| const HF = "https://huggingface.co/"; |
| const AUDIO_SPACE = HF + "spaces/danielrosehill/Audio-Dataset-Navigator"; |
| const SENTENCE_SPACE = HF + "spaces/danielrosehill/Sentence-Dataset-Navigator"; |
| const esc = (s) => String(s ?? "").replace(/[&<>"]/g, (c) => ({ "&": "&", "<": "<", ">": ">", '"': """ }[c])); |
| const ds = (id) => HF + "datasets/" + id; |
| const link = (url, text) => `<a href="${esc(url)}" target="_blank" rel="noopener">${esc(text)}</a>`; |
| const size = (b) => (b >= 1e9 ? (b / 1e9).toFixed(1) + " GB" : b >= 1e6 ? Math.round(b / 1e6) + " MB" : Math.max(1, Math.round(b / 1e3)) + " KB"); |
| const hours = (h) => (h >= 1 ? h.toFixed(1) + " hours" : Math.max(1, Math.round(h * 60)) + " min"); |
| const short = (id) => id.split("/")[1].replace(/-/g, " "); |
|
|
| function who(d, kind) { |
| const org = d.owner === "Hebrish"; |
| const text = org |
| ? `Hebrish org · Daniel's own ${kind}` |
| : `danielrosehill · Daniel's own ${kind}`; |
| return `<div class="who"><img src="${org ? "hebrish.webp" : "daniel.webp"}" alt=""><span>${text}</span></div>`; |
| } |
|
|
| function card(d, kind, unit) { |
| const meta = []; |
| if (d.rows != null) meta.push(`${d.rows.toLocaleString("en-US")} ${unit}`); |
| if (d.hours != null) meta.push(hours(d.hours)); |
| meta.push(size(d.bytes)); |
| if ("hours" in d) meta.push("Updated " + d.modified.slice(0, 10)); |
| if (d.gated) meta.push('<span class="badge">Access request</span>'); |
| const pairs = (d.banks || []).map((b) => link(ds(b), "Text: " + short(b))); |
| if (d.pairs_with) pairs.push(link(ds(d.pairs_with), "Audio: " + short(d.pairs_with))); |
| return `<article class="card${d.group === "Master" ? " master" : ""}">${who(d, kind)} |
| <h3>${link(ds(d.id), d.name)}</h3><p>${esc(d.summary)}</p> |
| <div class="meta">${meta.join("<span>·</span>")}</div> |
| ${pairs.length ? `<div class="meta">${pairs.join("")}</div>` : ""}</article>`; |
| } |
|
|
| async function navigator({ kind, unit }) { |
| const data = await (await fetch("data.json")).json(); |
| const list = data.datasets; |
| const q = document.getElementById("q"), owner = document.getElementById("owner"), sort = document.getElementById("sort"), out = document.getElementById("list"); |
| document.getElementById("built").textContent = data.built; |
| const total = list.reduce((n, d) => n + (d.rows || 0), 0); |
| const master = list.find((d) => d.group === "Master"); |
| const stat = (v, l) => `<span><b>${v}</b> ${l}</span>`; |
| document.getElementById("stats").innerHTML = |
| stat(list.length, "datasets") + |
| ["danielrosehill", "Hebrish"].map((o) => [o, list.filter((d) => d.owner === o).length]).filter(([, n]) => n).map(([o, n]) => stat(n, "on " + o)).join("") + |
| stat((master ? master.rows : total).toLocaleString("en-US"), master ? "takes in the master" : "sentences") + |
| (master && master.hours != null ? stat(master.hours.toFixed(1), "hours in the master") : ""); |
| if (new Set(list.map((d) => d.owner)).size < 2) owner.hidden = true; |
| const draw = () => { |
| const term = q.value.trim().toLowerCase(); |
| const hit = list.filter((d) => (!owner.value || d.owner === owner.value) && (d.id + " " + d.summary + " " + d.group).toLowerCase().includes(term)); |
| const groups = [...new Set(hit.map((d) => d.group))]; |
| const order = sort && { takes: (a, b) => (b.rows ?? -1) - (a.rows ?? -1), updated: (a, b) => b.modified.localeCompare(a.modified) }[sort.value]; |
| if (order && hit.length) { |
| out.innerHTML = `<div class="grid">${[...hit].sort(order).map((d) => card(d, kind, unit)).join("")}</div>`; |
| return; |
| } |
| out.innerHTML = groups.map((g) => `<h2>${esc(g)}</h2><div class="grid">${hit.filter((d) => d.group === g).map((d) => card(d, kind, unit)).join("")}</div>`).join("") |
| || '<p class="empty">No datasets match.</p>'; |
| }; |
| q.addEventListener("input", draw); |
| owner.addEventListener("change", draw); |
| if (sort) sort.addEventListener("change", draw); |
| draw(); |
| } |
|
|
| async function thirdParty() { |
| const data = await (await fetch("data.json")).json(); |
| document.getElementById("built").textContent = data.built; |
| document.getElementById("collection").href = HF + "collections/" + data.collections.sentences_third_party; |
| document.getElementById("list").innerHTML = data.third_party.map((d) => `<article class="card"> |
| <div class="who"><span>${esc(d.id.split("/")[0])}</span></div><h3>${link(ds(d.id), short(d.id))}</h3><p>${esc(d.note)}</p></article>`).join(""); |
| } |
|
|
