flowmarg / core /xml_parser.py
akash
Deploy to Hugging Face (source code only, no binary files)
2379054
Raw History Blame Contribute Delete
14.5 kB
"""
XML Parser for Station Base and Geometry Reports
Network Rail Standards - Professional Edition
Complete Implementation - Your Exact Logic
"""
import pandas as pd
from lxml import etree
from typing import Optional, Tuple, List
from .engineering import num, station_of_point, valid, classify
# ============================================
# ALIGNMENT HELPERS
# ============================================
def alignment_names(df: pd.DataFrame) -> List[str]:
"""
Get unique alignment names from DataFrame.
YOUR EXACT LOGIC.
"""
if df is None or df.empty:
return []
if "offset_alignment" not in df.columns:
return []
return sorted([x for x in df["offset_alignment"].dropna().unique().tolist() if x])
def suggest_alignment(arr: List[str], kind: str, track_no: int) -> str:
"""
Suggest alignment name based on classification.
YOUR EXACT LOGIC.
"""
vals = [x for x in arr if classify(x) == kind]
return vals[min(max(track_no - 1, 0), len(vals) - 1)] if vals else (arr[0] if arr else "")
# ============================================
# STATION BASE PARSER (YOUR EXACT LOGIC)
# ============================================
def parse_station_base(xml_bytes: bytes) -> pd.DataFrame:
"""
Parse Station Base XML file.
YOUR EXACT LOGIC from parse_station_base().
Args:
xml_bytes: XML file content as bytes
Returns:
DataFrame with station data
"""
try:
root = etree.fromstring(xml_bytes, parser=etree.XMLParser(recover=True, huge_tree=True))
except Exception:
return pd.DataFrame()
rows = []
# Find all HorizontalAlignments
for ha in root.xpath(".//*[local-name()='HorizontalAlignment']"):
datum = ha.get("name", "")
# Find all StationOffsetPoints
for sop in ha.xpath(".//*[local-name()='StationOffsetPoint']"):
cl = sop.find("centerLinePoint")
ol = sop.find("offsetLinePoint")
cp = cl.find("point") if cl is not None else None
op = ol.find("point") if ol is not None else None
off = sop.find("offsetPoint")
# Get chainages
dch = station_of_point(cp)
och = station_of_point(op)
row = {
"datum_alignment": datum,
"offset_alignment": sop.get("offsetAlignmentName", ""),
"datum_chainage": dch,
"offset_own_chainage": och,
"chainage_creep_m": (och - dch) if dch is not None and och is not None else None,
"first_offset_m": num(sop.get("firstOffset")),
"center_radius_m": num(cl.get("instantaneousRadius")) if cl is not None else None,
"center_grade": num(cl.get("instantaneousGrade")) if cl is not None else None,
"center_versine_m": num(cl.get("versine")) if cl is not None else None,
"center_cant_m": num(cl.get("cant")) if cl is not None else None,
"offset_radius_m": num(ol.get("instantaneousRadius")) if ol is not None else None,
"offset_smoothing_radius_m": num(ol.get("smoothingRadius")) if ol is not None else None,
"offset_cant_m": num(ol.get("cant")) if ol is not None else None,
}
# Add point coordinates
for pt, prefix in [(cp, "center"), (op, "offset_line"), (off, "offset_point")]:
if pt is not None:
row[f"{prefix}_n"] = num(pt.get("northing"))
row[f"{prefix}_e"] = num(pt.get("easting"))
row[f"{prefix}_z"] = num(pt.get("elevation"))
# Get level - YOUR EXACT LOGIC
level_candidates = [
row.get("offset_line_z"),
row.get("offset_point_z"),
row.get("center_z")
]
row["level_m"] = next((x for x in level_candidates if x is not None and abs(x) > 1e-9), None)
if row["level_m"] is None:
row["level_m"] = row.get("center_z")
rows.append(row)
df = pd.DataFrame(rows)
# Sort - YOUR EXACT LOGIC
if not df.empty and "datum_chainage" in df.columns:
df = df.sort_values(["datum_alignment", "offset_alignment", "datum_chainage"])
df = df.reset_index(drop=True)
return df
def detected_summary(df: pd.DataFrame) -> pd.DataFrame:
"""
Generate summary of detected alignments.
YOUR EXACT LOGIC.
"""
if df is None or df.empty:
return pd.DataFrame()
# Ensure chainage_creep_m column exists to avoid KeyError in groupby
if "chainage_creep_m" not in df.columns:
df = df.copy()
df["chainage_creep_m"] = None
out = df.groupby(["datum_alignment", "offset_alignment"], dropna=False).agg(
Rows=("datum_chainage", "count"),
From_Ch=("datum_chainage", "min"),
To_Ch=("datum_chainage", "max"),
Min_Level=("level_m", "min"),
Max_Level=("level_m", "max"),
Min_Creep=("chainage_creep_m", "min"),
Max_Creep=("chainage_creep_m", "max")
).reset_index().rename(columns={
"datum_alignment": "Datum Alignment",
"offset_alignment": "Alignment"
})
out["Max Abs Creep (m)"] = out[["Min_Creep", "Max_Creep"]].abs().max(axis=1)
return out
def own_station_key(row) -> Optional[float]:
"""
Get own station key from a row.
YOUR EXACT LOGIC.
"""
if row is None:
return None
if valid(row.get("offset_own_chainage")) and abs(float(row.get("offset_own_chainage"))) > 1e-6:
return float(row.get("offset_own_chainage"))
return float(row.get("datum_chainage")) if valid(row.get("datum_chainage")) else None
def row_by_chainage(df: pd.DataFrame, alg: str, chainage: float, tol: float) -> Optional[pd.Series]:
"""
Find row by chainage.
YOUR EXACT LOGIC.
"""
w = rows_for_alignment(df, alg)
if w.empty or chainage is None:
return None
idx = (w.datum_chainage - float(chainage)).abs().idxmin()
return w.loc[idx] if abs(float(w.loc[idx, "datum_chainage"]) - float(chainage)) <= tol else None
def rows_for_alignment(df: pd.DataFrame, alg: str) -> pd.DataFrame:
"""
Get rows for a specific alignment.
YOUR EXACT LOGIC.
"""
if df is None or df.empty:
return pd.DataFrame()
if alg in df.get("offset_alignment", pd.Series(dtype=str)).dropna().unique().tolist():
return df[df.offset_alignment == alg].copy()
kind = classify(alg)
candidates = [x for x in alignment_names(df) if classify(x) == kind]
return df[df.offset_alignment == candidates[0]].copy() if candidates else pd.DataFrame()
# ============================================
# GEOMETRY REPORT PARSER (YOUR EXACT LOGIC)
# ============================================
def parse_geometry_report(xml_bytes: bytes) -> Tuple[pd.DataFrame, pd.DataFrame, pd.DataFrame]:
"""
Parse Geometry Report XML file.
YOUR EXACT LOGIC from parse_geometry_report().
Args:
xml_bytes: XML file content as bytes
Returns:
Tuple of (H_df, V_df, E_df) - Horizontal, Vertical, Event Points
"""
try:
root = etree.fromstring(xml_bytes, parser=etree.XMLParser(recover=True, huge_tree=True))
except Exception:
return pd.DataFrame(), pd.DataFrame(), pd.DataFrame()
H, V, E = [], [], []
for ha in root.xpath(".//*[local-name()='HorizontalAlignment']"):
alg = ha.get("name", "")
# ===== Horizontal Elements =====
he = ha.find("HorizontalElements")
if he is not None:
for el in he:
if not isinstance(el.tag, str):
continue
tag = etree.QName(el).localname
if tag not in ["HorizontalLine", "HorizontalCircle", "HorizontalSpiral"]:
continue
s, e, pi = el.find("Start"), el.find("End"), el.find("PI")
typ = {
"HorizontalLine": "Line",
"HorizontalCircle": "Curve",
"HorizontalSpiral": "Transition"
}.get(tag, "Unknown")
H.append({
"Alignment": alg,
"Element No": num(el.get("elementNumber")),
"Type": typ,
"Start Ch": station_of_point(s),
"End Ch": station_of_point(e),
"Length (m)": num(el.get("length")),
"Chainage Length (m)": (station_of_point(e) - station_of_point(s)) if station_of_point(s) is not None and station_of_point(e) is not None else None,
"Radius (m)": num(el.get("radius")),
"Start Radius (m)": num(el.get("startRadius")),
"End Radius (m)": num(el.get("endRadius")),
"Direction": el.get("rotationDirection", el.get("direction", "")),
"Design Speed": num(el.get("designSpeed")),
"Applied Cant (mm)": num(el.get("appliedCant")) or 0,
"A Constant": num(el.get("aConstant")),
"Start Level (m)": num(s.get("elevation")) if s is not None else None,
"End Level (m)": num(e.get("elevation")) if e is not None else None,
"PI Ch": station_of_point(pi),
"PI Level (m)": num(pi.get("elevation")) if pi is not None else None,
})
# ===== Vertical Elements =====
va = ha.find("VerticalAlignment")
if va is not None and va.find("VerticalElements") is not None:
for el in va.find("VerticalElements"):
if not isinstance(el.tag, str):
continue
tag = etree.QName(el).localname
if tag not in ["VerticalLine", "VerticalSymmetricalParabola"]:
continue
s, e, pvi = el.find("VerticalStart"), el.find("VerticalEnd"), el.find("PVI")
tp, sd = el.find("TurningPoint"), el.find("SightDistance")
if tag == "VerticalLine":
g = num(el.get("grade"))
V.append({
"Alignment": alg,
"Element Type": "Vertical Line",
"Start Ch": station_of_point(s),
"End Ch": station_of_point(e),
"Length (m)": num(el.get("length")),
"Start Level (m)": num(s.get("elevation")) if s is not None else None,
"End Level (m)": num(e.get("elevation")) if e is not None else None,
"Grade (%)": g * 100 if g is not None else None,
"Start Grade (%)": g * 100 if g is not None else None,
"End Grade (%)": g * 100 if g is not None else None,
"K Value": None,
"Vertical Curve Length (m)": None,
"PVI Ch": None,
"PVI Level (m)": None,
"Turning Point Type": "",
"Turning Point Ch": None,
"Turning Point Level (m)": None,
"Design Speed": num(sd.get("designSpeed")) if sd is not None else None
})
else:
sg, eg = num(el.get("startGrade")), num(el.get("endGrade"))
V.append({
"Alignment": alg,
"Element Type": "Vertical Curve",
"Start Ch": station_of_point(s),
"End Ch": station_of_point(e),
"Length (m)": num(el.get("verticalCurveLength")),
"Start Level (m)": num(s.get("elevation")) if s is not None else None,
"End Level (m)": num(e.get("elevation")) if e is not None else None,
"Start Grade (%)": sg * 100 if sg is not None else None,
"End Grade (%)": eg * 100 if eg is not None else None,
"K Value": num(el.get("kValue")),
"Vertical Curve Length (m)": num(el.get("verticalCurveLength")),
"PVI Ch": station_of_point(pvi),
"PVI Level (m)": num(pvi.get("elevation")) if pvi is not None else None,
"Turning Point Type": tp.get("pointType", "") if tp is not None else "",
"Turning Point Ch": station_of_point(tp),
"Turning Point Level (m)": num(tp.get("elevation")) if tp is not None else None,
"Design Speed": num(sd.get("designSpeed")) if sd is not None else None
})
# ===== Event Points =====
for ep in ha.xpath(".//*[local-name()='HorizontalComputedEventPoints']/*[local-name()='HorizontalEventPoint']"):
gp = ep.find("GeometryPoint")
g = num(ep.get("grade"))
E.append({
"Alignment": alg,
"Chainage": station_of_point(gp),
"Northing": num(gp.get("northing")) if gp is not None else None,
"Easting": num(gp.get("easting")) if gp is not None else None,
"Level (m)": num(gp.get("elevation")) if gp is not None else None,
"Radius (m)": num(ep.get("radius")),
"Grade (%)": g * 100 if g is not None else None,
"Point Type": gp.get("pointType", "") if gp is not None else ""
})
H_df = pd.DataFrame(H)
V_df = pd.DataFrame(V)
E_df = pd.DataFrame(E)
# Sort (fixed: previous version reassigned the loop variable `df`
# instead of H_df/V_df/E_df, so nothing was actually sorted)
if not H_df.empty and "Start Ch" in H_df.columns:
H_df = H_df.sort_values(["Alignment", "Start Ch"], ignore_index=True)
if not V_df.empty and "Start Ch" in V_df.columns:
V_df = V_df.sort_values(["Alignment", "Start Ch"], ignore_index=True)
if not E_df.empty and "Chainage" in E_df.columns:
E_df = E_df.sort_values(["Alignment", "Chainage"], ignore_index=True)
return H_df, V_df, E_df