feat: collect basic statistics

README.md

@@ -7,8 +7,10 @@
 - [X] Parse obsidian wiki links
 - [X] Register total links
 - [X] Register links per note
+- [ ] Read config
 - [ ] Expose prometheus metrics endpoint
 - [ ] Handle uncreated links
+- [ ] Collect backlinks
 - [ ] Configurable ignore patterns
 - [ ] Parse markdown links
 - [ ] Get zettelkasten from git url

cmd/zettelkasten-exporter/main.go

@@ -0,0 +1,36 @@
+package main
+
+import (
+	"fmt"
+	"log/slog"
+	"os"
+	"path/filepath"
+
+	"github.com/luissimas/zettelkasten-exporter/internal/collector"
+)
+
+func main() {
+	if len(os.Args) != 2 {
+		slog.Error("Wrong arguments")
+		os.Exit(1)
+	}
+
+	dir := os.Args[1]
+	absolute_path, err := filepath.Abs(dir)
+	if err != nil {
+		slog.Error("Error getting absolute file path", slog.Any("error", err))
+		os.Exit(1)
+	}
+
+	metrics, err := collector.CollectMetrics(collector.CollectorConfig{Path: absolute_path})
+	if err != nil {
+		slog.Error("Error collecting note metrics", slog.Any("error", err))
+		os.Exit(1)
+	}
+
+	fmt.Printf("There are %d markdown files in %s\n", metrics.NoteCount, absolute_path)
+	fmt.Printf("There is a total of %d links in%s\n", metrics.LinkCount, absolute_path)
+	for name, metrics := range metrics.Notes {
+		fmt.Printf("\t%s: %v\n", name, metrics)
+	}
+}

internal/collector/collector.go

@@ -0,0 +1,43 @@
+package collector
+
+import (
+	"log/slog"
+	"os"
+	"path/filepath"
+)
+
+type Metrics struct {
+	NoteCount int
+	LinkCount int
+	Notes     map[string]NoteMetrics
+}
+
+type CollectorConfig struct {
+	Path string
+}
+
+func CollectMetrics(config CollectorConfig) (Metrics, error) {
+	pattern := filepath.Join(config.Path, "**/*.md")
+	files, err := filepath.Glob(pattern)
+	if err != nil {
+		slog.Error("Error getting files", slog.Any("error", err))
+		return Metrics{}, err
+	}
+
+	noteCount := len(files)
+	linkCount := 0
+	notes := make(map[string]NoteMetrics)
+
+	for _, file := range files {
+		content, err := os.ReadFile(file)
+		if err != nil {
+			slog.Error("Error reading file", slog.Any("error", err))
+			continue
+		}
+		metrics := CollectNoteMetrics(content)
+		notes[file] = metrics
+		linkCount += metrics.LinkCount
+	}
+
+	return Metrics{NoteCount: noteCount, LinkCount: linkCount, Notes: notes}, nil
+}

internal/collector/note.go

@@ -0,0 +1,19 @@
+package collector
+
+import (
+	"regexp"
+)
+
+type NoteMetrics struct {
+	LinkCount int
+}
+
+func CollectNoteMetrics(content []byte) NoteMetrics {
+	return NoteMetrics{LinkCount: collectLinkCount(content)}
+}
+
+func collectLinkCount(content []byte) int {
+	r, _ := regexp.Compile(`\[\[[^]]+\]\]`)
+	matches := r.FindAll(content, -1)
+	return len(matches)
+}

internal/collector/note_test.go

@@ -0,0 +1,25 @@
+package collector
+
+import "testing"
+
+func TestCollectNoteMetrics(t *testing.T) {
+	data := []struct {
+		name     string
+		content  string
+		expected NoteMetrics
+	}{
+		{"empty file", "", NoteMetrics{LinkCount: 0}},
+		{"file with a single link", "[[Link]]", NoteMetrics{LinkCount: 1}},
+		{"file with multiple links", "[[Link]]aksdjf[[anotherlink]]\n[[link]]", NoteMetrics{LinkCount: 3}},
+		{"wikilink dividers", "[[something|another]]\n\n[[link]]\n[[382dlk djfs link|yeah]]", NoteMetrics{LinkCount: 3}},
+	}
+
+	for _, d := range data {
+		t.Run(d.name, func(t *testing.T) {
+			result := CollectNoteMetrics([]byte(d.content))
+			if result != d.expected {
+				t.Errorf("Expected %v, got %v", d.expected, result)
+			}
+		})
+	}
+}