парсер

This commit is contained in:
Artem Tsyrulnikov
2026-01-06 21:04:42 +03:00
parent b77e4fd0b6
commit 2c0ae2dfc3
45 changed files with 2202 additions and 51 deletions

View File

@@ -0,0 +1,11 @@
package usecase
import (
"context"
"github.com/TelegramExchange/tgex-parser/internal/domain"
)
func (uc *UseCase) FetchChannelMeta(ctx context.Context, username string) (domain.Channel, error) {
return uc.telegram.ParseChannelMeta(ctx, username)
}

View File

@@ -0,0 +1,112 @@
package usecase
import (
"context"
"fmt"
"github.com/TelegramExchange/tgex-parser/pkg/transaction"
"github.com/rs/zerolog/log"
"github.com/TelegramExchange/tgex-parser/internal/domain"
)
func (uc *UseCase) FetchChannels(ctx context.Context) error {
channels := uc.database.GetChannels(ctx)
for _, c := range channels {
if c.Pts == 0 {
updatedChannel, err := uc.initializeChannel(ctx, c)
if err != nil {
return fmt.Errorf("uc.initializeChannel: %w", err)
}
c = updatedChannel
}
err := uc.processChannel(ctx, c)
if err != nil {
return fmt.Errorf("uc.processChannel: %w", err)
}
}
return nil
}
func (uc *UseCase) processChannel(ctx context.Context, channel domain.Channel) error {
diff, err := uc.telegram.GetChannelDiff(ctx, channel, 20)
if err != nil {
return fmt.Errorf("telegram.GetChannelDiff: %w", err)
}
for _, p := range diff.NewPosts {
log.Info().Msgf("New post: %s - Views: %d, Text: %.10s...", p.Link, p.Views, p.Text)
err = uc.database.CreatePost(ctx, p)
if err != nil {
return fmt.Errorf("database.CreatePost: %w", err)
}
}
for _, p := range diff.DeletedPosts {
log.Info().Msgf("Deleted post: %s - Views: %d", p.Link, p.Views)
err = uc.database.DeletePost(ctx, p)
if err != nil {
return fmt.Errorf("database.DeletePost: %w", err)
}
}
// Update channel metadata if changed
if diff.UpdatedChannel != nil {
diff.UpdatedChannel.Pts = diff.NewPts
err = uc.database.UpdateChannel(ctx, *diff.UpdatedChannel)
if err != nil {
return fmt.Errorf("database.UpdateChannel: %w", err)
}
} else {
channel.Pts = diff.NewPts
err = uc.database.UpdateChannel(ctx, channel)
if err != nil {
return fmt.Errorf("database.UpdateChannel: %w", err)
}
}
return nil
}
func (uc *UseCase) initializeChannel(ctx context.Context, channel domain.Channel) (domain.Channel, error) {
c, err := uc.telegram.ParseChannelMeta(ctx, channel.Username)
if err != nil {
return domain.Channel{}, fmt.Errorf("telegram.ParseChannelMeta: %w", err)
}
// Preserve persistent channel ID from the database so foreign keys remain valid.
c.ID = channel.ID
posts, err := uc.telegram.GetChannelHistory(ctx, c, 20)
if err != nil {
return domain.Channel{}, fmt.Errorf("telegram.GetChannelHistory: %w", err)
}
err = transaction.Wrap(ctx, func(ctx context.Context) error {
err = uc.database.UpdateChannel(ctx, c)
if err != nil {
return fmt.Errorf("database.UpdateChannel: %w", err)
}
for _, p := range posts {
log.Info().Msgf("New post (init): %s - Views: %d", p.Link, p.Views)
err = uc.database.CreatePost(ctx, p)
if err != nil {
return fmt.Errorf("database.CreatePost: %w", err)
}
}
return nil
})
if err != nil {
return domain.Channel{}, fmt.Errorf("transaction.Wrap: %w", err)
}
return c, nil
}

View File

@@ -0,0 +1,46 @@
package usecase
import (
"context"
"time"
"github.com/rs/zerolog/log"
"github.com/TelegramExchange/tgex-parser/internal/domain"
)
func (uc *UseCase) FetchViews(ctx context.Context) error {
channels, err := uc.database.GetChannelsWithTrackedPosts(ctx)
if err != nil {
return err
}
for _, channel := range channels {
posts, err := uc.database.GetTrackedPosts(ctx, channel)
if err != nil {
return err
}
err = uc.telegram.UpdatePostsViews(ctx, channel, posts)
if err != nil {
return err
}
for _, p := range posts {
v := domain.ViewsSnapshot{
ViewsCount: p.Views,
FetchedAt: time.Now().UTC(),
PostID: p.ID,
}
err = uc.database.CreateViewsSnapshot(ctx, v)
if err != nil {
return err
}
log.Info().Msgf("New ViewsSnapshot: %s - Post ID: %d, Views: %d", p.Link, p.MessageID, p.Views)
}
}
return nil
}

View File

@@ -0,0 +1,38 @@
package usecase
import (
"context"
"github.com/TelegramExchange/tgex-parser/internal/domain"
)
type Telegram interface {
ParseChannelMeta(ctx context.Context, username string) (domain.Channel, error)
GetChannelHistory(ctx context.Context, channel domain.Channel, limit int) ([]domain.Post, error)
GetChannelDiff(ctx context.Context, channel domain.Channel, limit int) (domain.ChannelDiff, error)
UpdatePostsViews(ctx context.Context, channel domain.Channel, posts []domain.Post) error
}
type Database interface {
CreatePost(ctx context.Context, post domain.Post) error
CreateViewsSnapshot(ctx context.Context, snapshot domain.ViewsSnapshot) error
GetChannels(ctx context.Context) []domain.Channel
UpdateChannel(ctx context.Context, channel domain.Channel) error
GetChannelsWithTrackedPosts(ctx context.Context) ([]domain.Channel, error)
GetTrackedPosts(ctx context.Context, channel domain.Channel) ([]domain.Post, error)
DeletePost(ctx context.Context, p domain.Post) error
}
type UseCase struct {
telegram Telegram
database Database
}
func New(telegram Telegram, database Database) *UseCase {
return &UseCase{
telegram: telegram,
database: database,
}
}