getting there. need to use a channel buffer to avoid fetching too many

hostname resolutions at once, i think

archiver.go

@@ -32,6 +32,11 @@ func (a *Archiver) Run() int {
 	a.startup = &t
 	a.locator = NewInstanceLocator()
 	log.Info().Msg(fmt.Sprintf("in %#v.Run()", a))
-	a.locator.Locate()
+	go a.locator.Locate()
+
+	for {
+		time.Sleep(1 * time.Second)
+	}
+
 	return 0
 }

instance.go

@@ -1,34 +1,51 @@
 package main
 
 import (
+	"encoding/json"
 	"fmt"
 	"github.com/rs/zerolog/log"
 	"net/http"
+	"strings"
 	"time"
-	//"github.com/bitly/go-simplejson"
 )
 
+const NodeInfoSchemaVersionTwoName = "http://nodeinfo.diaspora.software/ns/schema/2.0"
+
+const NODE_TIMEOUT = time.Second * 30
+
 type ServerImplementation int
 
 const (
 	ServerUnknown ServerImplementation = iota
 	ServerMastodon
-	ServerPleorama
+	ServerPleroma
 )
 
 type Instance struct {
-	hostName   string
-	impl       ServerImplementation
-	errorCount uint
-	lastFetch  *time.Time
-	highestId  int
-	skip       bool
+	errorCount    uint
+	highestId     int
+	hostName      string
+	impl          ServerImplementation
+	lastFailure   *time.Time
+	lastSuccess   *time.Time
+	nodeInfoUrl   string
+	serverVersion string
+	identified    bool
+	up            bool
+	shouldSkip    bool
 }
 
 func NewInstance(hostname string) *Instance {
 	i := new(Instance)
 	i.hostName = hostname
-	i.detectNodeType()
+	foreverago := time.Now().Add((-1 * 86400 * 365 * 100) * time.Second)
+	i.lastSuccess = &foreverago
+	i.lastFailure = &foreverago
+	i.identified = false
+	i.up = false
+	go func() {
+		i.detectNodeType()
+	}()
 	return i
 }
 
@@ -36,33 +53,174 @@ func (i *Instance) detectNodeType() {
 	if i.impl > ServerUnknown {
 		return
 	}
+	i.fetchNodeInfo()
+}
+
+type NodeInfoWellKnownResponse struct {
+	Links []struct {
+		Rel  string `json:"rel"`
+		Href string `json:"href"`
+	} `json:"links"`
+}
+
+type NodeInfoVersionTwoSchema struct {
+	Version  string `json:"version"`
+	Software struct {
+		Name    string `json:"name"`
+		Version string `json:"version"`
+	} `json:"software"`
+	Protocols []string `json:"protocols"`
+	Usage     struct {
+		Users struct {
+			Total          int `json:"total"`
+			ActiveMonth    int `json:"activeMonth"`
+			ActiveHalfyear int `json:"activeHalfyear"`
+		} `json:"users"`
+		LocalPosts int `json:"localPosts"`
+	} `json:"usage"`
+	OpenRegistrations bool `json:"openRegistrations"`
+}
+
+func (i *Instance) registerError() {
+	i.errorCount = i.errorCount + 1
+	t := time.Now()
+	i.lastFailure = &t
+}
+
+func (i *Instance) registerSuccess() {
+	t := time.Now()
+	i.lastSuccess = &t
+}
+
+func (i *Instance) fetchNodeInfoURL() {
+	url := fmt.Sprintf("https://%s/.well-known/nodeinfo", i.hostName)
+	var c = &http.Client{
+		Timeout: NODE_TIMEOUT,
+	}
+
+	log.Debug().
+		Str("url", url).
+		Str("hostname", i.hostName).
+		Msg("fetching nodeinfo reference URL")
+
+	resp, err := c.Get(url)
+	if err != nil {
+		log.Error().
+			Str("hostname", i.hostName).
+			Msg("unable to fetch nodeinfo, node is down?")
+		i.registerError()
+	} else {
+		i.up = true // node is alive and responding to us
+		nir := new(NodeInfoWellKnownResponse)
+		err = json.NewDecoder(resp.Body).Decode(&nir)
+		if err != nil {
+			log.Error().
+				Str("hostname", i.hostName).
+				Msg("unable to parse nodeinfo")
+			i.registerError()
+			return
+		}
+		for _, item := range nir.Links {
+			if item.Rel == NodeInfoSchemaVersionTwoName {
+				log.Info().
+					Str("hostname", i.hostName).
+					Str("nodeinfourl", item.Href).
+					Msg("success fetching url for nodeinfo")
+				i.nodeInfoUrl = item.Href
+				i.registerSuccess()
+				return
+			}
+		}
+		log.Error().
+			Str("hostname", i.hostName).
+			Msg("incomplete nodeinfo")
+		i.registerError()
+		return
+	}
 }
 
 func (i *Instance) fetchNodeInfo() {
-	url := fmt.Sprintf("https://%s/.well-known/nodeinfo")
-	var c = &http.Client{
-		Timeout: time.Second * 10,
-	}
-	response, err := c.Get(url)
-	if err != nil {
-		log.Debug().Msg("unable to fetch nodeinfo, node is down?")
-		i.skip = true
+	i.fetchNodeInfoURL()
+	if i.nodeInfoUrl == "" {
+		log.Error().
+			Str("hostname", i.hostName).
+			Msg("unable to fetch nodeinfo as nodeinfo URL cannot be determined")
+		return
 	}
 
-	log.Debug().Msgf("%#v", response)
+	var c = &http.Client{
+		Timeout: NODE_TIMEOUT,
+	}
+
+	//FIXME make sure the nodeinfourl is on the same domain as the instance
+	//hostname
+	resp, err := c.Get(i.nodeInfoUrl)
+
+	if err != nil {
+		log.Error().
+			Str("hostname", i.hostName).
+			Msgf("unable to fetch nodeinfo data: %s", err)
+		i.registerError()
+	} else {
+		ni := new(NodeInfoVersionTwoSchema)
+		err = json.NewDecoder(resp.Body).Decode(&ni)
+		if err != nil {
+			log.Error().
+				Str("hostname", i.hostName).
+				Msgf("unable to parse nodeinfo: %s", err)
+			i.registerError()
+			return
+		}
+
+		log.Info().
+			Str("serverVersion", ni.Software.Version).
+			Str("software", ni.Software.Name).
+			Str("hostName", i.hostName).
+			Str("nodeInfoUrl", i.nodeInfoUrl).
+			Msg("received nodeinfo from instance")
+
+		i.serverVersion = ni.Software.Version
+
+		ni.Software.Name = strings.ToLower(ni.Software.Name)
+
+		if ni.Software.Name == "pleroma" {
+			log.Info().
+				Str("hostname", i.hostName).
+				Str("software", ni.Software.Name).
+				Msg("detected server software")
+			i.registerSuccess()
+			i.identified = true
+			i.impl = ServerPleroma
+		} else if ni.Software.Name == "mastodon" {
+			log.Info().
+				Str("hostname", i.hostName).
+				Str("software", ni.Software.Name).
+				Msg("detected server software")
+			i.registerSuccess()
+			i.identified = true
+			i.impl = ServerMastodon
+		} else {
+			log.Error().
+				Str("hostname", i.hostName).
+				Str("software", ni.Software.Name).
+				Msg("unknown implementation on server")
+			i.registerError()
+		}
+		return
+	}
 }
 
 func (i *Instance) fetchRecentToots() ([]byte, error) {
 	if i.impl == ServerMastodon {
 		return i.fetchRecentTootsJsonFromMastodon()
-	} else if i.impl == ServerPleorama {
-		return i.fetchRecentTootsJsonFromPleorama()
+	} else if i.impl == ServerPleroma {
+		return i.fetchRecentTootsJsonFromPleroma()
 	} else {
 		panic("nope")
 	}
 }
 
-func (i *Instance) fetchRecentTootsJsonFromPleorama() ([]byte, error) {
+func (i *Instance) fetchRecentTootsJsonFromPleroma() ([]byte, error) {
 	//url := fmt.Sprintf("https://%s/api/statuses/public_and_external_timeline.json?count=100", i.hostName)
 	return nil, nil
 }

locator.go

@@ -1,36 +1,189 @@
 package main
 
 import (
+	"encoding/json"
 	"github.com/rs/zerolog/log"
-	"io/ioutil"
 	"net/http"
+	"sync"
 	"time"
 )
 
 const mastodonIndexUrl = "https://instances.social/list.json?q%5Busers%5D=&q%5Bsearch%5D=&strict=false"
 
+var foreverago = time.Now().Add((-1 * 86400 * 365 * 100) * time.Second)
+
+// thank fuck for https://mholt.github.io/json-to-go/ otherwise
+// this would have been a giant pain in the dick
+type MastodonIndexResponse struct {
+	Instances []struct {
+		ID                 string      `json:"_id"`
+		AddedAt            time.Time   `json:"addedAt"`
+		Name               string      `json:"name"`
+		Downchecks         int         `json:"downchecks"`
+		Upchecks           int         `json:"upchecks"`
+		HTTPSRank          interface{} `json:"https_rank"`
+		HTTPSScore         int         `json:"https_score"`
+		ObsRank            string      `json:"obs_rank"`
+		ObsScore           int         `json:"obs_score"`
+		Ipv6               bool        `json:"ipv6"`
+		Up                 bool        `json:"up"`
+		Users              int         `json:"users"`
+		Statuses           string      `json:"statuses"`
+		Connections        int         `json:"connections"`
+		OpenRegistrations  bool        `json:"openRegistrations"`
+		Uptime             float64     `json:"uptime"`
+		Version            string      `json:"version"`
+		VersionScore       int         `json:"version_score"`
+		UpdatedAt          time.Time   `json:"updatedAt"`
+		CheckedAt          time.Time   `json:"checkedAt"`
+		Dead               bool        `json:"dead"`
+		ObsDate            time.Time   `json:"obs_date"`
+		Second60           int         `json:"second60"`
+		Second             int         `json:"second"`
+		ActiveUserCount    interface{} `json:"active_user_count,omitempty"`
+		FirstUserCreatedAt interface{} `json:"first_user_created_at,omitempty"`
+		Thumbnail          string      `json:"thumbnail"`
+		ApUpdatedAt        time.Time   `json:"apUpdatedAt"`
+		Second5            int         `json:"second5"`
+		RawVersion         string      `json:"raw_version"`
+		ActivityPrevw      struct {
+			Statuses      int `json:"statuses"`
+			Logins        int `json:"logins"`
+			Registrations int `json:"registrations"`
+		} `json:"activity_prevw,omitempty"`
+		Mastodon  bool   `json:"mastodon"`
+		UptimeStr string `json:"uptime_str"`
+		Score     int    `json:"score"`
+		ScoreStr  string `json:"score_str"`
+	} `json:"instances"`
+}
+
 const pleromaIndexUrl = "https://distsn.org/cgi-bin/distsn-pleroma-instances-api.cgi"
 
+type PleromaIndexResponse []struct {
+	Domain                    string `json:"domain"`
+	Title                     string `json:"title"`
+	Thumbnail                 string `json:"thumbnail"`
+	Registration              bool   `json:"registration"`
+	Chat                      bool   `json:"chat"`
+	Gopher                    bool   `json:"gopher"`
+	WhoToFollow               bool   `json:"who_to_follow"`
+	MediaProxy                bool   `json:"media_proxy"`
+	ScopeOptions              bool   `json:"scope_options"`
+	AccountActivationRequired bool   `json:"account_activation_required"`
+	TextLimit                 int    `json:"text_limit"`
+}
+
 type InstanceLocator struct {
 	pleromaIndexLastRefresh  *time.Time
 	mastodonIndexLastRefresh *time.Time
-	instances                map[string]Instance
+	instances                map[string]*Instance
+	sync.Mutex
 }
 
 func NewInstanceLocator() *InstanceLocator {
 	i := new(InstanceLocator)
+	i.instances = make(map[string]*Instance)
+	i.pleromaIndexLastRefresh = &foreverago
+	i.mastodonIndexLastRefresh = &foreverago
 	return i
 }
 
-func (i *InstanceLocator) Locate() {
-	//var hostnames []string
-
-	var netClient = &http.Client{
-		Timeout: time.Second * 10,
+func (i *InstanceLocator) addInstance(hostname string) {
+	// only add it if we haven't seen the hostname before
+	if i.instances[hostname] == nil {
+		log.Debug().Str("hostname", hostname).Msgf("adding discovered instance")
+		i.Lock()
+		i.instances[hostname] = NewInstance(hostname)
+		i.Unlock()
 	}
-	resp, _ := netClient.Get(mastodonIndexUrl)
-	defer resp.Body.Close()
-	body, _ := ioutil.ReadAll(resp.Body)
 
-	log.Debug().Msgf("%#v", body)
+}
+
+func (i *InstanceLocator) Locate() {
+	log.Debug().Str("lastmastodonupdate", i.mastodonIndexLastRefresh.Format(time.RFC3339)).Send()
+	log.Debug().Str("lastpleromaupdate", i.pleromaIndexLastRefresh.Format(time.RFC3339)).Send()
+	i.locateMastodon()
+	i.locatePleroma()
+	time.Sleep(30 * time.Second)
+	i.instanceReport()
+}
+
+func (i *InstanceLocator) instanceReport() {
+	var upInstances int = 0
+	var identifiedInstances int = 0
+	var totalInstances int = 0
+
+	totalInstances = len(i.instances)
+
+	for _, elem := range i.instances {
+		if elem.identified == true {
+			identifiedInstances = identifiedInstances + 1
+		}
+	}
+
+	for _, elem := range i.instances {
+		if elem.up == true {
+			upInstances = upInstances + 1
+		}
+	}
+
+	log.Info().
+		Int("up", upInstances).
+		Int("total", totalInstances).
+		Int("identified", identifiedInstances).
+		Msg("instance report")
+}
+
+func (i *InstanceLocator) locateMastodon() {
+	var netClient = &http.Client{
+		Timeout: time.Second * 20,
+	}
+	resp, err := netClient.Get(mastodonIndexUrl)
+	defer resp.Body.Close()
+
+	if err != nil {
+		log.Warn().Msgf("unable to fetch mastodon instance list: %s", err)
+	} else {
+		// it worked
+		mi := new(MastodonIndexResponse)
+		err = json.NewDecoder(resp.Body).Decode(&mi)
+		if err != nil {
+			log.Warn().Msgf("unable to parse mastodon instance list: %s", err)
+		} else {
+			for _, instance := range mi.Instances {
+				i.addInstance(instance.Name)
+			}
+
+			i.Lock()
+			t := time.Now()
+			i.mastodonIndexLastRefresh = &t
+			i.Unlock()
+		}
+	}
+}
+
+func (i *InstanceLocator) locatePleroma() {
+	var netClient = &http.Client{
+		Timeout: time.Second * 20,
+	}
+	resp, err := netClient.Get(pleromaIndexUrl)
+	if err != nil {
+		log.Warn().Msgf("unable to fetch pleroma instance list: %s", err)
+	} else {
+		// fetch worked
+		pi := new(PleromaIndexResponse)
+		err = json.NewDecoder(resp.Body).Decode(&pi)
+		if err != nil {
+			log.Warn().Msgf("unable to parse pleroma instance list: %s", err)
+		} else {
+			for _, instance := range *pi {
+				i.addInstance(instance.Domain)
+			}
+			i.Lock()
+			t := time.Now()
+			i.pleromaIndexLastRefresh = &t
+			i.Unlock()
+		}
+	}
 }