whatsviewer/parser.go

package main

import (
	"regexp"
	"strings"
	"time"
)

// Match date/time prefix: M/DD/YY, H:MM[narrow-nbsp]AM/PM -
// The narrow no-break space (\u202f) appears before AM/PM in WhatsApp exports.
var messageStartRe = regexp.MustCompile(`^(\d{1,2}/\d{1,2}/\d{2}, \d{1,2}:\d{2}[\s\x{202f}]*[APap][Mm]) - (.*)`)

// Time layouts to try parsing (with both regular space and narrow no-break space)
var timeLayouts = []string{
	"1/2/06, 3:04\u202fPM",
	"1/2/06, 3:04 PM",
	"1/2/06, 3:04\u202fam",
	"1/2/06, 3:04 am",
}

func parseTimestamp(s string) (time.Time, bool) {
	// Normalize: replace narrow no-break space with regular space for parsing
	normalized := strings.ReplaceAll(s, "\u202f", " ")
	// Also try with the original
	for _, layout := range []string{"1/2/06, 3:04 PM", "1/2/06, 3:04 pm"} {
		t, err := time.Parse(layout, normalized)
		if err == nil {
			return t, true
		}
		// Try uppercase
		t, err = time.Parse(layout, strings.ToUpper(normalized))
		if err == nil {
			return t, true
		}
	}
	return time.Time{}, false
}

func classifyAttachment(filename string) string {
	lower := strings.ToLower(filename)
	if strings.HasPrefix(lower, "stk-") && strings.HasSuffix(lower, ".webp") {
		return "sticker"
	}
	if strings.HasPrefix(lower, "ptt-") {
		return "audio"
	}
	for _, ext := range []string{".jpg", ".jpeg", ".png", ".webp", ".gif"} {
		if strings.HasSuffix(lower, ext) {
			return "image"
		}
	}
	for _, ext := range []string{".mp4", ".3gp", ".mov"} {
		if strings.HasSuffix(lower, ext) {
			return "video"
		}
	}
	for _, ext := range []string{".opus", ".ogg", ".m4a", ".mp3", ".aac"} {
		if strings.HasSuffix(lower, ext) {
			return "audio"
		}
	}
	return "document"
}

func parseChat(content string) []Message {
	// Strip UTF-8 BOM
	content = strings.TrimPrefix(content, "\xef\xbb\xbf")
	// Normalize line endings
	content = strings.ReplaceAll(content, "\r\n", "\n")
	content = strings.ReplaceAll(content, "\r", "\n")

	lines := strings.Split(content, "\n")
	var messages []Message

	for i := 0; i < len(lines); i++ {
		line := lines[i]
		if line == "" {
			continue
		}

		m := messageStartRe.FindStringSubmatch(line)
		if m == nil {
			// Continuation line — append to previous message
			if len(messages) > 0 {
				prev := &messages[len(messages)-1]
				if prev.Text != "" {
					prev.Text += "\n"
				}
				prev.Text += line
				// Check if continuation line has attachment
				checkAttachment(prev, line)
			}
			continue
		}

		timestampStr := m[1]
		rest := m[2]

		ts, ok := parseTimestamp(timestampStr)
		if !ok {
			// If we can't parse timestamp, treat as continuation
			if len(messages) > 0 {
				prev := &messages[len(messages)-1]
				if prev.Text != "" {
					prev.Text += "\n"
				}
				prev.Text += line
			}
			continue
		}

		msg := Message{
			Timestamp: ts,
		}

		// Check if it's a system message (no colon after sender) or a user message
		colonIdx := strings.Index(rest, ": ")
		if colonIdx == -1 {
			// System message
			msg.IsSystem = true
			msg.Text = rest
		} else {
			msg.Sender = rest[:colonIdx]
			msg.Text = rest[colonIdx+2:]
		}

		checkAttachment(&msg, msg.Text)
		messages = append(messages, msg)
	}

	return messages
}

func checkAttachment(msg *Message, text string) {
	const suffix = " (file attached)"
	// Check each line of text for attachments
	for _, line := range strings.Split(text, "\n") {
		line = strings.TrimSpace(line)
		if strings.HasSuffix(line, suffix) {
			filename := strings.TrimSuffix(line, suffix)
			// Don't add duplicate attachments
			found := false
			for _, a := range msg.Attachments {
				if a.Filename == filename {
					found = true
					break
				}
			}
			if !found {
				msg.Attachments = append(msg.Attachments, Attachment{
					Filename: filename,
					Type:     classifyAttachment(filename),
				})
			}
			// Remove the "(file attached)" text from displayed text
			msg.Text = strings.Replace(msg.Text, line, filename, 1)
		}
	}
}