
-- [ ] Collect backlinks
 - [ ] Configurable ignore file patterns
 - [ ] Get zettelkasten from git url
 - [ ] Support private repositories (Maybe with Github's PAT?)
+- [ ] Exclude links to non existing files
+- [ ] Collect backlinks
+- [ ] Github actions CI
+- [ ] Document usage in README
+- [ ] Build image and push to OCI registry
+- [ ] Deploy on K8s
 
 https://prometheus.io/docs/instrumenting/writing_exporters/

internal/collector/collector.go

@@ -51,7 +51,7 @@ func (c *Collector) CollectMetrics() error {
 
 	metrics.TotalNoteCount.Set(float64(collected.NoteCount))
 	for name, metric := range collected.Notes {
-		metrics.LinkCount.WithLabelValues(name).Set(float64(metric.LinkCount))
+		metrics.LinkCount.WithLabelValues(name).Set(float64(len(metric.Links)))
 	}
 
 	elapsed := time.Since(started)
@@ -87,7 +87,7 @@ func (c *Collector) collectMetrics() (Metrics, error) {
 		}
 		metrics := CollectNoteMetrics(content)
 		notes[path] = metrics
-		linkCount += metrics.LinkCount
+		linkCount += len(metrics.Links)
 		noteCount += 1
 
 		slog.Info("collected metrics from file", slog.String("path", path), slog.Any("d", d), slog.Any("err", err))

internal/collector/note.go

@@ -1,6 +1,7 @@
 package collector
 
 import (
+	"log/slog"
 	"slices"
 
 	"github.com/yuin/goldmark"
@@ -16,23 +17,36 @@ var md = goldmark.New(
 )
 
 type NoteMetrics struct {
-	LinkCount int
+	Links map[string]int
 }
 
 func CollectNoteMetrics(content []byte) NoteMetrics {
-	return NoteMetrics{LinkCount: collectLinkCount(content)}
+	return NoteMetrics{Links: collectLinks(content)}
 }
 
-func collectLinkCount(content []byte) int {
+func collectLinks(content []byte) map[string]int {
 	linkKinds := []ast.NodeKind{ast.KindLink, wikilink.Kind}
 	reader := text.NewReader(content)
 	root := md.Parser().Parse(reader)
-	linkCount := 0
+	links := make(map[string]int)
 	ast.Walk(root, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
 		if entering && slices.Contains(linkKinds, n.Kind()) {
-			linkCount += 1
+			var target string
+			switch v := n.(type) {
+			case *ast.Link:
+				target = string(v.Destination)
+			case *wikilink.Node:
+				target = string(v.Target)
+			}
+
+			v, ok := links[target]
+			if !ok {
+				links[target] = 0
+			}
+			links[target] = v + 1
 		}
 		return ast.WalkContinue, nil
 	})
-	return linkCount
+	slog.Info("Collected links", slog.Any("links", links))
+	return links
 }

internal/collector/note_test.go

@@ -1,25 +1,47 @@
 package collector
 
-import "testing"
+import (
+	"maps"
+	"testing"
+)
 
 func TestCollectNoteMetrics(t *testing.T) {
 	data := []struct {
 		name     string
 		content  string
 		expected NoteMetrics
 	}{
-		{"empty file", "", NoteMetrics{LinkCount: 0}},
-		{"file with a single wiki link", "[[Link]]", NoteMetrics{LinkCount: 1}},
-		{"file with multiple wiki links", "[[Link]]aksdjf[[anotherlink]]\n[[link]]", NoteMetrics{LinkCount: 3}},
-		{"wikilink dividers", "[[something|another]]\n\n[[link]]\n[[382dlk djfs link|yeah]]", NoteMetrics{LinkCount: 3}},
-		{"file with markdown link", "[Link](target.md)", NoteMetrics{LinkCount: 1}},
-		{"file with multiple links", "okok[Link](target.md)\n**ddk**[[linked]]`test`[[anothe|link]]\n\n[test](yet-another.md)", NoteMetrics{LinkCount: 4}},
+		{
+			name:     "empty file",
+			content:  "",
+			expected: NoteMetrics{Links: map[string]int{}},
+		},
+		{
+			name:     "wiki links",
+			content:  "[[Link]]aksdjf[[something|another]]\n[[link]]",
+			expected: NoteMetrics{Links: map[string]int{"Link": 1, "something": 1, "link": 1}},
+		},
+		{
+			name:     "markdown link",
+			content:  "[Link](target.md)",
+			expected: NoteMetrics{Links: map[string]int{"target.md": 1}},
+		},
+		{
+			name:     "mixed links",
+			content:  "okok[Link](target.md)\n**ddk**[[linked]]`test`[[another|link]]\n\n[test](yet-another.md)",
+			expected: NoteMetrics{Links: map[string]int{"target.md": 1, "linked": 1, "another": 1, "yet-another.md": 1}},
+		},
+		{
+			name:     "repeated links",
+			content:  "[[target.md|link]]\n[link](target.md)\n[[link]]",
+			expected: NoteMetrics{Links: map[string]int{"target.md": 2, "link": 1}},
+		},
 	}
 
 	for _, d := range data {
 		t.Run(d.name, func(t *testing.T) {
 			result := CollectNoteMetrics([]byte(d.content))
-			if result != d.expected {
+			if !maps.Equal(result.Links, d.expected.Links) {
 				t.Errorf("Expected %v, got %v", d.expected, result)
 			}
 		})