CsPTools

CsPTools is a package of perl utilities (developed in the course of the project on auxiliary selection in the history of English that I was a part of at the University of Stuttgart) which assist in the processing of corpus files and especially the files output by queries with Beth Randall's CorpusSearch program (now available for download from this SourceForge page). It consists of a perl module and ten scripts that depend on it. A lot of the scripts presuppose a certain way of organizing searches and thus may not be of interest for everyone, but the most important ones should be useful for just about anyone working with CorpusSearch. CsPTools is free, open source software released under the Gnu General Public License. A recurring feature in the scripts, which was particularly simple to implement in perl, is that most kinds of user input for which it would make sense can be specified as regular expressions. Brief descriptions of the scripts follow. For more detailed information see the manual.

Documentation

The CsPTools manual is available in [html] and [pdf] formats, and is included in the full distribution (also in both formats). It contains full descriptions of all the scripts along with examples of their use, as well as more extensive installation instructions.

The components of CsPtools

• analyzer
  analyzer processes and reports on the coding strings in a CS output file. It is especially designed to deal with multicolumn codes and has several options for filtering and formatting the output, even producing simple tables.
• autocs
  autocs is purely a front end to CorpusSearch. It allows you to use one command to run separate searches on multiple files, enforcing strong naming conventions so that you can tell from the name of an output file where it came from and what queries were run to obtain it.
• codefinder
  codefinder gives you a quick and flexible way to get information about the coding string used in a corpus file. In distinction to analyzer, it is designed primarily to search for errors and monitor progress while hand-coding files.
• tagfinder
  tagfinder allows the user to search for the occurrences of words with a given part-speech-tag. The results are listed by form, either alphabetically or in order of frequency of occurrence, for each reporting either the number of occurrences or the line numbers of the actual occurrences. It has largely been superseded by the lexicon functions of CS, but still does a few things differently that might make it useful under certain circumstances (e.g. if you need regular expressions).
• editcode
  You give editcode a regular expression describing coding strings, and a list of coding files. It then finds all the examples in those files with a coding string that the regex matches, and opens them for editing one by one in emacs.
• integratecodes
  integratecodes is a tool for transferring coding strings between corpus files. This makes it possible, e.g., to use CorpusSearch to extract a subset of the sentences from some (set of) coded corpus file(s), hand-edit the coding strings that appear there, and then integrate the changes back into the original files. See the manual for some examples demonstrating how this can be useful.
• ipcoding
  ipcoding is a simple script that converts coding files output by earlier versions of CorpusSearch to the format used in CS version 2. Specifically, it moves the CODING node inside the IP node.
• progress
  progress is a very simple script that uses facilities originally created for codefinder to measure and report on progress in hand-editing automatically generated coding strings in a file or set of files.
• next
  next is just a time and typo saver for hand editing coding files. Using the same facilities as progress, it determines which file in a directory is the next to be hand edited and opens it in emacs. When the user closes emacs, next runs codefinder on it, reporting the statistics to the user, offering to clean up back up files, and then opening the next file.
• mvcodh
  mvcodh is a very specialized tool that renames the files in a given directory in order to facilitate the use of file-naming conventions developed for progress and next.

Requirements/dependencies

CsPTools requires perl v5.6 or later. editcodes and next require emacs, as do the editing functions of analyzer, though the basic functions of the latter are independent. The scripts have been developed and used on Linux and Mac OSX, and should also work on other Unix-like systems. I would be frankly surprised if everything worked as is on a DOS/Windows system or Mac OS 9.x or earlier. However, I would be interested in making them work on such systems in the future, so if anybody out there actually tries it out, let me know how it turns out.

Downloads and Installation

CsPTools is available as a zip archive containing the entire package, which consists of the ten tools, the perl module CsPTools.pm which all of them depend on, and the manual formatted as PDF and HTML. At the moment there is no automated installation program, but installing by hand should be reasonably painless, and there are instructions for doing so in the manual.


CsPTools v0.3.1 (July 26th, 2005): [CsPTools.zip]