pyGAEB 1.3.0

Python parser for GAEB DA XML construction data exchange files, with LLM-powered item classification

pyGAEB

Python parser for GAEB DA XML construction data exchange files, with LLM-powered item classification.

pyGAEB parses, validates, classifies, and writes GAEB DA XML files (versions 2.0 through 3.3), producing a unified Pydantic v2 domain model from all inputs. An optional LLM classification layer enriches each item with a semantic construction element type via LiteLLM (100+ providers).

Installation

# Core parser + writer + export (zero LLM dependencies)
pip install pyGAEB

# With LLM classification (supports 100+ providers via LiteLLM)
pip install pyGAEB[llm]

Quick Start

Parse any GAEB file

from pygaeb import GAEBParser

doc = GAEBParser.parse("tender.X83")    # DA XML 3.x
doc = GAEBParser.parse("old.D83")       # DA XML 2.x — same call

print(doc.source_version)               # SourceVersion.DA_XML_33
print(doc.exchange_phase)               # ExchangePhase.X83
print(doc.grand_total)                  # Decimal("1234567.89")

Iterate items

for item in doc.award.boq.iter_items():
    print(item.oz)              # "01.02.0030"
    print(item.short_text)      # "Mauerwerk der Innenwand…"
    print(item.qty)             # Decimal("1170.000")
    print(item.unit)            # "m2"
    print(item.unit_price)      # Decimal("45.50")
    print(item.total_price)     # Decimal("53235.00")
    print(item.item_type)       # ItemType.NORMAL

Validation

from pygaeb import GAEBParser, ValidationMode

# Lenient (default) — collect warnings, keep parsing
doc = GAEBParser.parse("tender.X83")
for issue in doc.validation_results:
    print(issue.severity, issue.message)

# Strict — raise on first ERROR
doc = GAEBParser.parse("tender.X83", validation=ValidationMode.STRICT)

Write / Round-trip

from pygaeb import GAEBWriter, ExchangePhase
from decimal import Decimal

doc = GAEBParser.parse("tender.X83")
item = doc.award.boq.get_item("01.02.0030")
item.unit_price = Decimal("48.00")

GAEBWriter.write(doc, "bid.X84", phase=ExchangePhase.X84)

Export to JSON / CSV

from pygaeb.convert import to_json, to_csv

to_json(doc, "boq.json")     # full nested BoQ tree
to_csv(doc, "items.csv")     # flat item table with classification columns

LLM Classification

from pygaeb import LLMClassifier

# Default: in-memory cache (no disk I/O, session-scoped)
classifier = LLMClassifier(model="anthropic/claude-sonnet-4-6")
# classifier = LLMClassifier(model="gpt-4o")
# classifier = LLMClassifier(model="ollama/llama3")  # local, free, private

# Opt-in: persistent SQLite cache (survives across runs)
from pygaeb import SQLiteCache
classifier = LLMClassifier(model="anthropic/claude-sonnet-4-6", cache=SQLiteCache("~/.pygaeb/cache"))

# Check cost before running
estimate = await classifier.estimate_cost(doc)
print(f"Will classify {estimate.items_to_classify} items for ~${estimate.estimated_cost_usd:.2f}")

# Classify all items
await classifier.enrich(doc)

# Or synchronous
classifier.enrich_sync(doc)

for item in doc.award.boq.iter_items():
    if item.classification:
        print(item.oz, item.classification.element_type, item.classification.confidence)

Structured Extraction — Custom Schemas

After classification, extract typed attributes into your own Pydantic schema:

from pydantic import BaseModel, Field
from typing import Optional
from pygaeb import StructuredExtractor

class DoorSpec(BaseModel):
    door_type: str = Field("", description="single, double, sliding")
    width_mm: Optional[int] = Field(None, description="Width in mm")
    fire_rating: Optional[str] = Field(None, description="T30, T60, T90")
    glazing: bool = Field(False, description="Has glass panels")
    material: str = Field("", description="wood, steel, aluminium")

extractor = StructuredExtractor(model="anthropic/claude-sonnet-4-6")

# Extract from all items classified as "Door"
doors = await extractor.extract(doc, schema=DoorSpec, element_type="Door")
for item, spec in doors:
    print(item.oz, spec.door_type, spec.fire_rating, spec.width_mm)

# Filter by trade (broad) or sub_type (narrow)
pipes = await extractor.extract(doc, schema=PipeSpec, trade="MEP-Plumbing")
fire_doors = await extractor.extract(doc, schema=DoorSpec, sub_type="Fire Door")

# Or synchronous
doors = extractor.extract_sync(doc, schema=DoorSpec, element_type="Door")

Built-in starter schemas: DoorSpec, WindowSpec, WallSpec, PipeSpec — or define your own.

Custom Cache Backend

from pygaeb import CacheBackend, InMemoryCache, SQLiteCache

# Default: in-memory (no disk, session-scoped)
classifier = LLMClassifier()

# Persistent: SQLite
classifier = LLMClassifier(cache=SQLiteCache("~/.pygaeb/cache"))

# Share one backend between classifier and extractor
shared = SQLiteCache("/tmp/project-cache")
classifier = LLMClassifier(cache=shared)
extractor = StructuredExtractor(cache=shared)

# Bring your own: implement CacheBackend protocol
class RedisCache:
    def get(self, key: str) -> str | None: ...
    def put(self, key: str, value: str) -> None: ...
    def delete(self, key: str) -> None: ...
    def keys(self) -> list[str]: ...
    def clear(self) -> None: ...
    def close(self) -> None: ...

classifier = LLMClassifier(cache=RedisCache())

Cross-Phase Validation

from pygaeb import GAEBParser, CrossPhaseValidator

tender = GAEBParser.parse("tender.X83")
bid = GAEBParser.parse("bid.X84")

issues = CrossPhaseValidator.check(source=tender, response=bid)
for issue in issues:
    print(issue.severity, issue.message)

Supported Versions

Version	Parser Track	Status
DA XML 2.0	Track A (German elements)	✅ v1.0
DA XML 2.1	Track A (German elements)	✅ v1.0
DA XML 3.0	Track B (English elements)	✅ v1.0
DA XML 3.1	Track B (English elements)	✅ v1.0
DA XML 3.2	Track B (English elements)	✅ v1.0
DA XML 3.3	Track B (English elements)	✅ v1.0
GAEB 90	Track C (fixed-width)	🔜 v1.1

Configuration

# Environment variables
export PYGAEB_DEFAULT_MODEL=ollama/llama3
export PYGAEB_XSD_DIR=/opt/gaeb-schemas

# Or programmatic
from pygaeb import PyGAEBSettings
settings = PyGAEBSettings(default_model="gpt-4o", classifier_concurrency=10)

License

MIT — see LICENSE for details.