Skip to main content

Modern Spreadsheets in Stata

Stata is a statistical software package. It offers a robust C-based extension system.

SheetJS is a JavaScript library for reading and writing data from spreadsheets.

This demo uses SheetJS to pull data from a spreadsheet for further analysis within Stata. We'll create a Stata native extension that loads the Duktape JavaScript engine and uses the SheetJS library to read data from spreadsheets and converts to a Stata-friendly format.

flowchart LR
ofile[(workbook\nXLSB file)]
nfile[(clean file\nXLSX)]
data[[Stata\nVariables]]
ofile --> |Stata Extension\nSheetJS + Duktape| nfile
nfile --> |Stata command\nimport excel|data

The demo will read a Numbers workbook and generate variables for each column. A sample Stata session is shown below:

Stata commands

This demo covers Stata extensions. For directly processing Stata DTA files, the "Stata DTA Codec" works in the browser or NodeJS.

Tested Deployments

This demo was tested by SheetJS users in the following deployments:

ArchitectureDate
darwin-x642024-04-10
win10-x642024-04-10
linux-x642024-04-25

Stata has limited support for processing spreadsheets through the import excel command1. At the time of writing, it lacked support for XLSB, NUMBERS, and other common spreadsheet formats.

SheetJS libraries help fill the gap by normalizing spreadsheets to a form that Stata can understand.

Integration Details​

The current recommendation involves a native plugin that reads arbitrary files and generates clean XLSX files that Stata can import.

The extension function ultimately pairs the SheetJS read2 and write3 methods to read data from the old file and write a new file:

var wb = XLSX.read(original_file_data, {type: "buffer"});
var new_file_data = XLSX.write(wb, {type: "array", bookType: "xlsx"});

The extension function cleanfile will take one or two arguments:

plugin call cleanfile, "pres.numbers" will generate sheetjs.tmp.xlsx from the first argument ("pres.numbers") and print instructions to load the file.

plugin call cleanfile, "pres.numbers" verbose will additionally print CSV contents of each worksheet in the workbook.

flowchart LR
ofile{{File\nName}}
subgraph JS Operations
ojbuf[(Buffer\nFile Bytes)]
wb(((SheetJS\nWorkbook)))
njbuf[(Buffer\nXLSX bytes)]
end
obuf[(File\nbytes)]
nbuf[(New file\nbytes)]
nfile[(XLSX\nFile)]
ofile --> |C\nRead File| obuf
obuf --> |Duktape\nBuffer Ops| ojbuf
ojbuf --> |SheetJS\n`read`| wb
wb --> |SheetJS\n`write`| njbuf
njbuf --> |Duktape\nBuffer Ops| nbuf
nbuf --> |C\nWrite File| nfile

C Extensions​

Stata C extensions are shared libraries or DLLs that use special Stata methods for parsing arguments and returning values.

Structure​

Arguments are passed to the stata_call function in the plugin.4 The function receives the argument count and an array of C strings:

stata_call declaration
STDLL stata_call(int argc, char *argv[]);

For example, argc is 2 and argv has two C strings in the following command:

plugin call cleanfile, "pres.numbers" verbose
* arguments start
* argv[0] ^^^^^^^^^^^^
* argv[1] ^^^^^^^
* argc = 2

Communication​

SF_display and SF_error display text and error messages respectively.

Message text follows the "Stata Markup and Control Language"5.

{stata ...} is a special directive that displays the arguments and creates a clickable link. Clicking the link will run the string.

For example, a plugin may attempt to print a link:

SF_display C plugin example
SF_display("{stata import excel \"sheetjs.tmp.xlsx\", firstrow} will read the first sheet and use headers\n");

The function will print the following text to the terminal:

import excel "sheetjs.tmp.xlsx", firstrow will read the first sheet and use headers

The blue text is clickable. When a user clicks the text, the command import excel "sheetjs.tmp.xlsx", firstrow will be executed.

Duktape JS Engine​

This demo uses the Duktape JavaScript engine. The SheetJS + Duktape demo covers engine integration details in more detail.

The SheetJS Standalone scripts can be loaded in Duktape by reading the source from the filesystem.

Complete Demo​

This demo was tested in Windows x64 and macOS x64. The path names and build commands will differ in other platforms and operating systems.

The cleanfile.c extension defines one plugin function. It can be chained with import excel:

program cleanfile, plugin
plugin call cleanfile, "pres.numbers" verbose
program drop cleanfile
import excel "sheetjs.tmp.xlsx", firstrow

Create Plugin​

  1. Ensure a compatible C compiler (Xcode on macOS) is installed.

  2. Open Stata and run the following command:

pwd

The output will be the default data directory. On macOS this is typically ~/Documents/Stata

  1. Open a terminal window and create a project folder sheetjs-stata within the Stata data directory:
# `cd` to the Stata data directory
cd ~/Documents/Stata
mkdir sheetjs-stata
cd sheetjs-stata
  1. Download stplugin.c and stplugin.h from the Stata website:
curl -LO https://www.stata.com/plugins/stplugin.c
curl -LO https://www.stata.com/plugins/stplugin.h
  1. Download Duktape. In Windows, the following commands should be run in WSL. In macOS, the commands should be run in the same Terminal session.
curl -LO https://duktape.org/duktape-2.7.0.tar.xz
tar -xJf duktape-2.7.0.tar.xz
mv duktape-2.7.0/src/*.{c,h} .
  1. Download cleanfile.c.

In Windows, the following commands should be run in WSL. In macOS, the commands should be run in the same Terminal session.

curl -LO https://docs.sheetjs.com/stata/cleanfile.c
  1. Observe that macOS does not need a "Linux Subsystem" and move to Step 7.

  2. Build the plugin:

gcc -shared -fPIC -DSYSTEM=APPLEMAC stplugin.c duktape.c cleanfile.c -lm -std=c99 -Wall -ocleanfile.plugin

Install Plugin​

  1. Copy the plugin to the Stata data directory:
cp cleanfile.plugin ../

Download SheetJS Scripts​

  1. Move to the Stata data directory:
cd ..
  1. Observe that macOS does not need a "Linux Subsystem" and move to Step 11.
  1. Download SheetJS scripts and the test file.

In Windows, the following commands should be run in WSL. In macOS, the commands should be run in the same Terminal session.

curl -LO https://cdn.sheetjs.com/xlsx-0.20.3/package/dist/shim.min.js
curl -LO https://cdn.sheetjs.com/xlsx-0.20.3/package/dist/xlsx.full.min.js
curl -LO https://docs.sheetjs.com/pres.numbers

Stata Test​

The screenshot in the introduction shows the result of steps 13 - 19

  1. If it is not currently running, start the Stata application.
  1. Run the following command in Stata:
dir

Inspect the output and confirm that cleanfile.plugin is listed.

  1. Load the cleanfile plugin:
program cleanfile, plugin
  1. Read the pres.numbers test file:
plugin call cleanfile, "pres.numbers" verbose

The result will show the data from pres.numbers:

. plugin call cleanfile, "pres.numbers" verbose Worksheet 0 Name: Sheet1 Name,Index Bill Clinton,42 GeorgeW Bush,43 Barack Obama,44 Donald Trump,45 Joseph Biden,46

Saved to sheetjs.tmp.xlsx import excel "sheetjs.tmp.xlsx", firstrow will read the first sheet and use headers for more help, see import excel

  1. Close the plugin:
program drop cleanfile
  1. Clear the current session:
clear
  1. In the result of Step 16, click the link on import excel "sheetjs.tmp.xlsx", firstrow

Alternatively, manually type the command:

import excel "sheetjs.tmp.xlsx", firstrow

The output will show the import result:

. import excel "sheetjs.tmp.xlsx", firstrow (2 vars, 5 obs)

  1. Open the Data Editor (in Browse or Edit mode) and compare to the screenshot:
browse Name Index

Data Editor showing data from the file

In the terminal version of Stata, browse does not work:

. browse Name Index
command browse is unrecognized
r(199);

The codebook command will display details.

Expected Output (click to show)
-------------------------------------------------------------------------------
Name Name
-------------------------------------------------------------------------------

Type: String (str12)

Unique values: 5 Missing "": 0/5

Tabulation: Freq. Value
1 "Barack Obama"
1 "Bill Clinton"
1 "Donald Trump"
1 "GeorgeW Bush"
1 "Joseph Biden"

Warning: Variable has embedded blanks.

-------------------------------------------------------------------------------
Index Index
-------------------------------------------------------------------------------

Type: Numeric (byte)

Range: [42,46] Units: 1
Unique values: 5 Missing .: 0/5

Tabulation: Freq. Value
1 42
1 43
1 44
1 45
1 46

Footnotes​

  1. Run help import excel in Stata or see "import excel" in the Stata documentation. ↩

  2. See read in "Reading Files" ↩

  3. See write in "Writing Files" ↩

  4. See "Creating and using Stata plugins" in the Stata website ↩

  5. run help smcl in Stata or see "smcl" in the Stata documentation. ↩