LibreOffice Calc — Python Manipulation Guide

This skill teaches setup-gen (create/modify xlsx) and reward-gen (read/verify xlsx) how to work with spreadsheet files using pure Python code.

• Library: openpyxl (+ pandas for bulk data)
• Install: pip3 install openpyxl pandas

---

0. GUI Startup on VM (for setup-gen)

After generating /home/user/<task_id>_initial.xlsx, setup-gen should leave the task in a GUI-ready state by opening Calc (and any additional required app windows).

CRITICAL VM LIMIT: GUI apps must use DISPLAY=:0, otherwise open commands may fail silently or open nowhere.

import os
import shlex
import subprocess
import time

def launch_gui(command: str, delay_sec: float = 1.0):
    env = os.environ.copy()
    env["DISPLAY"] = ":0"
    subprocess.Popen(
        shlex.split(command),
        stdout=subprocess.DEVNULL,
        stderr=subprocess.DEVNULL,
        env=env,
    )
    time.sleep(delay_sec)

# Open the initial workbook for the GUI agent
launch_gui(f'libreoffice --calc "{output_path}"', delay_sec=2.0)

# Optional multi-app startup example
# launch_gui('nautilus "/home/user"', delay_sec=1.0)

Guidelines:

• Open *_initial.xlsx, never *_golden.xlsx.
• Use non-blocking launch (Popen) so initial_setup.py can exit.
• Add short delays between app launches for stability.

---

1. Creating & Writing Files (setup-gen)

Basic Workbook

import openpyxl
from openpyxl.styles import Font, PatternFill, Alignment, Border, Side

wb = openpyxl.Workbook()
ws = wb.active
ws.title = "Sales"

# Write cell by cell
ws.cell(row=1, column=1, value="Name")
ws["B1"] = "Revenue"

# Write rows in bulk
data = [["Alice", 50000], ["Bob", 72000], ["Carol", 61000]]
for r, row in enumerate(data, 2):
    for c, val in enumerate(row, 1):
        ws.cell(row=r, column=c, value=val)

# Add sheets
ws2 = wb.create_sheet("Summary")
ws3 = wb.create_sheet("Charts")

# Reorder sheets (optional)
wb.move_sheet("Summary", offset=-1)  # move left

wb.save("/home/user/output.xlsx")

Styling Cells

# --- Font ---
cell.font = Font(name="Arial", size=12, bold=True, italic=False,
                 color="FF0000")  # 6-char RGB for Font constructor

# --- Background Color (PatternFill) ---
# CRITICAL: use start_color/end_color, openpyxl maps this to fgColor internally
cell.fill = PatternFill(start_color="FF4472C4", end_color="FF4472C4",
                        fill_type="solid")
# 8-char ARGB: FF=alpha, 44=R, 72=G, C4=B

# --- Alignment ---
cell.alignment = Alignment(horizontal="center", vertical="center",
                           wrap_text=True)

# --- Borders ---
thin = Side(style="thin", color="000000")
cell.border = Border(left=thin, right=thin, top=thin, bottom=thin)

# --- Number Formats ---
cell.number_format = '#,##0.00'       # 1,234.56
cell.number_format = '0.00%'          # 12.50%
cell.number_format = 'yyyy-mm-dd'     # 2024-01-15
cell.number_format = '$#,##0.00'      # $1,234.56
cell.number_format = '0'              # integer

Color Gotchas (CRITICAL)

openpyxl stores colors in ARGB hex (8 characters):

FF4472C4 = alpha(FF) + R(44) + G(72) + B(C4)
FFFF0000 = opaque red
FF00FF00 = opaque green
FF0000FF = opaque blue

• PatternFill(start_color="4472C4") → openpyxl prepends 00 → stored as 004472C4 (transparent!)
• Always use 8-char ARGB: PatternFill(start_color="FF4472C4", ...)
• Font(color="FF0000") → stored as 00FF0000. For Font, 6-char is fine since alpha doesn't matter visually.
• When reading back, cell.fill.fgColor.rgb returns the 8-char ARGB string.

Merged Cells

ws.merge_cells("A1:D1")
ws["A1"] = "Quarterly Report"
ws["A1"].font = Font(size=16, bold=True)
ws["A1"].alignment = Alignment(horizontal="center")
# Only A1 holds data; B1,C1,D1 become MergedCell (value=None)

Formulas

ws["C2"] = "=A2+B2"
ws["D10"] = "=SUM(D2:D9)"
ws["E2"] = "=VLOOKUP(A2,Sheet2!A:B,2,FALSE)"

# GOTCHA: openpyxl does NOT evaluate formulas.
# The .value will be the formula string, not the computed result.
# To get computed values, you need LibreOffice to open & save the file first.

Charts

from openpyxl.chart import BarChart, LineChart, PieChart, ScatterChart, Reference, Series

# --- Bar/Column Chart ---
chart = BarChart()
chart.type = "col"          # "col"=vertical columns, "bar"=horizontal bars
chart.title = "Monthly Sales"
chart.y_axis.title = "Revenue ($)"
chart.x_axis.title = "Month"
data = Reference(ws, min_col=2, min_row=1, max_col=3, max_row=13)
cats = Reference(ws, min_col=1, min_row=2, max_row=13)
chart.add_data(data, titles_from_data=True)
chart.set_categories(cats)
ws.add_chart(chart, "F2")

# --- Line Chart ---
line = LineChart()
line.title = "Trend"
line.add_data(Reference(ws, min_col=2, min_row=1, max_row=13), titles_from_data=True)
line.set_categories(Reference(ws, min_col=1, min_row=2, max_row=13))
ws.add_chart(line, "F18")

# --- Pie Chart ---
pie = PieChart()
pie.title = "Distribution"
pie.add_data(Reference(ws, min_col=2, min_row=1, max_row=6), titles_from_data=True)
pie.set_categories(Reference(ws, min_col=1, min_row=2, max_row=6))
ws.add_chart(pie, "F34")

# --- Scatter Chart ---
scatter = ScatterChart()
scatter.title = "X vs Y"
x_vals = Reference(ws, min_col=1, min_row=2, max_row=10)
y_vals = Reference(ws, min_col=2, min_row=2, max_row=10)
series = Series(y_vals, x_vals, title="Series 1")
scatter.series.append(series)
ws.add_chart(scatter, "F50")

Freeze Panes

ws.freeze_panes = "A2"    # freeze row 1 (header row)
ws.freeze_panes = "B1"    # freeze column A
ws.freeze_panes = "C3"    # freeze rows 1-2 and columns A-B
ws.freeze_panes = None     # no freeze

Row/Column Dimensions

ws.row_dimensions[1].height = 30       # row height in points
ws.row_dimensions[5].hidden = True     # hide row 5
ws.column_dimensions["A"].width = 25   # column width in characters
ws.column_dimensions["C"].hidden = True # hide column C

Zoom

ws.sheet_view.zoomScale = 150  # 150%

Auto-Filter

ws.auto_filter.ref = "A1:F20"
# NOTE: openpyxl can define the filter range but cannot actually
# apply/execute the filter. The filtering only takes effect when
# the file is opened in LibreOffice/Excel.

Data Validation (Dropdowns)

from openpyxl.worksheet.datavalidation import DataValidation

dv = DataValidation(
    type="list",
    formula1='"High,Medium,Low"',
    allow_blank=True,
    showDropDown=False,  # COUNTERINTUITIVE: False = show the dropdown!
)
dv.error = "Invalid priority"
dv.errorTitle = "Error"
dv.prompt = "Select priority level"
dv.promptTitle = "Priority"
dv.add("C2:C100")
ws.add_data_validation(dv)

Conditional Formatting

from openpyxl.formatting.rule import CellIsRule, FormulaRule
from openpyxl.styles.differential import DifferentialStyle

# Highlight cells > 100 in red
red_fill = PatternFill(start_color="FFFF0000", end_color="FFFF0000", fill_type="solid")
ws.conditional_formatting.add("B2:B50",
    CellIsRule(operator="greaterThan", formula=["100"],
              fill=red_fill))

# Formula-based rule
green_fill = PatternFill(start_color="FF00FF00", end_color="FF00FF00", fill_type="solid")
ws.conditional_formatting.add("A2:A50",
    FormulaRule(formula=['ISBLANK(A2)'], fill=green_fill))

Sheet Visibility

ws.sheet_state = "visible"      # default
ws.sheet_state = "hidden"       # hidden from UI, accessible via menu
ws.sheet_state = "veryHidden"   # only accessible via macro

Pivot Tables

openpyxl cannot create pivot tables from scratch. It can only preserve existing ones. If your task needs a pivot table, you must:

1. Create one in a template file using LibreOffice GUI
2. Use that file as the initial file, then copy-and-modify

---

2. Reading & Verifying Files (reward-gen)

Loading and Inspecting

import openpyxl
from openpyxl.cell.cell import MergedCell

wb = openpyxl.load_workbook("/path/to/file.xlsx")

# Sheet names
print(wb.sheetnames)  # ['Sales', 'Summary', 'Charts']

# Access sheet
ws = wb["Sales"]
# or by index: ws = wb.worksheets[0]

Reading Cell Values

# Direct access
val = ws["A1"].value        # returns raw value or formula string
val = ws.cell(row=1, column=1).value

# Read a range
for row in ws.iter_rows(min_row=2, max_row=10, min_col=1, max_col=3):
    for cell in row:
        print(cell.value)

# IMPORTANT: for formula cells, .value returns the formula string "=SUM(...)".
# To get computed values, load with data_only=True:
wb2 = openpyxl.load_workbook("file.xlsx", data_only=True)
val = wb2["Sales"]["C2"].value  # returns cached computed value (or None if never opened by Calc)

Verifying Cell Styles

cell = ws["A1"]

# Font
cell.font.bold        # True/False
cell.font.italic      # True/False
cell.font.underline   # 'single', 'double', None
cell.font.size        # 11, 12, etc.
cell.font.name        # "Arial", "Calibri", etc.
cell.font.color.rgb   # "00FF0000" (ARGB string) — may be None if theme color

# Background color
cell.fill.fgColor.rgb   # "FF4472C4" (ARGB) — THIS is the visible background
cell.fill.fill_type     # "solid", "none", etc. (via .patternType or .tagname)

# Alignment
cell.alignment.horizontal  # "center", "left", "right", "general"
cell.alignment.vertical    # "center", "top", "bottom"
cell.alignment.wrap_text   # True/False

# Border
cell.border.left.style   # "thin", "medium", "thick", None
cell.border.left.color   # Color object

# Number format
cell.number_format  # '#,##0.00', '0.00%', 'yyyy-mm-dd', 'General'

# Merged cell check
isinstance(cell, MergedCell)  # True if part of a merge (not the top-left)

Common Verification Patterns for reward.py

def check_value(ws, coord, expected, tolerance=0.01):
    """Check cell value with numeric tolerance."""
    val = ws[coord].value
    if val is None:
        return False
    if isinstance(expected, (int, float)):
        try:
            return abs(float(val) - expected) <= tolerance
        except (ValueError, TypeError):
            return False
    return str(val).strip() == str(expected).strip()

def check_bold(ws, coord):
    """Check if cell has bold font."""
    return ws[coord].font.bold == True

def check_bgcolor(ws, coord, expected_argb):
    """Check cell background color. expected_argb like 'FF4472C4'."""
    try:
        return ws[coord].fill.fgColor.rgb == expected_argb
    except:
        return False

def check_font_color(ws, coord, expected_argb):
    """Check font color."""
    try:
        return ws[coord].font.color.rgb == expected_argb
    except:
        return False

def check_merged(ws, coord):
    """Check if a cell is part of a merged range (not the top-left)."""
    return isinstance(ws[coord], MergedCell)

def check_formula(ws, coord, expected_formula):
    """Check if cell contains expected formula (case-insensitive)."""
    val = ws[coord].value
    if not isinstance(val, str):
        return False
    return val.upper().replace(" ", "") == expected_formula.upper().replace(" ", "")

Verifying Sheet Structure

# Sheet count and names
assert wb.sheetnames == ["Sales", "Summary"]
assert len(wb.sheetnames) == 2
assert "Sales" in wb.sheetnames

# Row/column count (using pandas for data shape)
import pandas as pd
df = pd.read_excel("file.xlsx", sheet_name="Sales")
assert len(df) >= 10          # at least 10 data rows
assert "Revenue" in df.columns # has expected column

Verifying Charts

ws = wb["Charts"]
charts = ws._charts  # list of chart objects

assert len(charts) >= 1

chart = charts[0]
print(chart.type)      # e.g., "barChart", "lineChart"
print(chart.title)     # chart title string or None
print(chart.y_axis.title)
print(chart.x_axis.title)
print(len(chart.series))  # number of data series

Verifying Freeze Panes

assert ws.freeze_panes == "A2"  # or "B1", "C3", None

Verifying Data Validation

validations = ws.data_validations.dataValidation
assert len(validations) >= 1

dv = validations[0]
print(dv.type)       # "list", "whole", "decimal", etc.
print(dv.formula1)   # '"High,Medium,Low"'
print(dv.sqref)      # CellRange of where it applies

Verifying Row/Column Properties

# Hidden rows
assert ws.row_dimensions[5].hidden == True

# Row height
assert ws.row_dimensions[1].height == 30

# Hidden columns
assert ws.column_dimensions["C"].hidden == True

# Column width
assert ws.column_dimensions["A"].width >= 20

Verifying Filters

if ws.auto_filter.ref:
    print(ws.auto_filter.ref)  # e.g., "A1:F20"

Verifying Conditional Formatting

cf_rules = ws.conditional_formatting
for cf in cf_rules:
    print(cf)           # cell range
    for rule in cf.rules:
        print(rule.type, rule.operator, rule.formula)

Bulk Data Comparison (pandas)

import pandas as pd

df_result = pd.read_excel("result.xlsx", sheet_name="Sales")
df_golden = pd.read_excel("golden.xlsx", sheet_name="Sales")

# Exact match
assert df_result.round(4).equals(df_golden.round(4))

# Or partial: just check specific columns exist and have right shape
assert list(df_result.columns) == list(df_golden.columns)
assert len(df_result) == len(df_golden)

---

3. Bitter Lessons

1. Formula values are NOT computed by openpyxl. cell.value returns the formula string "=SUM(A1:A10)", not the result. Use data_only=True to get the last-cached value (requires the file to have been opened in Calc/Excel at least once). For reward scripts that check formula outputs: either verify the formula string itself, or use pandas read_excel which reads cached values.

2. Always use 8-char ARGB for colors. PatternFill(start_color="4472C4") silently becomes 004472C4 (alpha=00, transparent). Always write "FF4472C4". When reading back, compare against the 8-char form.

3. fgColor is the visible background, not bgColor. This is counterintuitive. cell.fill.fgColor.rgb gives you the background color you see. bgColor is rarely what you want.

4. Merged cells: only top-left has data. After merge_cells("A1:D1"), cells B1/C1/D1 become MergedCell objects with value=None. Style the top-left cell only.

5. Copy-then-modify for golden files. Never create the golden file from scratch if an initial file exists. shutil.copy(initial, golden) then open and modify. This preserves metadata, print settings, and other invisible properties that a from-scratch file would lack.

6. Pivot tables cannot be created by openpyxl. openpyxl can only read/preserve existing pivot tables. If you need to create one, use a template file with the pivot table already built.

7. Filters are definition-only. ws.auto_filter.ref = "A1:D20" sets the range, but rows are NOT actually hidden/filtered. Filtering only takes effect when the file is opened in LibreOffice.

8. showDropDown=False means SHOW the dropdown. In DataValidation, this boolean is inverted from what you'd expect. Set False to display the dropdown arrow.

9. Chart type naming. BarChart(type="col") = vertical column chart. BarChart(type="bar") = horizontal bar chart. This does not match LibreOffice's menu names.

10. Styles are immutable after assignment. You cannot do cell.font.bold = True. You must create a new Font object: cell.font = Font(bold=True, ...). Same for Fill, Alignment, Border.

11. Theme colors may return None for .rgb. If a cell uses a theme color instead of explicit RGB, cell.font.color.rgb can be None or a theme index. Always wrap color reads in try/except.

12. data_only=True loses formulas. Loading with data_only=True replaces formulas with their cached values. If you need both the formula and the value, load the file twice — once normally, once with data_only=True.