fedilogue/fedilogue/retrieve.go

package main

import (
	"context"
	"encoding/json"
	//"errors"
	"html"
	"io/ioutil"
	"net/http"
	"strings"
	"time"
	"regexp"
	"github.com/microcosm-cc/bluemonday"
)

var p *bluemonday.Policy
var spaceReg *regexp.Regexp
var removeHTMLReg *regexp.Regexp
var re *regexp.Regexp
var matchurl *regexp.Regexp

type ImageType struct {
	Url string `json:"url"`
}

type PublicKeyType struct {
	PublicKeyPem string `json:"publicKeyPem"`
}

type ActorJson struct {
	id                int
	uri               string        `json:"id"`
	Type              string        `json:"type"`
	Inbox             string        `json:"inbox"`
	Outbox            string        `json:"outbox"`
	Followers         string        `json:"followers"`
	Following         string        `json:"following"`
	Url               string        `json:"url"`
	PreferredUsername string        `json:"preferredUsername"`
	Name              string        `json:"name"`
	Summary           string        `json:"summary"`
	Icon              ImageType     `json:"icon"`
	Image             ImageType     `json:"image"`
	PublicKey         PublicKeyType `json:"publicKey"`

	instance string
}

type PostJson struct {
	id            int
	uri       string `json:"id"`
	InReplyTo string `json:"inReplyTo"`

	normalized string
	receivedAt time.Time `json:"created_at"`

	Content      string    `json:"content"`
	Conversation string    `json:"conversation"`
	Published    time.Time `json:"published"`
	Source       string    `json:"source"`
	Summary      string    `json:"summary"`
	//	Ignoring tag for now
	To   []string `json:"to"`
	Type string   `json:"type"`

	Actor        string `json:"actor"`
	AttributedTo string `json:"attributedTo"`

	instance string
}

func check_activity(uri string) {
	var activityjson PostJson

	// Ignore banned
	for _, banned := range settings.Banned {
		if strings.Index(uri, "https://"+banned+"/") == 0 {
			return
		}
	}

	// Ignore invalid URIs
	endslash := strings.Index(uri[8:], "/")
	if endslash == -1 {
		return
	}
	activityjson.instance = uri[8 : endslash+8]

	o, _ := GetRunner(activityjson.instance)

	// Check if there were any recent requests on this
	o.recentactivities.Mu.Lock()
	if o.recentactivities.Add(uri) == true {
		o.recentactivities.Mu.Unlock()
		return
	}

	o.recentactivities.Mu.Unlock()
	var jsondocument string

	selectRet := pool.QueryRow(context.Background(), "SELECT FROM activities WHERE document->>'id' = $1", uri)
	err := selectRet.Scan()
	if err == nil {
		return
	}

	req, _ := http.NewRequest("GET", uri, nil)
	req.Header.Set("User-Agent", "Tusky")
	req.Header.Add("Accept", "application/ld+json")

	resp, err := DoTries(&o, req)
	if err != nil {
		return
	}

	body, err := ioutil.ReadAll(resp.Body)
	if err != nil {
		return
	}
	resp.Body.Close()
	jsondocument = string(body)
	err = json.Unmarshal(body, &activityjson)
	if err != nil {
		return
	}

	if activityjson.InReplyTo != "" && activityjson.InReplyTo != uri {
		if activityjson.InReplyTo != uri {
			go check_activity(activityjson.InReplyTo)
		}
	}

	// If AttributedTo is blank, this is likely an authentication failure
	// For now, skip it...
	if activityjson.AttributedTo == "" {
		return
	}

	// This must be done BEFORE the `INSERT INTO activities'` below
	go check_actor(activityjson.AttributedTo)

	activityjson.normalized = removeHTMLReg.ReplaceAllString(activityjson.Content, " ")
	activityjson.normalized = html.UnescapeString(strings.ToLower(p.Sanitize(activityjson.normalized)))
	activityjson.normalized = matchurl.ReplaceAllString(activityjson.normalized, "")
	activityjson.normalized = spaceReg.ReplaceAllString(activityjson.normalized, " ")

	_, err = pool.Exec(context.Background(), "INSERT INTO activities (document, normalized, instance) VALUES($1, $2, $3)", jsondocument, activityjson.normalized, activityjson.instance)
	if err != nil {
		logWarn("Error inserting %s into `activities`: "+ uri, err)
		return
	}

	for _, to := range activityjson.To {
		if to != "https://www.w3.org/ns/activitystreams#Public" && to != "" {
			if strings.HasSuffix(to, "/followers") == true {
				// This check is very much a bad solution, may consider removing the entire for-loop
				continue
			}
			go check_actor(to)
		}
	}

}

/* Test: TestCheck_actor */
func check_actor(uri string) int {
	var actorjson ActorJson

	if len(uri) <= 7 {
		return 400 // Bad actor
	}

	endslash := strings.Index(uri[8:], "/")
	if endslash == -1 {
		return 400 // Bad actor
	}
	actorjson.instance = uri[8 : endslash+8]
	for _, banned := range settings.Banned {
		if strings.Index(uri, "https://"+banned+"/") == 0 {
			return 401 // Banned actor
		}
	}

	// Check if there were any recent requests on this
	o, _ := GetRunner(actorjson.instance)
	o.recentactors.Mu.Lock()
	if o.recentactors.Add(uri) == true {
		o.recentactors.Mu.Unlock()
		return 402 // Actor in recent queue, good!
	}
	o.recentactors.Mu.Unlock()

	selectRet := pool.QueryRow(context.Background(), "SELECT FROM actors WHERE document->>'id' = $1", uri)
	err := selectRet.Scan()
	if err == nil {
		return 403 // Actor already in database, good!
	}

	req, _ := http.NewRequest("GET", uri, nil)
	req.Header.Set("User-Agent", "Tusky")
	req.Header.Add("Accept", "application/ld+json")

	var resp *http.Response
	tries := 0
	for {
		resp, err = o.client.Do(req)
		if err != nil {
			if tries > 10 {
				logErr("Unable to connect to "+uri+" attempt 10/10, giving up.")
				return 404 // Unable to connect to host after 10 attempts
			}
			logWarn("Unable to connect to "+uri+", attempt ",tries+1,"+/10 sleeping for 30 seconds.")
			time.Sleep(time.Second * 30)
			tries = tries + 1
			continue
		}
		break
	}

	body, err := ioutil.ReadAll(resp.Body)
	if err != nil {
		return 405 // Unable to read body of message
	}
	resp.Body.Close()

	jsondocument := string(body)

	err = json.Unmarshal(body, &actorjson)
	if err != nil {
		return 406 // Unable to unmarshal body of message
	}

	_, err = pool.Exec(context.Background(), "INSERT INTO actors (document, instance) VALUES($1, $2)", jsondocument, actorjson.instance)
	if err != nil {
		logWarn("Error inserting %s into `actors`: "+uri, err)
		return 407 // Unable to insert actor
	}

	return 0 // Successful
}