Where everything lives

Everything Parseh has is in its own folder: the software, everything you put into it, and everything it makes. Outside it there are only:

• the models of the Chinese word analyzer, in ~/.pkuseg;
• the environment, when you made it with your own conda;
• on a TeX Live that has no Italian hyphenation, the Italian patterns the server adds to your own TeX folders as it starts (TEXMFHOME and TEXMFCONFIG, with the XeLaTeX format rebuilt to take them), so that the PDF of a studio document written in Italian breaks its words where Italian breaks them;
• with ./install.sh --pdf, the TeX packages it adds to TeX Live itself;
• in each browser, a few of your preferences.

Deleting the Parseh folder leaves these behind: Removing it says where each of them is.

1. The folder#

Parseh/  books/                 the books  youtube/videos/        the videos  youtube/anki/          the Anki card store  markdown/library/      the studio's documents  exercises/             the exercise decks  clips/                 the clip tray  dict/  corpus/  mt/  components/                         what the reading help fetches  stt/                   speech to text: its program and its two models  arasaac/               the ARASAAC pictograms and the words that name them  texmf/                 the TeX packages Parseh got for its drawings  markdown/latex/        the LaTeX drawings, made again when missing  config/                your settings, the devices let in  lib/  markdown/app/  markdown/exlex/  youtube/lib/                         the software  html-guide/            this guide  docs/                  notes on the design  .runtime/              the environment the installer made  .tls/                  the certificate  .parseh-release.json   the list of the release's own files  .parseh-update/        the backups of the last updates

What one of each is made of:

• books/<language>/<slug>/ — a book: book.json, main.tex and its chapters, source/ (the text it was made from), markdown/ (the notes written beside it), audio/ (its narration), reader/ (its reader, built) and main.pdf.
• youtube/videos/<language>/<id>/ — a video: video.json, annotations.json (the glossed transcript), its transcript and parts, media.* (the film, for a video of your own), markdown/ (its notes).
• youtube/anki/<language>/<deck>/ — a deck of the Anki card store.
• markdown/library/<language>/<id>/ — a studio document: source.md, meta.json, its pictures and recordings, build/ (its PDF).
• exercises/<language>/<slug>/ — an exercise deck: deck.json, items/, schedule/ (your answers), images/, audio/.
• clips/ — the clip tray: the recordings and frames cut for cards.
• dict/, corpus/, mt/, components/ — what the reading help fetches (Settings → Reading help): dictionaries, translated sentences, translation models, character packs.
• stt/ — speech to text, fetched on Settings → Speech to text: runtime/ (the program, in a folder named for the pinned list and the Python it was made for), models/ (the two Whisper models) and tmp/ (the sound of a video while it is being transcribed, deleted afterwards).
• arasaac/ — the ARASAAC pictograms, fetched on Settings → Pictograms (ARASAAC) (Pictograms): pictograms/ (the pictures), index.<language>.json (the words that name them), a manifest, and LICENSE-ARASAAC.txt, which carries the credit their licence asks for.
• texmf/ — the TeX packages Parseh got on Settings → LaTeX drawings, for the latex blocks of a theme, with parseh-packages.json, the list of them.
• markdown/latex/ — the drawings the latex blocks became, each once: made again from its block when it is missing.
• config/ — your settings: prefs.json, each book’s reading place, the settings that follow you from device to device, the names you gave the levels and the chunks you flagged to review later; network.json, who may reach Parseh, its port, and every device you let in, each with the secret token it carries — keep it to yourself, as you would a key; languages.json, the languages you added, when you have; updates.json, whether Parseh looks for a new version once a day; prompts.json, the prompts you wrote for a chatbot; and digests.json with wheres.json, what a phone checks the things it keeps against.

The files at the top:

Every kind of content is filed under its language, in a folder named after the language in English, in lower case: persian, arabic, italian, japanese, french, german, turkish, english, hindi, spanish, chinese. A book, a video or a note lying directly under books/, youtube/videos/ or markdown/library/ — the way they were kept before Parseh had languages — is still read, as Persian, with a note to move it.

You never need to touch these folders: every door adds, changes, backs up and removes what is behind it from its own pages. They are described here so that you know what to keep safe, and what a folder you see is for.

2. What a release brings#

Parseh comes as a release: one zip, parseh-<version>.zip, holding the software, and not your content (Installing Parseh). A fresh install has:

• all the code, the fonts of Persian, Arabic and Hindi (with their licences), the language registry, and the installers;
• the studio’s starting pages — what a new document opens with, one for each language — with their pictures and their recording;
• this guide, its pages already compiled;
• the list of its own files, .parseh-release.json;
• and empty shelves: every folder of content holds nothing but a placeholder until you fill it.

An update from Settings replaces the files on that list and nothing else. What you add is not on it, so no update touches it.

3. What never leaves your computer#

What you read and write is yours, not the software’s. No release carries it, no update touches it, and none of it leaves the computer unless you take it — a download, a backup, a book kept on your phone:

• your books, with their narrations, hundreds of megabytes each: books/;
• your videos, and the films of your own: youtube/videos/;
• your studio documents: markdown/library/;
• your exercise decks, with every answer you gave in them: exercises/;
• your Anki cards and decks: youtube/anki/;
• the clip tray: clips/;
• the dictionaries, sentences, translation models and character packs — somebody else’s work, under somebody else’s licence: dict/, corpus/, mt/, components/;
• the TeX packages Parseh got for its drawings: texmf/;
• speech to text’s program and models — somebody else’s work, and gigabytes: stt/;
• the ARASAAC pictograms — somebody else’s pictures, under a licence that asks for a credit: arasaac/;
• your settings and the devices you let in: config/;
• the environment, the certificate, the log: .runtime/, .tls/, serve.log;
• the backups the last updates kept: .parseh-update/.

So the backups are how your content moves to another computer: Backups at a glance has the four shelves, and a single book, video, document or deck travels as its own zip from its own page. Moving a whole Parseh from before it could update itself into a fresh install is a few folders moved once.

A copy of the source code, made with git, holds the same software and more — the tests, and what GitHub needs. Git ignores most of the list above, and shows your books, videos and documents as files it has never been told to keep, so none of it is committed unless you commit it. Such a copy is for working on Parseh itself, not for using it: it cannot update itself from Settings.

4. Kept in the browser#

A few things are not in the folder at all, because they are yours in one browser rather than the toolbox’s: the language picked on the chips, the Browser | Mobile choice, the switches of the reader and the player, the settings of the ⚙ panels — the sizes, the levels shown, Diacritics, the page zoom —, the kanji you marked as known. Each browser keeps its own, per address, and a backup does not carry them. Where you are in each book, the narration’s speed and its gap, the theme, the names of the levels and the chunks you flagged to review later are the computer’s, in config/prefs.json, so a phone goes on where the desk stopped.

5. Pages that open from the disk#

Three kinds of page are plain files that a browser opens without the server: a book’s reader (books/<language>/<slug>/reader/index.html), the books’ library page (books/index.html), and this guide (html-guide/index.html). They read as they always do; whatever needs the server — saving, uploading, building, a backup — needs the page opened from Parseh’s own address, and the library page’s buttons say so when it is not.