--- pretty_name: Warcraft III Community Map Archive license: other license_name: community-authored-maps license_link: https://huggingface.co/datasets/magicwenli/war3-maps#rights language: - zh - en tags: - games - archive - warcraft-iii - warcraft - w3x - mpq - digital-preservation size_categories: - 10K/.`: immutable map or campaign objects - `covers//.webp`: minimap / preview thumbnail, when one could be extracted; the same SHA-256 as the object it belongs to - `catalog/maps.json`: versioned website catalog - `catalog/maps.jsonl`: one record per line for downstream processing - `catalog/failures.json`: inputs that could not be extracted or parsed Catalog records point at both files by path and by resolve URL (`dataset_path` / `download_url`, `cover_path` / `cover_url`), so the catalog stays small enough to fetch directly from a browser. ## Dataset viewer The preview and `load_dataset` both read **only** `catalog/maps.jsonl` — one row per archived object. The map files themselves are binary MPQ archives and are not loadable as tabular data; fetch them from `download_url` (or `dataset_path` via `hf_hub_download`) once you have selected rows from the catalog. ```python from datasets import load_dataset catalog = load_dataset("magicwenli/war3-maps", split="train") melee = catalog.filter(lambda row: row["content_type"] == "map" and row["max_players"] == 2) ``` The record schema is pinned in this card's metadata, so a new catalog field is invisible to the viewer until it is added there as well. ### Catalog fields | Field | Meaning | |-------|---------| | `sha256`, `size`, `filename`, `extension`, `format`, `content_type` | object identity; `content_type` is `map` or `campaign` | | `name`, `author`, `description`, `category`, `collection`, `collections` | presentation metadata; `name_source` records where the title came from (`w3i`, filename, …) | | `recommended_players`, `max_players`, `player_count`, `playable_width`, `playable_height`, `tileset` | gameplay metadata parsed from `war3map.w3i` | | `format_version`, `editor_version`, `build_version` | map format and authoring-tool versions; `build_version` is a 4-part list when known | | `parse_status`, `parse_error` | `ok`, `carved` (salvaged from raw sectors), `metadata_error`, or `metadata_unavailable` (campaign packs are not parsed for metadata) | | `modification` | third-party script modification detected in the map, when any: tool, label, variant, activation strings and evidence | | `cover_status`, `cover_source`, `cover_path`, `cover_url` | thumbnail availability and origin: `preview`, `map` (minimap), or `salvage` — recovered from an unreadable archive's raw member data, correct in about 95% of cases | | `dataset_path`, `download_url`, `source_paths` | where the object lives here, and the original paths it was ingested from | ## Acknowledgements Special thanks to Bilibili creator [关先生丶的游戏实况](https://space.bilibili.com/2534568), whose collection and sharing work is the source of a substantial part of the classic maps here. ## Rights The original maps remain works of their respective authors. Inclusion in this archive does not claim ownership or grant rights beyond those provided by each author. The `other` license tag on this dataset refers to this section, not to a single license covering the contents. Source provenance is retained in the catalog. For attribution fixes, copyright concerns, or takedown requests, open an issue at https://github.com/war3-archive/war3-maps/issues with the SHA-256 and supporting details. Confirmed requests will be handled by removing the public object and recording a tombstone in the catalog.