From 5f00e5efb66b95c42f7fdbf9542e8aef15761abc Mon Sep 17 00:00:00 2001
From: Christine Dodrill <me@christine.website>
Date: Wed, 25 Jan 2017 10:40:15 -0800
Subject: [PATCH] initial commit

---
 .gitignore   |   4 ++
 main.go      | 175 +++++++++++++++++++++++++++++++++++++++++++++++++++
 main_test.go |  37 +++++++++++
 3 files changed, 216 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 main.go
 create mode 100644 main_test.go

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..4e736c8
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+their
+*.log
+*.csv
+lines.txt
diff --git a/main.go b/main.go
new file mode 100644
index 0000000..6e9edf0
--- /dev/null
+++ b/main.go
@@ -0,0 +1,175 @@
+package main
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"log"
+	"os"
+	"sort"
+	"time"
+)
+
+type LogLine struct {
+	Time     time.Time
+	TimeTook time.Duration
+	Host     string
+}
+
+type Bucket struct {
+	Time    time.Time // only resolved to the minute
+	Entries []*LogLine
+}
+
+const dateFormat = "2006-01-02T15:04:05.999999Z07:00"
+const outDateFormat = "2006-01-02T15:04:05"
+const outDateFormatNoSeconds = "2006-01-02T15:04"
+
+func ParseLogLine(line []byte) (*LogLine, error) {
+	sl := bytes.Split(line, []byte(" "))
+
+	date := sl[0]
+	lTime, err := time.Parse(dateFormat, string(date))
+	if err != nil {
+		return nil, err
+	}
+
+	var sdur time.Duration
+	var host string
+
+	for i, section := range sl {
+		switch i {
+		case 0, 1:
+			continue // not key->value pairs
+		}
+
+		set := bytes.Split(section, []byte("="))
+		if len(set) != 2 {
+			log.Printf("invalid: %v", set)
+			continue
+		}
+
+		k := string(set[0])
+		v := set[1]
+
+		if k == "service" {
+			dur, err := time.ParseDuration(string(v))
+			if err != nil {
+				return nil, err
+			}
+
+			sdur = dur
+		}
+
+		if k == "host" {
+			host = Shuck(string(v))
+		}
+	}
+
+	ll := &LogLine{
+		Time:     lTime,
+		TimeTook: sdur,
+		Host:     host,
+	}
+
+	return ll, nil
+}
+
+// Shuck removes the first and last character of a string, analogous to
+// shucking off the husk of an ear of corn.
+func Shuck(victim string) string {
+	return victim[1 : len(victim)-1]
+}
+
+func main() {
+	scanner := bufio.NewScanner(os.Stdin)
+
+	var lastTime time.Time
+
+	// active buckets, short var name because it's going to be referenced a lot
+	ab := map[string]*Bucket{}
+
+	for scanner.Scan() {
+		line := scanner.Bytes()
+
+		ll, err := ParseLogLine(line)
+		if err != nil {
+			log.Fatal(err)
+		}
+
+		if lastTime.IsZero() {
+			year, month, day := ll.Time.Date()
+			hour, minute, _ := ll.Time.Clock()
+			lastTime = time.Date(year, month, day, hour, minute, 0, 0, time.UTC)
+		}
+
+		// last line minutes
+		_, llm, _ := ll.Time.Clock()
+		_, ltm, _ := lastTime.Clock()
+
+		if llm != ltm {
+			processBuckets(ab)
+			ab = map[string]*Bucket{}
+
+			lastTime = lastTime.Add(5 * time.Minute)
+		}
+
+		b := ab[ll.Host]
+		if b == nil {
+			year, month, day := ll.Time.Date()
+			hour, minute, _ := ll.Time.Clock()
+			b = &Bucket{
+				Time: time.Date(year, month, day, hour, minute, 0, 0, time.UTC),
+			}
+			ab[ll.Host] = b
+		}
+
+		b.Entries = append(b.Entries, ll)
+
+		ab[ll.Host] = b
+	}
+
+	processBuckets(ab)
+}
+
+// sees if the minute and hour field of a == the minute and hour field of b
+func cmpTime(a, b time.Time) bool {
+	_, am, _ := a.Clock()
+	_, bm, _ := b.Clock()
+
+	return am == bm
+}
+
+func toMS(dur time.Duration) int64 {
+	return dur.Nanoseconds() / 1000000
+}
+
+func processBuckets(set map[string]*Bucket) {
+	hosts := []string{}
+	for host, _ := range set {
+		hosts = append(hosts, host)
+	}
+
+	sort.Sort(sort.StringSlice(hosts))
+
+	for _, host := range hosts {
+		bucket := set[host]
+		var longest time.Duration
+		var shortest time.Duration
+		shortest = 9999 * time.Minute // make this absurdly large so everything is smaller
+		var total time.Duration
+
+		for _, entry := range bucket.Entries {
+			switch true {
+			case entry.TimeTook > longest:
+				longest = entry.TimeTook
+			case entry.TimeTook < shortest:
+				shortest = entry.TimeTook
+			}
+
+			total += entry.TimeTook
+		}
+
+		fmt.Printf("%s,%s,%d,%d,%d,%d\n", bucket.Time.Format(outDateFormat), host, len(bucket.Entries), toMS(total), toMS(shortest), toMS(longest))
+	}
+}
diff --git a/main_test.go b/main_test.go
new file mode 100644
index 0000000..555a08c
--- /dev/null
+++ b/main_test.go
@@ -0,0 +1,37 @@
+package main
+
+import (
+	"testing"
+	"time"
+
+	"github.com/kr/pretty"
+)
+
+func TestParseLogLine(t *testing.T) {
+	const line = `2016-05-07T09:07:00.001490+00:00 heroku[router]: at=info method=GET path="/blog" host="brs.org" request_id=fc693802-8851-484e-aab4-1d013714b68b fwd="10.29.10.29" dyno=web.3 connect=2ms service=994ms status=200 bytes=552`
+
+	ll, err := ParseLogLine([]byte(line))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	ms := (ll.TimeTook.Nanoseconds() / 1000000) // ns -> ms == div 1e6
+	if ms != 996 {
+		t.Fatalf("invalid time took for this line")
+	}
+
+	if ll.Host != "brs.org" {
+		pretty.Println(ll)
+
+		t.Fatal("invalid host")
+	}
+}
+
+func TestCmpTime(t *testing.T) {
+	now := time.Date(2016, time.January, 1, 13, 37, 0, 0, time.UTC)
+	then := now.Add(5 * time.Minute)
+
+	if cmpTime(then, now) {
+		t.Fatal("cmpTime error")
+	}
+}