Book Analyzer

Book Analyzer

Extract structured bible files from source books for the writing framework.

Quick Start

# Analyze a single book
Analyze analysis/src/[book].txt

Output structure in analysis/output/[book-slug]/:

├── style.md
├── structure.md
├── characters/
│   └── [character].md
└── universe/
    └── [location].md

Workflow

1. Read source completely

Load the full text, note overall structure and length.

2. Extract style

• Identify POV (first/third limited/omniscient)
• Identify tenses (primary, secondary)
• Sample 50 sentences for average length
• List 20-30 characteristic vocabulary items
• List dialogue tags used
• Estimate dialogue-to-prose ratio
• Note description techniques

3. Extract characters

For each significant character:

• Identity (name, age, role)
• Physical description
• 3-5 core personality traits with evidence
• 5-10 dialogue quotes showing voice
• Relationships mapped

4. Extract structure

• Map acts (setup/development/climax/resolution)
• Note chapter count and length range
• Analyze chapter openings and endings
• Identify recurring patterns
• Track mystery mechanics if applicable

5. Extract universe

For each significant location:

• Physical description with quotes
• Atmosphere and sensory details
• Story function

6. Generate files

Create all output files following templates in bible/.

Evidence-Based Extraction

Every trait, pattern, or rule must be supported by:

• Direct quotes (with chapter reference)
• Quantitative data (averages, counts, percentages)

Bad: "Claude is stubborn" Good: "Claude is stubborn (Ch.2: refuses to answer to 'Claudine'; Ch.5: insists on leading despite danger)"

Output Format

Follow templates exactly:

• bible/style.md → output/[book]/style.md
• bible/characters/_template.md → output/[book]/characters/[name].md
• bible/universe/_template.md → output/[book]/universe/[name].md
• bible/structure.md → output/[book]/structure.md

For detailed extraction checklists, see references/extraction-guide.md.

Output Language

French (for extracted content and quotes) English (for metadata and structural annotations)