Frequently Asked Questions

Common questions about PageTurner's translation platform, features, and troubleshooting.

General Questions

How is this different from Google Translate?

Short answer: Google Translate processes sentences in isolation. PageTurner uses a 5-phase pipeline that ensures technical term consistency across your entire documentation.

Key differences:

Feature	PageTurner	Google Translate
Term Consistency	99%+ (same term = same translation everywhere)	~70% (inconsistent across pages)
Context Awareness	Full document context	Sentence-by-sentence
Technical Accuracy	94%	~68%
MDX/React Preservation	Native support, 100% preserved	Breaks components
Quality Score	91.3/100	65-70/100

Example problem with Google Translate:

Page 1: "repository" → "repositorio"
Page 50: "repository" → "repo"
Page 100: "repository" → "almacén"

PageTurner solution: "repository" → "repositorio" on ALL pages.

What frameworks are supported?

Currently supported (Production):

✅ Docusaurus - Full production support with MDX, React components, i18n

Coming soon:

🚧 Next.js (Q2 2025) - App router, MDX support
🚧 Nextra (Q3 2025) - Next.js-based docs framework
🚧 Hugo (Q3 2025) - Static site generator
🚧 VitePress (Q4 2025) - Vue-powered docs

Want another framework? Contact us to request support or vote on our roadmap.

How much does it cost?

Current Beta Pricing:

Open Source Program (FREE):

Free translations for qualifying open source projects
Up to 500 pages
All languages supported
Community support

Commercial Beta:

Pay-as-you-go: ~$0.10 per page per language
Typical costs:
- 100 pages × 3 languages = $30
- 200 pages × 5 languages = $100
Updates: 60-80% cheaper with translation memory

See full details: Beta Program

Note: Beta pricing may change at general availability. Early adopters get locked-in pricing.

How long does translation take?

Typical translation times:

Documentation Size	Languages	Time
50 pages	3	10-15 minutes
100 pages	3	20-30 minutes
200 pages	3	40-60 minutes
100 pages	10	60-90 minutes

Updates with translation memory: 2-5 minutes for typical changes (only retranslates changed content)

Factors affecting speed:

Number of pages
Content complexity
Number of target languages
LLM provider rate limits

Can I customize translations?

Yes! Several ways to customize:

1. Translation Memory (Recommended)

# Provide preferred translations for specific terms
from transaurus import GlobalTMManager

tm = GlobalTMManager()
tm.add_translation(
    source_text="webhook",
    target_language="es",
    translation="webhook"  # Keep as-is, don't translate
)

2. Source Content Improvements

Use consistent terminology in English
Provide context for abbreviations
Add glossary sections

3. Post-Translation Editing

Edit translated files directly in GitHub
Translation memory remembers your edits
Future updates preserve your customizations

Enterprise plans (coming Q2 2025) include:

Custom terminology glossaries
Domain-specific tuning
Human review workflows
Translation approval processes

What languages are supported?

100+ languages powered by Claude AI.

Most popular languages:

Spanish (es), French (fr), German (de)
Japanese (ja), Korean (ko)
Chinese Simplified (zh-Hans), Chinese Traditional (zh-Hant)
Portuguese (pt), Russian (ru), Arabic (ar)
Italian (it), Dutch (nl), Polish (pl)

Quality by language:

Major languages (es, fr, de, ja, zh, ko): 91-95/100
European languages (it, pt, nl, pl, sv, da): 88-92/100
Asian languages (th, vi, id): 85-90/100
Other languages: 80-88/100

See Configuration → Supported Languages for complete list.

Is my source code safe?

Yes. PageTurner takes security seriously:

Data handling:

✅ Code is processed via secure APIs (GitHub, Anthropic)
✅ No code stored on PageTurner servers
✅ LLM providers (Anthropic) have enterprise security
✅ All connections use HTTPS/TLS encryption

Access control:

✅ You control GitHub permissions (read-only access is sufficient)
✅ Translations created in repositories you own
✅ No third-party access to your code

Open source:

✅ Translation engine is open source (review code yourself)
✅ No proprietary black boxes

Note: LLM providers (like Anthropic) process your content for translation but don't use it for model training (per their enterprise agreements).

Technical Questions

What gets created when I run a translation?

GitHub repositories:

Mirror repositories (one per target language)
- Example: docs-es, docs-fr, docs-zh-Hans
Full copy of source repository with translations
GitHub Actions workflow for auto-updates
Security settings configured

Files created/modified:

Translated MDX/Markdown files (in docs/, blog/, etc.)
Translated JSON configuration files
Updated docusaurus.config.js with i18n settings
Updated sidebars with translated labels
Translation memory files (.tm/ directory)

Vercel deployment (if token provided):

Vercel project linked to GitHub repo
Automatic build and deployment
Production URLs (e.g., https://docs-es.vercel.app)
Automatic deployments on future commits

Database records:

Repository configuration in Supabase
Translation memory entries
Status tracking and metrics

How do I handle updates to my documentation?

Simple: Just re-run the same command.

# Initial translation
result = master.run_full_translation_pipeline()

# ... time passes, you update your docs ...

# Update translations (only changed content gets retranslated)
result = master.run_full_translation_pipeline()

What happens:

Change detection: SHA256 hashing identifies changed content
Selective translation: Only changed segments get retranslated
Cost savings: 60-80% reduction (only pay for what changed)
Auto-deployment: Changes automatically deployed to Vercel

Example:

First run: 100 pages → 300 translation requests
Update 5 pages: 5 pages → 15 translation requests
Savings: 95% cost reduction

Automation (optional): Set up GitHub Actions to auto-translate on every commit to main branch.

Will my React components break?

No. PageTurner was built specifically for Docusaurus and preserves all JSX/MDX components.

What gets preserved:

✅ React component tags (<Tabs>, <TabItem>, custom components)
✅ JSX attributes and props
✅ Import statements
✅ Code blocks (never translated)
✅ Inline code expressions
✅ Component nesting and hierarchy

What gets translated:

✅ Text content inside components
✅ String attributes (like label="Click here")
✅ YAML frontmatter
✅ Markdown text

Example:

Before:

<Tabs>
  <TabItem value="js" label="JavaScript">
    Install the package:
    (code block: npm install watermelondb)
  </TabItem>
</Tabs>

After (Spanish):

<Tabs>
  <TabItem value="js" label="JavaScript">
    Instala el paquete:
    (code block: npm install watermelondb)
  </TabItem>
</Tabs>

If components DO break: This is a bug! Contact support immediately with your repository URL.

Can I translate only specific pages or sections?

Yes! Use file allow/disallow patterns.

master = DocusaurusMaster(
    source_repo_url="https://github.com/your-org/docs",
    target_languages=["es", "fr"],

    # Only translate docs/ folder
    file_allow_patterns=[r"docs/.*\.mdx?$"],

    # Exclude internal documentation
    file_disallow_patterns=[
        r"docs/internal/.*",
        r"docs/draft/.*"
    ],
)

Use cases:

Translate only public documentation
Skip draft/WIP pages
Exclude internal notes
Translate blog separately from docs

What happens if translation fails mid-process?

PageTurner is resilient to interruptions:

Automatic recovery:

✅ Translation memory saves progress continuously
✅ Completed translations are preserved
✅ Re-running continues from last successful state
✅ Failed segments automatically retried

What gets saved:

Completed file translations
Translation memory entries
Repository state
Deployment status

To recover: Simply re-run the same command - it will skip completed work and resume.

Troubleshooting

"GitHub repository creation failed"

Possible causes:

1. Insufficient permissions

Error: Resource not accessible by integration

Fix: Ensure GitHub PAT has repo scope:

Go to GitHub Settings → Tokens
Click your token
Check repo is selected
Regenerate if needed

2. Target organization doesn't exist

Error: Organization 'your-org-i18n' not found

Fix: Either create the organization or omit github_target_org parameter:

# Use default (creates in same org as source)
master = DocusaurusMaster(
    source_repo_url="https://github.com/your-org/docs",
    # Don't specify github_target_org
)

3. Repository name already exists

Error: Repository 'docs-es' already exists

Fix: Either delete the existing repo or use a different target org.

"LLM translation error" or "Anthropic API error"

Possible causes:

1. Invalid API key

Error: Invalid API key

Fix: Verify your Anthropic API key:

Visit Anthropic Console
Check key is active
Copy key exactly (starts with sk-ant-)

2. Insufficient credits

Error: Insufficient credits

Fix: Add credits to Anthropic account:

Go to Billing in Anthropic Console
Add $20+ (covers ~50,000 pages)

3. Rate limit exceeded

Error: 429 Too Many Requests

Fix: Rate limits vary by account tier:

Free tier: 50 requests/minute
Paid tier: 1000 requests/minute

PageTurner automatically retries with backoff. If persists, reduce concurrency or upgrade Anthropic tier.

"Vercel deployment shows 404" or "Build failed"

Possible causes:

1. Missing vercel.json

Error: Build failed - No build command specified

Fix: Ensure source repository has proper Docusaurus configuration. PageTurner copies build configuration from source repo.

2. Build errors in translated content Check Vercel build logs:

Go to Vercel dashboard
Select deployment
View build logs

Common issues:

Broken Markdown syntax (usually preserved correctly)
Missing imports (check if source builds)
Configuration errors

Fix: If PageTurner introduced errors, contact support with build logs.

3. Wrong branch deployed Fix: Check Vercel project settings:

Go to Vercel project settings
Git → Production Branch
Should be main or master

"Translation quality is poor" or "Doesn't sound natural"

Causes and fixes:

1. Source content quality

Problem: Source docs use inconsistent terminology or unclear writing
Fix: Improve source documentation first
- Use consistent terms in English
- Define abbreviations on first use
- Write clearly and concisely

2. Missing context

Problem: Technical terms without explanation
Fix: Add brief explanations or glossary in source docs

3. Wrong terminology

Problem: Technical term translated incorrectly
Fix: Use translation memory to provide preferred translation:

from transaurus import GlobalTMManager

tm = GlobalTMManager()
tm.add_translation(
    source_text="webhook",
    target_language="es",
    translation="webhook"  # Keep as-is
)

4. Cultural mismatch

Problem: Some expressions don't translate well culturally
Fix: Rephrase source content or post-edit translation

Expected quality:

Technical documentation: 91-95/100 (excellent)
Marketing content: 85-90/100 (good, may need human review)
Creative content: 80-85/100 (acceptable, likely needs editing)

"Translation is taking too long"

Normal durations:

100 pages, 3 languages: 20-30 minutes
200 pages, 5 languages: 60-90 minutes

If significantly slower:

1. Check LLM provider rate limits

Free tier: 50 req/min → slower processing
Paid tier: 1000 req/min → faster

2. Large content segments

Very long pages take longer to translate
Consider breaking up mega-pages

3. Network issues

Check internet connection
Verify API endpoints accessible

4. High concurrency

PageTurner processes up to 100 files in parallel
May occasionally hit rate limits and back off

To monitor progress: Check console output for current translation status.

"Cost is higher than expected"

Understanding costs:

Token usage depends on:

Content length (longer pages = more tokens)
Number of languages
Translation memory reuse (first run vs. updates)

To estimate costs BEFORE running:

# Use DRY_RUN mode
result = master.run_full_translation_pipeline(run_mode="DRY_RUN")
print(f"Estimated cost: ${result['estimated_cost']}")

To reduce costs:

Use translation memory (60-80% savings on updates)
Translate incrementally (start with high-priority pages)
Exclude unnecessary files (use file_allow_patterns)
Share TM across projects (reuse translations)

Still Need Help?

Get Support

Documentation:

Quickstart Guide - Get started in 10 minutes
How It Works - Understand the pipeline
Configuration Reference - All settings explained

Contact us:

Email: support@pageturner.ai
Contact form: pageturner.ai/contact
Book a demo: calendly.com/pageturner/demo

Community:

GitHub Issues: github.com/pageturner/issues
Discord: discord.gg/pageturner (coming soon)

Response times:

Beta users: 24-48 hours
Open source projects: 2-5 business days
Enterprise: 4-hour SLA (coming Q2 2025)

Didn't Find Your Question?

We're continuously updating this FAQ based on real user questions.

Submit your question:

Contact us with your question
We'll answer personally
We'll add it to this FAQ for others

Help us improve: If you found this FAQ helpful (or not), let us know!

General Questions​

How is this different from Google Translate?​

What frameworks are supported?​

How much does it cost?​

How long does translation take?​

Can I customize translations?​

What languages are supported?​

Is my source code safe?​

Technical Questions​

What gets created when I run a translation?​

How do I handle updates to my documentation?​

Will my React components break?​

Can I translate only specific pages or sections?​

What happens if translation fails mid-process?​

Troubleshooting​

"GitHub repository creation failed"​

"LLM translation error" or "Anthropic API error"​

"Vercel deployment shows 404" or "Build failed"​

"Translation quality is poor" or "Doesn't sound natural"​

"Translation is taking too long"​

"Cost is higher than expected"​

Still Need Help?​

Get Support​

Didn't Find Your Question?​

General Questions

How is this different from Google Translate?

What frameworks are supported?

How much does it cost?

How long does translation take?

Can I customize translations?

What languages are supported?

Is my source code safe?

Technical Questions

What gets created when I run a translation?

How do I handle updates to my documentation?

Will my React components break?

Can I translate only specific pages or sections?

What happens if translation fails mid-process?

Troubleshooting

"GitHub repository creation failed"

"LLM translation error" or "Anthropic API error"

"Vercel deployment shows 404" or "Build failed"

"Translation quality is poor" or "Doesn't sound natural"

"Translation is taking too long"

"Cost is higher than expected"

Still Need Help?

Get Support

Didn't Find Your Question?