wip: cdx lookup

1 files changed, 68 insertions, 5 deletions

diff --git a/skate/cdx.go b/skate/cdx.go
index f98b781..ed394ed 100644
--- a/skate/cdx.go
+++ b/skate/cdx.go
@@ -5,15 +5,17 @@ import (
 	"io/ioutil"
 	"log"
 	"net/http"
+	"sort"
 	"strings"
+	"time"
 
 	"github.com/sethgrid/pester"
 )
 
 const cdxApi = "http://web.archive.org/cdx/search/cdx"
 
-// CDX line, might add more fields later.
-type CDX struct {
+// CDXLine line, might add more fields later.
+type CDXLine struct {
 	Surt        string
 	Date        string
 	Link        string
@@ -23,13 +25,74 @@ type CDX struct {
 	Size        string
 }
 
+type CDX []CDXLine
+
+type ByDate CDX
+
+func (b ByDate) Len() int {
+	return len(b)
+}
+
+func (b ByDate) Swap(i, j int) {
+	b[i], b[j] = b[j], b[i]
+}
+
+func (b ByDate) Less(i, j int) bool {
+	return b[i].Date < b[j].Date
+}
+
+func (c CDX) Summary() string {
+	var (
+		dateLast   = "NA"
+		dateLastOK = "NA"
+		delta      = "NA"
+	)
+	if len(c) == 0 {
+		return fmt.Sprintf("last=%s ok=%s", dateLast, dateLastOK)
+	}
+	sort.Sort(sort.Reverse(ByDate(c)))
+	dateLast = c[0].Date
+	for _, cdx := range c {
+		if cdx.StatusCode == "200" {
+			dateLastOK = cdx.Date
+			break
+		}
+	}
+	d, err := tsDiff(dateLast, dateLastOK)
+	if err == nil {
+		if d.Hours()/24 > 365 {
+			delta = fmt.Sprintf("\033[31;1;4m%0.0f\033[0m", d.Hours()/24)
+		} else {
+			delta = fmt.Sprintf("%0.0f", d.Hours()/24)
+		}
+	}
+	return fmt.Sprintf("last=%s ok=%s delta=%v", dateLast, dateLastOK, delta)
+}
+
+// tsDiff returns the duration between two timestamps, like: 20140304124333.
+func tsDiff(a, b string) (time.Duration, error) {
+	ta, err := time.Parse("20060102150405", a)
+	if err != nil {
+		return 0, err
+	}
+	tb, err := time.Parse("20060102150405", b)
+	if err != nil {
+		return 0, err
+	}
+	if ta.Before(tb) {
+		return tb.Sub(ta), nil
+	} else {
+		return ta.Sub(tb), nil
+	}
+}
+
 // LookupCDX asks CDX API. Result will be like:
 // net,ijmse)/uploadfile/2016/1214/20161214052559646.pdf 20170516210333
 // http://www.ijmse.net:80/uploadfile/2016/1214/20161214052559646.pdf
 // application/pdf 200 PBPHE2OILTB43TAOUO33GBWLE2SS4LQX 2079755
 //
 // Also returns the raw response body.
-func LookupCDX(link string) (result []CDX, b []byte, err error) {
+func LookupCDX(link string) (result CDX, b []byte, err error) {
 	link = prependSchema(link)
 	cdxlink := fmt.Sprintf("%s?url=%s", cdxApi, link)
 	log.Printf("[lookup] %s", cdxlink)
@@ -50,7 +113,7 @@ func LookupCDX(link string) (result []CDX, b []byte, err error) {
 	return result, b, err
 }
 
-func ParseCDX(b []byte) (result []CDX, err error) {
+func ParseCDX(b []byte) (result CDX, err error) {
 	for _, line := range strings.Split(string(b), "\n") {
 		var fields = strings.Fields(line)
 		if len(fields) == 0 {
@@ -60,7 +123,7 @@ func ParseCDX(b []byte) (result []CDX, err error) {
 			log.Printf("short line: %s", line)
 			continue
 		}
-		cdx := CDX{
+		cdx := CDXLine{
 			Surt:        fields[0],
 			Date:        fields[1],
 			Link:        fields[2],