Packagesearch: module to scan Stata .do files and identify SSC packages used by the code

Installation

To install, type the following command into Stata.

net install packagesearch, from("https://lydreiner.github.io/Statapackagesearch/")

Note: There are two required ancillary files (signalcommands.txt and stopwords.txt) that are automatically pulled from the repository after the first use of the packagesearch command.

Syntax: (also available in the help file)

packagesearch, codedir(directorytoscan)[ filesave excelsave econstats falsepos installfounds]

codedir(directorytoscan) is required. It specifies the directory that contains the .do files to be scanned for SSC packages.

filesave outputs a list of all files that were parsed during the scanning process.

excelsave saves the results of the scan into an Excel spreadsheet titled candidatepackages.xlsx.

• This file is saved in the specified directorytoscan and will include a list of parsed programs if filesave is also indicated as an option.

econstats matches the scanned .do files against a list of packages commonly found in replication packages for AEA economics research articles (rather than a list of all packages in existence at SSC)

• This option will improve the accuracy (reduce # of false positives) when scanning Stata code that is used in economics research

falsepos removes packages that were frequently found to be false positives during beta testing.

• Presently this includes the following packages: white, missing, index, dash, title, cluster, pre, bys.
• This option becomes slightly redundant with the econstats option, so one or the other is recommended

installfounds installs all SSC packages found during the scanning process into the current working directory.

Description:

The code begins by either collecting a list of all packages hosted at SSC using the whatshot command, or pulling a list of common SSC packages used in economics research (if option econstats is specified).
Next, it identifies all .do files in the specified codedir directory and subdirectories, then parses each .do file into individual words using the txttool command. Finally, it matches the individual words against the list of SSC packages and outputs a list of candidate packages that were (likely) used when the Stata code was run.

Questions?

Contact:
Lydia Reiner (lr397@cornell.edu)
Lars Vilhuber (lars.vilhuber@cornell.edu)