spectra-lexer 5.0.0

Stroke and rule analyzer for stenography

The Spectra Steno Lexer is an experimental tool for analyzing and matching patterns of steno keys against the text they produce using various rules from steno theories (mostly Plover theory). It also has advanced search functions for steno dictionaries which have been mostly lacking in Plover up to this point.

Installation and Operation

Spectra is primarily designed as a plugin for Plover. If you have installed the latest binary release of Plover (4.0.0.dev8 as of this writing), the plugins manager should be able to find this program in the PyPI database and set it up automatically for you. When opened from the main toolbar, Spectra will automatically load Plover’s dictionaries for manual searching and will also attempt to analyze strokes sent from Plover as you type. Configuration of various features is possible through the menu bar.

Perhaps the most interesting, yet resource-intensive part of this software is the index feature. On first start, you will be asked if you want to create an index of steno rules from your Plover dictionary. With an index, not only will you be able to see the rules that make up any given translation, but you will also be able to search for other translations that use the same rule from a link on the steno board diagram. The initial index is restricted in size in order to keep load times down, but if you want rule comparisons against more complicated words, you can create a larger index from the Tools menu if you want. Please be aware that this isn’t without tradeoffs; a large index will increase startup time significantly.

Advanced Installation

If you have built and installed Plover from source, it is not likely to have the plugins manager by default. In this case it is possible to install Spectra through pip like any other Python package. Plover searches through all available Python paths to find plugins; so long as it ends up in the same general place that Plover looks for its other dependencies, it should find it just fine.

To run this software on its own from source, you must have a correctly installed Python distribution (3.6 or greater). Download or clone the source into a free directory, change to this directory in a terminal and type:

python3 setup.py install

This will install it to your Python distribution with all features.

Advanced Operation (from console)

From the console, you can execute the main GUI program on its own by typing:

spectra_lexer gui

The standalone mode operates identically to the plugin in all respects except that it cannot decipher strokes in real-time. By default, the program will look for your Plover dictionaries in the default user app data directory for your platform. It may not find them if you have them somewhere else, or have a strange user configuration. In any case, you can load dictionaries manually from the menu bar in standalone mode.

There are also batch modes available to analyze entire dictionaries, such as:

spectra_lexer analyze [--translations-files=FILE_IN1 [FILE_IN2 ...]] [--rules-out=FILE_OUT]

Each FILE_IN is a path to a JSON file containing a dictionary of steno translations, and FILE_OUT is a path to a (new) JSON file that will store the output. The lexer will run on each translation and store the output in the same format as the rules dictionary files.

The most advanced stuff is available in the console, by typing:

spectra_lexer console

This starts a Python interpreter with Spectra’s engine commands directly exposed as top-level functions. Documentation on this is currently sparse, so it is difficult to operate without direct knowledge of the Python source code.

Details

This software is currently experimental with many rules unaccounted for, so do not rely on it to figure out the rules of stenography with 100% accuracy. Depending on the config setting, if it cannot match every single steno key to letters in the word, it may not return a result at all (to avoid guessing wrong) or may return an incomplete guess on the first part of the word. Inversions and asterisks are particularly troublesome here; inversions of steno order violate the strict left-to-right parsing that lexers rely on, and oftentimes there is not enough context to figure out the meaning of an asterisk from just a stroke and the word it makes in the absence of other information. Briefs are often constructed by keeping only the most important parts or sounds of a word, and Spectra can usually match these, but briefs relying on strange phonetics or arbitrary sequences of keys simply cannot be matched without pre-programmed custom rules (which are included for some of the most common briefs, but not many).

When searching from the lookup tool, if a word is chosen and there is more than one stroke entry for it, the lexer will analyze each one and select the one that has the best possibility of being “correct” (i.e. not a misstroke). This choice is based on, among other things: the number of steno keys matched to rules, the number of letters those rules cover, the number of “unusual” rules involved, and if all else fails, choosing shorter strokes over longer ones to break ties.