---

name: Aggregation and Export description: Aggregate flat building elements into hierarchical construction documents with 4-level cost breakdown structure.

Aggregation and Export

Maps flat JSON input to hierarchical DocumentConstructionDashboard schema with automatic parent section generation and total calculations.

When to Use This Skill

Use this skill when the user asks about:

• Saving estimation results
• Generating final construction documents
• Creating cost breakdowns
• Exporting project summaries
• Organizing elements into construction sections

Workflow Overview

This is the final step in the estimation pipeline:

CopyIngestion → Estimation Strategy → Aggregation & Export

Purpose: Take flat pricing results and:

1. Parse building IDs and hierarchical structure
2. Generate missing parent sections automatically
3. Calculate totals bottom-up through the hierarchy
4. Create DocumentConstructionDashboard with pricing lines
5. Save final output to JSON file

Processing Architecture

Hierarchy Building

The script analyzes building IDs (e.g., "1", "1.1", "1.1.1", "1.1.1.1") to:

• Extract parent sections that need to be created
• Determine element types (titre, section, subsection, line)
• Calculate indent levels based on depth
• Build section title mapping from Roman numerals

Type Determination

Based on indent level and whether an element has children:

• Level 0 ("1", "2"): titre or line
• Level 1 ("1.1", "2.1"): section or line
• Level 2 ("1.1.1"): subsection or line
• Level 3+ ("1.1.1.1"): line

Total Calculation

Calculates totals bottom-up:

1. Start with leaf elements (actual building elements)
2. Sum direct children for each parent
3. Propagate totals up through hierarchy

Usage

CLI Command

Copypython aggregation_and_export.py --path_temp_file /path/to/input.json --project_id project_123

Arguments:

• --path_temp_file (required): Path to input JSON file
• --project_id (optional): Project ID to use in output (overrides project.id from input JSON)

Output: aggregation_ui_<uuid>.json saved in same directory as input file

The script will print the output path and format information when complete.

Input Format

Flat array of building elements with project information:

Copy{
  "project": {
    "name": "Building A",
    "id": "project_123"
  },
  "elements": [
    {
      "building_id": "1.1.1",
      "building_name": "External Wall - North",
      "building_elm_section": "I - Préparation de chantier",
      "building_elm_unit": "m²",
      "building_elm_quantity": 50.0,
      "building_elm_unit_price": 150.0,
      "building_elm_total_ht": 7500.0,
      "project_id": "project_123",
      "project_name": "Building A"
    }
  ]
}

Key Fields:

• building_id: Hierarchical ID (e.g., "1", "1.1", "1.1.1")
• building_name: Element description
• building_elm_section: Section title with Roman numeral (e.g., "I - Title")
• building_elm_unit, building_elm_quantity, building_elm_unit_price, building_elm_total_ht: Pricing data

Output Format

Dashboard-ready JSON with hierarchical pricing lines:

Structure:

• projectId: Project identifier
• projectName: Project name
• pricingLines: Array of pricing lines in hierarchical order

Pricing Line Types:

• titre: Top-level section (indent 0)
• section: Second-level section (indent 1)
• subsection: Third-level section (indent 2)
• line: Leaf element with pricing details (indent 3+)

Parent Lines (titre, section, subsection):

• No unit, quantity, or unitPrice
• totalPrice calculated from children
• isExpanded = true (collapsible UI)
• confidence = null

Leaf Lines:

• Full pricing details (unit, quantity, unitPrice, totalPrice)
• Confidence level based on data completeness
• isExpanded = null (not collapsible)

The script will print the output file path when complete.

Schema Validation

All outputs follow Pydantic schemas defined in schemas.py:

• DocumentConstructionDashboard: Dashboard format with pricing lines
• PricingLine: Individual line with type, indentLevel, and optional pricing
• PricingLineType: "titre" | "section" | "subsection" | "line"
• ConfidenceLevel: "low" | "medium" | "high"

Confidence Calculation

Confidence level is determined by pricing data completeness:

• Low: unit_price or quantity is None, or total_ht is 0.0
• Medium: All pricing data present and valid

Error Handling

• Missing input file → Error and exit
• Invalid JSON → Error and exit with traceback
• Processing errors → Logged with full traceback

Key Features

• Automatic Hierarchy: Generates parent sections from building IDs
• Bottom-up Totals: Calculates totals by summing children
• Roman Numeral Parsing: Extracts section titles from input data
• Type Detection: Determines element type based on hierarchy position
• Confidence Scoring: Evaluates pricing data completeness
• Unique IDs: Generates stable IDs for each pricing line

Related Files

• schemas.py: Pydantic model definitions
• aggregation_and_export.py: Main script

---

Version: 2.0 (Simplified - No LLM) Last Updated: November 2025