NameChanger

One popular interview question for system administrators is what is the easiest way to give a set of files a new extension. Take a directory of *.log files, for example: How do you rename them all in one go to *.log.old? It has reportedly happened that candidates suggested the shell command mv *.log *.log.old for this – however, they were then not hired.

There are already quite a few tools lurking around on GitHub that handle such tasks, such as the Renamer tool written in Rust [1]. But such simple utilities make for great illustrative examples, so I wanted to explore some Go techniques for bulk renaming. Paying tribute to the original, the Go variant presented below will also go by the name of Renamer. For example, to rename an entire set of logfiles ending in .log to .log.bak, just use the call shown in line 1 of Listing 1.

01 $ renamer -v '.log$/.log.bak' *.log
02 out.log -> out.log.bak
03 [...]
04 $ renamer -v '/hawaii2020-{seq}.jpg' *.JPG
05 IMG_8858.JPG -> hawaii2020-0001.jpg
06 IMG_8859.JPG -> hawaii2020-0002.jpg

Or how about renaming vacation photos currently named IMG_8858.JPG through IMG_9091.JPG to hawaii-2020-0001.jpg through hawaii-2020-0234.jpg? My Go program does that too with the call from line 4, replacing the placeholder {seq} with a counter incremented by one for each renamed file, which it pads with leading zeros to four digits.

Mass Production

The renamer main program (Listing 2) processes its command-line options -d for a test run without consequences (dryrun) and -v for chatty (verbose) status messages in lines 19 and 20. The standard flag package used for this purpose not only assigns the dryrun and verbose pointer variables the values true or false, respectively, but it also jumps to a Usage() function defined in the Usage attribute if the user tries to slip in an option that the program doesn't know.

01 package main
02
03 import (
04   "flag"
05   "fmt"
06   "os"
07   "path"
08 )
09
10 func usage() {
11   fmt.Fprintf(os.Stderr,
12     "Usage: %s 'search/replace' file ...\n",
13     path.Base(os.Args[0]))
14   flag.PrintDefaults()
15   os.Exit(1)
16 }
17
18 func main() {
19   dryrun := flag.Bool("d", false, "dryrun only")
20   verbose := flag.Bool("v", false, "verbose mode")
21   flag.Usage = usage
22   flag.Parse()
23
24   if *dryrun {
25     fmt.Printf("Dryrun mode\n")
26   }
27
28   if len(flag.Args()) < 2 {
29     usage()
30   }
31
32   cmd := flag.Args()[0]
33   files := flag.Args()[1:]
34   modifier, err := mkmodifier(cmd)
35   if err != nil {
36     fmt.Fprintf(os.Stderr,
37       "Invalid command: %s\n", cmd)
38     usage()
39   }
40
41   for _, file := range files {
42     modfile := modifier(file)
43     if file == modfile {
44       continue
45     }
46     if *verbose || *dryrun {
47       fmt.Printf("%s -> %s\n", file, modfile)
48     }
49     if *dryrun {
50       continue
51     }
52     err := os.Rename(file, modfile)
53     if err != nil {
54       fmt.Printf("Renaming %s -> %s failed: %v\n",
55         file, modfile, err)
56       break
57     }
58   }
59 }

In any case, the program expects a command to manipulate the file names and one or more files to rename later. Line 12 informs the user of the correct call syntax of the renamer binary compiled from the source code.

The array slice arithmetic assigns the first command-line parameter with index number   to the cmd variable. This is followed by one or more file names, which the shell is also welcome to expand using wildcards before passing them to the program. The arguments from the second to last position are fetched from the array slice by the expression [1:]; line 33 assigns the list of files to the variable files.

The instruction passed in at the command line to manipulate the file names (e.g., '.log$/.log.old') gets sent to the mkmodifier() function defined further down in Listing 3. This turns the instruction into a Go function that manipulates input file names according to the user's instructions and returns a modified name.

01 package main
02
03 import (
04   "errors"
05   "fmt"
06   "regexp"
07   "strings"
08 )
09
10 func mkmodifier(cmd string) (func(string) string, error) {
11   parts := strings.Split(cmd, "/")
12   if len(parts) != 2 {
13     return nil, errors.New("Invalid repl command")
14   }
15   search := parts[0]
16   repltmpl := parts[1]
17   seq := 1
18
19   var rex *regexp.Regexp
20
21   if len(search) == 0 {
22     search = ".*"
23   }
24
25   rex = regexp.MustCompile(search)
26
27   modifier := func(org string) string {
28     repl := strings.Replace(repltmpl,
29       "{seq}", fmt.Sprintf("%04d", seq), -1)
30     seq++
31     res := rex.ReplaceAllString(org, repl)
32     return string(res)
33   }
34
35   return modifier, nil
36 }

Function Returns Function

You've read that correctly: The mkmodifier() function actually returns a function in line 34 of Listing 2, which is assigned to the modifier variable there. A few lines down, in the for loop that iterates over all the files to be manipulated, the main program simply calls this function by referencing modifier. With every call, the main program passes the returned file name the original name of the file and, in line 42, picks up the new name and stores it in modfile.

If the user has chosen dryrun mode (-d), line 47 simply prints the intended rename action, and line 50 rings in the next round of the for loop with continue, skipping the call of rename() in line 52.

In production mode, however, line 52 calls the Unix system rename() function from the standard os package and renames the file to the new name from modfile. If access rights prevent this, the function fails and os.Rename() returns an error, which line 53 fields. The associated if block prints a message and breaks the for loop with break, because in that case the end of the world is nigh.

Regular Expressions

Instead of requesting a plain vanilla string replacement, the user can also specify regular expressions to remodel file names. For example, the .log$ search expression illustrated earlier specifies that the .log suffix must actually be at the end of the name – it would ignore foo.log.bak. To enable this, Listing 3 draws on the standard regexp package and compiles the regular expression from the user input to create a rex variable of the *regexp.Regexp type using MustCompile() in line 25. After that, the modifier defined in line 27 can call the ReplaceAllString() function. It replaces all matches that match the expression in the original name org with the replacement string stored in repl.

Attentive readers may wonder about the mkmodifier() function in Listing 3: It returns a function to the main program, to be called multiple times, but this function actually seems to maintain state between calls. For example, take a look at the function's local variable seq: Each new call to the function injects a value incremented by one into the modified file name. How is this possible?