fedilogue/poll/instance.go

package main

import (
	"github.com/microcosm-cc/bluemonday"
	"encoding/json"
	"crypto/sha1"
	"io/ioutil"
	"net/http"
	"strings"
	"html"
	"time"
	"fmt"
	"log"
)

var p *bluemonday.Policy

func PollMastodonPleroma(endpoint string, reportPostChan chan ReportPost) {
	newposts := make([]ReportPost, 0)

	min_id := ""

	http_client := http.Client{}

	for {
		ri_mutex.Lock()
		m := runninginstances[endpoint]
		ri_mutex.Unlock()

		api_timeline := "https://" + endpoint + "/api/v1/timelines/public?limit=40&min_id=" + min_id
		resp, err := http_client.Get(api_timeline)
		m.LastRun = time.Now().Format("2006.01.02-15:04:05")
		if err != nil {
			ri_mutex.Lock()
			m.Status = CLIENT_ISSUE
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			log.Fatal("Failure here", err.Error())
			return
		}

		if resp.StatusCode == UNAUTHORIZED || resp.StatusCode == UNPROCESSABLE_ENTITY {
			// Apparently you just need to do this to throw away the body
			_, _ = ioutil.ReadAll(resp.Body)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = UNAUTHORIZED
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			return
		} else if resp.StatusCode == TOOMANYREQUESTS {
			// Apparently you just need to do this to throw away the body
			_, _ = ioutil.ReadAll(resp.Body)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = TOOMANYREQUESTS
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			time.Sleep(time.Second * 30)
			continue
		} else if resp.StatusCode == FORBIDDEN {
			// Apparently you just need to do this to throw away the body
			_, _ = ioutil.ReadAll(resp.Body)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = FORBIDDEN
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			return
		} else if resp.StatusCode == INTERNAL_ERROR {
			// Apparently you just need to do this to throw away the body
			_, _ = ioutil.ReadAll(resp.Body)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = FORBIDDEN
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			time.Sleep(time.Second * 3600)
			continue
		} else if resp.StatusCode == NOT_FOUND { // 404
			// Apparently you just need to do this to throw away the body
			_, _ = ioutil.ReadAll(resp.Body)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = FORBIDDEN
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			time.Sleep(time.Second * 3600)
			continue
		}

		err = json.NewDecoder(resp.Body).Decode(&newposts)
		if err != nil {
			fmt.Println("-----------------")
			fmt.Println("Error Message 1:", resp.StatusCode, err, err.Error, endpoint, resp.Status)
			fmt.Println("-----------------")
			q, err := ioutil.ReadAll(resp.Body)
			fmt.Println(string(q))
			fmt.Println(q)
			fmt.Println(err)
			resp.Body.Close() // Release as soon as done
			ri_mutex.Lock()
			m.Status = BAD_RESPONSE
			runninginstances[endpoint] = m
			ri_mutex.Unlock()
			log.Fatal("Exiting")
			continue
		}
		resp.Body.Close() // Release as soon as done

		ri_mutex.Lock()
		m.Status = RUNNING
		runninginstances[endpoint] = m
		ri_mutex.Unlock()

		for _, newpost := range newposts {
			if newpost.Account.Acct == "" {
				continue
			}
			posthash := sha1.New()

			at_sign := strings.Index(newpost.Account.Acct, "@")

			if at_sign == -1 {
				at_sign = len(newpost.Account.Acct)
				newpost.Account.Acct += "@" + endpoint
			}

			// Calculate the post hash
			fmt.Fprint(posthash, newpost.Url)
			fmt.Fprint(posthash, newpost.normalized)
			fmt.Fprint(posthash, newpost.Account.Acct)
			fmt.Fprint(posthash, newpost.Account.Display_name)
			newpost.posthash = posthash.Sum(nil)

			newpost.normalized = html.UnescapeString(strings.ToLower(p.Sanitize(newpost.Content)))

			// Validate time
			//t, err := time.Parse("2006-01-02T15:04:05.999Z", newpost.Created_at)
			t, err := time.Parse(time.RFC3339, newpost.Created_at)
			if err != nil {
				log.Print("Time was: " + newpost.Created_at)
				newpost.Created_at = time.Now().Format(time.RFC3339)
				log.Print("Set to  : " + newpost.Created_at)
			}
			if t.Unix() < 0 {
				log.Print("Time was: " + newpost.Created_at)
				newpost.Created_at = time.Now().Format(time.RFC3339)
				log.Print("Set to  : " + newpost.Created_at)
			}

			//t, err = time.Parse("2006-01-02T15:04:05.999Z", newpost.Account.Created_at)
			t, err = time.Parse(time.RFC3339, newpost.Created_at)
			if err != nil {
				log.Print("Time was: " + newpost.Account.Created_at)
				newpost.Account.Created_at = time.Now().Format(time.RFC3339)
				log.Print("Set to  : " + newpost.Account.Created_at)
			}
			if t.Unix() < 0 {
				log.Print("Time was: " + newpost.Account.Created_at)
				newpost.Account.Created_at = time.Now().Format(time.RFC3339)
				log.Print("Set to  : " + newpost.Account.Created_at)
			}

			reportPostChan <- newpost

			// Check min_id
			if newpost.Id > min_id {
				min_id = newpost.Id
			}
			newinstance := newpost.Account.Acct[at_sign+1:]
			ri_mutex.Lock()
			_, exists := runninginstances[newinstance]
			if exists == false {
				m := RunningInstance{}
				runninginstances[newinstance] = m
				go StartInstance(newinstance, reportPostChan)
			}

			ri_mutex.Unlock()
		}
		time.Sleep(time.Second * 10)
	}
}

// Change this to return a proper "err"
func GetNodeInfo(endpoint string) (NodeInfo) {
	/* Checking order
	 * Mastodon/Pleroma
	 * Um..nothing else yet
	*/
	pleromastodon_nodeinfo_url := "https://" + endpoint + "/nodeinfo/2.0.json"
	//http_client := http.Client{Timeout: 10 * time.Second}
	http_client := http.Client{}
	pleromastodon_api_resp, err := http_client.Get(pleromastodon_nodeinfo_url)
	if err != nil {
		return NodeInfo{}
	} else {
		defer pleromastodon_api_resp.Body.Close()
	}

	if pleromastodon_api_resp.StatusCode == 200 {
		var nodeinfo NodeInfo
		err = json.NewDecoder(pleromastodon_api_resp.Body).Decode(&nodeinfo)
		if err != nil {
			log.Println("Was not unmarshalled, continuing...")
		}
		return nodeinfo
	}

	// Check the front page
	index_url := "https://" + endpoint + "/"
	resp_index, err := http_client.Get(index_url)
	if err != nil {
		log.Fatal("Error Message 2b:", resp_index.StatusCode, err, endpoint, resp_index.Status)
	}
	defer resp_index.Body.Close()
	indexbin, err := ioutil.ReadAll(resp_index.Body)
	if err != nil {
		log.Fatal("Error Message 2c:", resp_index.StatusCode, err, endpoint, resp_index.Status)
	}
	indexstr := string(indexbin)
	nodeinfo := NodeInfo{}
	if strings.Contains(indexstr, "Pleroma") || strings.Contains(indexstr, "Soapbox") {
		log.Print("Manual view: Pleroma" + endpoint)
		nodeinfo.Software.Name = "pleroma"
		nodeinfo.Software.Version = "guess"
	} else if strings.Contains(indexstr, "Mastodon") {
		log.Print("Manual view: Mastodon" + endpoint)
		nodeinfo.Software.Name = "mastodon"
		nodeinfo.Software.Version = "guess"
	} else if strings.Contains(indexstr, "Gab") {
		log.Print("Manual view: Gab" + endpoint)
		nodeinfo.Software.Name = "gab"
		nodeinfo.Software.Version = "guess"
	}

	return nodeinfo
}

func StartInstance(endpoint string, reportPostChan chan ReportPost) {
	nodeinfo := GetNodeInfo(endpoint)
	if nodeinfo.Software.Name == "" {
		var m = runninginstances[endpoint]
		m.Software = ""
		m.LastRun = time.Now().Format(time.RFC3339)
		m.Status = UNSUPPORTED_INSTANCE
		ri_mutex.Lock()
		runninginstances[endpoint] = m
		ri_mutex.Unlock()
		return
	}

	if nodeinfo.Software.Name == "pleroma" || nodeinfo.Software.Name == "mastodon" {
		log.Print("Starting " + endpoint + " as Mastodon/Pleroma instance")
		go PollMastodonPleroma(endpoint, reportPostChan)
	}

}