Victoria II Tracker

A small Python program that watches Victoria II autosaves and records the world-market good prices, the player country's unlocked technologies, its westernisation, and its active inventions to .csv files suitable for import into LibreOffice Calc.

Requirements

• Python 3.9+ (uses pathlib, builtin generics, and f-strings).
• No third-party packages (stdlib only).
• A working Victoria II installation. Note the save directory is not under the game install: saves live in your Documents folder at Documents\Paradox Interactive\Victoria II\save games\ (e.g. C:\Users\<you>\Documents\Paradox Interactive\Victoria II\save games\).

Setup

1. Clone (or copy) this repository into the Victoria II save directory:

   Documents\Paradox Interactive\Victoria II\save games\tracker\
   

2. (Optional) create a virtual environment — the script is stdlib-only, but a venv keeps your system Python clean:

   cd "Documents\Paradox Interactive\Victoria II\save games\tracker"
   python -m venv .venv
   .venv\Scripts\activate

3. (One-off) build the reference data from your game install (already committed for vanilla; re-run only for modded installs):

   # Edit GAME_DIR in core/config.py first, or pass --game-dir explicitly.
    python -m setup.initialise --check-save example_saves/example_japan_1836.v2

   This builds the invention ID → name mapping (data/vanilla/inventions_map.json), the exhaustive modifier-name list (data/vanilla/modifiers_from_research_list.txt, every numeric tech/invention/reform effect, e.g. factory_input, artillery_defence, rgo_goods_output_iron), the 24 per-goods modifier matrices (data/vanilla/{tech,invention, westernisation}_modifiers/<group>/{rgo_goods,factory_goods}_modifiers.csv), and one <category>_modifiers.csv per group directory for each of the colonial, prestige, population, diplomacy, other, research, economic, military and rebels categories. Every matrix holds one value per source, with 0.0 for no effect within a row, and lists only the rows that group actually grants — so a file for a group with no effect in a category holds just its header. data/vanilla/matrix_index.csv lists every generated matrix (file,kind,rows,columns,nonzero_cells), so you can find the ~110 populated files without scanning all of them.

   Reference data lives in one directory per game variant (data/vanilla/; modded installs generate data/<mod>/ siblings). Like the tracking output directories, the directory must already exist — create it first, then point the run at it:

   mkdir data\my_mod
    python -m setup.initialise --game-dir <modded-install> --output-dir data\my_mod --source my_mod

   python -m setup.initialise --list shows the available setup tasks.

Usage

Both modes require an output directory that already exists. This directory is the "history" for a specific save game: it holds that world's goods_prices.csv, technology_changes.csv, westernisation_changes.csv, invention_changes.csv, and processed_dates.json. Point the script at the directory matching the save you are about to play.

One-shot

Record the save file(s) you name and exit. You are responsible for picking the files that belong to the save game being tracked. Each file must be a .v2 file name in the save directory (no paths):

python main.py --once --files autosave.v2 saves\france

# Backfill from the previous two autosaves (only correct if all three
# files are from the save game being tracked).
python main.py --once --files autosave.v2,oldautosave.v2,olderautosave.v2 saves\france

# Record a manual save (stop any watcher first, and point at the output
# directory for that world).
python main.py --once --files mysave.v2 saves\france

Watch mode

Poll the live autosave.v2 continuously; each new autosave is processed as soon as it is detected (after a short delay to let the game finish writing). Press Ctrl+C to stop.

python main.py --watch saves\france

Only autosave.v2 is watched. The rotated files (oldautosave.v2, olderautosave.v2) are deliberately ignored: when a new save game autosaves for the first time, the game cascades the previous session's autosaves into those slots, and recording them would pollute the watch with a different world's data. Since every date first appears in autosave.v2, no current-world observation is missed. Tradeoff: an autosave missed while the watcher is down is not backfilled automatically — recover it manually with --once --files (see above).

If you load a different save game, stop the watcher and restart it with the output directory for that world.

Output

The program writes four CSV files (created on first run) inside the output directory you specify. In all files, date is the in-game date (YYYY-MM-DD), not the real-world date.

goods_prices.csv: for each in-game date, a row for every good found in worldmarket.price_pool:

date,good,price
1836-01-02,coal,2.33002
1836-01-02,iron,3.53003
1836-02-01,coal,2.60028
1836-02-01,iron,3.61029

technology_changes.csv: the player country's unlocked technologies as a change log — the full set on the first date tracked, then only acquisitions and losses (0 = absent, 1 = present):

date,technology,old_value,new_value
1836-01-02,flintlock_rifles,0,1
1836-05-03,clean_coal,0,1

Replaying technology_changes.csv in date order reconstructs the unlocked set at any date; dates with no changes add no rows.

westernisation_changes.csv: the player country's westernisation levels as a change log — the full set on the first date tracked, then only changes (0 = no reform / missing / civilised, 1 = first enacted step, 2 = second enacted step, today only finance_reform_two):

date,westernisation,old_value,new_value
1845-01-01,land_reform,0,1
1846-02-01,finance_reform,1,2
1850-10-04,land_reform,1,0

Replaying it in date order reconstructs the levels at any date; dates with no changes add no rows. A civilised player (civilized=yes) tracks as no westernisation, so a fresh civilised history holds just the header and the westernisation date records every previously tracked reform returning to 0. Stale reform lines kept by the game after westernisation are ignored. Histories written before the numeric 0/1/2 scheme hold raw level names and must not be mixed: start a fresh output directory.

invention_changes.csv: the player country's active inventions as a change log — the full set on the first date tracked, then only activations and deactivations (0 = absent, 1 = present). Save-file IDs in active_inventions={ ... } map through data/vanilla/inventions_map.json to names; a missing block means no inventions:

date,invention,old_value,new_value
1836-01-02,post_napoleonic_army_doctrine,0,1
1836-05-03,flintlock_rifle_armament,0,1

Replaying invention_changes.csv in date order reconstructs the active set at any date; dates with no changes add no rows.

The good list is taken straight from the save, so late-game goods (for example automobiles, aeroplanes, radio) appear automatically as they are unlocked.

Importing into LibreOffice Calc

This section is planned and not yet ready.

Dedup

A processed_dates.json file is maintained automatically in the output directory. It stores in-game dates that have already been exported so that repeated polls never produce duplicate rows. A date is recorded for all categories together or not at all — processed_dates.json is the single source of truth for what has been tracked.

If this file becomes corrupt, simply delete it — the script will recreate it on the next run (with a warning on stderr).

Save file location

The script looks for autosave.v2 in the parent directory of the repository. In watch mode that is the only file it reads; in one-shot mode it reads whatever .v2 files you name via --files (plain file names in the save directory, no paths):

save games\
  autosave.v2
  oldautosave.v2
  olderautosave.v2
  mysave.v2
  tracker\
    main.py
    core\
      config.py
      parsing.py
      ledger.py
    domains\
      goods.py
      technologies.py
      westernisation.py
      inventions.py
    setup\
      initialise.py
      build_inventions_map.py
      research_modifiers\        player bonuses from in-game research:
        build_modifiers_from_research_list.py
        build_per_goods_matrices.py
        build_per_unit_matrices.py
        build_{colonial,prestige,population,diplomacy,other}_matrices.py
      build_{research,economic}_matrices.py
        build_military_matrices.py
        build_rebels_matrices.py
        build_matrix_index.py
    data\
      vanilla\                    <- committed vanilla reference data
        inventions_map.json
        modifiers_from_research_list.txt
        matrix_index.csv           one row per generated matrix (discovery)
        tech_modifiers\<type>\      <- matrices: one dir per tech/invention
        invention_modifiers\<type>\    type (army, commerce, culture,
        westernisation_modifiers\      industry, navy) or reform group
          <group>\                     (economic, military), each holding
            rgo_goods_modifiers.csv    rgo/factory per-goods files, land/naval
            factory_goods_modifiers.csv  per-unit files, plus one
            per_unit_land_modifiers.csv  <category>_modifiers.csv per
            per_unit_naval_modifiers.csv   implemented category (colonial,
            colonial_modifiers.csv       prestige, population, diplomacy,
            prestige_modifiers.csv       other, research, economic)
            population_modifiers.csv
            diplomacy_modifiers.csv
            other_modifiers.csv
            research_modifiers.csv
            economic_modifiers.csv
            military_modifiers.csv
            rebels_modifiers.csv
      <mod>\                     <- per-mod reference data (generated, not committed)
    example_saves\
      example_japan_1836.v2
      example_japan_1845.v2
    saves\france\
      goods_prices.csv
      technology_changes.csv
      westernisation_changes.csv
      invention_changes.csv
      processed_dates.json