getsentry / getsentry/sentry-java

Generate low-cardinality SQL summaries for database spans

Ouverte
#5,981 1 commentaire 0 réactions 0 personnes assignées Voir sur GitHub
Feature Java Platform: Java Spans
Langage dominant
Kotlin
Étoiles
1.4k
Forks
478
Merge moyen
2 j 23 h
PR mergées (30 j)
67

Description

## Problem

When `dataCollection.databaseQueryData` disables raw SQL collection, database spans lose their SQL description entirely. Keeping the raw statement is not safe because inline literals and other query values can contain sensitive data and create high-cardinality span descriptions.

We should provide useful, low-cardinality descriptions without retaining query values, for example:

- `SELECT users`
- `UPDATE orders`
- `INSERT products`

This is follow-up work from #5801 and supports #5666.

## Proposed solution

Add a SQL summarizer that extracts only a safe operation and target from a statement. Use the summary for JDBC and Android SQLite span descriptions when raw database query data is disabled. Preserve the full statement only when the effective Data Collection policy allows it.

Consider adapting OpenTelemetry Java's Apache-2.0 SQL query analyzer instead of depending on its incubating API. If code is adapted, include the required source attribution and update `THIRD_PARTY_NOTICES.md`.

The summarizer must fail closed: malformed or unsupported SQL must never fall back to the raw statement.

## Acceptance criteria

- JDBC and Android SQLite spans use low-cardinality summaries when raw query collection is disabled.
- Summaries do not contain inline literals, bound values, comments, or other query values.
- Parsing covers common operations such as `SELECT`, `INSERT`, `UPDATE`, and `DELETE`.
- Malformed and unsupported SQL produces a generic description or no description, never the raw statement.
- Tests cover inline literals, placeholders, mixed literal-and-bound queries, malformed SQL, and representative dialect-specific syntax.
- Any adapted third-party code includes complete license attribution.

Guide de contribution

Ouvrir le guide de contribution

Piste de recherche

Start by locating the JDBC and Android SQLite span-description entry points and how the effective dataCollection.databaseQueryData policy is applied. Trace the existing tests, then add coverage for supported operations, literals, placeholders, malformed SQL, and dialect-specific syntax. Done means both integrations use safe summaries when raw collection is disabled, never expose raw SQL on failure, and any adapted analyzer code is attributed in THIRD_PARTY_NOTICES.md.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Évaluation

Stack technique
android, java, kotlin, sqlite
Domaine
backend, databases, mobile
Type d'issue
Fonctionnalité
Difficulté
4/5
Temps estimé
3-5 jours
Activité
Active
Clarté
Plutôt claire
Accessibilité débutants
55/100

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.