Diogenes

Diogenes is a tool for searching, browsing and reading the databases of ancient texts, primarily in Latin and Greek, that were once published by the Packard Humanities Institute and the Thesaurus Linguae Graecae.

If you just want to install and run the program, go to the Diogenes webpage, and download a pre-packaged version for your operating system (Windows, Mac and many flavours of Linux) from there:

https://d.iogen.es/d

The detailed technical information contained here is only intended for people who want to look at the source code of Diogenes and build it for themselves.

Building

I would prefer not to have Diogenes packaged as part of a larger (e.g. Linux) distribution; I would rather that users who are not building the application themselves download the installer from the Diogenes website. This is because I use the download statistics to help justify spending part of my publicly funded time on this project.

Diogenes can either be built and run as a HTTP server application, or as a standalone application that seamlessly combines both server and browser by using the Electron application framework.

Diogenes uses several dictionaries, as well as pre-computed morphology tables, which need to be built or downloaded before use. If you want to skip building these, execute this command, which will download the pre-computed data from Github:

make -f mk.prebuilt-data

If you would prefer to build the morphology data and dictionaries yourself, see the instructions below.

Platform-Specific Requirements

For the most part, Diogenes and its installers can be built for all platforms on either Linux or Mac, but there are some steps that need to be done on a specific OS. In particular, with respect to Windows:

• The Windows installer is created with a Windows application, Inno Setup, which is currently only available as a 32-bit app. OS X will not run these anymore, even under emulation via wine. So the Windows installer currently must be made on Linux, and requires wine and innoextract.

• Creating the Diogenes icon file for Windows requires some utility programs that can be installed on Linux (the Debian packages are librsvg2-bin and icoutils) or OS X (the Homebrew packages are librsvg and icoutils).

• Integrating the icon with the Windows .exe file requires a Windows utility (rcedit.exe) running under wine.

• Modifying the Windows Perl executable to use utf-8 for system calls requires another Windows utility (mt.exe), which also runs under wine.

In light of these issues, building the Windows app and installer ought currently to be done under Linux.

Creating the Mac icon file needs to be done on OS X and requires installing the png2icns package for Node.js: install Node via Homebrew and then run npm install png2icns -g. (There is an entirely different png2icns program that also runs on Linux, but it looks obsolescent.)

Regarding the issues with compiling Morpheus on Linux and OS X, see below.

Building the Electron app

Once you have installed the utility programs for building the icons as mentioned above, you can run this command, which will create the icons and collect a number of other required files:

make

To build the standalone Diogenes application which has the server and client browser integrated via Electron, use one of these make commands according to the platform you're building for:

make linux64   # for linux (64 bit) build
make w32       # for windows build
make mac-x64   # for mac osx build (Intel)
make mac-arm64 # for mac osx build (Apple silicon)

Gatekeeper will prevent the unsigned Apple silicon app from running unless you remove its quarantine attribute: xattr -cr Diogenes.app.

Building the installers

Apart from Windows, the other installers can be built on either Linux or OS X. To create the Linux installers you will need to install fpm, which is done via the Ruby package manager (see instructions), and for the RPM installer you will also need to install rpm, (Linux or Homebrew).

There is a target in the Makefile to create an OS X pkg, but I don't recommend using it: if another version of Diogenes with the same version number is already installed, the Mac package installer will leave it untouched, will not install the new package, and will nonetheless report success. The Diogenes.app for OS X is distributed in a simple zip file, which seems to work fine for users.

To build an installer for your target platform, run one of the commands below.

make installer-w32        # Make a Windows installer
make installer-mac        # Make zip files of the Mac app for both Intel and ARM
make installer-deb64      # Make a Debian package
make installer-rpm64      # Make an RPM Linux package
make installer-arch64     # Make a pacman package for Arch Linux

Running the server

Instead of running the standalone, integrated app, you may prefer to run Diogenes as a server and to connect to it via an ordinary web browser. The server can be started using the script:

server/diogenes-server.pl

For full usage details run it like this:

server/diogenes-server.pl -?

Additional features

Diogenes has a number of other features which predate the development of the standalone app and which are no longer fully supported but which may still work. These include a command-line interface (diogenes-cli.pl), LaTeX output, and support for a wide variety of pre-Unicode encodings for Ancient Greek.

Various options for XML export are available by running the server/xml-export.pl script from the command line.

Building the morphology data & dictionaries

Instead of downloading the pre-built lexical data via make -f mk.prebuilt-data, you can build it from scratch. The following steps will download the lexica and the Morpheus parser, which have been provided by the Perseus project, and repackage them for Diogenes.

1. Wordlists

   The first step is to generate Greek and Latin wordlists, which for Greek are derived from the Perseus corpus and the TLG wordlist; and for Latin from the PHI, Perseus and DigiLibLT corpora. The DigiLibLT corpus has to be downloaded first by hand after making an account on their website, but the Perseus corpora are downloaded automatically. Run this command, specifying the location of the non-Perseus databases on the command line:

    make -f mk.wordlists PHIDIR=/path/to/phi TLGDIR=/path/to/tlg_e DIGILIBDIR=~/path/to/digilib
   

2. Morphology

   The next step is to generate the morphological data by running Morpheus over the wordlists. Diogenes still uses an older, known-good version of Morpheus, which is very old code written in C. It does not compile on OS X or with recent version of GCC. It does compile on Linux with GCC version 6; to do so, run this command:

    make -f mk.morpheus-old
   

   There are newer versions of Morpheus on Github that will compile on OS X, but some of these have a serious bug that causes incorrect output. This bug has been fixed in the Morpheus repository of Johan Winge, who has also improved the marking of vowel lengths. Judging by the smaller size of the output files, this version may have less coverage, but I have not had time to test it properly. If you want to try it, run this command:

    make -f mk.morpheus-alatius
   

   If you don't want to compile Morpheus, you can just download and use the morphological data from version 3 of Diogenes, which still works fine with version 4. Run the following command:

    make -f mk.morpheus-v3
   

3. Lexica

   The next step is to download the LSJ Greek lexicon and the L-S Latin lexicon, which were originally digitized by the Perseus project and have subsequently been corrected by the Logeion project. To get the lexica from Logeion, run:

    make -f mk.lexica-logeion
   

   Alternatively, you can get the Perseus version of the lexica by running:

    make -f mk.lexica-perseus
   

4. Integration

   The next step is to integrate the morphological data with the lexica, and package all this in the form that Diogenes requires. To do this, run:

    make -f mk.data
   

   The intermediate files generated in the course of all of the steps above are put the build/ directory, and the final lexical data which is used used by Diogenes at runtime is put in the dependencies/data directory, whence it is read by diogenes-server.pl.

5. PDFs of Lexica

   There is one more step, which is to integrate information on where words can be found in the print versions of the Thesaurus Linguae Latinae and the first edition of the Oxford Latin Dictionary. The PDFs of the TLL can be downloaded from the website of the Bayerische Akademie der Wissenschaften by hand, via a menu item in the Diogenes Electron application, or by running on the command line:

    server/tll-pdf-download.pl path/to/destination/folder
   

   If you also have a PDF of the first edition of the OLD that has the running heads as bookmarks, you can extract the necessary information from that as well. To generate the bookmarks for both the TLL and OLD, run:

    make -f mk.pdf-data TLLDIR=/path/to/tll/directory OLDFILE=/path/to/old/file