5 files changed, 570 insertions, 0 deletions
diff --git a/internal/processor/audio.go b/internal/processor/audio.go
new file mode 100644
index 0000000..98aedcf
--- /dev/null
+++ b/internal/processor/audio.go
@@ -0,0 +1,49 @@
+package processor
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+)
+
+// processAudio copies an .mp3 file into destDir and returns an HTML <audio> snippet.
+// The audio element has controls enabled so visitors can play it inline.
+func processAudio(srcPath, destDir, postID string) (filename, htmlContent string, err error) {
+	outName := filepath.Base(srcPath)
+	outPath := filepath.Join(destDir, outName)
+
+	if err := copyFile(srcPath, outPath); err != nil {
+		return "", "", err
+	}
+
+	// The src attribute is relative to the site root.
+	src := fmt.Sprintf("posts/%s/%s", postID, outName)
+	html := fmt.Sprintf(
+		`<audio controls class="post-audio"><source src="%s" type="audio/mpeg">Your browser does not support audio.</audio>`,
+		src,
+	)
+
+	return outName, html, nil
+}
+
+// copyFile copies the file at src to dst, creating dst if it does not exist.
+func copyFile(src, dst string) error {
+	in, err := os.Open(src)
+	if err != nil {
+		return fmt.Errorf("open source %s: %w", src, err)
+	}
+	defer in.Close()
+
+	out, err := os.Create(dst)
+	if err != nil {
+		return fmt.Errorf("create dest %s: %w", dst, err)
+	}
+	defer out.Close()
+
+	if _, err := io.Copy(out, in); err != nil {
+		return fmt.Errorf("copy %s → %s: %w", src, dst, err)
+	}
+
+	return nil
+}
diff --git a/internal/processor/image.go b/internal/processor/image.go
new file mode 100644
index 0000000..9a7d769
--- /dev/null
+++ b/internal/processor/image.go
@@ -0,0 +1,116 @@
+package processor
+
+import (
+	"fmt"
+	"image"
+	"image/gif"
+	"image/jpeg"
+	"image/png"
+	"os"
+	"path/filepath"
+
+	"golang.org/x/image/draw"
+)
+
+const (
+	maxImageWidth  = 1024
+	jpegQuality    = 80
+)
+
+// processImage reads the source image, resizes it if wider than maxImageWidth,
+// encodes it as JPEG at jpegQuality, and writes the result to destDir.
+// Returns the output filename (always a .jpg) and an HTML <img> snippet.
+func processImage(srcPath, destDir, postID string) (filename, htmlContent string, err error) {
+	img, err := decodeImage(srcPath)
+	if err != nil {
+		return "", "", err
+	}
+
+	img = resizeIfNeeded(img)
+
+	outName := "image.jpg"
+	outPath := filepath.Join(destDir, outName)
+
+	if err := writeJPEG(img, outPath); err != nil {
+		return "", "", err
+	}
+
+	// The <img> src is relative to the site root, pointing into the posts dir.
+	src := fmt.Sprintf("posts/%s/%s", postID, outName)
+	html := fmt.Sprintf(`<img src="%s" alt="" class="post-image">`, src)
+
+	return outName, html, nil
+}
+
+// decodeImage decodes a JPEG, PNG, or GIF (first frame) from srcPath.
+func decodeImage(srcPath string) (image.Image, error) {
+	f, err := os.Open(srcPath)
+	if err != nil {
+		return nil, fmt.Errorf("open image %s: %w", srcPath, err)
+	}
+	defer f.Close()
+
+	ext := filepath.Ext(srcPath)
+	switch ext {
+	case ".jpg", ".jpeg":
+		img, err := jpeg.Decode(f)
+		if err != nil {
+			return nil, fmt.Errorf("decode JPEG %s: %w", srcPath, err)
+		}
+		return img, nil
+
+	case ".png":
+		img, err := png.Decode(f)
+		if err != nil {
+			return nil, fmt.Errorf("decode PNG %s: %w", srcPath, err)
+		}
+		return img, nil
+
+	case ".gif":
+		// Use only the first frame of animated GIFs.
+		g, err := gif.Decode(f)
+		if err != nil {
+			return nil, fmt.Errorf("decode GIF %s: %w", srcPath, err)
+		}
+		return g, nil
+
+	default:
+		return nil, fmt.Errorf("unsupported image format: %s", ext)
+	}
+}
+
+// resizeIfNeeded returns a resized copy of img if its width exceeds maxImageWidth,
+// preserving aspect ratio. Otherwise the original is returned unchanged.
+func resizeIfNeeded(img image.Image) image.Image {
+	bounds := img.Bounds()
+	w := bounds.Dx()
+
+	if w <= maxImageWidth {
+		return img
+	}
+
+	h := bounds.Dy()
+	newW := maxImageWidth
+	newH := (h * newW) / w
+
+	dst := image.NewRGBA(image.Rect(0, 0, newW, newH))
+	draw.BiLinear.Scale(dst, dst.Bounds(), img, bounds, draw.Over, nil)
+
+	return dst
+}
+
+// writeJPEG encodes img as JPEG at the configured quality level and writes to path.
+func writeJPEG(img image.Image, path string) error {
+	f, err := os.Create(path)
+	if err != nil {
+		return fmt.Errorf("create JPEG %s: %w", path, err)
+	}
+	defer f.Close()
+
+	opts := &jpeg.Options{Quality: jpegQuality}
+	if err := jpeg.Encode(f, img, opts); err != nil {
+		return fmt.Errorf("encode JPEG %s: %w", path, err)
+	}
+
+	return nil
+}
diff --git a/internal/processor/markdown.go b/internal/processor/markdown.go
new file mode 100644
index 0000000..8d69bfe
--- /dev/null
+++ b/internal/processor/markdown.go
@@ -0,0 +1,68 @@
+package processor
+
+import (
+	"bytes"
+	"fmt"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+
+	"github.com/yuin/goldmark"
+	"github.com/yuin/goldmark/extension"
+	"github.com/yuin/goldmark/renderer/html"
+)
+
+// imageRefPattern matches Markdown image syntax: ![alt](filename)
+// We use it to discover local asset references that must be copied.
+var imageRefPattern = regexp.MustCompile(`!\[[^\]]*\]\(([^)]+)\)`)
+
+// processMd converts a Markdown file to an HTML snippet.
+// Returns the HTML and a list of local image filenames referenced in the document.
+// Referenced images that exist alongside the source file are returned so the
+// caller can copy them into the post asset directory.
+func processMd(path string) (htmlContent string, localImages []string, err error) {
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return "", nil, fmt.Errorf("read markdown %s: %w", path, err)
+	}
+
+	// Collect local image references so the caller can copy them as assets.
+	localImages = findLocalImages(string(data), filepath.Dir(path))
+
+	md := goldmark.New(
+		goldmark.WithExtensions(extension.GFM),
+		goldmark.WithRendererOptions(
+			html.WithUnsafe(), // Allow raw HTML in markdown (user-controlled content).
+		),
+	)
+
+	var buf bytes.Buffer
+	if err := md.Convert(data, &buf); err != nil {
+		return "", nil, fmt.Errorf("convert markdown %s: %w", path, err)
+	}
+
+	return buf.String(), localImages, nil
+}
+
+// findLocalImages returns image filenames referenced in markdown that actually
+// exist in sourceDir. Remote URLs (http/https) are ignored.
+func findLocalImages(mdContent, sourceDir string) []string {
+	matches := imageRefPattern.FindAllStringSubmatch(mdContent, -1)
+	var locals []string
+
+	for _, m := range matches {
+		ref := m[1]
+		// Skip remote URLs.
+		if strings.HasPrefix(ref, "http://") || strings.HasPrefix(ref, "https://") {
+			continue
+		}
+
+		candidate := filepath.Join(sourceDir, ref)
+		if _, err := os.Stat(candidate); err == nil {
+			locals = append(locals, filepath.Base(ref))
+		}
+	}
+
+	return locals
+}
diff --git a/internal/processor/processor.go b/internal/processor/processor.go
new file mode 100644
index 0000000..077cc5b
--- /dev/null
+++ b/internal/processor/processor.go
@@ -0,0 +1,234 @@
+// Package processor scans the input directory for new source files and converts
+// each one into a self-contained post directory under outdir/posts/.
+// Supported formats: .txt, .md, .png, .jpg, .jpeg, .gif, .mp3.
+// Each processed source file is deleted from the input directory afterward.
+package processor
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"codeberg.org/snonux/snonux/internal/config"
+	"codeberg.org/snonux/snonux/internal/post"
+)
+
+// Run scans cfg.InputDir and processes every eligible file into a post directory
+// under cfg.OutputDir/posts/. Returns the number of posts created.
+//
+// Images referenced by a .md file in the same input directory are consumed by
+// that markdown post and are not processed as independent image posts.
+func Run(cfg *config.Config) (int, error) {
+	entries, err := os.ReadDir(cfg.InputDir)
+	if err != nil {
+		return 0, fmt.Errorf("read input dir %s: %w", cfg.InputDir, err)
+	}
+
+	postsDir := filepath.Join(cfg.OutputDir, "posts")
+	if err := os.MkdirAll(postsDir, 0o755); err != nil {
+		return 0, fmt.Errorf("create posts dir: %w", err)
+	}
+
+	// Pre-scan markdown files to discover which image filenames they claim.
+	// Claimed images are excluded from independent processing.
+	claimed := claimedByMarkdown(entries, cfg.InputDir)
+
+	count := 0
+
+	for _, entry := range entries {
+		if entry.IsDir() || strings.HasPrefix(entry.Name(), ".") {
+			continue
+		}
+		if claimed[entry.Name()] {
+			continue // consumed by a .md post — skip independent processing
+		}
+
+		srcPath := filepath.Join(cfg.InputDir, entry.Name())
+		if err := processFile(srcPath, postsDir); err != nil {
+			return count, fmt.Errorf("process %s: %w", entry.Name(), err)
+		}
+
+		count++
+	}
+
+	return count, nil
+}
+
+// claimedByMarkdown scans all .md entries in inputDir and returns a set of
+// image filenames that are referenced within those markdown files.
+// Those images should be embedded in the markdown post, not processed alone.
+func claimedByMarkdown(entries []os.DirEntry, inputDir string) map[string]bool {
+	claimed := make(map[string]bool)
+
+	for _, entry := range entries {
+		if entry.IsDir() || strings.ToLower(filepath.Ext(entry.Name())) != ".md" {
+			continue
+		}
+
+		mdPath := filepath.Join(inputDir, entry.Name())
+		data, err := os.ReadFile(mdPath)
+		if err != nil {
+			continue
+		}
+
+		for _, imgName := range findLocalImages(string(data), inputDir) {
+			claimed[imgName] = true
+		}
+	}
+
+	return claimed
+}
+
+// processFile processes a single input file into a new post directory.
+// The source file is removed from the input dir on success.
+func processFile(srcPath, postsDir string) error {
+	now := time.Now().UTC()
+	id := uniqueID(postsDir, now)
+
+	postDir := filepath.Join(postsDir, id)
+	if err := os.MkdirAll(postDir, 0o755); err != nil {
+		return fmt.Errorf("create post dir %s: %w", id, err)
+	}
+
+	p, err := buildPost(srcPath, postDir, id)
+	if err != nil {
+		// Clean up the half-created directory to avoid partial state.
+		_ = os.RemoveAll(postDir)
+		return err
+	}
+
+	if err := p.Save(postDir); err != nil {
+		_ = os.RemoveAll(postDir)
+		return err
+	}
+
+	// Delete the source file only after the post has been successfully persisted.
+	return os.Remove(srcPath)
+}
+
+// buildPost dispatches to the appropriate sub-processor based on file extension
+// and returns a populated Post ready to be saved.
+func buildPost(srcPath, postDir, id string) (*post.Post, error) {
+	ext := strings.ToLower(filepath.Ext(srcPath))
+
+	switch ext {
+	case ".txt":
+		return buildTextPost(srcPath, id)
+
+	case ".md":
+		return buildMarkdownPost(srcPath, postDir, id)
+
+	case ".png", ".jpg", ".jpeg", ".gif":
+		return buildImagePost(srcPath, postDir, id)
+
+	case ".mp3":
+		return buildAudioPost(srcPath, postDir, id)
+
+	default:
+		return nil, fmt.Errorf("unsupported file type: %s", ext)
+	}
+}
+
+func buildTextPost(srcPath, id string) (*post.Post, error) {
+	html, err := processTxt(srcPath)
+	if err != nil {
+		return nil, err
+	}
+
+	return &post.Post{
+		ID:        id,
+		Timestamp: time.Now().UTC(),
+		PostType:  post.TypeText,
+		Content:   html,
+	}, nil
+}
+
+func buildMarkdownPost(srcPath, postDir, id string) (*post.Post, error) {
+	html, localImages, err := processMd(srcPath)
+	if err != nil {
+		return nil, err
+	}
+
+	sourceDir := filepath.Dir(srcPath)
+
+	assets, err := copyLocalImages(localImages, sourceDir, postDir)
+	if err != nil {
+		return nil, err
+	}
+
+	// Delete the referenced image files from the input dir so they are not
+	// processed again as independent posts.
+	for _, name := range localImages {
+		_ = os.Remove(filepath.Join(sourceDir, name))
+	}
+
+	return &post.Post{
+		ID:        id,
+		Timestamp: time.Now().UTC(),
+		PostType:  post.TypeMarkdown,
+		Content:   html,
+		Assets:    assets,
+	}, nil
+}
+
+func buildImagePost(srcPath, postDir, id string) (*post.Post, error) {
+	filename, html, err := processImage(srcPath, postDir, id)
+	if err != nil {
+		return nil, err
+	}
+
+	return &post.Post{
+		ID:        id,
+		Timestamp: time.Now().UTC(),
+		PostType:  post.TypeImage,
+		Content:   html,
+		Assets:    []string{filename},
+	}, nil
+}
+
+func buildAudioPost(srcPath, postDir, id string) (*post.Post, error) {
+	filename, html, err := processAudio(srcPath, postDir, id)
+	if err != nil {
+		return nil, err
+	}
+
+	return &post.Post{
+		ID:        id,
+		Timestamp: time.Now().UTC(),
+		PostType:  post.TypeAudio,
+		Content:   html,
+		Assets:    []string{filename},
+	}, nil
+}
+
+// copyLocalImages copies referenced image files from sourceDir into postDir.
+// Returns the list of filenames that were successfully copied.
+func copyLocalImages(filenames []string, sourceDir, postDir string) ([]string, error) {
+	var copied []string
+
+	for _, name := range filenames {
+		src := filepath.Join(sourceDir, name)
+		dst := filepath.Join(postDir, name)
+
+		if err := copyFile(src, dst); err != nil {
+			return nil, fmt.Errorf("copy image asset %s: %w", name, err)
+		}
+
+		copied = append(copied, name)
+	}
+
+	return copied, nil
+}
+
+// uniqueID generates a post ID for the given time that does not already exist
+// as a directory under postsDir. Appends a numeric suffix if needed.
+func uniqueID(postsDir string, t time.Time) string {
+	for i := 0; ; i++ {
+		id := post.NewID(t, i)
+		if _, err := os.Stat(filepath.Join(postsDir, id)); os.IsNotExist(err) {
+			return id
+		}
+	}
+}
diff --git a/internal/processor/txt.go b/internal/processor/txt.go
new file mode 100644
index 0000000..8381271
--- /dev/null
+++ b/internal/processor/txt.go
@@ -0,0 +1,103 @@
+package processor
+
+import (
+	"fmt"
+	"html"
+	"os"
+	"regexp"
+	"strings"
+)
+
+// urlPattern matches http/https URLs in plain text.
+// Trailing sentence punctuation is stripped separately by stripURLTrailing.
+var urlPattern = regexp.MustCompile(`https?://\S+`)
+
+// processTxt reads a plain-text file and wraps each non-empty paragraph in <p> tags.
+// URLs are automatically converted to clickable <a> links.
+// Non-URL text is HTML-escaped to prevent XSS.
+func processTxt(path string) (string, error) {
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return "", fmt.Errorf("read txt %s: %w", path, err)
+	}
+
+	raw := strings.TrimSpace(string(data))
+	if raw == "" {
+		return "<p></p>", nil
+	}
+
+	// Split on blank lines to get logical paragraphs.
+	paragraphs := strings.Split(raw, "\n\n")
+	var sb strings.Builder
+
+	for _, para := range paragraphs {
+		trimmed := strings.TrimSpace(para)
+		if trimmed == "" {
+			continue
+		}
+		fmt.Fprintf(&sb, "<p>%s</p>\n", formatParagraph(trimmed))
+	}
+
+	return sb.String(), nil
+}
+
+// formatParagraph formats a single paragraph: auto-links URLs, escapes non-URL
+// text, and converts single newlines to <br> line breaks.
+func formatParagraph(para string) string {
+	lines := strings.Split(para, "\n")
+	formatted := make([]string, 0, len(lines))
+
+	for _, line := range lines {
+		if t := strings.TrimSpace(line); t != "" {
+			formatted = append(formatted, autolinkLine(t))
+		}
+	}
+
+	return strings.Join(formatted, "<br>\n")
+}
+
+// autolinkLine escapes non-URL text and wraps detected URLs in <a> tags.
+// Opens in a new tab with rel="noopener noreferrer" for security.
+func autolinkLine(line string) string {
+	locs := urlPattern.FindAllStringIndex(line, -1)
+	if len(locs) == 0 {
+		return html.EscapeString(line)
+	}
+
+	var sb strings.Builder
+	prev := 0
+
+	for _, loc := range locs {
+		sb.WriteString(html.EscapeString(line[prev:loc[0]]))
+
+		rawURL := line[loc[0]:loc[1]]
+		cleanURL := stripURLTrailing(rawURL)
+		trailing := rawURL[len(cleanURL):]
+
+		fmt.Fprintf(&sb, `<a href="%s" target="_blank" rel="noopener noreferrer">%s</a>`,
+			html.EscapeString(cleanURL), html.EscapeString(cleanURL))
+
+		if trailing != "" {
+			sb.WriteString(html.EscapeString(trailing))
+		}
+
+		prev = loc[1]
+	}
+
+	sb.WriteString(html.EscapeString(line[prev:]))
+
+	return sb.String()
+}
+
+// stripURLTrailing removes common sentence-ending punctuation from the end of a
+// URL match. These characters are valid in URLs but almost never appear there
+// at the end in prose (e.g. "Visit https://foo.com." — the "." ends the sentence).
+func stripURLTrailing(u string) string {
+	const cutset = ".,;:!?\"')>]}"
+
+	for len(u) > 0 && strings.ContainsRune(cutset, rune(u[len(u)-1])) {
+		u = u[:len(u)-1]
+	}
+
+	return u
+}