A Plaine and Easie Code to ABC music notation converter
Both Plaine and Easie Code (from now, PAE) and ABC were born to allow musical notation using just alphanumeric characters. PAE's manifesto by Barry S. Brook, a musicologist, had the title "Notating Music with Ordinary Typewriter Characters" (1964), with the stated goal to identify otherwise ambiguous works in musical library catalogs. The idea is that the few first bars of the musical composition, the musical incipit, should be part of the library catalog, so works with a identical name could be differentiated if the musical incipit is known and available. On the other hand, ABC was designed by Chris Walshaw, a (at that time) "musically illiterate" hobbyist that was hitch-hicking around Europe with his rucksack and carrying a flute, and its first implementation was a computer program released in 1983. Both are based on the German and Anglo-Saxon tradition of naming the notes using the first letters of the alphabet, A to G, and share similar, although not identical, solutions for common situations, like using the comma (,) and the apostroph (') for designing lower or upper octaves, or using a similar visual symbol for the bar line: the slash (/) for PAE and the vertical bar or pipe (|) for ABC. They also share another common characteristic: their initial goal is for transcribing short fragments, and monodic, or single-voice, melodies. Although ABC has clearly overcome its initial limitations (most notably abcm2ps), its software implementations naturaly lead to more compact engraving representations than other, full-score oriented, packages.
PAE has always been used in a niche context, that is, in music library catalogs, and formalised in scholarly journals and meetings by musicologists and librarians, while ABC, born in the Internet era by music aficionados and computer enthusiasts, has been discussed in the open, using mailing lists, web pages, tutorials and forums. ABC enjoys a very lively community and a myriad of software packages that PAE lacks. Using the classical Eric S. Raymond dichotomy, PAE would be the cathedral, and ABC the bazaar.
ABC software implementations include editors, converters, engravers, sound generators, web plugins, etc. For PAE, only a few converters exist, and all of them to convert from PAE to some other notation that can produce a graphical (pentagram) output.
In 2003, Rainer Typke released pae2xml, that converts PAE to, unsurprisingly, MusicXML notation. It was developed for converting music incipits as they can be found in the RISM database. Written in Perl, it relies heavily on regular expreesions. Available at http://rainer.typke.org/pae2xml.html.
Reinhold Kainhofer took pae2xml's maintenance, updated and polished it, and in 2010 published version 1.0. Besides general improvements and fixes, it has gained input file format flexibility and a test suite. However, to my taste it is still somewhat unstructured and difficult to contribute to. It can be found at http://repo.or.cz/w/pae2xml.git.
In 2007, Michal Žbodák wrote another converter as part of his master thesis. Also written in Perl, incipity.pl converts PAE to Lilypond. Unfortunately, the variable names and comments are in Slovak, the program is not particularly well structured, it is written using a CGI interface, and there are a few hardcoded addresses. It is available at http://is.muni.cz/th/60712.
In 2011, Ignacio José Lizarán combined Michal Žbodák's script with Typke's and Kainhofer's pae2xml.pl, published pae2ly.pl. It requires the input file to be coded in Marc21, using 031 subfields. This requirement is unsurprising given the niche field PAE is used. However, the program is written using a mix of English and Spanish, poorly commented, also with a heavy use of long regular expressions and, again, not easy to follow. It is available at http://sourceforge.net/projects/paetolilypond/.
There is also the pae2kern converter, that takes a PAE notation file, using a special syntax to code the incipit part, and converts it to Humdrum. For printing, Humdrum relies on ABC format, using abcm2ps. pae2kern can be found at http://extras.humdrum.org/man/pae2kern/.
Finally, after pae2abc.py was initially written, as set of tools for creating and rendering PAE under the name of Verovio (https://www.verovio.org/), from an editor (https://www.verovio.org/pae-editor.xhtml) to several rendering tools (https://www.verovio.org/pae-examples.xhtml or https://www.verovio.org/command-line.xhtml).
pae2abc converts PAE musical incipits to ABC, because, as far as I've seen, ABC tools are perfectly capable to express all the musical elements defined in the PAE standard, and its output more appropiate for those incipits that full score oriented software like Lilypond. For example, ABC optional information fields allow abcm2ps to add some identification fields to the incipit, like the title, the author, origin, literary incipits or even notes, all in a compact form that is suitable for music catalogs. Even more, the style can be changed via ABC stylesheets. And, although gregorian notation is not yet supported, at least the four-line staff is, easily.
On the other side, I've tried to write pae2abc as clear and as explicit as possible, easy to follow, and to contribute to, and to debug. I've seen that it is perfectly possible to write such a converter without resorting to regular expressions, and I hope that the code legibility benefits from that.
TODO: test suite, output formats, etc.
There is not such a think as a standard PAE file format. This program allows several formats.
In it simplest form, each line contains a single string starting with the % character that define the key, followed by a $ that codes the key signature, a '@' with the time signature, a space and the tune.
In its minimal expression, a file with those contents:
%G-2$xFC@3/4 4''C+/{6C'B''CEDx'A}8B-''4C+/{6C'B''CDExE}8F-'xB/
Gets expanded and converted to:
X: 1
L: 1/4
M: 3/4
K: D treble2
(c | c/4)B/4c/4e/4d/4^A/4 B/2 z/2 (c | c/4)B/4c/4d/4e/4^e/4 f/2 z/2 ^B/2 |
A slightly more enriched on is inspired by ABC, where the tune may be preceded by several ABC compatible fields, that get passed as-is to abcm2ps to be converted to graphical format. For simplicity, pae2abc only recognizes them preceding the PAE string; that is, when the line starts with %, the clef sign, the PAE string is converted to ABC and all preceding input information fields are reordered so they are valid and can be passed to abcm2ps; a missing X field is automatically added. For example,
C: Liszt, Franz
T: Du bist wie eine Blume
O: 1842
P: 455009044
w: Du bist wie eine Blume
%G-2$xFCGDAE@3/4 4-/=1/4-'AD/F{8A''D}C'F/2A4G/F-F/FBB/BAD/G--/
Gets converted to:
X: 1
T: Du bist wie eine Blume
C: Liszt, Franz
O: 1842
P: 455009044
L: 1/4
M: 3/4
K: F# treble2
z | Z | z A D | F A/2d/2 c/2 F/2 | A2 G | F z F | F B B | B A D | G z z |
w: Du bist wie eine Blume
A third format is the one that was more recently introduced by Verovio (https://www.verovio.org), where the file consists on pairs of key-values, as seen in this example:
@clef:G-2
@keysig:
@timesig:c
@data:2''C'B/=/''CC/=/2-4DE/2-8{'B''C+C'B}/2''C-//
It gets converted to this abc compatible output:
X: 1
L: 1/4
M: C
K: C treble2
c2 B2 | Z | c c | Z | z2 d e | z2 B/2(c/2c/2)B/2 | c2 z2 ||
Finally, there should be a Marc21 input format, as PAE is often embedded in Marc21 records, 031 tag, subfields $g, $n, $o and $p (http://www.loc.gov/marc/bibliographic/bd031.html). This is in the TODO list.