db_query / apps /fnb_parser.py
DavMelchi's picture
adding sample file to fn4b parser
a4525dd
"""
Streamlit application for extracting site and sector information from .docx design files.
The logic is adapted from `Sector Stacked.py` but provides an interactive UI where users can
upload one or many Word documents and instantly visualise / download the results.
"""
import io
import os
import re
from typing import List
import pandas as pd
import plotly.express as px
import streamlit as st
from docx import Document
from streamlit.commands.execution_control import rerun
###############################################################################
# --------------------------- Core extract logic -------------------------- #
###############################################################################
def extract_info_from_docx_separated_sectors(
docx_bytes: bytes, filename: str
) -> List[dict]:
"""Extract the site-level and sector-level information from a Word design file.
Parameters
----------
docx_bytes : bytes
Raw bytes of the `.docx` file – read directly from the Streamlit uploader.
filename : str
Original filename. Used only for reference in the output.
Returns
-------
list[dict]
A list containing up to three dictionaries – one for each sector.
"""
# python-docx can open a file-like object, so we wrap the bytes in BytesIO
doc = Document(io.BytesIO(docx_bytes))
# Shared site information
site_shared = {
"File": filename,
"Code": None,
"Site Name": None,
"Localité": None,
"Adresse": None,
"X": None,
"Y": None,
"Z": None,
"UTM_Zone": None,
}
# Per-sector placeholders (we assume max 3 sectors)
sector_data = {
"Azimuth": [None] * 3,
"Height": [None] * 3,
"MechTilt": [None] * 3,
"ElecTilt": [None] * 3,
}
# Iterate tables / rows / cells once, filling the data structures
for table in doc.tables:
for row in table.rows:
# Drop empty cells and overspaces
cells = [cell.text.strip() for cell in row.cells if cell.text.strip()]
if not cells:
continue
row_text_lower = " | ".join(cells).lower()
# Code (assumes pattern "T00" / "N01" typical of site codes)
if site_shared["Code"] is None and any("code" in c.lower() for c in cells):
for val in cells:
if ("t00" in val.lower()) or ("n01" in val.lower()):
site_shared["Code"] = val.replace(" ", "").strip()
break
# Site Name – same heuristic as original script
if site_shared["Site Name"] is None and any(
"nom" in c.lower() for c in cells
):
for val in cells:
if ("t00" in val.lower()) or ("n01" in val.lower()):
site_shared["Site Name"] = val.strip()
break
# UTM Zone
if site_shared["UTM_Zone"] is None:
utm_match = re.search(r"utm\s*(\d+)", row_text_lower)
if utm_match:
site_shared["UTM_Zone"] = f"UTM{utm_match.group(1)}"
# Localité and Adresse
if site_shared["Localité"] is None and any(
"localité" in c.lower() for c in cells
):
for val in cells:
if val.lower() != "localité:":
site_shared["Localité"] = val.strip()
break
if site_shared["Adresse"] is None and any(
"adresse" in c.lower() for c in cells
):
for val in cells:
if val.lower() != "adresse:":
site_shared["Adresse"] = val.strip()
break
# Coordinates (X, Y, Z)
if {"X", "Y", "Z"}.intersection(cells):
for i, cell_text in enumerate(cells):
text = cell_text.strip()
# X coordinate
if text == "X" and i + 1 < len(cells):
site_shared["X"] = cells[i + 1].strip()
# Y coordinate – could be in same cell e.g. "Y 123" or split
elif re.search(r"Y\s*[0-9]", text):
match = re.search(r"Y\s*([0-9°'\.\sWE]+)", text)
if match:
site_shared["Y"] = match.group(1).strip()
elif text == "Y" and i + 1 < len(cells):
site_shared["Y"] = cells[i + 1].strip()
# Z / Elevation
elif re.search(r"Z\s*[0-9]", text):
match = re.search(r"Z\s*([0-9]+)", text)
if match:
site_shared["Z"] = match.group(1).strip()
elif text == "Z" and i + 1 < len(cells):
z_val = re.search(r"([0-9]+)", cells[i + 1])
if z_val:
site_shared["Z"] = z_val.group(1).strip()
# Sector-specific lines
first_cell = cells[0].lower()
if first_cell == "azimut":
for i in range(min(3, len(cells) - 1)):
sector_data["Azimuth"][i] = cells[i + 1]
elif "hauteur des aériens" in first_cell:
for i in range(min(3, len(cells) - 1)):
sector_data["Height"][i] = cells[i + 1]
elif "tilt mécanique" in first_cell:
for i in range(min(3, len(cells) - 1)):
sector_data["MechTilt"][i] = cells[i + 1]
elif "tilt électrique" in first_cell:
for i in range(min(3, len(cells) - 1)):
sector_data["ElecTilt"][i] = cells[i + 1]
# Convert to per-sector rows
rows: List[dict] = []
for sector_id in range(3):
if sector_data["Azimuth"][sector_id]:
rows.append(
{
**site_shared,
"Sector ID": sector_id + 1,
"Azimuth": sector_data["Azimuth"][sector_id],
"Height": sector_data["Height"][sector_id],
"MechTilt": sector_data["MechTilt"][sector_id],
"ElecTilt": sector_data["ElecTilt"][sector_id],
}
)
return rows
def convert_coord_to_decimal(coord: str, default_direction: str | None = None):
"""Convert coordinate strings containing degrees/minutes/seconds to decimal degrees.
Handles various formats, e.g. "3° 33' 12.4\" W", "3 33 12.4 O", "-3.5534", "3.5534E".
West (W/O) or South (S) are returned as negative values.
Returns None if conversion fails.
"""
if coord is None or (isinstance(coord, float) and pd.isna(coord)):
return None
# Normalise the string – unify decimal separator and strip spaces
text = str(coord).replace(",", ".").strip()
if not text:
return None
# Detect hemisphere / direction letters
direction = None
match_dir = re.search(r"([NSEWnsewOo])", text)
if match_dir:
direction = match_dir.group(1).upper()
text = text.replace(match_dir.group(1), "") # remove letter for numeric parsing
else:
# No explicit letter – use supplied default if provided
if default_direction is not None:
direction = default_direction.upper()
# Grab all numeric components
nums = re.findall(r"[-+]?(?:\d+\.?\d*)", text)
if not nums:
return None
# Convert strings to float
nums_f = [float(n) for n in nums]
# Determine decimal value depending on how many components we have
if len(nums_f) >= 3:
deg, minute, sec = nums_f[0], nums_f[1], nums_f[2]
dec = deg + minute / 60 + sec / 3600
elif len(nums_f) == 2:
deg, minute = nums_f[0], nums_f[1]
dec = deg + minute / 60
else: # Already decimal degrees
dec = nums_f[0]
# Apply sign for West/Ouest/South
if direction in {"W", "O", "S"}: # West/Ouest or South => negative
dec = -abs(dec)
return dec
def process_files_to_dataframe(uploaded_files) -> pd.DataFrame:
"""Run extraction on the uploaded files and return a concatenated dataframe."""
all_rows: List[dict] = []
for uploaded in uploaded_files:
rows = extract_info_from_docx_separated_sectors(uploaded.read(), uploaded.name)
all_rows.extend(rows)
df = pd.DataFrame(all_rows)
# Add decimal conversion for X and Y
if not df.empty and {"X", "Y"}.issubset(df.columns):
df["X_decimal"] = df["X"].apply(
lambda c: convert_coord_to_decimal(c, default_direction="N")
)
df["Y_decimal"] = df["Y"].apply(
lambda c: convert_coord_to_decimal(c, default_direction="W")
)
return df
###############################################################################
# ----------------------------- Streamlit UI ------------------------------ #
###############################################################################
def main() -> None:
st.set_page_config(
page_title="F4NB Extractor to Excel", page_icon="📄", layout="wide"
)
st.title("📄 F4NB Extractor to Excel")
st.markdown(
"Convert F4NB Word documents into a tidy Excel / DataFrame containing site & sector information.\n"
"Upload one or many F4NB `.docx` files and hit **Process**."
)
# Download Sample file
fnb_sample_file_path = "samples/FN4B.docx"
# Create a download button
st.download_button(
label="Download FNB Sample File",
data=open(fnb_sample_file_path, "rb").read(),
file_name="fnb.docx",
mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
)
st.subheader("Upload Files")
uploaded_files = st.file_uploader(
"Select one or more F4NB `.docx` files",
type=["docx"],
accept_multiple_files=True,
)
process_btn = st.button("Process", type="primary", disabled=not uploaded_files)
if process_btn and uploaded_files:
with st.spinner("Extracting information…"):
df = process_files_to_dataframe(uploaded_files)
if df.empty:
st.warning(
"No data extracted. Check that the files conform to the expected format."
)
return
st.success(
f"Processed {len(uploaded_files)} file(s) – extracted {len(df)} sector rows."
)
st.dataframe(df, use_container_width=True)
st.markdown("---")
# Offer download as Excel
buffer = io.BytesIO()
with pd.ExcelWriter(buffer, engine="xlsxwriter") as writer:
df.to_excel(writer, index=False, sheet_name="Extract")
st.download_button(
label="💾 Download Excel",
data=buffer.getvalue(),
file_name="extracted_fnb.xlsx",
mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
on_click="ignore",
type="primary",
)
st.markdown("---")
# Interactive map of extracted coordinates using Plotly
if {"Y_decimal", "X_decimal"}.issubset(df.columns):
geo_df = (
df[["Y_decimal", "X_decimal", "Site Name", "Code"]]
.dropna()
.rename(columns={"Y_decimal": "Longitude", "X_decimal": "Latitude"})
.assign(
Size=lambda d: (
pd.to_numeric(d["Height"], errors="coerce").fillna(10)
if "Height" in d.columns
else 10
)
)
)
if not geo_df.empty:
st.subheader("🗺️ Site Locations")
fig = px.scatter_map(
geo_df,
lat="Latitude",
lon="Longitude",
hover_name="Site Name",
hover_data={"Code": True},
size="Size",
size_max=10,
zoom=6,
height=500,
)
fig.update_layout(
mapbox_style="open-street-map",
margin={"r": 0, "t": 0, "l": 0, "b": 0},
)
st.plotly_chart(fig, use_container_width=True)
if __name__ == "__main__": # pragma: no cover
main()