luftdaten-prom/main.go

package main

import (
	"flag"
	"fmt"
	"net/http"
	"strconv"
	"strings"
	"time"

	"github.com/golang/glog"
	"golang.org/x/net/html"
)

var (
	flagLuftdaten      string
	flagScrapeInterval int
	flagListen         string
)

func init() {
	flag.Set("logtostderr", "true")
}

type sensorData struct {
	// Whether the data is fresh.
	stale bool
	// PM2.5 in ug/m3
	pm25 float64
	// PM10 in ug/m3
	pm10 float64
	// Temperature in degrees centigrade
	temperature float64
	// Humidity in percent
	humidity float64
}

func getSensorData() (sensorData, error) {
	data := sensorData{}

	resp, err := http.Get(flagLuftdaten)
	if err != nil {
		return data, fmt.Errorf("when GETing data: %v", err)
	}

	root, err := html.Parse(resp.Body)
	if err != nil {
		return data, fmt.Errorf("when parsing HTML: %v", err)
	}

	// Find html.
	htmlN := root.FirstChild
	if htmlN == nil || htmlN.Type != html.ElementNode || htmlN.Data != "html" {
		return data, fmt.Errorf("unexpected root node")
	}
	// Find body.
	bodyN := htmlN.FirstChild.NextSibling
	if bodyN == nil || bodyN.Type != html.ElementNode || bodyN.Data != "body" {
		return data, fmt.Errorf("unexpected html[1] node")
	}
	// Find div class=content
	contentN := bodyN.FirstChild.NextSibling
	if contentN == nil || contentN.Type != html.ElementNode || contentN.Data != "div" || contentN.Attr[0].Key != "class" || contentN.Attr[0].Val != "content" {
		return data, fmt.Errorf("unexpected body[1] node")
	}
	// Find table.
	var tableN *html.Node = nil
	for c := contentN.FirstChild; c != nil; c = c.NextSibling {
		if c.Type == html.ElementNode && c.Data == "table" {
			tableN = c
			break
		}
	}
	if tableN == nil {
		return data, fmt.Errorf("could not find table")
	}
	// Find tbody.
	tbodyN := tableN.FirstChild
	if tbodyN == nil || tbodyN.Type != html.ElementNode || tbodyN.Data != "tbody" {
		return data, fmt.Errorf("unexpected table[0] node")
	}

	got := make(map[string]bool)
	// Iterate throught rows.
	for row := tbodyN.FirstChild; row != nil; row = row.NextSibling {
		td1 := row.FirstChild
		if td1 == nil {
			continue
		}
		td2 := td1.NextSibling
		if td2 == nil {
			continue
		}
		td3 := td2.NextSibling
		if td3 == nil {
			continue
		}
		if td1.Data != "td" || td2.Data != "td" || td3.Data != "td" {
			continue
		}

		if td2.FirstChild.Data == "PM2.5" {
			// This is not an ASCII space.
			parts := strings.Split(td3.FirstChild.Data, " ")
			if len(parts) != 2 {
				continue
			}
			pm25, err := strconv.ParseFloat(parts[0], 64)
			if err != nil {
				continue
			}
			data.pm25 = pm25
			got["PM2.5"] = true
		}
		if td2.FirstChild.Data == "PM10" {
			// This is not an ASCII space.
			parts := strings.Split(td3.FirstChild.Data, " ")
			if len(parts) != 2 {
				continue
			}
			pm10, err := strconv.ParseFloat(parts[0], 64)
			if err != nil {
				continue
			}
			data.pm10 = pm10
			got["PM10"] = true
		}
		if td2.FirstChild.Data == "Temperatur" {
			// This is not an ASCII space.
			parts := strings.Split(td3.FirstChild.Data, " ")
			if len(parts) != 2 {
				continue
			}
			temp, err := strconv.ParseFloat(parts[0], 64)
			if err != nil {
				continue
			}
			data.temperature = temp
			got["Temperature"] = true
		}
		if td2.FirstChild.Data == "rel. Luftfeuchte" {
			// This is not an ASCII space.
			parts := strings.Split(td3.FirstChild.Data, " ")
			if len(parts) != 2 {
				continue
			}
			hum, err := strconv.ParseFloat(parts[0], 64)
			if err != nil {
				continue
			}
			data.humidity = hum
			got["Humidity"] = true
		}
	}

	for _, want := range []string{"PM2.5", "PM10", "Temperature", "Humidity"} {
		if !got[want] {
			return data, fmt.Errorf("could not find %v in HTML", want)
		}
	}
	return data, nil
}

func scraper(req chan int, resp chan sensorData) error {
	data := sensorData{
		stale: true,
	}

	feed := func() {
		newData, err := getSensorData()
		if err != nil {
			glog.Error("Could not get sensor data: %v", err)
			data.stale = true
		} else {
			data = newData
			glog.Errorf("Got sensor data: %f/%f/%f/%f", data.pm25, data.pm10, data.temperature, data.humidity)
		}
	}

	feed()

	ticker := time.NewTicker(time.Duration(flagScrapeInterval) * time.Second)
	for {
		select {
		case <-req:
			resp <- data
		case <-ticker.C:
			feed()
		}
	}
}

func main() {
	flag.StringVar(&flagLuftdaten, "luftdaten", "http://127.0.0.1:2140/values", "Luftdaten sensor endpoint")
	flag.IntVar(&flagScrapeInterval, "scrape_interval", 60, "Scrape interval of Luftdaten, in seconds")
	flag.StringVar(&flagListen, "listen", "0.0.0.0:2141", "Address to listen at for Prometheus requests")
	flag.Parse()

	glog.Info("Starting scraper...")
	req := make(chan int, 1)
	res := make(chan sensorData, 1)
	go scraper(req, res)

	http.HandleFunc("/metrics", func(w http.ResponseWriter, r *http.Request) {
		req <- 0
		data := <-res
		if data.stale {
			fmt.Fprintf(w, "# data is stale :(\n")
			return
		}

		fmt.Fprintf(w, "# HELP pm25 The current PM2.5 concentration in micrograms per cubic meter\n")
		fmt.Fprintf(w, "# TYPE pm25 gauge\n")
		fmt.Fprintf(w, "pm25 %f\n", data.pm25)
		fmt.Fprintf(w, "# HELP pm10 The current PM10 concentration in micrograms per cubic meter\n")
		fmt.Fprintf(w, "# TYPE pm10 gauge\n")
		fmt.Fprintf(w, "pm10 %f\n", data.pm10)
		fmt.Fprintf(w, "# HELP temperature The current temperature in degrees centigrate\n")
		fmt.Fprintf(w, "# TYPE temperature gauge\n")
		fmt.Fprintf(w, "temperature %f\n", data.temperature)
		fmt.Fprintf(w, "# HELP humidity The current relative humidity in percent\n")
		fmt.Fprintf(w, "# TYPE humidity gauge\n")
		fmt.Fprintf(w, "humidity %f\n", data.humidity)
	})
	http.ListenAndServe(flagListen, nil)
}