惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

M
MIT News - Artificial intelligence
罗磊的独立博客
Hugging Face - Blog
Hugging Face - Blog
Apple Machine Learning Research
Apple Machine Learning Research
Last Week in AI
Last Week in AI
S
SegmentFault 最新的问题
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
美团技术团队
人人都是产品经理
人人都是产品经理
WordPress大学
WordPress大学
The Cloudflare Blog
IT之家
IT之家
雷峰网
雷峰网
小众软件
小众软件
博客园 - 叶小钗
博客园 - 聂微东
爱范儿
爱范儿
博客园 - 司徒正美
博客园 - 三生石上(FineUI控件)
V
Visual Studio Blog
博客园 - 【当耐特】
V
V2EX
博客园_首页
T
Tailwind CSS Blog

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Why Our Android Build Was Signed with the Wrong Key (A Re...
Anand Rathna · 2026-05-22 · via DEV Community

Anand Rathnas

This article was originally published on Jo4 Blog.

"Your Android App Bundle is not signed with the correct key."

That was the Google Play Console rejection after what should have been a routine release. The SHA1 fingerprint on the uploaded AAB didn't match the expected upload key. We'd been deploying to the Play Store for weeks with no issues. What changed?

Nothing, as it turned out. The signing had been broken the whole time -- we just hadn't noticed until Google tightened its fingerprint check.

The Setup

We use Expo with expo prebuild --clean to generate the android/ directory before each build. Because it's regenerated every time, the entire android/ folder is gitignored. This means any customization to build.gradle needs to happen through a post-prebuild injection script.

Our scripts/release.sh runs after prebuild and uses Node.js to patch the generated build.gradle with the release signing configuration. It finds the release buildType block and replaces signingConfig signingConfigs.debug with signingConfig signingConfigs.release.

Sounds straightforward. But the regex doing this work had a subtle, devastating bug.

The Symptom

After release.sh ran, we expected build.gradle to look like this:

buildTypes {
    debug {
        signingConfig signingConfigs.debug
    }
    release {
        signingConfig signingConfigs.release  // patched
    }
}

Enter fullscreen mode Exit fullscreen mode

Instead, it looked like this:

buildTypes {
    debug {
        signingConfig signingConfigs.release  // WRONG
    }
    release {
        signingConfig signingConfigs.debug    // WRONG
    }
}

Enter fullscreen mode Exit fullscreen mode

The configs were swapped. Debug was using the release keystore. Release was using the debug keystore. The build succeeded (both keystores are valid), but the release AAB was signed with the debug key.

The Buggy Regex

Here's the regex our script used:

const patched = buildGradle.replace(
  /release \{[\s\S]*?signingConfig signingConfigs\.debug/,
  match => match.replace('signingConfig signingConfigs.debug', 'signingConfig signingConfigs.release')
);

Enter fullscreen mode Exit fullscreen mode

The intent: find release { followed (lazily) by signingConfig signingConfigs.debug, then replace that debug with release.

The problem: release { appears twice in the generated build.gradle.

signingConfigs {
    debug {
        // ...
    }
    release {        // <-- FIRST occurrence of "release {"
        storeFile file("release.keystore")
        storePassword "..."
        keyAlias "..."
        keyPassword "..."
    }
}

buildTypes {
    debug {
        signingConfig signingConfigs.debug   // <-- target line
    }
    release {        // <-- SECOND occurrence of "release {"
        signingConfig signingConfigs.debug   // <-- intended target
    }
}

Enter fullscreen mode Exit fullscreen mode

The lazy quantifier [\s\S]*? matched the first release { (inside signingConfigs), then expanded minimally until it found signingConfig signingConfigs.debug. The first signingConfig signingConfigs.debug it encountered was inside the debug buildType. So the regex matched from signingConfigs.release { all the way down to the debug buildType's signing config -- and replaced it.

This is the core misunderstanding: lazy quantifiers don't find the closest release { to the target. They find the first release { in the file, then minimize the gap from there. If the first match is in the wrong block, the lazy expansion crosses block boundaries to reach the target string.

The Fix

Anchor the regex to the buildTypes section:

const patched = buildGradle.replace(
  /(buildTypes\s*\{[\s\S]*?release\s*\{[\s\S]*?)signingConfig signingConfigs\.debug/,
  '$1signingConfig signingConfigs.release'
);

Enter fullscreen mode Exit fullscreen mode

By requiring buildTypes { before release {, the regex skips the signingConfigs.release block entirely. The capture group grabs everything from buildTypes { through release { and any content before the signing config line. Then we replace just the signingConfig reference while preserving the surrounding structure.

The key difference:

BEFORE: /release \{[\s\S]*?signingConfig signingConfigs\.debug/
AFTER:  /(buildTypes\s*\{[\s\S]*?release\s*\{[\s\S]*?)signingConfig signingConfigs\.debug/

Enter fullscreen mode Exit fullscreen mode

The buildTypes\s*\{ anchor ensures we're in the right block before we ever look for release {.

Bonus Bug: versionCode Stuck at 1

While debugging the signing issue, we found a second problem. Expo prebuild defaults versionCode to 1 in every generated build.gradle. Google Play requires versionCode to be strictly increasing -- you can never upload a version code equal to or lower than a previously uploaded one.

Our fix: auto-generate versionCode from epoch minutes in the release script:

VERSION_CODE=$(($(date +%s) / 60))

Enter fullscreen mode Exit fullscreen mode

This produces a value like 29432517 that increases every minute. No manual tracking, no CI state to maintain, and no collisions as long as you don't release twice in the same minute.

The Node.js injection then patches this into build.gradle:

buildGradle = buildGradle.replace(
  /versionCode \d+/,
  `versionCode ${process.env.VERSION_CODE}`
);

Enter fullscreen mode Exit fullscreen mode

Lessons Learned

1. Lazy quantifiers aren't always lazy enough.

The *? quantifier minimizes the match after fixing the start position. If your start anchor (release {) appears multiple times, the regex locks onto the first occurrence and expands from there. It doesn't backtrack to try the second occurrence unless the first one fails entirely.

2. When a pattern appears in multiple blocks, anchor to the surrounding context.

Don't match release { when you mean buildTypes { ... release {. The extra context eliminates ambiguity. This applies to any structured text you're patching with regex -- Gradle files, XML, YAML, anything with nested blocks.

3. Gitignored generated files need persistent injection scripts -- and those scripts need tests.

We tested the app. We tested the build. We never tested release.sh in isolation. A simple assertion -- "after running the script, the release buildType should have signingConfigs.release" -- would have caught this immediately.

4. Always verify build artifacts before pushing to a store.

A one-line check would have saved hours:

# Verify the AAB is signed with the correct key
jarsigner -verify -verbose -certs app-release.aab | grep "SHA1:"

Enter fullscreen mode Exit fullscreen mode

If the fingerprint doesn't match your expected upload key, stop. Don't submit and hope for the best.

The Meta-Lesson

Regex on structured data is inherently fragile. Every time expo prebuild changes the generated build.gradle format, our regex could break in new and creative ways. The real long-term fix is to use Expo's config plugins to inject signing configuration declaratively, removing the regex entirely. We're migrating to that approach now.

But until then -- anchor your patterns, test your scripts, and verify your artifacts.


Have you been burned by regex matching across block boundaries? What's your approach to patching generated build files? We'd love to hear about it in the comments.

Building jo4.io -- a URL shortener with analytics, QR codes, and a mobile app that is now correctly signed.