- {categoryOrder.map(([categoryId, category]) => {
+ {filteredCategoryOrder.map(([categoryId, category]) => {
const manual = category.tags.filter((tagId) => !isDerivedTag(tagId));
if (manual.length === 0) {
return
;
diff --git a/src/elements/tag-selector.tsx b/src/elements/tag-selector.tsx
index 6c2621a8..48d12d0b 100644
--- a/src/elements/tag-selector.tsx
+++ b/src/elements/tag-selector.tsx
@@ -56,9 +56,10 @@ export type TagSelectorStyle = 'simple' | 'advanced';
export interface TagSelectorProps {
selection: TagSelection;
onChange: (selection: TagSelection, style: TagSelectorStyle) => void;
+ context?: 'models' | 'datasets';
}
-export function TagSelector({ selection, onChange }: TagSelectorProps) {
+export function TagSelector({ selection, onChange, context = 'models' }: TagSelectorProps) {
const [simple, setSimple] = useState(true);
const { editMode } = useWebApi();
@@ -66,22 +67,28 @@ export function TagSelector({ selection, onChange }: TagSelectorProps) {
useEffect(() => {
if (simple) {
- const reduced = reduceToSimple(tagCategoryData.values(), selection);
+ const filteredCategories =
+ context === 'datasets'
+ ? [...tagCategoryData.entries()].filter(([id]) => id === 'dataset').map(([, c]) => c)
+ : [...tagCategoryData.entries()].filter(([id]) => id !== 'dataset').map(([, c]) => c);
+ const reduced = reduceToSimple(filteredCategories, selection);
if (reduced !== selection) {
setSimple(false);
}
}
- }, [simple, tagData, tagCategoryData, selection]);
+ }, [simple, tagData, tagCategoryData, selection, context]);
return (
{simple ? (
) : (
@@ -92,7 +99,15 @@ export function TagSelector({ selection, onChange }: TagSelectorProps) {
onClick={() => {
setSimple(!simple);
if (!simple) {
- const reduced = reduceToSimple(tagCategoryData.values(), selection);
+ const filteredCategories =
+ context === 'datasets'
+ ? [...tagCategoryData.entries()]
+ .filter(([id]) => id === 'dataset')
+ .map(([, c]) => c)
+ : [...tagCategoryData.entries()]
+ .filter(([id]) => id !== 'dataset')
+ .map(([, c]) => c);
+ const reduced = reduceToSimple(filteredCategories, selection);
if (reduced !== selection) onChange(reduced, 'simple');
}
}}
@@ -128,13 +143,20 @@ export function TagSelector({ selection, onChange }: TagSelectorProps) {
);
}
-function AdvancedTagSelector({ selection, onChange }: TagSelectorProps) {
+function AdvancedTagSelector({ selection, onChange, context = 'models' }: TagSelectorProps) {
const { tagData, categoryOrder } = useTags();
const { editMode } = useWebApi();
+ const filteredCategoryOrder = useMemo(() => {
+ if (context === 'datasets') {
+ return categoryOrder.filter(([id]) => id === 'dataset');
+ }
+ return categoryOrder.filter(([id]) => id !== 'dataset');
+ }, [categoryOrder, context]);
+
return (
- {categoryOrder.map(([categoryId, category]) => {
+ {filteredCategoryOrder.map(([categoryId, category]) => {
if (category.tags.length === 0 || (category.editOnly && !editMode))
return
;
@@ -166,11 +188,18 @@ function AdvancedTagSelector({ selection, onChange }: TagSelectorProps) {
);
}
-function SimpleTagSelector({ selection, onChange }: TagSelectorProps) {
+function SimpleTagSelector({ selection, onChange, context = 'models' }: TagSelectorProps) {
const { tagData, categoryOrder } = useTags();
+ const filteredCategoryOrder = useMemo(() => {
+ if (context === 'datasets') {
+ return categoryOrder.filter(([id]) => id === 'dataset');
+ }
+ return categoryOrder.filter(([id]) => id !== 'dataset');
+ }, [categoryOrder, context]);
+
const tags = useMemo(() => {
- return categoryOrder
+ return filteredCategoryOrder
.map(([, category]) => category)
.filter((category) => category.simple)
.flatMap(({ tags }) => {
@@ -183,7 +212,7 @@ function SimpleTagSelector({ selection, onChange }: TagSelectorProps) {
})
.filter(isNonNull);
});
- }, [categoryOrder, tagData]);
+ }, [filteredCategoryOrder, tagData]);
const selected: TagId | undefined = useMemo(() => {
const required = tags.filter(([tagId]) => selection.get(tagId) === SelectionState.Required);
diff --git a/src/lib/license.ts b/src/lib/license.ts
index 0a95e6ef..960c181c 100644
--- a/src/lib/license.ts
+++ b/src/lib/license.ts
@@ -21,6 +21,12 @@ const KNOWN_LICENSES_: Record
= {
conditions: ['include-copyright', 'state-changes'],
limitations: ['liability', 'trademark-use', 'warranty'],
},
+ 'Academic-Research-Only': {
+ name: 'Academic Research Purpose Only',
+ permissions: ['private-use', 'modifications'],
+ conditions: ['include-copyright'],
+ limitations: ['liability', 'warranty'],
+ },
'BSD-3-Clause': {
name: 'BSD 3-Clause "New" or "Revised" License',
permissions: ['commercial-use', 'distribution', 'modifications', 'private-use'],
diff --git a/src/lib/util.ts b/src/lib/util.ts
index 9608dabe..fba6a9f5 100644
--- a/src/lib/util.ts
+++ b/src/lib/util.ts
@@ -138,7 +138,7 @@ export function compareTagId(a: TagId, b: TagId): number {
return compareString(getTagCategory(a) ?? '', getTagCategory(b) ?? '') || compareString(a, b);
}
export function isDerivedTag(id: TagId): boolean {
- return id.includes(':');
+ return id.includes(':') && !id.startsWith('dataset:');
}
export function getColorMode(numberOfChannels: number) {
diff --git a/src/pages/add-dataset.tsx b/src/pages/add-dataset.tsx
new file mode 100644
index 00000000..16b719e8
--- /dev/null
+++ b/src/pages/add-dataset.tsx
@@ -0,0 +1,280 @@
+import { useRouter } from 'next/router';
+import React, { useEffect, useMemo, useState } from 'react';
+import { TextLink } from '../elements/components/link';
+import { HeadCommon } from '../elements/head-common';
+import { PageContainer } from '../elements/page';
+import { useArchitectures } from '../lib/hooks/use-architectures';
+import { useDatasets } from '../lib/hooks/use-datasets';
+import { useModels } from '../lib/hooks/use-models';
+import { useTags } from '../lib/hooks/use-tags';
+import { useWebApi } from '../lib/hooks/use-web-api';
+import { withImpliedTags } from '../lib/implied-tags';
+import { ParseResult, parseDiscordMessage } from '../lib/parse-discord-message';
+import { Dataset, DatasetId, Tag, TagId } from '../lib/schema';
+import { canonicalizeDatasetId } from '../lib/schema-util';
+import { IS_DEPLOYED } from '../lib/site-data';
+
+function guessDatasetTags(name: string, description: string, tagData: ReadonlyMap): TagId[] {
+ const tags = new Set();
+
+ if (/\b(?:realistic|photo|photography|real)\b/i.test(`${name} ${description}`)) {
+ tags.add('dataset:realistic' as TagId);
+ }
+ if (/\b(?:anime|cartoon)\b/i.test(`${name} ${description}`)) {
+ tags.add('dataset:anime' as TagId);
+ }
+ if (/\b(?:manga)\b/i.test(`${name} ${description}`)) {
+ tags.add('dataset:manga' as TagId);
+ }
+ if (/\b(?:game[- ]textures?|textures?)\b/i.test(`${name} ${description}`)) {
+ tags.add('dataset:game-textures' as TagId);
+ }
+
+ return withImpliedTags(tags, tagData);
+}
+
+const EMPTY_PARSE_RESULT: ParseResult = { failed: [], parsed: {} };
+
+const discordMessageTemplate = `
+**Name:** DatasetNameThatIsCreative
+**License:** GNU GPL3 for example
+**Link:**
+**Description:** Your Description
+`.trim();
+
+function PageContent() {
+ const { datasetData } = useDatasets();
+ const { modelData } = useModels();
+ const { archData } = useArchitectures();
+ const { tagData } = useTags();
+ const router = useRouter();
+ const { webApi, editMode } = useWebApi(IS_DEPLOYED);
+
+ const [processing, setProcessing] = useState(false);
+ const [name, setName] = useState('Unknown');
+ const [partialId, setPartialId] = useState();
+ const [url, setUrl] = useState('');
+ const [description, setDescription] = useState('');
+
+ let fullId = canonicalizeDatasetId(partialId ?? name);
+ const partialIdFromFull = fullId;
+
+ const [parseMessageTemplate, setParseMessageTemplate] = useState(false);
+ const [messageTemplate, setMessageTemplate] = useState('');
+ const parsedMessage = useMemo((): ParseResult => {
+ if (!parseMessageTemplate) {
+ return EMPTY_PARSE_RESULT;
+ }
+ return parseDiscordMessage(messageTemplate, modelData, archData);
+ }, [parseMessageTemplate, messageTemplate, modelData, archData]);
+
+ useEffect(() => {
+ if (parsedMessage.parsed.name) {
+ setName(parsedMessage.parsed.name.replace(/[\s_\-]+/g, ' '));
+ setPartialId(undefined);
+ }
+ }, [parsedMessage.parsed.name]);
+
+ useEffect(() => {
+ if (parsedMessage.parsed.link) {
+ setUrl(parsedMessage.parsed.link);
+ }
+ }, [parsedMessage.parsed.link]);
+
+ useEffect(() => {
+ if (parsedMessage.parsed.description) {
+ setDescription(parsedMessage.parsed.description);
+ }
+ }, [parsedMessage.parsed.description]);
+
+ if (!editMode) return null;
+
+ const addDataset = async () => {
+ if (datasetData.has(fullId)) {
+ alert(`Dataset ${fullId} already exists`);
+ return;
+ }
+
+ const combinedDescription = [
+ parsedMessage.parsed.purpose ? `Purpose: ${parsedMessage.parsed.purpose}` : '',
+ description || parsedMessage.parsed.description || '',
+ ]
+ .join('\n\n')
+ .trim();
+
+ const dataset: Dataset = {
+ name,
+ author: [],
+ license: parsedMessage.parsed.license ?? null,
+ tags: guessDatasetTags(name, combinedDescription, tagData),
+ description: combinedDescription,
+ date: new Date().toISOString().split('T')[0],
+ url,
+ images: [],
+ };
+
+ setProcessing(true);
+
+ if (IS_DEPLOYED) {
+ sessionStorage.setItem('dummy-datasetId', fullId);
+ sessionStorage.setItem('dummy-dataset', JSON.stringify(dataset));
+ fullId = 'OMDB_ADDDATASET_DUMMY' as DatasetId;
+ }
+
+ await webApi.datasets.update([[fullId, dataset]]);
+
+ // fetch before navigating to ensure the dataset page is available
+ const page = `/datasets/${fullId}`;
+ await fetch(page);
+ await router.push(`/datasets/${fullId}`);
+ };
+
+ let inputError;
+ if (name.trim() === '') {
+ inputError = 'Name cannot be empty';
+ } else if (fullId.trim() === '') {
+ inputError = 'ID cannot be empty';
+ }
+
+ const canAddDataset = !inputError && !processing;
+
+ return (
+ <>
+ Add Dataset
+
+
{
+ setParseMessageTemplate(e.target.checked);
+ }}
+ />
+
Parse Discord Message Template?
+
+ {parseMessageTemplate && (
+
+
+ How to use: Paste a message from the{' '}
+
+ model-releases
+ {' '}
+ channel (or any message following the message template).
+