logging and cleanup

This commit is contained in:
David Janowski 2023-05-24 23:11:49 +02:00
parent 690201433d
commit 8113adba53

View File

@ -3,6 +3,7 @@ package services
import (
"context"
"e621_to_neo4j/e621"
"e621_to_neo4j/e621/models"
"e621_to_neo4j/neo4jAPI"
"e621_to_neo4j/utils"
"github.com/neo4j/neo4j-go-driver/v5/neo4j"
@ -18,136 +19,223 @@ func ScrapeUser(ctx context.Context, driver neo4j.DriverWithContext, e621Client
return err
}
if e621User.IsBanned {
log.Printf("User %s is banned from e621!", e621User.Name)
return nil
}
log.Printf("Processing user: %s with id %d", e621User.Name, e621User.ID)
err = neo4jAPI.CreateUserNode(ctx, driver, e621User)
if err != nil {
log.Fatal(err)
}
log.Printf("Getting favorites for user %s with id %d", e621User.Name, e621User.ID)
start := time.Now()
userFavorites, err := e621Client.GetFavorites(e621User.Name)
if err != nil {
log.Fatal(err)
}
log.Printf("User %s with id %d has %d favorites. Time took to scrape: %v", e621User.Name, e621User.ID, len(userFavorites), time.Since(start))
startUploadPosts := time.Now()
// Uploads all Tags, Posts as Nodes to Neo4j
for i, post := range userFavorites {
start = time.Now()
err = uploadNodes(ctx, driver, post)
if err != nil {
return err
}
log.Printf("Uploading post for user %s with id %d, %d of %d with ID: %d took: %v", e621User.Name, e621User.ID, i, len(userFavorites), post.ID, time.Since(start))
}
// Makes relationships between different nodes
for i, post := range userFavorites {
start := time.Now()
uniqueGeneralTags := make([]string, 0)
uniqueCharacterTags := make([]string, 0)
uniqueCopyrightTags := make([]string, 0)
uniqueArtistTags := make([]string, 0)
allGeneralTags := make([]string, 0)
allCharacterTags := make([]string, 0)
allCopyrightTags := make([]string, 0)
allArtistTags := make([]string, 0)
allGeneralTags = append(allGeneralTags, post.Tags.General...)
allCharacterTags = append(allCharacterTags, post.Tags.Character...)
allCopyrightTags = append(allCopyrightTags, post.Tags.Character...)
allArtistTags = append(allArtistTags, post.Tags.Artist...)
uniqueGeneralTags = utils.UniqueNonEmptyElementsOf(allGeneralTags)
uniqueCharacterTags = utils.UniqueNonEmptyElementsOf(allCharacterTags)
uniqueCopyrightTags = utils.UniqueNonEmptyElementsOf(allCopyrightTags)
uniqueArtistTags = utils.UniqueNonEmptyElementsOf(allArtistTags)
// Uploads post to database
err := neo4jAPI.CreatePostNode(ctx, driver, post.ID)
err = uploadPostToUserRelationship(ctx, driver, post, e621User)
if err != nil {
log.Fatal(err)
return err
}
// Uploads the source to the database
for _, source := range post.Sources {
err := neo4jAPI.CreateSourceNode(ctx, driver, source)
if err != nil {
return err
}
}
for _, uniqueGeneralTag := range uniqueGeneralTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueGeneralTag, "general")
if err != nil {
return err
}
}
for _, uniqueCharacterTag := range uniqueCharacterTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueCharacterTag, "character")
if err != nil {
return err
}
}
for _, uniqueCopyrightTag := range uniqueCopyrightTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueCopyrightTag, "copyright")
if err != nil {
return err
}
}
for _, uniqueArtistTag := range uniqueArtistTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueArtistTag, "artist")
if err != nil {
return err
}
}
log.Printf("Uploading post %d of %d with ID: %d took: %v", i, len(userFavorites), post.ID, time.Since(start))
}
for _, post := range userFavorites {
id := post.ID
err := neo4jAPI.UserToPostRelationship(ctx, driver, post.ID, e621User.ID)
err = uploadSourceTagRelationship(ctx, driver, post)
if err != nil {
log.Fatal(err)
return err
}
log.Printf("Created UserToPostRelationship for user: %s to post: %d", username, post.ID)
postID := post.ID
for _, source := range post.Sources {
err := neo4jAPI.PostToSourceRelationship(ctx, driver, postID, source)
if err != nil {
return err
}
log.Printf("Created PostToSourceRelationship for Post: %s to source: %d", post.ID, source)
err = uploadGeneralTagRelationship(ctx, driver, post)
if err != nil {
log.Fatal(err)
return err
}
for _, generalTag := range post.Tags.General {
err := neo4jAPI.PostToTagRelationship(ctx, driver, id, generalTag)
if err != nil {
log.Println(err)
}
log.Printf("Created PostToTagRelationship for post: %d to tag: %s", post.ID, generalTag)
err = uploadCharacterTagtRelationship(ctx, driver, post)
if err != nil {
log.Fatal(err)
return err
}
for _, characterTag := range post.Tags.Character {
err := neo4jAPI.PostToTagRelationship(ctx, driver, id, characterTag)
if err != nil {
log.Println(err)
}
log.Printf("Created PostToTagRelationship for post: %d to tag: %s", post.ID, characterTag)
err = uploadCopyrightTagRelationship(ctx, driver, post)
if err != nil {
log.Fatal(err)
return err
}
for _, copyrightTag := range post.Tags.Copyright {
err := neo4jAPI.PostToTagRelationship(ctx, driver, id, copyrightTag)
if err != nil {
log.Println(err)
}
log.Printf("Created PostToTagRelationship for post: %d to tag: %s", post.ID, copyrightTag)
err = uploadArtistTagRelationship(ctx, driver, post)
if err != nil {
log.Fatal(err)
return err
}
for _, artistTag := range post.Tags.Artist {
err := neo4jAPI.PostToTagRelationship(ctx, driver, id, artistTag)
if err != nil {
log.Println(err)
}
log.Printf("Created PostToTagRelationship for post: %d to tag: %s", post.ID, artistTag)
log.Printf("Making relationship for user %s with id %d, %d for Post: %d with ID: %d took: %v", e621User.Name, e621User.ID, i, len(userFavorites), post.ID, time.Since(start))
}
}
log.Printf("Uploading posts for user %s took: %v", username, time.Since(startUploadPosts))
log.Printf("Uploading all posts for user %s took: %v", username, time.Since(startUploadPosts))
return nil
}
// uploadNodes uploads the post to the database and creates the nodes
func uploadNodes(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
uniqueGeneralTags := make([]string, 0)
uniqueCharacterTags := make([]string, 0)
uniqueCopyrightTags := make([]string, 0)
uniqueArtistTags := make([]string, 0)
allGeneralTags := make([]string, 0)
allCharacterTags := make([]string, 0)
allCopyrightTags := make([]string, 0)
allArtistTags := make([]string, 0)
allGeneralTags = append(allGeneralTags, post.Tags.General...)
allCharacterTags = append(allCharacterTags, post.Tags.Character...)
allCopyrightTags = append(allCopyrightTags, post.Tags.Character...)
allArtistTags = append(allArtistTags, post.Tags.Artist...)
uniqueGeneralTags = utils.UniqueNonEmptyElementsOf(allGeneralTags)
uniqueCharacterTags = utils.UniqueNonEmptyElementsOf(allCharacterTags)
uniqueCopyrightTags = utils.UniqueNonEmptyElementsOf(allCopyrightTags)
uniqueArtistTags = utils.UniqueNonEmptyElementsOf(allArtistTags)
// Uploads post to database
err := neo4jAPI.CreatePostNode(ctx, driver, post.ID)
if err != nil {
return err
}
// Uploads the source to the database
for _, source := range post.Sources {
err := neo4jAPI.CreateSourceNode(ctx, driver, source)
if err != nil {
return err
}
}
for _, uniqueGeneralTag := range uniqueGeneralTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueGeneralTag, "general")
if err != nil {
return err
}
}
for _, uniqueCharacterTag := range uniqueCharacterTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueCharacterTag, "character")
if err != nil {
return err
}
}
for _, uniqueCopyrightTag := range uniqueCopyrightTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueCopyrightTag, "copyright")
if err != nil {
return err
}
}
for _, uniqueArtistTag := range uniqueArtistTags {
err := neo4jAPI.CreateTagNode(ctx, driver, uniqueArtistTag, "artist")
if err != nil {
return err
}
}
return nil
}
// uploadPostToUserRelationship creates a relationship between the user and the post
func uploadPostToUserRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post, e621User models.E621User) error {
err := neo4jAPI.UserToPostRelationship(ctx, driver, post.ID, e621User.ID)
if err != nil {
return err
}
// log.Printf("Created UserToPostRelationship for user: %s to post: %d", e621User.Name, post.ID)
return nil
}
// uploadSourceTagRelationship creates a relationship between the post and the source
func uploadSourceTagRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
for _, source := range post.Sources {
err := neo4jAPI.PostToSourceRelationship(ctx, driver, post.ID, source)
if err != nil {
return err
}
// log.Printf("Created PostToSourceRelationship for Post: %d to source: %s", post.ID, source)
}
return nil
}
// uploadGeneralTagRelationship creates a relationship between the post and the general tag
func uploadGeneralTagRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
for _, generalTag := range post.Tags.General {
err := neo4jAPI.PostToTagRelationship(ctx, driver, post.ID, generalTag)
if err != nil {
return err
}
// log.Printf("Created PostToTagRelationship for post: %d to general tag: %s", post.ID, generalTag)
}
return nil
}
// uploadCharacterTagtRelationship creates a relationship between the post and the character tag
func uploadCharacterTagtRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
for _, characterTag := range post.Tags.Character {
err := neo4jAPI.PostToTagRelationship(ctx, driver, post.ID, characterTag)
if err != nil {
return err
}
// log.Printf("Created PostToTagRelationship for post: %d to character tag: %s", post.ID, characterTag)
}
return nil
}
// uploadCopyrightTagRelationship creates a relationship between the post and the copyright tag
func uploadCopyrightTagRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
for _, copyrightTag := range post.Tags.Copyright {
err := neo4jAPI.PostToTagRelationship(ctx, driver, post.ID, copyrightTag)
if err != nil {
return err
}
// log.Printf("Created PostToTagRelationship for post: %d to copyrigh tag: %s", post.ID, copyrightTag)
}
return nil
}
// uploadArtistTagRelationship creates a relationship between the post and the artist tag
func uploadArtistTagRelationship(ctx context.Context, driver neo4j.DriverWithContext, post models.Post) error {
for _, artistTag := range post.Tags.Artist {
err := neo4jAPI.PostToTagRelationship(ctx, driver, post.ID, artistTag)
if err != nil {
return err
}
// log.Printf("Created PostToTagRelationship for post: %d to artist tag: %s", post.ID, artistTag)
}
return nil
}
//11min für Selloo, simultan mit mutt_jake