feat: allow import of newer pocket data export files in csv format (#1023)

* feat: allow import of newer pocket data export files in csv format

Signed-off-by: Mel <einebeere@gmail.com>

* fix: linter error

Signed-off-by: Mel <einebeere@gmail.com>

---------

Signed-off-by: Mel <einebeere@gmail.com>
Co-authored-by: Felipe Martin <812088+fmartingr@users.noreply.github.com>

Author: melnary

internal/cmd/pocket.go

@@ -1,22 +1,29 @@
 package cmd
 
 import (
+	"context"
+	"encoding/csv"
+	"errors"
 	"fmt"
 	"os"
+	"path/filepath"
+	"regexp"
+	"slices"
 	"strconv"
 	"strings"
 	"time"
 
 	"github.com/PuerkitoBio/goquery"
 	"github.com/go-shiori/shiori/internal/core"
+	"github.com/go-shiori/shiori/internal/database"
 	"github.com/go-shiori/shiori/internal/model"
 	"github.com/spf13/cobra"
 )
 
 func pocketCmd() *cobra.Command {
 	cmd := &cobra.Command{
 		Use:   "pocket source-file",
-		Short: "Import bookmarks from Pocket's exported HTML file",
+		Short: "Import bookmarks from Pocket's data export file",
 		Args:  cobra.ExactArgs(1),
 		Run:   pocketHandler,
 	}
@@ -25,17 +32,43 @@ func pocketCmd() *cobra.Command {
 }
 
 func pocketHandler(cmd *cobra.Command, args []string) {
-	_, deps := initShiori(cmd.Context(), cmd)
+	ctx := cmd.Context()
+	_, deps := initShiori(ctx, cmd)
 
 	// Open pocket's file
-	srcFile, err := os.Open(args[0])
+	filePath := args[0]
+	srcFile, err := os.Open(filePath)
 	if err != nil {
 		cError.Println(err)
 		os.Exit(1)
 	}
 	defer srcFile.Close()
 
-	// Parse pocket's file
+	var bookmarks []model.BookmarkDTO
+	switch filepath.Ext(filePath) {
+	case ".html":
+		bookmarks = parseHtmlExport(ctx, deps.Database, srcFile)
+	case ".csv":
+		bookmarks = parseCsvExport(ctx, deps.Database, srcFile)
+	default:
+		cError.Println("Invalid file format. Only HTML and CSV are supported.")
+		os.Exit(1)
+	}
+
+	// Save bookmark to database
+	bookmarks, err = deps.Database.SaveBookmarks(ctx, true, bookmarks...)
+	if err != nil {
+		cError.Printf("Failed to save bookmarks: %v\n", err)
+		os.Exit(1)
+	}
+
+	// Print imported bookmarks
+	fmt.Println()
+	printBookmarks(bookmarks...)
+}
+
+// Parse bookmarks from HTML file
+func parseHtmlExport(ctx context.Context, db database.DB, srcFile *os.File) []model.BookmarkDTO {
 	bookmarks := []model.BookmarkDTO{}
 	mapURL := make(map[string]struct{})
 
@@ -49,69 +82,137 @@ func pocketHandler(cmd *cobra.Command, args []string) {
 		// Get metadata
 		title := a.Text()
 		url, _ := a.Attr("href")
-		strTags, _ := a.Attr("tags")
-		strModified, _ := a.Attr("time_added")
-		intModified, _ := strconv.ParseInt(strModified, 10, 64)
-		modified := time.Unix(intModified, 0)
-
-		// Clean up URL
-		var err error
-		url, err = core.RemoveUTMParams(url)
+		tagsStr, _ := a.Attr("tags")
+		timeAddedStr, _ := a.Attr("time_added")
+
+		title, url, timeAdded, tags, err := verifyMetadata(title, url, timeAddedStr, tagsStr)
 		if err != nil {
-			cError.Printf("Skip %s: URL is not valid\n", url)
+			cError.Printf("Skip %s: %v\n", url, err)
 			return
 		}
 
-		// Make sure title is valid Utf-8
-		title = validateTitle(title, url)
-
-		// Check if the URL already exist before, both in bookmark
-		// file or in database
-		if _, exist := mapURL[url]; exist {
-			cError.Printf("Skip %s: URL already exists\n", url)
+		if err = handleDuplicates(ctx, db, mapURL, url); err != nil {
+			cError.Printf("Skip %s: %v\n", url, err)
 			return
 		}
 
-		_, exist, err := deps.Database.GetBookmark(cmd.Context(), 0, url)
-		if err != nil {
-			cError.Printf("Skip %s: Get Bookmark fail, %v", url, err)
-			return
+		// Add item to list
+		bookmark := model.BookmarkDTO{
+			URL:        url,
+			Title:      title,
+			ModifiedAt: timeAdded.Format(model.DatabaseDateFormat),
+			CreatedAt:  timeAdded.Format(model.DatabaseDateFormat),
+			Tags:       tags,
 		}
 
-		if exist {
-			cError.Printf("Skip %s: URL already exists\n", url)
-			mapURL[url] = struct{}{}
-			return
-		}
+		mapURL[url] = struct{}{}
+		bookmarks = append(bookmarks, bookmark)
+	})
+
+	return bookmarks
+}
+
+// Parse bookmarks from CSV file
+func parseCsvExport(ctx context.Context, db database.DB, srcFile *os.File) []model.BookmarkDTO {
+	bookmarks := []model.BookmarkDTO{}
+	mapURL := make(map[string]struct{})
 
-		// Get bookmark tags
-		tags := []model.Tag{}
-		for _, strTag := range strings.Split(strTags, ",") {
-			if strTag != "" {
-				tags = append(tags, model.Tag{Name: strTag})
+	reader := csv.NewReader(srcFile)
+	records, err := reader.ReadAll()
+	if err != nil {
+		cError.Println(err)
+		os.Exit(1)
+	}
+
+	for i, cols := range records {
+		// Check and skip header
+		if i == 0 {
+			expected := []string{"title", "url", "time_added", "cursor", "tags", "status"}
+			if slices.Compare(cols, expected) != 0 {
+				cError.Printf("Invalid CSV format. Header must be: %s\n", strings.Join(expected, ","))
+				os.Exit(1)
 			}
+			continue
+		}
+
+		// Get metadata
+		title, url, timeAdded, tags, err := verifyMetadata(cols[0], cols[1], cols[2], cols[4])
+		if err != nil {
+			cError.Printf("Skip %s: %v\n", url, err)
+			continue
+		}
+
+		if err = handleDuplicates(ctx, db, mapURL, url); err != nil {
+			cError.Printf("Skip %s: %v\n", url, err)
+			continue
 		}
 
 		// Add item to list
 		bookmark := model.BookmarkDTO{
 			URL:        url,
 			Title:      title,
-			ModifiedAt: modified.Format(model.DatabaseDateFormat),
+			ModifiedAt: timeAdded.Format(model.DatabaseDateFormat),
+			CreatedAt:  timeAdded.Format(model.DatabaseDateFormat),
 			Tags:       tags,
 		}
 
 		mapURL[url] = struct{}{}
 		bookmarks = append(bookmarks, bookmark)
-	})
+	}
 
-	// Save bookmark to database
-	bookmarks, err = deps.Database.SaveBookmarks(cmd.Context(), true, bookmarks...)
+	return bookmarks
+}
+
+// Parse metadata and verify it's validity
+func verifyMetadata(title, url, timeAddedStr, tags string) (string, string, time.Time, []model.Tag, error) {
+	// Clean up URL
+	var err error
+	url, err = core.RemoveUTMParams(url)
 	if err != nil {
-		cError.Printf("Failed to save bookmarks: %v\n", err)
-		os.Exit(1)
+		err = fmt.Errorf("URL is not valid, %w", err)
+		return "", "", time.Time{}, nil, err
 	}
 
-	// Print imported bookmark
-	fmt.Println()
-	printBookmarks(bookmarks...)
+	// Make sure title is valid Utf-8
+	title = validateTitle(title, url)
+
+	// Parse time added
+	timeAddedInt, err := strconv.ParseInt(timeAddedStr, 10, 64)
+	if err != nil {
+		err = fmt.Errorf("Invalid time added, %w", err)
+		return "", "", time.Time{}, nil, err
+	}
+	timeAdded := time.Unix(timeAddedInt, 0)
+
+	// Get bookmark tags
+	tagsList := []model.Tag{}
+	// We need to split tags by both comma or pipe,
+	// because Pocket's CSV export use pipe as separator,
+	// while HTML export use comma.
+	for _, tag := range regexp.MustCompile(`[,|]`).Split(tags, -1) {
+		if tag != "" {
+			tagsList = append(tagsList, model.Tag{Name: tag})
+		}
+	}
+
+	return title, url, timeAdded, tagsList, nil
+}
+
+// Checks if the URL already exist, both in bookmark
+// file or in database
+func handleDuplicates(ctx context.Context, db database.DB, mapURL map[string]struct{}, url string) error {
+	if _, exists := mapURL[url]; exists {
+		return errors.New("URL already exists")
+	}
+
+	_, exists, err := db.GetBookmark(ctx, 0, url)
+	if err != nil {
+		return fmt.Errorf("Failed getting bookmark, %w", err)
+	}
+
+	if exists {
+		return errors.New("URL already exists")
+	}
+
+	return nil
 }