fixrbc

commit a4f4ed2822641a131cc605bf8c45be7ef571e654
parent 80d5a3a3a9c9e458441bf33bc78e55b246d2fe8c
Author: Chris Bracken <chris@bracken.jp>
Date:   Thu, 30 Sep 2021 23:34:22 -0700

Better error handling, add tests

Skips files that don't match the expected RBC filename format. Handle
files with weird capitalisation.

Also adds some minimal tests.

Diffstat:
M
fixrbc.go
 | 
57
+++++++++++++++++++++++++++++++++++++--------------------
A
fixrbc_test.go
 | 
25
+++++++++++++++++++++++++

2 files changed, 62 insertions(+), 20 deletions(-)
diff --git a/fixrbc.go b/fixrbc.go
@@ -3,30 +3,47 @@
 package main
 
 import (
+	"errors"
 	"fmt"
 	"io/ioutil"
 	"log"
 	"os"
+	"regexp"
 	"strings"
 )
 
+func IsValidFilename(n string) bool {
+	r := regexp.MustCompile(`\d+-\d{4}[A-Za-z]{3}\d\d-\d{4}[A-Za-z]{3}\d\d\.pdf`)
+	return r.MatchString(n)
+}
+
+// SanitizedFilename returns a filename in YYYY-mm-dd.pdf format. As input, it
+// takes a filename in default RBC format (ACCTNO-YYYYmmmDD-YYYYmmmDD.pdf).
+func SanitizedFilename(n string) (string, error) {
+	if !IsValidFilename(n) {
+		return "", errors.New(fmt.Sprintf("invalid filename %s", n))
+	}
+	date := strings.Split(n, "-")[2][0:9]
+	y := date[0:4]
+	m, err := getMonthNumber(strings.ToLower(date[4:7]))
+	if err != nil {
+		return "", err
+	}
+	d := date[7:9]
+	return fmt.Sprintf("%s-%s-%s.pdf", y, m, d), nil
+}
+
 func main() {
 	files, err := ioutil.ReadDir("./")
 	if err != nil {
 		log.Fatal(err)
 	}
 	for _, f := range files {
-		if strings.HasSuffix(f.Name(), ".pdf") {
-			// Format: ACCTNO-YYYYmmmDD-YYYYmmmDD.pdf
-			date := strings.Split(f.Name(), "-")[2][0:9]
-			y := date[0:4]
-			m, err := getMonthNumber(date[4:7])
+		if IsValidFilename(f.Name()) {
+			newFname, err := SanitizedFilename(f.Name())
 			if err != nil {
 				log.Fatal(err)
 			}
-			d := date[7:9]
-
-			newFname := fmt.Sprintf("%s-%s-%s.pdf", y, m, d)
 			os.Rename(f.Name(), newFname)
 		}
 	}
@@ -34,29 +51,29 @@ func main() {
 
 func getMonthNumber(m string) (string, error) {
 	switch {
-	case m == "Jan":
+	case m == "jan":
 		return "01", nil
-	case m == "Feb":
+	case m == "feb":
 		return "02", nil
-	case m == "Mar":
+	case m == "mar":
 		return "03", nil
-	case m == "Apr":
+	case m == "apr":
 		return "04", nil
-	case m == "May":
+	case m == "may":
 		return "05", nil
-	case m == "Jun":
+	case m == "jun":
 		return "06", nil
-	case m == "Jul":
+	case m == "jul":
 		return "07", nil
-	case m == "Aug":
+	case m == "aug":
 		return "08", nil
-	case m == "Sep":
+	case m == "sep":
 		return "09", nil
-	case m == "Oct":
+	case m == "oct":
 		return "10", nil
-	case m == "Nov":
+	case m == "nov":
 		return "11", nil
-	case m == "Dec":
+	case m == "dec":
 		return "12", nil
 	}
 	return "", fmt.Errorf("Bad month name: %s", m)
diff --git a/fixrbc_test.go b/fixrbc_test.go
@@ -0,0 +1,25 @@
+package main
+
+import "testing"
+
+func TestSanitizedFilename(t *testing.T) {
+	cases := []struct {
+		desc     string
+		in       string
+		expected string
+	}{
+		{"TestJanFeb", "123456-2021jan01-2021feB02.pdf", "2021-02-02.pdf"},
+		{"TestMarApr", "123456-2021mAr01-2021aPR02.pdf", "2021-04-02.pdf"},
+		{"TestMayJun", "123456-2021May01-2021JuN02.pdf", "2021-06-02.pdf"},
+		{"TestJulAug", "123456-2021Jul01-2021AUG02.pdf", "2021-08-02.pdf"},
+		{"TestSepOct", "123456-2021sep01-2021ocT02.pdf", "2021-10-02.pdf"},
+		{"TestNovDec", "123456-2021nOv01-2021dEC02.pdf", "2021-12-02.pdf"},
+	}
+
+	for _, tc := range cases {
+		actual, _ := SanitizedFilename(tc.in)
+		if actual != tc.expected {
+			t.Errorf("%s: expected: %s got %s for input: %s", tc.desc, tc.expected, actual, tc.in)
+		}
+	}
+}