latexgrambot: inline LaTeX bot for Telegram in Go

Renders LaTeX expressions into PNG pictures, PDF documents and native
Telegram rich text messages, entirely locally with TeX Live and poppler.

- Chat mode with /image, /pdf, /rich plus per-user preamble and DPI
- Inline mode delivering picture, PDF and rich text results
- Broad default package set preloaded into a TeX format at image build
- MongoDB persistence: settings, users, groups, render statistics
- Private S3-compatible bucket with short-lived presigned URLs
- OpenTelemetry traces over OTLP
- Kubernetes manifests and Argo CD application, Docker Compose, GitHub
  Actions CI, BSD 3-Clause license
This commit is contained in:
corrado.mulas
2026-10-04 03:38:08 +02:00
commit 98a3614d58
29 changed files with 5140 additions and 0 deletions

145
internal/files/files.go Normal file
View File

@@ -0,0 +1,145 @@
// Package files stores rendered outputs in a private S3-compatible bucket
// and issues short-lived presigned URLs for Telegram inline-query results.
package files
import (
"bytes"
"context"
"fmt"
"strings"
"time"
"github.com/aws/aws-sdk-go-v2/aws"
awsconfig "github.com/aws/aws-sdk-go-v2/config"
"github.com/aws/aws-sdk-go-v2/credentials"
"github.com/aws/aws-sdk-go-v2/service/s3"
"github.com/google/uuid"
)
// Config points the store at the internal S3-compatible API and the public
// API endpoint used only when signing download URLs.
type Config struct {
Endpoint string
Region string
Bucket string
AccessKeyID string
SecretAccessKey string
Prefix string
PublicEndpoint string
PresignTTL time.Duration
HTTPClient aws.HTTPClient
}
// Store keeps content-addressed rendered assets in S3.
type Store struct {
client *s3.Client
presigner *s3.PresignClient
bucket string
prefix string
presignTTL time.Duration
}
// New builds an S3-compatible object store.
func New(ctx context.Context, cfg Config) (*Store, error) {
awsCfg, err := awsconfig.LoadDefaultConfig(ctx,
awsconfig.WithRegion(cfg.Region),
awsconfig.WithCredentialsProvider(
credentials.NewStaticCredentialsProvider(cfg.AccessKeyID, cfg.SecretAccessKey, ""),
),
)
if err != nil {
return nil, fmt.Errorf("load aws config: %w", err)
}
if cfg.HTTPClient != nil {
awsCfg.HTTPClient = cfg.HTTPClient
}
client := s3.NewFromConfig(awsCfg, func(options *s3.Options) {
options.BaseEndpoint = aws.String(cfg.Endpoint)
options.UsePathStyle = true
})
publicClient := s3.NewFromConfig(awsCfg, func(options *s3.Options) {
options.BaseEndpoint = aws.String(cfg.PublicEndpoint)
options.UsePathStyle = true
})
return &Store{
client: client,
presigner: s3.NewPresignClient(publicClient),
bucket: cfg.Bucket,
prefix: strings.Trim(cfg.Prefix, "/"),
presignTTL: cfg.PresignTTL,
}, nil
}
// Put uploads the data under a fresh UUIDv4 object name and returns a
// short-lived presigned GET URL.
func (s *Store) Put(ctx context.Context, ext, contentType string, data []byte) (string, error) {
key := s.key(uuid.NewString(), ext)
_, err := s.client.PutObject(ctx, &s3.PutObjectInput{
Bucket: aws.String(s.bucket),
Key: aws.String(key),
Body: bytes.NewReader(data),
ContentType: aws.String(contentType),
CacheControl: aws.String("private, max-age=86400"),
})
if err != nil {
return "", fmt.Errorf("put %s object: %w", ext, err)
}
request, err := s.presigner.PresignGetObject(ctx, &s3.GetObjectInput{
Bucket: aws.String(s.bucket),
Key: aws.String(key),
}, func(options *s3.PresignOptions) {
options.Expires = s.presignTTL
})
if err != nil {
return "", fmt.Errorf("presign %s object: %w", ext, err)
}
return request.URL, nil
}
// DeleteExpired removes objects older than age and returns how many were
// deleted. Telegram downloads the file when an inline result is picked and
// the presigned link expires after the presign TTL; objects past their
// lifetime can no longer be fetched, so they are garbage collected.
func (s *Store) DeleteExpired(ctx context.Context, age time.Duration) (int, error) {
cutoff := time.Now().Add(-age)
prefix := ""
if s.prefix != "" {
prefix = s.prefix + "/"
}
deleted := 0
var token *string
for {
list, err := s.client.ListObjectsV2(ctx, &s3.ListObjectsV2Input{
Bucket: aws.String(s.bucket),
Prefix: aws.String(prefix),
ContinuationToken: token,
})
if err != nil {
return deleted, fmt.Errorf("list objects: %w", err)
}
for _, object := range list.Contents {
if object.LastModified == nil || !object.LastModified.Before(cutoff) {
continue
}
if _, err := s.client.DeleteObject(ctx, &s3.DeleteObjectInput{
Bucket: aws.String(s.bucket),
Key: object.Key,
}); err != nil {
return deleted, fmt.Errorf("delete %s: %w", aws.ToString(object.Key), err)
}
deleted++
}
if list.IsTruncated == nil || !*list.IsTruncated {
return deleted, nil
}
token = list.NextContinuationToken
}
}
func (s *Store) key(name, ext string) string {
if s.prefix == "" {
return name + "." + ext
}
return s.prefix + "/" + name + "." + ext
}

View File

@@ -0,0 +1,176 @@
package files
import (
"context"
"io"
"net/http"
"net/url"
"os"
"strings"
"sync/atomic"
"testing"
"time"
"github.com/google/uuid"
)
func TestPutUploadsUniqueUUIDObjectsWithPresignedURLs(t *testing.T) {
var puts atomic.Int32
keys := make(chan string, 4)
client := &http.Client{Transport: roundTripFunc(func(request *http.Request) (*http.Response, error) {
status := http.StatusOK
switch request.Method {
case http.MethodPut:
puts.Add(1)
keys <- request.URL.Path
}
return &http.Response{
StatusCode: status,
Status: http.StatusText(status),
Header: make(http.Header),
Body: io.NopCloser(strings.NewReader("")),
Request: request,
}, nil
})}
store, err := New(context.Background(), Config{
Endpoint: "https://s3.internal",
PublicEndpoint: "https://s3.example",
Region: "us-east-1",
Bucket: "latexgrambot",
AccessKeyID: "access",
SecretAccessKey: "secret",
Prefix: "content",
PresignTTL: time.Hour,
HTTPClient: client,
})
if err != nil {
t.Fatalf("New: %v", err)
}
seen := map[string]bool{}
for range 2 {
signed, err := store.Put(context.Background(), "jpg", "image/jpeg", []byte("image"))
if err != nil {
t.Fatalf("Put: %v", err)
}
parsed, err := url.Parse(signed)
if err != nil {
t.Fatalf("parse signed URL: %v", err)
}
base := strings.TrimSuffix(strings.TrimPrefix(parsed.Path, "/latexgrambot/content/"), ".jpg")
if _, err := uuid.Parse(base); err != nil || uuid.MustParse(base).Version() != 4 {
t.Errorf("object name %q is not a UUIDv4", base)
}
if seen[parsed.Path] {
t.Errorf("object name %q was reused", parsed.Path)
}
seen[parsed.Path] = true
if parsed.Query().Get("X-Amz-Signature") == "" || parsed.Query().Get("X-Amz-Expires") != "3600" {
t.Errorf("URL is not signed for one hour: %s", signed)
}
}
if got := puts.Load(); got != 2 {
t.Fatalf("PUT requests = %d, want 2 (one object per request)", got)
}
}
func TestDeleteExpired(t *testing.T) {
var deleted []string
client := &http.Client{Transport: roundTripFunc(func(request *http.Request) (*http.Response, error) {
header := make(http.Header)
body := ""
status := http.StatusOK
switch request.Method {
case http.MethodGet: // ListObjectsV2
header.Set("Content-Type", "application/xml")
body = `<?xml version="1.0" encoding="UTF-8"?>
<ListBucketResult xmlns="http://s3.amazonaws.com/doc/2006-03-01/">
<Name>latexgrambot</Name><Prefix>content/</Prefix><KeyCount>2</KeyCount><MaxKeys>1000</MaxKeys><IsTruncated>false</IsTruncated>
<Contents><Key>content/old.jpg</Key><LastModified>2000-01-01T00:00:00.000Z</LastModified><ETag>"a"</ETag><Size>1</Size><StorageClass>STANDARD</StorageClass></Contents>
<Contents><Key>content/fresh.pdf</Key><LastModified>2099-01-01T00:00:00.000Z</LastModified><ETag>"b"</ETag><Size>1</Size><StorageClass>STANDARD</StorageClass></Contents>
</ListBucketResult>`
case http.MethodDelete:
deleted = append(deleted, request.URL.Path)
status = http.StatusNoContent
}
return &http.Response{
StatusCode: status,
Status: http.StatusText(status),
Header: header,
Body: io.NopCloser(strings.NewReader(body)),
Request: request,
}, nil
})}
store, err := New(context.Background(), Config{
Endpoint: "https://s3.internal",
PublicEndpoint: "https://s3.example",
Region: "us-east-1",
Bucket: "latexgrambot",
AccessKeyID: "access",
SecretAccessKey: "secret",
Prefix: "content",
PresignTTL: time.Hour,
HTTPClient: client,
})
if err != nil {
t.Fatalf("New: %v", err)
}
removed, err := store.DeleteExpired(context.Background(), time.Hour)
if err != nil {
t.Fatalf("DeleteExpired: %v", err)
}
if removed != 1 {
t.Fatalf("removed = %d, want 1", removed)
}
if len(deleted) != 1 || !strings.HasSuffix(deleted[0], "/content/old.jpg") {
t.Fatalf("deleted = %v, want the expired object only", deleted)
}
}
func TestS3Integration(t *testing.T) {
endpoint := os.Getenv("LATEXGRAMBOT_S3_TEST_ENDPOINT")
accessKey := os.Getenv("LATEXGRAMBOT_S3_TEST_ACCESS_KEY")
secretKey := os.Getenv("LATEXGRAMBOT_S3_TEST_SECRET_KEY")
if endpoint == "" || accessKey == "" || secretKey == "" {
t.Skip("S3 integration environment is not configured")
}
store, err := New(context.Background(), Config{
Endpoint: endpoint,
PublicEndpoint: endpoint,
Region: "us-east-1",
Bucket: "latexgrambot",
AccessKeyID: accessKey,
SecretAccessKey: secretKey,
Prefix: "integration",
PresignTTL: 5 * time.Minute,
})
if err != nil {
t.Fatalf("New: %v", err)
}
const payload = "latexgrambot S3 integration probe\n"
signed, err := store.Put(context.Background(), "txt", "text/plain", []byte(payload))
if err != nil {
t.Fatalf("Put: %v", err)
}
response, err := http.Get(signed)
if err != nil {
t.Fatalf("GET presigned URL: %v", err)
}
defer response.Body.Close()
body, err := io.ReadAll(response.Body)
if err != nil {
t.Fatalf("read presigned response: %v", err)
}
if response.StatusCode != http.StatusOK || string(body) != payload {
t.Fatalf("presigned GET = %s, body %q", response.Status, body)
}
}
type roundTripFunc func(*http.Request) (*http.Response, error)
func (f roundTripFunc) RoundTrip(request *http.Request) (*http.Response, error) {
return f(request)
}