qdas-parser

A Python parser for the Q-DAS ASCII Transfer Format (V12).
Reads .dfd description files paired with .dfx (process) or .dfb (batch) value files, decodes K-Field metadata, and returns structured pandas DataFrames ready for downstream quality analysis.

---

Features

• Parse Q-DAS .dfd + .dfx/.dfb file pairs into tidy pandas DataFrames
• Full K-Field decoding — field names and coded values translated via typed Python registries (REQUIRED, DEFINED, SUPPORTED, CATALOG)
• MultiIndex columns (module, feature) for easy slicing across assembly lines
• Production order normalisation via format_order() / ProductionOrder
• Optional test-cell detection through an injected module map
• Pure Python — no compiled extensions required
• Typed API with __slots__ throughout for low-memory footprint

---

Requirements

Dependency	Version
Python	≥ 3.13
pandas	≥ 3.0
numpy	≥ 2.4
python-dateutil	≥ 2.8

---

Installation

pip install qdas-parser

---

Quick start

from pathlib import Path
from qdas_parser import QDASFileParser

qdas = QDASFileParser(
    Path("data/bd/m01/2603160002.dfd"),
    product="ACT1",
)
qdas.parse_description()
qdas.parse_values()

df  = qdas.dataframe()   # measurement data as MultiIndex DataFrame
dfm = qdas.metadata()    # feature metadata as MultiIndex DataFrame

The DataFrame index is ['Auftragsnummer', 'Seriennummer'].
Columns are a two-level MultiIndex: (module_name, feature_label).

---

File format overview

A Q-DAS measurement set consists of two files sharing the same stem:

File	Role
<stem>.dfd	Description file — K-Field header and feature metadata
<stem>.dfx	Value file for process data (.dfx)
<stem>.dfb	Value file for batch data (.dfb)

Each line in the .dfd file is a K-Field:

K0100 3               ← number of features
K1001 1234567         ← part number (Teilenummer)
K2002/1 Diameter      ← feature 1 label
K2142/1 mm            ← feature 1 unit

---

Public API

QDASFileParser

Main entry point.

from qdas_parser import QDASFileParser

parser = QDASFileParser(
    description_file="2603160002.dfd",
    product="ACT1",
    kind="bd",                          # 'bd' | 'pc' — inferred from path if omitted
    tc_modules={"ACT1": {"as1": ["tc1_bd", "tc2_bd"]}},
    tc_shortcut="TC",
    index_columns=["Auftragsnummer", "Seriennummer"],  # default; customise as needed
)
parser.parse_description()
parser.parse_values()

df  = parser.dataframe()
dfm = parser.metadata()

format_order

Normalise raw order numbers from any source to a 12-digit zero-padded string:

from qdas_parser import format_order

df["order"] = df["order"].map(format_order)
# '1234567'  →  '000001234567'
# 1234567    →  '000001234567'
# ''         →  ''

ProductionOrder

A lightweight value object wrapping a formatted order number:

from qdas_parser import ProductionOrder

po = ProductionOrder("1234567")
str(po)       # '000001234567'
int(po)       # 1234567
bool(po)      # True
po == 1234567 # True  — comparison normalises both sides

KField

Low-level K-Field model, useful for custom description-file processing:

from qdas_parser import KField

kf = KField.from_line("K2002/1 Diameter\n")
kf.key            # 'K2002'
kf.value          # 'Diameter'
kf.feature_number # 1
kf.feature_index  # 0  (zero-based)

name, value = kf.decode()

QDAS constants

from qdas_parser import QDAS

QDAS.SEP_F          # feature separator character (ASCII 15)
QDAS.SEP_E          # extension separator character (ASCII 20)
QDAS.TIMESTAMP      # 'Zeitstempel'
QDAS.ORDER          # 'Auftragsnummer'
QDAS.PART_ID        # 'Seriennummer'

---

Running the tests

pip install pytest pytest-benchmark
pytest tests/

Benchmarks only:

pytest tests/bench_fast.py --benchmark-only -v

---

License

MIT — see LICENSE.