elixcee

English | 日本語 | 中文

Run, test, and diagnose Excel VBA macros without requiring Microsoft Excel. elixcee is a Rust-powered headless VBA runtime for Linux, macOS, and Windows, with static checks, property-based workbook testing, and structured diagnostics for VBA and workbook-operation failures.

The core engine is written in Rust; Python bindings are provided via pyo3 + maturin.

Name

elixcee = Excel + elixir + C

An elixir that cures your Excel dependency — running at C-level speed via Rust.

Comparison with similar tools

Feature	elixcee	xlwings	LibreOffice UNO	openpyxl	xlcalculator
Runs VBA macros	Yes	Yes	Yes (subset)	No	No
Requires Excel	No	Yes	No	No	No
Requires LibreOffice	No	No	Yes	No	No
Evaluates formulas	Yes	Yes	Yes	No	Yes
macOS/Linux/Windows	Yes	partial	Yes	Yes	Yes
Simple Python API	Yes	Yes	No	Yes	Yes
Read .xlsx	Yes	Yes	Yes	Yes	Yes
Read .ods	Yes	Yes	Yes	No	No
Write .xlsx	Yes	Yes	Yes	Yes	No
Write .ods	Yes	Yes	Yes	No	No
Execution speed	Rust (native)	COM/IPC (slow)	IPC (slow)	—	Python

Notes:

xlwings requires Excel for Mac on macOS (via AppleScript) and Excel on Windows (via COM). Linux support requires a running Excel instance or a cloud bridge.
LibreOffice UNO has a slow startup (≥ 1 s process launch) and a complex API. It runs VBA via LibreOffice's own interpreter, which may not match Excel's behavior exactly.
openpyxl reads cached formula values from .xlsx files but does not re-evaluate formulas at runtime.
xlcalculator re-evaluates Excel formulas in Python but has no VBA support.
elixcee's VBA interpreter covers the subset of VBA used in typical data-processing macros (loops, conditionals, cell read/write, string/math functions, multi-sheet access). Most Excel UI operations such as charting and formatting are unsupported or no-ops. MsgBox is handled specially: depending on the mode, it's printed to stdout, collected in JSON output, or raised as an error.

Installation

pip install elixcee

Development build (from source):

python3 -m venv .venv && source .venv/bin/activate
maturin develop

CLI (Windows / Linux / macOS)

Pre-built binaries are available on the Releases page — no Python required.

Download	Platform
elixcee-x86_64-windows.exe	Windows x64
elixcee-x86_64-linux	Linux x64
elixcee-aarch64-macos	macOS Apple Silicon

Usage

elixcee <vba_file>... <MacroName> [OPTIONS]

Arguments:
  <vba_file>...  One or more VBA source files (.vbs / .bas / .txt). With
                 more than one, use Module.Sub to disambiguate same-named
                 Subs/Functions across modules.
  <MacroName>    Name of the Sub to execute (last argument)

Options:
  --file <path>    Load cell data from spreadsheet (.xlsx / .xlsm / .ods)
  --sheet <name>   Active sheet name (default: first sheet in --file)
  --output <path>  Save result cells to spreadsheet (.xlsx / .ods)
  --json           Emit a single JSON object (result or error) instead of plain text

Examples

Run a VBA file and print results to stdout:

elixcee macro.vbs ProcessData

Load data from an Excel file, run a macro, and save the output:

elixcee macro.vbs ProcessData --file input.xlsx --output result.xlsx

Output format — one line per non-empty cell, tab-separated address and value:

A1    Hello
B1    42
A2    3.14

MsgBox calls are printed to stdout.

Multiple files (multi-module projects)

Pass more than one source file to run a project spanning several modules. Sub/Function names are shared project-wide — use Module.Sub to pick a specific one if the bare name exists in more than one module (module names come from Attribute VB_Name if present, else the filename):

elixcee Helpers.bas Main.bas Main.ProcessData

There's no project manifest yet (see docs/agent-contract.md for exactly what is/isn't supported, including how cross-module name collisions are handled).

JSON output (for scripts / AI agents)

Add --json for a single machine-readable JSON object instead of plain text:

elixcee macro.vbs ProcessData --json

{"schema_version":1,"ok":true,"entrypoint":"ProcessData","duration_ms":0.42,"cells":[{"sheet":"sheet1","address":"A1","value":42}],"messages":[]}

Full contract — error codes, exit codes, messages semantics: docs/agent-contract.md.

Static analysis without running the macro

elixcee check inspects one or more .bas files without executing them: parse errors, whether the entrypoint macro exists, undefined Sub/Function calls anywhere in the body, and interactive MsgBox calls. Every positional argument is a file; the entrypoint (if any) is always --entry, never positional — so elixcee check *.bas checks every module in a project without asserting any particular entrypoint.

elixcee check macro.vbs --entry ProcessData --json

{"schema_version":1,"ok":true,"diagnostics":[]}

Workbook snapshot

elixcee snapshot reads a .xlsx/.xlsm/.ods file directly — no VBA execution — and prints every sheet's non-empty cells as Markdown by default, or JSON with --json:

elixcee snapshot Book1.xlsx --json

{"schema_version":1,"ok":true,"file":"Book1.xlsx","sheets":[{"name":"Sheet1","sheet_id":"1","stable_id":"sheet1","cells":[{"address":"A1","value":42}]}]}

stable_id is derived from the file's own sheetId when available (else a positional fallback) — it is not VBA's CodeName property. See docs/agent-contract.md for the full rationale.

Property-based workbook testing

elixcee test-workbook reruns a macro against a starting workbook many times with generated boundary-value inputs (blank, 0, 1, -1, near overflow, empty/short/long strings), checking every run for panics, runtime errors, timeouts, and Excel error values — each case starts from a completely fresh workbook state:

# fixture.toml
name = "order calculation"
workbook = "orders.xlsx"
vba_files = ["Main.bas"]
macro = "Main.Process"
cases = 100
seed = 42

[[inputs]]
range = "Input!B2:B10"
strategy = "boundary_numeric"

[[assertions]]
range = "Result!A1:F100"
rule = "no_excel_errors"

elixcee test-workbook fixture.toml --json

A failing case reports its seed and case index so it can be reproduced exactly: elixcee test-workbook fixture.toml --seed 42 --case 17. Full schema, strategies, and assertion rules: docs/agent-contract.md.

Excel operation diagnostics

elixcee diagnose runs a macro once and explains why Excel would reject it — a missing worksheet, a missing workbook, an out-of-bounds array index, a Copy/Paste shape mismatch, a write to a protected sheet, or a Copy/Paste that conflicts with a merged-cell layout — with evidence, instead of a bare error string:

elixcee diagnose Main.bas --file report.xlsx --json Main.Run

{
  "schema_version": 1,
  "ok": false,
  "message": "Sheet 'Sales2025' not found",
  "location": {"file": "Main.bas", "line": 2, "column": 5},
  "root_causes": [
    {
      "code": "WORKSHEET_NOT_FOUND",
      "certainty": "definite",
      "expression": "Worksheets(\"Sales2025\")",
      "requested": "Sales2025",
      "available": ["input", "sales2026", "summary"],
      "suggested": "sales2026",
      "suggestions": ["did you mean 'sales2026'?"]
    }
  ],
  "messages": []
}

Range("A1:C10").Copy followed by Range("E1:F10").PasteSpecial reports both the shape mismatch and where each statement is:

{
  "code": "PASTE_SHAPE_MISMATCH",
  "source_addr": "A1:C10", "source_rows": 10, "source_cols": 3,
  "dest_addr": "E1:F10", "dest_rows": 10, "dest_cols": 2,
  "copy_location": {"file": "Main.bas", "line": 2, "column": 5},
  "suggestions": [
    "resize the destination to E1:G10",
    "or specify only the top-left cell E1"
  ]
}

Writing to a .Protected sheet reports which sheet and how to fix it:

{
  "code": "SHEET_PROTECTED",
  "sheet": "sheet1",
  "suggestions": ["unprotect the sheet first: Worksheets(\"sheet1\").Unprotect"]
}

Pasting A1:C10 into E1:G10 when the destination's first row is merged (E1:G1) but the source's isn't reports the layout conflict and both locations:

{
  "code": "PASTE_MERGE_LAYOUT_MISMATCH",
  "source_addr": "A1:C10", "dest_addr": "E1:G10",
  "conflicts": ["E1:G1"],
  "copy_location": {"file": "Main.bas", "line": 2, "column": 5},
  "suggestions": [
    "unmerge E1:G1 before pasting",
    "or make the source and destination merge layouts identical"
  ]
}

Full classification rules and JSON schema: docs/agent-contract.md.

Diagnosing across generated inputs

elixcee diagnose-workbook combines the two features above: it reruns a macro across test-workbook's generated cases and classifies whichever failures it finds, instead of only reporting a bare error string. It's most useful for input-dependent failures like array-bounds errors, where only some drawn values trigger the bug — a single diagnose call already finds structural issues (shape mismatches, merged-cell conflicts, sheet protection) in one shot, since those don't depend on the input at all:

elixcee diagnose-workbook fixture.toml --json

{
  "schema_version": 1,
  "ok": false,
  "seed": 42,
  "case_index": 3,
  "inputs": [{"address": "sheet1!B2", "value": 999999999}],
  "failure": {
    "rule": "no_runtime_error",
    "message": "Array 'arr': index 999999999 out of bounds (len=6)"
  },
  "root_causes": [
    {
      "code": "ARRAY_INDEX_OUT_OF_BOUNDS",
      "name": "arr", "index": 999999999, "lower": 0, "upper": 5,
      "suggestions": ["check that 'arr' is large enough for index 999999999 (valid range is 0 To 5)"]
    }
  ]
}

Same fixture format and --seed/--case replay as test-workbook, plus --cases N to override the fixture's own case count for one run. Full schema: docs/agent-contract.md.

Multi-area ranges

Range("A1:A10,C1:C10") — a disjoint, multi-area range — is now recognized by .Copy, but pasting it is diagnose-only: diagnose/ diagnose-workbook classify why, instead of silently doing nothing:

{
  "code": "MULTI_AREA_TO_SINGLE_AREA_PASTE",
  "source_areas": [
    {"address": "A1:A10", "rows": 10, "columns": 1},
    {"address": "C1:C10", "rows": 10, "columns": 1}
  ],
  "destination_areas": [
    {"address": "E1:F10", "rows": 10, "columns": 2}
  ],
  "suggestions": [
    "paste each source area separately",
    "copy a contiguous rectangular range",
    "use destination areas with matching count and shapes"
  ]
}

Union(), Areas, Dim rng As Range/Set object variables, and matching-shape multi-area paste are now implemented — see "VBA object model" below. The 4 classified codes above still apply to every multi-area shape that doesn't match exactly (different area counts or shapes, or either side single-area): full scope in docs/agent-contract.md.

Hidden row/column evidence

diagnose/diagnose-workbook now report when a .Copy'd range overlaps hidden rows/columns (read from real XLSX hidden="1" metadata) — not an error, just a new observations field, present alongside (or instead of) root_causes:

{
  "code": "RANGE_CONTAINS_HIDDEN_CELLS",
  "certainty": "observed",
  "range": {"sheet": "sheet1", "address": "A1:C100", "rows": 100, "columns": 3},
  "visibility": {
    "hidden_rows": ["11:14", "30:39"],
    "hidden_columns": ["B:B"],
    "total_cells": 300,
    "visible_cells": 172
  },
  "message": "The range contains hidden rows or columns. Excel operations using visible cells only may produce a multi-area range."
}

This is what SpecialCells(xlCellTypeVisible) (below) builds on — plain Copy/Paste itself is unaffected (hidden cells still copy/paste exactly as before). XLSX only; ODS is deferred. Full scope: docs/agent-contract.md.

VBA object model

Dim rng As Range
Set rng = Range("A1:B2")
rng.Value = 5                        ' real Set reference semantics — an alias, not a copy

Dim u As Range
Set u = Union(Range("A1"), Range("D1"))
Range("C1").Value = u.Areas.Count    ' 2

Dim ws As Worksheet
Set ws = ActiveSheet
ws.Range("A1").Value = 1

Range("F1").Value = 7 Mod 3          ' 1
Range("F2").Value = 2 ^ 3            ' 8
Range("F3").Value = 7 \ 3            ' 2 (integer division)
If Not (a And b) Then MsgBox "ok"

With Cells(r, c)                     ' any target expression, evaluated once
  .Value = 5
  If .Value > 0 Then .Value = .Value + 1   ' .member works at any nesting depth
End With

Set rng = Range("A1"): Set rng2 = rng: Set rng = Nothing
rng2.Value = 1                       ' aliases survive Set ... = Nothing on the original
rng.Value = 2                        ' raises "Object variable or With block variable not set"

Dim n
n = Null
If IsNull(n + 5) Then MsgBox "Null propagates through +"   ' True

Function DoubleIt(x As Integer) As Integer
  DoubleIt = x * 2
End Function

Set-assigned Range/Worksheet/Workbook object variables with real reference semantics — including a genuine unset/Nothing state (member access through a never-Set or explicitly-Nothing variable raises real VBA's "Object variable or With block variable not set"; Set x = Nothing clears only x, not any alias made from it earlier) — Union/Areas, SpecialCells(xlCellTypeVisible) (built on the hidden row/column evidence above), matching-shape multi-area Copy/Paste, Mod/\/^, infix And/Or/Xor/Not (real bitwise semantics on non-Boolean operands), a runtime With stack (any target expression — including a computed one like With Cells(r, c) — evaluated once, with .member resolving correctly at any nesting depth inside If/For/Do/Select Case), Variant's Null (documented VBA propagation through +/&/comparisons, distinct from Empty), the : multi-statement-per-line separator, typed Function parameters/return types, comma-separated multi-declarator Dim (Dim a As Integer, b As Range), and single-line If cond Then stmt [Else stmt] are all supported.

Known gaps: multi-area Paste only executes when both sides are multi-area with matching Areas.Count and per-area shapes — every other combination stays diagnose-only (see above).

XLSX.read()/write() — `@elixcee/xlsx` (npm, prepared but not yet published)

A synchronous, WebAssembly-backed XLSX.read(bytes) — no await init() required — is implemented in the @elixcee/xlsx npm package (see docs/xlsx-architecture.md for the compatibility initiative and the sync-bridge design), along with readFile()/readFileSync() (Node-only; the browser entry point throws rather than faking a filesystem). They return sheet names, !ref, !merges, !rows/!cols (hidden rows/columns), and per-cell {t, v, f, w, z} — values, formula text, formatted display strings, and date-typed cells, resolved via real styles.xml/number-format parsing. Differential-tested against the real xlsx@0.18.5 package: 33/33 MATCH, 0 disclosed (the src/reader.rs xml:space="preserve" trimming defect noted in earlier rounds is fixed; see CHANGELOG.md). Works in Node (CJS/ESM) and the browser: a "browser" export condition routes to the inlined-bytes/initSync WASM artifact, verified not just by Node simulating that export condition but by an actual headless Chrome process loading a real bundle and reading XLSX.read()'s result back out of the page's own DOM (no Safari claim). The browser entry point still assumes bundled consumption — its shared code has a CJS require('ssf'), so it's not literal no-build <script type="module"> usage — but a real packed-npm-tarball install (not a relative import into this repo) and CJS/ESM bundling both round-trip cleanly with no manual asset copy step required anymore.

XLSX.write(wb, opts)/writeFile()/writeFileSync() — pure JS/XML/ZIP generation, no Rust writer needed — are implemented too (bookType: "xlsx" only), differential-tested both directions against the real oracle: 36 MATCH + 1 disclosed (bookType: "ods", not implemented). package.json's description was updated to match, but its version (0.0.0-development), private (true), and publishConfig (unset) were deliberately left untouched — no npm publish has actually run, and @elixcee scope ownership on npm is unconfirmed from this environment either way (see ROADMAP.md's "Known gaps").

Build from source

cargo build --release --bin elixcee
# binary: target/release/elixcee  (or elixcee.exe on Windows)

Quick Start

import elixcee

# Run a VBA macro and get all resulting cells
cells = elixcee.run_macro("""
Sub FillSquares()
    For i = 1 To 5
        Cells(i, 1).Value = i * i
    Next i
End Sub
""", "FillSquares")
# cells == {(1,1): 1, (2,1): 4, (3,1): 9, (4,1): 16, (5,1): 25}

# Pre-populate cells from Python, then run a macro
vm = elixcee.Vm()
vm.set_cell(1, 1, 100)
vm.set_cell(2, 1, 200)
vm.run("""
Sub CalcTotal()
    total = Cells(1,1).Value + Cells(2,1).Value
    Cells(3,1).Value = total
End Sub
""", "CalcTotal")
print(vm.get_cell(3, 1))   # 300
print(vm.variables())       # {"total": 300}

# Load cell data from an existing Excel file, then run a macro
vm = elixcee.load_workbook("data.xlsx")
vm.run(vba_code, "ProcessData")
result_cells = vm.cells()   # {(row, col): value, ...}

# Store a worksheet formula on a cell and evaluate it
vm.set_cell_formula(4, 1, "=SUM(A1:A3)")
print(vm.get_cell(4, 1))   # sum of rows 1-3 in column A

# Bulk range/row access -- no per-cell round trips needed
vm = elixcee.load_workbook("input.xlsx")
rows = vm.get_range("A1:C10", sheet="Data")
vm.set_range("E1:F2", [[1, 2], [3, 4]], sheet="Result")
vm.append_row(["Alice", 100], sheet="Result")
vm.save_workbook("output.xlsx")

# Sheet management and row/column edits
vm.rename_sheet("Result", "Summary")
vm.move_sheet("Summary", 0)          # move to the first tab
vm.insert_rows(1, sheet="Summary")   # shift everything down by one row
print(vm.merged_cells(sheet="Data")) # e.g. ["B1:C1"]

# Column iteration, sorting, and merge create/remove
cols = vm.iter_cols(max_col=3, sheet="Data")  # column-major, values only
vm.sort_range("A2:B10", key_col=1, sheet="Data")
vm.merge_cells("D1:E1", sheet="Data")
vm.unmerge_cells("B1:C1", sheet="Data")

# Hide/unhide rows and columns
vm.set_row_hidden(5, sheet="Data")
vm.set_column_hidden(4, hidden=False, sheet="Data")
print(vm.hidden_rows(sheet="Data"))  # e.g. [5]

# Copy a sheet
vm.copy_sheet("Data", "Data Backup")

# Read workbook-level defined names
print(vm.defined_names())  # e.g. {"MyRange": "Sheet1!$A$1:$A$3"}

# Read a sheet's whole-tab visibility
print(vm.sheet_state("Data"))  # "visible", "hidden", or "veryHidden"

# Read a row's height or a column's width, if explicitly set
print(vm.row_height(5, sheet="Data"))     # e.g. 30.5, or None
print(vm.column_width(2, sheet="Data"))   # e.g. 12.5, or None

# Control MsgBox behavior
vm = elixcee.Vm(on_msgbox="skip")   # silently ignore MsgBox calls (default)
vm = elixcee.Vm(on_msgbox="error")  # raise RuntimeError on MsgBox

Python API

Method	Description
`Vm(on_msgbox="skip")`	Create a new VM. `on_msgbox="error"` raises `RuntimeError` on `MsgBox`.
`vm.run(vba_code, macro_name)`	Parse and execute the named Sub.
`vm.set_cell(row, col, value)`	Write a value into a cell (1-based).
`vm.get_cell(row, col)`	Read a cell value. Returns `None` for empty cells.
`vm.cells()`	All non-empty cells as `{(row, col): value}`.
`vm.variables()`	All VBA variables as `{name: value}`.
`vm.set_cell_formula(row, col, formula)`	Store a formula (e.g. `"=SUM(A1:A3)"`) and evaluate it.
`vm.set_cell_formula_batch(formulas)`	Set multiple formulas at once: `{(row, col): formula_str}`.
`vm.recalculate()`	Re-evaluate all formula cells (useful after manual cell writes).
`vm.set_sheet(name)`	Switch the active sheet (creates it if absent).
`vm.active_sheet()`	Name of the currently active sheet.
`vm.sheet_names()`	List of all sheet names.
`vm.get_sheet(name)`	Cells of a named sheet as `{(row, col): value}`.
`vm.get_range(addr, sheet=None)`	Read a rectangular range (e.g. `"A1:C5"`) as a nested list.
`vm.set_range(addr, values, sheet=None)`	Write a rectangular range from a nested list.
`vm.append_row(values, sheet=None)`	Write one row just past the sheet's used range; returns the row number.
`vm.iter_rows(min_row=1, max_row=None, min_col=1, max_col=None, sheet=None)`	Values-only iteration over a rectangular region.
`vm.iter_cols(min_row=1, max_row=None, min_col=1, max_col=None, sheet=None)`	Column-major values-only iteration -- the transposed sibling of `iter_rows`.
`vm.max_row(sheet=None)` / `vm.max_column(sheet=None)`	Highest used row/column, or `None` if the sheet is empty.
`vm.calculate_dimension(sheet=None)`	Used range as an A1-style string (e.g. `"B2:D10"`), or `None` if empty.
`vm.sort_range(addr, key_col, descending=False, header=False, sheet=None)`	Sort a rectangular range in place by one column.
`vm.rename_sheet(old_name, new_name)`	Rename a sheet.
`vm.move_sheet(name, new_index)`	Move a sheet to an absolute 0-based tab position.
`vm.copy_sheet(source_name, new_name)`	Duplicate a sheet (cells, merges, hidden state, styles) into a new one.
`vm.defined_names()`	Workbook-level defined names as `{name: raw_formula_text}`. Read-only.
`vm.sheet_state(name)`	A sheet's whole-tab visibility: `"visible"`, `"hidden"`, or `"veryHidden"`. Read-only.
`vm.row_height(row, sheet=None)` / `vm.column_width(col, sheet=None)`	A row's height in points / column's width in characters, or `None` if never explicitly set. Read-only.
`vm.insert_rows(idx, amount=1, sheet=None)` / `vm.delete_rows(...)`	Insert/delete rows (values only -- doesn't shift merges/styles).
`vm.insert_cols(idx, amount=1, sheet=None)` / `vm.delete_cols(...)`	Insert/delete columns (same caveat as rows).
`vm.merged_cells(sheet=None)`	List a sheet's merged ranges as A1 strings, e.g. `["B1:C1"]`.
`vm.merge_cells(addr, sheet=None)` / `vm.unmerge_cells(addr, sheet=None)`	Create/remove a merge.
`vm.hidden_rows(sheet=None)` / `vm.hidden_columns(sheet=None)`	Sorted list of hidden row/column numbers.
`vm.set_row_hidden(row, hidden=True, sheet=None)` / `vm.set_column_hidden(col, ...)`	Hide or unhide a single row/column.
`vm.save_workbook(path)`	Save all sheets to `.xlsx` or `.ods`.
`vm.cells_df()`	Return the active sheet as a pandas DataFrame (requires pandas).
`elixcee.run_macro(vba, name)`	One-shot: run a macro and return `{(row, col): value}`.
`elixcee.load_workbook(path)`	Load an `.xlsx` or `.ods` file into a `Vm`.

Coverage

See FUNCTIONS.md for the complete function and VBA syntax reference, including Excel version for each function.

Highlights:

Classic (Excel 2003-): SUM, VLOOKUP, IF, PMT, FV, PV, NPER, RATE, IPMT, PPMT, NPV, IRR, MIRR, XNPV, XIRR, DGET, DSUM, DAVERAGE, DCOUNT, DCOUNTA, DMAX, DMIN, and 100+ core functions
2007–2019: IFERROR, COUNTIFS/SUMIFS, XOR, IFS, SWITCH, TEXTJOIN, MAXIFS/MINIFS
365/2021: XLOOKUP, XMATCH, FILTER, SORT, UNIQUE, SEQUENCE, LET, LAMBDA, MAP, REDUCE
2024/365: TEXTSPLIT, TEXTBEFORE, TEXTAFTER, VSTACK, HSTACK, TAKE, DROP, CHOOSECOLS, CHOOSEROWS
VBA: For/If/While/With/On Error/Function/Type...End Type/Named Ranges/Array of UDT

Named Ranges

Register a named range in VBA with Range("A1:B5").Name = "MyData", then use the name anywhere a range address is accepted:

Range("MyData").Value = 0          ' write to all cells in the range
For Each cell In Range("MyData")   ' iterate over cells
    total = total + cell
Next cell

Named ranges are stored on vm.named_ranges (a dict[str, str] mapping lowercase name → address).

Criteria Syntax (COUNTIF / SUMIF / SUMIFS / etc.)

Criteria	Example	Meaning
Number	`10`	Exact numeric match
String	`"apple"`	Case-insensitive string match
Comparison	`">5"`, `"<=10"`, `"<>"`	Numeric comparison
Wildcard	`"a*"`, `"?bc"`	`*` = any chars, `?` = one char

Application Object

Property / Method	Description	Behavior
`Application.Calculation = xlCalculationManual`	Disable auto-recalculation	Active
`Application.Calculation = xlCalculationAutomatic`	Enable auto-recalculation + re-evaluate all formula cells	Active
`Application.ScreenUpdating = False/True`	Suppress screen refresh	No-op (no screen)
`Application.EnableEvents = False/True`	Disable/enable event triggers	No-op (no events)
`Application.DisplayAlerts = False/True`	Suppress dialog boxes	No-op (no dialogs)
`Application.StatusBar = "..."` / `False`	Set/clear status bar text	No-op (no UI)
`Application.Cursor = xlWait` / `xlDefault`	Change cursor shape	No-op (no UI)
`Application.CutCopyMode = False`	Cancel clipboard mode	Active (clears the modeled clipboard)

No-op properties are parsed and accepted without error, but have no effect. This allows VBA macro performance patterns (e.g., Application.ScreenUpdating = False at the start of a macro) to run unchanged.

Microsoft Excel round-trip validation

elixcee's workbook save path has been validated using five sanitized, Microsoft Excel-authored .xlsm fixtures on Microsoft Excel for Mac.

Validated scope:

open an Excel-authored workbook
modify cells with elixcee
save-as and in-place save
reopen in Microsoft Excel without a repair warning
preserve formulas, existing cell styles, merged cells, hidden rows/columns, VBA project bytes, unknown ZIP parts, and surviving relationships

Not validated:

post-save VBA macro execution
tables, data validation, conditional formatting, hyperlinks, comments, defined names, charts, images, and print settings embedded in regenerated worksheet XML

See compat/oracle-excel-com/results/0.9.0-A_summary.md for the full results.

Not Yet Supported

See FUNCTIONS.md — Not Yet Supported for the full list.

Key gaps by category:

Statistical: NORM.S.DIST, T.INV, F.DIST, CHISQ.DIST, and more
Text: REPT, NUMBERVALUE, PHONETIC
Out of scope: IMAGE (URL image fetch), GROUPBY (pivot aggregation), TRIMRANGE

Status Legend

Mark	Meaning
Done	Implemented and tested
TBD	Not yet scheduled

Development Phases

Phase	Content	Status
Phase 1	Rust project setup + pyo3 Python bindings	Done
Phase 2	VBA parser MVP (Sub/End Sub, assignment, Cells)	Done
Phase 3	Virtual Excel VM (variables, cell storage, interpreter)	Done
Phase 3.5	Excel formula engine (SUM, IF, VLOOKUP, Application.Calculation, etc.)	Done
Phase 4	Control flow (For loop, If/Else, arithmetic expressions)	Done
Phase 5	Python interface (Vm class, run_macro, load_workbook, MsgBox)	Done
Phase 6	Formula function expansion (100+ Excel functions, 118 tests)	Done
Phase 7	Advanced VBA constructs (ElseIf, Exit, For Each, On Error, Function, arrays, While-Wend)	Done
Phase 8	Range API (ClearContents, Offset, Sheets.Cells, WorksheetFunction, multi-sheet)	Done
Phase 9	Multi-sheet support (Sheets HashMap, With Sheets, Python API, load_workbook all sheets)	Done
Phase 10	Worksheet function expansion (math, trig, stats, array/spill, lambda functions)	Done
Phase 11	User-defined types (`Type...End Type`), named ranges, `RANDARRAY`, pandas integration (`cells_df`), type stubs (`.pyi`)	Done
Phase D1	Remove rust_xlsxwriter, hand-written XLSX via zip (dependencies: 5→4)	Done
Phase D2	Remove pest/pest_derive, hand-written recursive descent VBA parser (dependencies: 4→3)	Done
Phase D3	Remove calamine from runtime, hand-written XLSX/ODS reader (dependencies: 3→2)	Done
Perf R4	SUM/AVERAGE/MIN/MAX fast path (skip `Vec<Variant>`), RangeWrite dirty-flag batching	Done
CLI	Standalone `elixcee` binary; pyo3 made optional; GitHub Actions release workflow	Done
Milestone A	JSON agent contract (`--json`), error classification, MsgBox message log	Done
Milestone A.1	JSON contract hardening (`serde_json`-verified tests, message-log lifecycle, error code docs)	Done
Milestone A.5	Source location tracking — line/column in parse and runtime errors	Done
Milestone B1	`check` subcommand — parse diagnostics, entrypoint check, `MsgBox`/interactive-call detection	Done
Milestone B1.1	`check`: undefined Sub/Function call detection, unsupported-construct (no-op) detection	Done
Milestone B2	Multi-module projects — multiple `.bas` files, `Module.Sub` qualified entrypoints, cross-module collision detection	Done
Milestone B3	Deterministic black-box tests (`tests/blackbox.rs`, declarative `.toml` fixtures)	Done
Milestone B4	`snapshot` subcommand — read a workbook's cells without executing VBA	Done
Milestone B5a	`test-workbook` subcommand — property-based testing with generated boundary-value inputs	Done
Milestone B6a	`diagnose` subcommand — missing sheet/workbook, array-out-of-bounds root causes	Done
Milestone B6b	`diagnose`: Copy/Paste shape mismatch + clipboard state	Done
Milestone B6c	`diagnose`: sheet protection (`Protect`/`Unprotect`)	Done
Milestone B6c2	`diagnose`: merged-cell-aware Copy/Paste diagnostics	Done
Milestone B6d	`diagnose-workbook` — root-cause diagnosis across generated test cases	Done
Milestone B7a	Multi-area `Range`/`Union`/`Areas` foundation for Copy/Paste diagnostics	Done
Milestone B7b	Hidden row/column metadata foundation for Copy/Paste diagnostics	Done
Phase 3A-1	`compat/vba-semantics/` value-correctness suite: 208 → 301 cases (6 new categories); fixed single-line-`If` statement dispatch, `Boolean` arithmetic (`True` = -1), `WorksheetFunction` Boolean coercion, `Empty` equality	Done
Phase 3A-2	CI `wasm` job: fresh `wasm-pack` build (Node + web targets) plus a Node/browser-condition smoke test, wired into GitHub Actions	Done
0.5.0	VBA structural semantics (`:` statement separator, `Variant::Null` with documented propagation, real object-`Nothing` state with alias safety, a runtime `With` target stack) merged with `@elixcee/xlsx` real-consumer/real-browser validation (packed-tarball install, headless-Chrome smoke, bundle-safe WASM loading, `readFile()`); `compat/vba-semantics/` 301 → 386 cases; `elixcee-types` bumped to 0.2.0 for the new public `Variant::Null` enum variant; published to crates.io, PyPI, and GitHub Releases	Done

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

elixcee-0.14.0.tar.gz (2.9 MB view details)

Uploaded Aug 30, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

elixcee-0.14.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 PyPymanylinux: glibc 2.17+ x86-64

elixcee-0.14.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 PyPymanylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp315-cp315t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.15tmanylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp315-cp315-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.15manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.14tmanylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.14tmanylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp314-cp314-win_amd64.whl (1.1 MB view details)

Uploaded Aug 30, 2026 CPython 3.14Windows x86-64

elixcee-0.14.0-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.14manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.14manylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp314-cp314-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl (2.4 MB view details)

Uploaded Aug 30, 2026 CPython 3.14macOS 10.12+ universal2 (ARM64, x86-64)macOS 10.12+ x86-64macOS 11.0+ ARM64

elixcee-0.14.0-cp313-cp313-win_amd64.whl (1.1 MB view details)

Uploaded Aug 30, 2026 CPython 3.13Windows x86-64

elixcee-0.14.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.13manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.13manylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp313-cp313-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl (2.4 MB view details)

Uploaded Aug 30, 2026 CPython 3.13macOS 10.12+ universal2 (ARM64, x86-64)macOS 10.12+ x86-64macOS 11.0+ ARM64

elixcee-0.14.0-cp312-cp312-win_amd64.whl (1.1 MB view details)

Uploaded Aug 30, 2026 CPython 3.12Windows x86-64

elixcee-0.14.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.12manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.12manylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp312-cp312-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl (2.4 MB view details)

Uploaded Aug 30, 2026 CPython 3.12macOS 10.12+ universal2 (ARM64, x86-64)macOS 10.12+ x86-64macOS 11.0+ ARM64

elixcee-0.14.0-cp311-cp311-win_amd64.whl (1.1 MB view details)

Uploaded Aug 30, 2026 CPython 3.11Windows x86-64

elixcee-0.14.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.11manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.11manylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp311-cp311-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl (2.4 MB view details)

Uploaded Aug 30, 2026 CPython 3.11macOS 10.12+ universal2 (ARM64, x86-64)macOS 10.12+ x86-64macOS 11.0+ ARM64

elixcee-0.14.0-cp310-cp310-win_amd64.whl (1.1 MB view details)

Uploaded Aug 30, 2026 CPython 3.10Windows x86-64

elixcee-0.14.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.10manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.10manylinux: glibc 2.17+ ARM64

elixcee-0.14.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.9manylinux: glibc 2.17+ x86-64

elixcee-0.14.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.3 MB view details)

Uploaded Aug 30, 2026 CPython 3.9manylinux: glibc 2.17+ ARM64

File details

Details for the file elixcee-0.14.0.tar.gz.

File metadata

Download URL: elixcee-0.14.0.tar.gz
Upload date: Aug 30, 2026
Size: 2.9 MB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for elixcee-0.14.0.tar.gz
Algorithm	Hash digest
SHA256	`8b1c5ac1f521b34ad56448172ff450dfb67fac218a382f4b77db717afe91850d`
MD5	`b0c4e0f6c45e23e5bb854542f47f4ab4`
BLAKE2b-256	`9b51429de4cc8ce8c051d41dc68579f3233ba2f434986dfa2ee6982ed898a6ef`

Algorithm	Hash digest
SHA256	`a443518125cd854a0248e01ce3c23dbfcdd2a4b482b71b95bcbb0c6ce77b1385`
MD5	`403618c354dd371f95f2d6cf44b44188`
BLAKE2b-256	`8df3bee7858e261adacd0a3016370c8548942a416d4a0becc3e3abb84c837fd0`

Algorithm	Hash digest
SHA256	`3780c4c4c4cbb7465c919caea32a46228e0476b9c94efdf107878e05dc2a1d5a`
MD5	`80e8cfd63187e95bc35b406fabf35782`
BLAKE2b-256	`f6a405e87b4bbe70dc4373633d3389a6d29705d2dcfd9148324eaccabda42a55`

Algorithm	Hash digest
SHA256	`faec74c77a1f7f0098802fda2f5647a059474caa844b972e92bcc6092819a70b`
MD5	`d6bb58cca2d4f88bdaf67c7180c9286d`
BLAKE2b-256	`b4c422b4625e79c50da1882f9f25ed2fd640a03a661b40d385a7aec10676e6ae`

Algorithm	Hash digest
SHA256	`e2d13bae2a409332073dd25d694a0d24d0108b612cce7ca164575f342109da60`
MD5	`6f4791e51b571f972b96ec2099c982c2`
BLAKE2b-256	`d654b4270e05b81afec7ff3feb067b07d713916ed2c1788fc6dee1c1e6b6d628`

Algorithm	Hash digest
SHA256	`58ac6b1af29cdc95833d589a614bb4fcb4db63b4b7a25c3be1d2479b5342eea6`
MD5	`97995f84088ecf51887d29fcd1d07bbc`
BLAKE2b-256	`baca8855c442b2992270cd747a83e0cbb415e1fab53f06b27d2f186d0560cc48`

Algorithm	Hash digest
SHA256	`24ecc1c0553df18250cdde129bce8e080efc4480643fd4154e90da3e9f475c1b`
MD5	`7d5c0ba2e984c8d16a26d666e7b91439`
BLAKE2b-256	`cbcd6b0a6579a29ac9f9b38b260363721dfca818325727153e972c54e4bbb7ac`

Algorithm	Hash digest
SHA256	`cfe079a0a72217e8c4c714fd5c01f8bb06752c4e8141e5cee62b394c85d0f792`
MD5	`5abbc7962139871349a2fa453d3205bd`
BLAKE2b-256	`e7a4261573700061adfc7362fb5be676b3ef023178d6a151dc9d4278bc3f628c`

Algorithm	Hash digest
SHA256	`1353b173c3bcec7ec3346c87640ed3f81b6faae3d20d95f3926a61c7bac2b4da`
MD5	`63b8651367ff7015d1ab0455403ca58b`
BLAKE2b-256	`58a368dd2c354dc64541b1c2d1eb866ba27f9f2f80c7919dce3c50b0ba1d6741`

Algorithm	Hash digest
SHA256	`4a6c10b055305c71db41f0585ab77887d59dc52478d0adccb9daf3a5d3ad69d0`
MD5	`6dce7b1f6229410d49f1fbf105abe970`
BLAKE2b-256	`f1a537e4f52429b1c07bf29245b2e5a36bce5ea50f45392077674a747c5080aa`

Algorithm	Hash digest
SHA256	`735ab48e85428e6808044625e376446f4776c939a3aa48d3158eb2c625c444e1`
MD5	`07bde47b6123c42382e2930748eaf52b`
BLAKE2b-256	`f652644a248a74e6e14b413e0c7ae1377e4e9f8ce6c8f1dde4431823e4c37cf1`

Algorithm	Hash digest
SHA256	`31a2d93987cc396b636551d815bfba76ac8a08db49e0d912e70becc80ad60413`
MD5	`a4807044782aed76f10d524651ebdb51`
BLAKE2b-256	`fa565cb1d2532a72222da17541181d5aeed36e466e47311f4719e97a2c70650e`

Algorithm	Hash digest
SHA256	`5face7b40e5b05ca1fac761c5a4761014a984ea32ebed7b7af3b6bd434e3e8db`
MD5	`e11d3afc7c16b8dfeff2c812f20e05ac`
BLAKE2b-256	`54ba07f43da90bf60983929dabfcdfb4aaed4fb8316d779e75bde4a419da8ffe`

Algorithm	Hash digest
SHA256	`b292d71088fed04acbece60c4eb69d9129807d44ce8b23e1293716f5d3f22bc2`
MD5	`92c53d30ba7b511c96df3697ee3886e9`
BLAKE2b-256	`93dda099a8a772841160c12bd993ac00c0a550e2eb6af5cc6f5a94654669ec25`

Algorithm	Hash digest
SHA256	`a1bdcc3cb5e09c9bf96ca4747a353ea87a28167ea10673264ffdb2751e15e9ec`
MD5	`339bcad365d90c54426cacfb66a380d8`
BLAKE2b-256	`d9865cd216e58f7aa09adcb75458ce87015ccce3e868f66608e41b43307e6555`

Algorithm	Hash digest
SHA256	`c4876c4fa8cf53585f8b12b64a5373ddafb0fc18a3c5c99db749e839f1a55926`
MD5	`d5b7b1ed087754155dbf7a6d45351f7f`
BLAKE2b-256	`c5740693702196462f6792a9e03bd71cffdfd4f9b565427b9da7f1aa5dc79774`

Algorithm	Hash digest
SHA256	`c30230eb34cc3e2fb9e60b615f7a6fef875ceabb6ef002fd58586a1f785f58f9`
MD5	`597342f2df1b7fa9ae662dafdf3155a6`
BLAKE2b-256	`12160161319bf6faf085ed7046ce0f6d7061e29607a03b516506aa6891cb9637`

Algorithm	Hash digest
SHA256	`3be8b1727bb7badb5ba6b417dc7313254f5a1db7607e3bf5894e97ab0add5c58`
MD5	`571d5a4e473027aa65a30b048ca8967b`
BLAKE2b-256	`1f6a58bbdc8ae72a06973c4debf5a9bae94980b754012041409f587db43f1abb`

Algorithm	Hash digest
SHA256	`af788016bfe9a3a76df74f4bd9a4e1611ab2d26b2eb67b3b971e639dc000640d`
MD5	`b4dc236078544ba4a67104c3f97721d6`
BLAKE2b-256	`8d9fb6b7299b734790a11730744581e9d6edd70240326683c13b4a3c3d9d09d8`

Algorithm	Hash digest
SHA256	`e85720af23e80eddf7242553d0d23c5ca371b43d21491e26f99786ffe99c73d7`
MD5	`709bd1f15de9775ca261be038b2e4460`
BLAKE2b-256	`3928ee048feacb06ba3a140dc41ac98ad2bfe509de4fce3472a3efbed08c87e1`

Algorithm	Hash digest
SHA256	`8d8d6b470bc553fc572b2942ad8bf90689cb365b014d4c3c626209e8e2920156`
MD5	`9d977fd79e81bedb81beff8cf7da4c4f`
BLAKE2b-256	`499cc53e342883be926e59200410db2f42f482876c0c4bacdae5a22b1e293b6b`

elixcee 0.14.0

elixcee

Name

Comparison with similar tools

Installation

CLI (Windows / Linux / macOS)

Usage

Examples

Multiple files (multi-module projects)

JSON output (for scripts / AI agents)

Static analysis without running the macro

Workbook snapshot

Property-based workbook testing

Excel operation diagnostics

Diagnosing across generated inputs

Multi-area ranges

Hidden row/column evidence

VBA object model

XLSX.read()/write() — @elixcee/xlsx (npm, prepared but not yet published)

Build from source

Quick Start

Python API

Coverage

Named Ranges

Criteria Syntax (COUNTIF / SUMIF / SUMIFS / etc.)

Application Object

Microsoft Excel round-trip validation

Not Yet Supported

Status Legend

Development Phases

Project links

Key dates

1 maintainer

GitHub Statistics

License

Requires

Tags

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

XLSX.read()/write() — `@elixcee/xlsx` (npm, prepared but not yet published)

Algorithm	Hash digest
SHA256	`c56a08ce84287bcb8ce7441d6b3c6308ef8228c22eed293da94012a5689fdfc4`
MD5	`8b9c6b642860e2cf750da8d367e05455`
BLAKE2b-256	`cbdbfa07f036b4aabd45b445d83d70f9be4c795907726d94913f107c2499dae4`

Algorithm	Hash digest
SHA256	`d546c499dc810517879dc402726c0b06b39350ce8409650d375dae131a602b25`
MD5	`8b6078e8f84570839ca37769482bb622`
BLAKE2b-256	`bbf6c410224c83a15a50f80da7a832c119fb1c8c840c742f1d1887451195ba43`

Algorithm	Hash digest
SHA256	`8efc5ed05fffdc4636ed695480c5a2a9257b8f59f05fd8c3254865f31b6178c3`
MD5	`47a1fafc8ee89e2b66e7d4baaa55a6c0`
BLAKE2b-256	`38ff45914620bfa38412112a00422da26561e257aedcc40579bf1a1b64b2a3be`

Algorithm	Hash digest
SHA256	`8f14f1f84e02a39674ba393dd7f38bc7816b9d0f7027c97c42aec712f8db1e44`
MD5	`0d2dd63b277186408ff2e36e062df433`
BLAKE2b-256	`0a17ad3783cbad792bc90ff040189af5a50a23067d72b08f1c1f2db52f6d1571`

Algorithm	Hash digest
SHA256	`2fb928ad0858739e8eb36bd203b693989fb29ab3956b030be62d8e4f9265b2d4`
MD5	`3e6516fe9b199216ad2d70961a6b75ff`
BLAKE2b-256	`0420d953e0a56f913d95bdf1283f2c6e4815703fcae44f7488d6874c742114e9`

Algorithm	Hash digest
SHA256	`531a715b56415c36e8371197c4214148526f005c49c223416ed42d8208f9d6c6`
MD5	`4cc5ff9559735f4630a3c655752ce528`
BLAKE2b-256	`eb5a243b86f435b1f1033468e8f76306935b9ca699b69469b0687101c8a49b5b`

Algorithm	Hash digest
SHA256	`62d67465867aab6407cfe1a007abdd9b57e1f6706cac0fb0a5b81ef9816c105e`
MD5	`cdd16a519807fbe19bbc2cbc783033c8`
BLAKE2b-256	`45b4e790ffb0802b6837b63a822f5641b972c8cf0162ab2bfb55fff6151ecfef`