Marking the Cards

In this way, Read() from line 54 reads the data from the cache file and turns it into a Go map that assigns photo paths to Boolean values. To do this, it uses os.Open() to open the file and calls in a scanner from the bufio package for a new scanner starting in line 62. It calls Scan() in line 63 to browse through every single line of the cache file and then calls Text() to fetch the matching text as a string, excluding the line break.

The assignment in line 65 creates a key in the cache map for each cache entry and assigns a value of true to it. The cache.cache map remains stored in the instance structure, where other functions like cache.Exists() or cache.Set() can access it later.

To update the cache file after completing the work, the Write() function starting in line 71 writes the modified map back. To do this, it calls OpenFile() to open the cache file in line 72 and iterates over the map entries to write them back to the cache file one by one with fmt.Fprintf, overwriting the old ones because of the O_TRUNC option.

Previously unseen SD cards do not have .uuid files in their root directories. The Init() function starting in line 34 checks for this and creates a new UUID with the uuid GitHub package from Google in line 40 if line 39 failed to find one previously. This 36-character string is guaranteed to be unique each time, so it will continue to uniquely identify cards marked with it in the future [2].

Date from Exif Headers

The date a photo was taken is determined by the function photoDate() starting in line 11 in Listing 2. The exif package from the goexif2 project on GitHub provides convenient functions that read the Exif header of a JPEG image, decode it, and return it as a Go time.Time type variable. Its functions Year(), Month(), and Day() convert the photo date into year, month, and day. importer relies on this later on to create the nested file structure for keeping the photos organized in storage.

01 package main
02
03 import (
04   "fmt"
05   exif "github.com/xor-gate/goexif2/exif"
06   "io"
07   "os"
08   "path"
09 )
10
11 func photoDate(path string) ([]int, error) {
12   dt := []int{}
13
14   f, err := os.Open(path)
15   if err != nil {
16     return dt, err
17   }
18
19   x, err := exif.Decode(f)
20   if err != nil {
21     return dt, err
22   }
23
24   t, err := x.DateTime()
25   if err != nil {
26     return dt, err
27   }
28
29   return []int{int(t.Year()), int(t.Month()), int(t.Day()),
30                int(t.Hour()), int(t.Minute()), int(t.Second())}, nil
31 }
32
33 func copy(src, dst string) (int64, error) {
34   sourceFileStat, err := os.Stat(src)
35   if err != nil {
36     return 0, err
37   }
38
39   if !sourceFileStat.Mode().IsRegular() {
40     return 0, fmt.Errorf("%s is not a regular file", src)
41   }
42
43   source, err := os.Open(src)
44   if err != nil {
45     return 0, err
46   }
47   defer source.Close()
48
49   dest, err := os.Create(dst)
50   if err != nil {
51     return 0, err
52   }
53   defer dest.Close()
54   nBytes, err := io.Copy(dest, source)
55   return nBytes, err
56 }
57
58 func targetDir() string {
59   homedir, err := os.UserHomeDir()
60   panicOnErr(err)
61   return path.Join(homedir, "/idb")
62 }

However, the Go standard library does not have a function for copying files. This is why copy() has to open the source and target files starting in line 33, read them block by block from the source, and write them to the target dest with io.Copy(). The archive directory for the importer is idb/ in the user's home directory; the path is determined and returned by the targetDir() function starting in line 58 of Listing 2.

In the main program in Listing 3, main() first checks whether the call also includes a directory for importing photos. After reading the cache file in line 35, the Walk() function from the standard filepath package plumbs the depths of the specified import directory and processes all the JPEG files it finds there.

01 package main
02
03 import (
04   "errors"
05   "flag"
06   "fmt"
07   "os"
08   "path"
09   "path/filepath"
10   rex "regexp"
11 )
12
13 func main() {
14   flag.Usage = func() {
15     fmt.Printf("Usage: %s dir\n", path.Base(os.Args[0]))
16     os.Exit(1)
17   }
18
19   flag.Parse()
20   if flag.NArg() < 1 {
21     flag.Usage()
22   }
23
24   idir := flag.Args()[0]
25
26   tDir := targetDir()
27   _, err := os.Stat(tDir)
28   if errors.Is(err, os.ErrNotExist) {
29     err := os.Mkdir(tDir, 0755)
30     panicOnErr(err)
31   }
32
33   cache := NewCache(idir)
34   cache.Init()
35   cache.Read()
36
37   filepath.Walk(idir, func(ipath string, f os.FileInfo, err error) error {
38     jpgMatch := rex.MustCompile(`(?i)^\w.*JPG$`)
39     dir, bpath := path.Split(ipath)
40     match := jpgMatch.MatchString(bpath)
41     if !match {
42       return nil
43     }
44
45     dir = path.Base(dir)
46     twoPath := path.Join(dir, bpath) // parent/file
47
48     ok := cache.Exists(twoPath)
49     if ok {
50       return nil // already archived
51     }
52
53     dt, err := photoDate(ipath)
54     if err != nil {
55       fmt.Printf("Error: %s: %s\n", ipath, err)
56       return nil
57     }
58     dstDir := fmt.Sprintf("%s/%d/%02d/%02d", tDir, dt[0], dt[1], dt[2])
59     os.MkdirAll(dstDir, 0755)
60     newFile := path.Base(ipath)
61     dst := fmt.Sprintf("%s/%d%02d%02d%02d%02d%02d-%s",
62       dstDir, dt[0], dt[1], dt[2], dt[3], dt[4], dt[5], newFile)
63     fmt.Printf("Copying %s to %s\n", ipath, dst)
64     _, err = copy(ipath, dst)
65     panicOnErr(err)
66
67     cache.Set(twoPath)
68     return nil
69   })
70
71   cache.Write()
72 }
73
74 func panicOnErr(err error) {
75   if err != nil {
76     panic(err)
77   }
78 }

JPEGs Only

The regular expression in line 38 of Listing 3 filters out all non-JPEGs and tells the walker to return without touching any foreign objects. If the file is obviously a regular photo, then line 39 breaks down the path into the parent directory and file name, while line 45 truncates everything but the last subpath from the directory. Based on this and the file name, line 46 in twoPath then creates the short path consisting of the parent directory and file name that the cache uses as a key later.

Line 48 checks if the short path already exists in the cache (i.e., if the file has been archived before). If so, the Walk() callback returns in line 50 without taking any further action. But if a previously unarchived photo is found, photoDate() in line 53 extracts the year, month, and day the photo was taken from the photo's Exif header. From this, the function determines the target directory in the archive, as idb/<Year>/<Month>/<Day>, and then creates it if it does not already exist.

Now it's time to copy the photo to the archive. Line 61 once again adds the date the photo was taken to the name of the target file in the archive directory. The reason for this apparent redundancy is the idb tool from the last issue [2], which uses the -xlink option to link all tagged photos to a directory. If we only used the original file name for archived photos, several photo files with the name DSC00001.JPG could end up in the search results because the consecutive numbers are used again and again by the camera on reformatted cards.

After completing the copying work, line 67 records the file and the card's UUID in the cache; line 71 writes the cache back to the disk at the end of the function.