seedGen

7 Days to Die Random Seed Generator

I use this project to automatically generate random seeds in 7 Days to Die, create and save a map preview for each seed, and analyse the results to search for insteresting maps.

Usage

There's multiple "entry points" in here, depending on what I want to do, the most common ones being:

• Generation & Preview
  • genWorlds.sh: generate multiple seeds, save preview and delete;
  • genSeed.sh: generate a seed, and preview;
  • drawMap.sh: generate a map preview without prefabs;
  • drawPrefabs.sh: overlay prefab previews on a map;
  • legend.png: this is the color legend for the map previews;
• Retrieving Seeds
  • uz.sh: Extracts a preview into its own directory;
  • prefabs.sh: Extracts prefab information of all previews of a size;
  • previews.sh: Extracts all previews of a size;
  • makeMontage.sh: Makes a montage with thumbnails of all extracted previews;
  • spawnpoints.sh: Extracts spawn points from all seeds (for debugging);
• Comparing Seeds
  • sortBySpecialPrefabs.sh: Sort prefabs by the number of unique "interesting" prefabs;
  • sortByPrefabs.sh: Sort prefabs by number of prefabs;
  • sortByTotalSpecialPrefabs.sh: Sort prefabs by number of special prefabs;
  • sortByUniquePrefabs.sh: Sort prefabs by number of unique prefabs;
  • sortByRate.sh: Sort prefabs by best base location;
  • allSorts.sh: Displays seeds sorted by the four different criteria above side by side;
  • allSpecials.sh: Displays seeds sorted by various special categories (work in progress);
  • speadsheet.sh: Saves various analysis as a csv file
  • highlight.sh: Highlights words (seeds) from stdin (use: allSorts.sh | highlight seedName);
  • greatest.sh: Highlights seeds that are first in multiple criteria (use: greatest.sh allSpecials.sh);
  • tops.sh: Highlights seeds that are first in at least one criteria (use: tops.sh allSorts.sh);
  • bestSeeds.sh: Sort top seeds by number of criteria they are top of;
  • withSpecial.sh: Select special list (example: withSpecial.sh -top5.txt allSorts.sh);
• Evaluating seed:
  • showSeed.sh: Opens seed preview image on windows;
  • rate.py: Shows best base location and prefabs around it;
  • prefabRules.sh: Shows prefab distribution by rule;
  • listSpecials.sh: Shows insteresting prefabs on a preview;
  • missingPrefabs.sh: Shows what prefabs are missing, optionally filtered by special list;
  • showBiomeDistribution.sh: Shows how prefabs are distributed among biomes;
  • showZoningDistribution.sh: Shows the relative proportions of residential, commercial, etc;
  • special/special.txt: List of prefabs deemed "interesting". Other lists can be found on the same folder;
• Helpful stuff:
  • listAllPrefabs.sh: Lists all prefabs that may appear in a world, including parts and tiles.
  • showPrefab.sh: displays the image preview of a prefab.
  • bash_completion.sh: auto-complete for these scripts;

There's more in here, some used by the above scripts, others used by me for more arcane purposes. Some, in fact, I no longer have use for at all.

Special prefabs come in many different types. By default, it includes all tier 4 and 5 prefabs plus traders. It can be changed by setting SPECIAL to one of the included categories (check special/ folder). You can point to your own separately maintained list of special prefabs with the SPECIAL_FOLDER, which is where files mentioned by SPECIAL will be looked for. One can also use withSpecial.sh to pick a special file to use from the special folder, and both listSpecials.sh and missingPrefabs.sh accept an optional parameter with the special list to use, from the special folder.

The best base location (rate.py) is checked by region. The region size and the radius of regions around is set by default to 512 and 1 respectively, meaning the best location depends on the prefabs found on a 1.5km by 1.5km square. The script accepts parameters to change these defaults, and sortByRate.ph will pass parameters found on the environment variable RATE_OPTS to it.

Many of these scripts rely on an environment variable called F7D2D. You have to assign it to the path to the folder where 7 Days to Die is installed. It does work with mods -- in fact, it was created to be used with Ravenhearst, though it doesn't depend on that.

This should work to point to 7 Days to Die for most people:

export F7D2D='/mnt/c/Program Files (x86)/Steam/steamapps/common/7 Days to Die'

Most scripts, when invoked, display a help message with the parameters it expects to receive.

Requirements

This is incomplete, most likely, but:

• AutoHotKey (not needed for seed generation with server, though)
• ImageMagick 6 (if using 7, you'll need to fix calls on the draw* scripts);
• XMLStarlet
• expect
• zip/unzip
• Python 3
• Python's scipy package (for rate.py, sortByRate.sh, allSorts.sh)

To make it easier to work with Python 3 you may want to switch Ubuntu to using it by default:

sudo update-alternatives --install /usr/bin/python python /usr/bin/python3 1

Note that ImageMagick comes with some rather restrictive (imho) limits by default. I fixed it doing something someone recommended on their forums, which is commenting out the policies in policy.xml (/etc/ImageMagick-6/policy.xml on Ubuntu). Like this:

  <!--
  <policy domain="resource" name="memory" value="256MiB"/>
  <policy domain="resource" name="map" value="512MiB"/>
  <policy domain="resource" name="width" value="16KP"/>
  <policy domain="resource" name="height" value="16KP"/>
  <policy domain="resource" name="area" value="128MB"/>
  <policy domain="resource" name="disk" value="1GiB"/>
  -->

How it works

This whole setup expects to be run under WSL on Windows 10, with 7 Days to Die for Windows. I have all of it on ~/seedGen, and I often add that to the path when working, though it is not necessary.

I run everything from a wsl session, using Ubuntu. I require a number of extra packages to be installed, though I honestly can't say what all of them are. See requirements above for an incomplete list. My scripts, however, will start a Windows version of 7 Days to Die.

Whenever a seed is generated, 7 Days to Die is started by startClient.sh, with seed and world size passed on the command line. Everything it writes is written to the folder UserData, which will be created inside $F7D2D. Not a very good location, I admit, but the install locations I actually used were all on folders in my desktop, which is fine.

To make the client generate seeds, I use an AutoHotKey script, which is basically a program that simulates clicking and typing. It does require hard-coding locations for mouse clicks, and I couldn't make it keyboard-only. The locations work for screen resolutions of 2560x1440. If your resolution is different, you'll need to change the locations on the two scripts: previewSeed.ahk and exitGame.ahk. The locations are all defined as X and Y coordinates at the beginning of the files.

The seed generatior, genSeed.sh, then looks at the log looking for a message that the generation has finished, or that an error occurred and it has aborted. Not all errors are detected, unfortunately, so sometimes you have to kill 7 Days to Die by hand and cancel the generation.

Once the generation has finished, savePreview.sh is called and it creates a zip file with a text file called "-.txt" with the "County" name (what 7d2d calls the seed when you look it up), the prefabs.xml file renamed to "-.xml", and a preview image of the map called "-.png". That zip file is then moved to ${F7D2D}/previews/.

By default the prefabs are draw in the map preview, but you can save time by creating a file called nodraw, and only the basic map will be created. It can still be slow, but I found I wanted the preview readily available more than I wanted to generate seeds faster.

The map preview is created by two scripts: drawMap.sh is the one that creates the basic map, and drawPrefabs.sh is the one that overlays a preview image of each prefab at it's location, and with that prefab's dimension. The preview image is "part" of the prefab, though it is optional and, therefore, might not be present. In such cases, a rectangle is drawn.

If you are using genSeed.sh directly, that's where it stops. If you are using genWorlds.sh, however, it will go on to the next seed until it creates all the seeds you asked for for that size, and then on to the next size. Often enough, however, I want to stop the generation, but given how long it can take to generate a single seed, I'm loath to just abort it. To avoid angst, you can create a file called "stop" in the directory you are calling genWorlds.sh from (not genWorlds.sh's directory), and it will finish as soon as the current seed has been generated. It will also delete "stop" at the end, so you don't risk starting a generation and leaving the computer, only to find out later that it stopped right after the first seed.

When analysing seeds, I start by going to the previews folder and using prefabs.sh to get all prefabs.xml files (which were renamed before being saved, as described above) for a particular seed size into a directory. For example, prefabs.sh 4096, run from the previews folder, will extract all these xml files into the previews/4096 folder.

Next I go inside the size folder and run commands like tops.sh allSorts.sh and greatest.sh allSpecials.sh to get a quick summary of what seeds stand out according to multiple criteria. Sometimes I'll see seeds that are high-rated on multiple criteria right away. Sometimes nothing will stand out, so I pick some of the highest rate in one criteria and another and check how they fare overall. For example, if X, Y and Z are the top three seeds in unique prefabs, I might do allSpecials.sh X Y Z to hightlight them on the lists by special prefabs and see if any one of them fares well enough.

Once I spot some candidate seeds, I use the seed evaluation commands to evaluate that seed. For example, I might do listSpecials.sh X skyscrapers.txt to see which skyscrapers seed X has, or missingPrefabs.sh X top7.txt to see which of my top 7 prefabs is missing on seed X. If a seed is really great, I might just do missingPrefabs.sh X to list all prefabs it does not have.

The final step for me is to look at what the maps look like. I run previews.sh which, similar to prefabs.sh, will extract the map previews of seeds of a size (or all sizes) into a folder. For example, preview.sh will created the folders 4096-previews and 8192-previews, assuming I have seeds of both 4096 and 8192 sizes. It also creates a montage with the thumbnails of every map, in case I'm looking for seeds based on what the map looks like.

I then go into that seed folder, look at the map (I have an "open" alias that makes it easy), and maybe do some more evaluation. Looking at maps for multiple seeds side by side is often useful, particularly when they each have different strengths and weaknesses.

My "open" alias, by the way, is this:

open() {
    cmd.exe /C start "" "$(wslpath -w "$1")"
    }

And I load my bash completions with this line in the file ~/.bash_completion:

source ~/seedGen/bash_completion.sh

Using Server for Seeds

This can use the server for seed generation, which is how it worked at first. But since alpha 18, seed generation on servers is different than on clients. It uses alpha 17 code, essentially. I don't believe this has changed on alpha 19.

You need to change genSeed.sh and uncomment startServer.sh, and comment startClient.sh. It is possible that other adjustments are necessary, since I haven't used that code in a long while.

Seed names

The random seeds are created by seed.sh. It can be configured with the following variables:

• SEED_DICTIONARY_FOLDER: folder containing word files to be used;
• SEED_FIRST_WORD_FILE: file inside the folder above containing the first word;
• SEED_SECOND_WORD_FILE: file inside the folder above containing the second word.

By default, the dictionary folder inside the folder containing the scripts is used for the seed dicitonary folder, the first word file is a file containing a list of adjectives, and the second file is a file containing a list of nouns.

The dictionary included comes from a "parts of speech files" file compiled by Ashley Bovan, which I found on the internet. See readme on that folder for more information. Also provided, and usable, are lists of adverbs and verbs. These lists are also broken down into 1, 2, 3 and 4 syllables, so if you want short seeds, you could set first and second word to dictionaries that are just one or two syllables.

Examples:

$ export SEED_FIRST_WORD_FILE='adjectives/1syllableadjectives.txt'
$ export SEED_SECOND_WORD_FILE='nouns/1syllablenouns.txt'
$ seed.sh
DenseBurk
$ export SEED_FIRST_WORD_FILE='adverbs/2syllableadverbs.txt'
$ export SEED_SECOND_WORD_FILE='verbs/2syllableverbs.txt'
$ seed.sh
MezzoConceal