Spaces:
Running
Running
Update server.js
Browse files
server.js
CHANGED
|
@@ -2,33 +2,26 @@ const express = require('express');
|
|
| 2 |
const app = express();
|
| 3 |
const PORT = 7860;
|
| 4 |
|
| 5 |
-
// Environment Variables from HF Space Settings
|
| 6 |
const HF_TOKEN = process.env.HF_TOKEN;
|
| 7 |
const REPO_ID = process.env.SIMBA_DATA;
|
| 8 |
|
| 9 |
-
// In-memory cache
|
| 10 |
let CACHED_DATA = null;
|
| 11 |
|
| 12 |
-
// Serve static files from 'public' folder
|
| 13 |
app.use(express.static('public'));
|
| 14 |
|
| 15 |
-
// --- Helper
|
| 16 |
async function fetchHFFile(filename) {
|
| 17 |
const url = `https://huggingface.co/datasets/${REPO_ID}/resolve/main/${filename}`;
|
| 18 |
console.log(`Downloading: ${filename}...`);
|
| 19 |
const res = await fetch(url, { headers: { "Authorization": `Bearer ${HF_TOKEN}` } });
|
| 20 |
-
if (!res.ok)
|
| 21 |
-
console.warn(`Failed to fetch ${filename}: ${res.statusText}`);
|
| 22 |
-
return [];
|
| 23 |
-
}
|
| 24 |
const text = await res.text();
|
| 25 |
-
// Parse JSONL
|
| 26 |
return text.split('\n').filter(line => line.trim()).map(line => {
|
| 27 |
try { return JSON.parse(line); } catch(e) { return null; }
|
| 28 |
}).filter(x => x);
|
| 29 |
}
|
| 30 |
|
| 31 |
-
// ---
|
| 32 |
async function listHFFiles(folder) {
|
| 33 |
const url = `https://huggingface.co/api/datasets/${REPO_ID}/tree/main/${folder}`;
|
| 34 |
const res = await fetch(url, { headers: { "Authorization": `Bearer ${HF_TOKEN}` } });
|
|
@@ -37,52 +30,53 @@ async function listHFFiles(folder) {
|
|
| 37 |
return files.filter(f => f.path.endsWith('.jsonl')).map(f => f.path);
|
| 38 |
}
|
| 39 |
|
| 40 |
-
// ---
|
| 41 |
async function loadAndProcessData() {
|
| 42 |
console.log("Starting Data Sync...");
|
| 43 |
|
| 44 |
-
// 1.
|
| 45 |
const metaList = await fetchHFFile("final_results/metadata.jsonl");
|
| 46 |
const langMap = {};
|
|
|
|
| 47 |
metaList.forEach(m => { langMap[m.lang_code_key] = m; });
|
| 48 |
|
| 49 |
-
// 2. Load
|
| 50 |
const ttsData = await fetchHFFile("final_results/tts_results.jsonl");
|
| 51 |
const slidData = await fetchHFFile("final_results/slid_results.jsonl");
|
| 52 |
|
| 53 |
-
// 3. Load ASR
|
| 54 |
const files = await listHFFiles("final_results");
|
| 55 |
let asrRecords = [];
|
| 56 |
for (const file of files) {
|
| 57 |
-
// Skip metadata/tts/slid if they appear in list
|
| 58 |
if (file.includes('metadata') || file.includes('tts_') || file.includes('slid_')) continue;
|
| 59 |
const records = await fetchHFFile(file);
|
| 60 |
asrRecords.push(...records);
|
| 61 |
}
|
| 62 |
|
| 63 |
-
// 4. Process ASR
|
| 64 |
const families = new Set();
|
| 65 |
const models = new Set();
|
| 66 |
-
const familyData = {};
|
| 67 |
|
| 68 |
asrRecords.forEach(r => {
|
| 69 |
if (r.task !== 'asr') return;
|
| 70 |
|
| 71 |
-
//
|
| 72 |
-
const meta = langMap[r.lang_code] || { name: r.lang_code, family: 'Unknown' };
|
| 73 |
const fam = meta.family;
|
| 74 |
const mod = r.model_name;
|
| 75 |
|
| 76 |
families.add(fam);
|
| 77 |
models.add(mod);
|
| 78 |
|
| 79 |
-
// Init structure
|
| 80 |
if (!familyData[fam]) familyData[fam] = {};
|
| 81 |
if (!familyData[fam][mod]) familyData[fam][mod] = {};
|
| 82 |
|
|
|
|
| 83 |
familyData[fam][mod][meta.name] = {
|
| 84 |
wer: (r.datasets_avg_wer || 0) * 100,
|
| 85 |
-
cer: (r.datasets_avg_cer || 0) * 100
|
|
|
|
| 86 |
};
|
| 87 |
});
|
| 88 |
|
|
@@ -91,18 +85,28 @@ async function loadAndProcessData() {
|
|
| 91 |
Array.from(families).sort().forEach(fam => {
|
| 92 |
const modData = familyData[fam];
|
| 93 |
|
| 94 |
-
//
|
| 95 |
-
const
|
| 96 |
-
Object.values(modData).forEach(m =>
|
| 97 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
|
| 99 |
-
// Build Rows
|
| 100 |
const rows = Object.keys(modData).map(modName => {
|
| 101 |
const scores = modData[modName];
|
| 102 |
let sW=0, sC=0, n=0;
|
| 103 |
|
| 104 |
-
|
| 105 |
-
if(scores[l]) { sW += scores[l].wer; sC += scores[l].cer; n++; }
|
| 106 |
});
|
| 107 |
|
| 108 |
const row = {
|
|
@@ -111,19 +115,17 @@ async function loadAndProcessData() {
|
|
| 111 |
Avg_CER: n ? sC/n : 0
|
| 112 |
};
|
| 113 |
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
row[`
|
| 118 |
-
row[`CER_${l}`] = scores[l].cer;
|
| 119 |
}
|
| 120 |
});
|
| 121 |
return row;
|
| 122 |
});
|
| 123 |
|
| 124 |
-
// Sort by WER
|
| 125 |
rows.sort((a,b) => a.Avg_WER - b.Avg_WER);
|
| 126 |
-
asrByFamily[fam] = { data: rows, languages:
|
| 127 |
});
|
| 128 |
|
| 129 |
// 6. Format ASR By Model
|
|
@@ -132,9 +134,10 @@ async function loadAndProcessData() {
|
|
| 132 |
const rows = [];
|
| 133 |
asrRecords.forEach(r => {
|
| 134 |
if(r.model_name === mod && r.task === 'asr') {
|
| 135 |
-
const meta = langMap[r.lang_code] || { name: r.lang_code, family: 'Unknown' };
|
| 136 |
rows.push({
|
| 137 |
Language: meta.name,
|
|
|
|
| 138 |
Family: meta.family,
|
| 139 |
WER: (r.datasets_avg_wer || 0) * 100,
|
| 140 |
CER: (r.datasets_avg_cer || 0) * 100
|
|
@@ -151,6 +154,9 @@ async function loadAndProcessData() {
|
|
| 151 |
ttsData.forEach(r => {
|
| 152 |
ttsModels.add(r.model);
|
| 153 |
if(!ttsGrouped[r.model]) ttsGrouped[r.model] = [];
|
|
|
|
|
|
|
|
|
|
| 154 |
ttsGrouped[r.model].push(r);
|
| 155 |
});
|
| 156 |
|
|
@@ -164,10 +170,9 @@ async function loadAndProcessData() {
|
|
| 164 |
tts: ttsGrouped,
|
| 165 |
slid: slidData
|
| 166 |
};
|
| 167 |
-
console.log("Data
|
| 168 |
}
|
| 169 |
|
| 170 |
-
// API Route
|
| 171 |
app.get('/api/data', async (req, res) => {
|
| 172 |
if (!CACHED_DATA) {
|
| 173 |
try { await loadAndProcessData(); }
|
|
@@ -176,8 +181,7 @@ app.get('/api/data', async (req, res) => {
|
|
| 176 |
res.json(CACHED_DATA);
|
| 177 |
});
|
| 178 |
|
| 179 |
-
// Start
|
| 180 |
app.listen(PORT, '0.0.0.0', () => {
|
| 181 |
-
console.log(`Server
|
| 182 |
-
loadAndProcessData();
|
| 183 |
});
|
|
|
|
| 2 |
const app = express();
|
| 3 |
const PORT = 7860;
|
| 4 |
|
|
|
|
| 5 |
const HF_TOKEN = process.env.HF_TOKEN;
|
| 6 |
const REPO_ID = process.env.SIMBA_DATA;
|
| 7 |
|
|
|
|
| 8 |
let CACHED_DATA = null;
|
| 9 |
|
|
|
|
| 10 |
app.use(express.static('public'));
|
| 11 |
|
| 12 |
+
// --- Fetch Helper ---
|
| 13 |
async function fetchHFFile(filename) {
|
| 14 |
const url = `https://huggingface.co/datasets/${REPO_ID}/resolve/main/${filename}`;
|
| 15 |
console.log(`Downloading: ${filename}...`);
|
| 16 |
const res = await fetch(url, { headers: { "Authorization": `Bearer ${HF_TOKEN}` } });
|
| 17 |
+
if (!res.ok) return [];
|
|
|
|
|
|
|
|
|
|
| 18 |
const text = await res.text();
|
|
|
|
| 19 |
return text.split('\n').filter(line => line.trim()).map(line => {
|
| 20 |
try { return JSON.parse(line); } catch(e) { return null; }
|
| 21 |
}).filter(x => x);
|
| 22 |
}
|
| 23 |
|
| 24 |
+
// --- List Files Helper ---
|
| 25 |
async function listHFFiles(folder) {
|
| 26 |
const url = `https://huggingface.co/api/datasets/${REPO_ID}/tree/main/${folder}`;
|
| 27 |
const res = await fetch(url, { headers: { "Authorization": `Bearer ${HF_TOKEN}` } });
|
|
|
|
| 30 |
return files.filter(f => f.path.endsWith('.jsonl')).map(f => f.path);
|
| 31 |
}
|
| 32 |
|
| 33 |
+
// --- Processor ---
|
| 34 |
async function loadAndProcessData() {
|
| 35 |
console.log("Starting Data Sync...");
|
| 36 |
|
| 37 |
+
// 1. Metadata
|
| 38 |
const metaList = await fetchHFFile("final_results/metadata.jsonl");
|
| 39 |
const langMap = {};
|
| 40 |
+
// Map code key to full metadata object
|
| 41 |
metaList.forEach(m => { langMap[m.lang_code_key] = m; });
|
| 42 |
|
| 43 |
+
// 2. Load Others
|
| 44 |
const ttsData = await fetchHFFile("final_results/tts_results.jsonl");
|
| 45 |
const slidData = await fetchHFFile("final_results/slid_results.jsonl");
|
| 46 |
|
| 47 |
+
// 3. Load ASR
|
| 48 |
const files = await listHFFiles("final_results");
|
| 49 |
let asrRecords = [];
|
| 50 |
for (const file of files) {
|
|
|
|
| 51 |
if (file.includes('metadata') || file.includes('tts_') || file.includes('slid_')) continue;
|
| 52 |
const records = await fetchHFFile(file);
|
| 53 |
asrRecords.push(...records);
|
| 54 |
}
|
| 55 |
|
| 56 |
+
// 4. Process ASR
|
| 57 |
const families = new Set();
|
| 58 |
const models = new Set();
|
| 59 |
+
const familyData = {};
|
| 60 |
|
| 61 |
asrRecords.forEach(r => {
|
| 62 |
if (r.task !== 'asr') return;
|
| 63 |
|
| 64 |
+
// Use metadata to get Family and ISO
|
| 65 |
+
const meta = langMap[r.lang_code] || { name: r.lang_code, family: 'Unknown', lang_code_key: r.lang_code };
|
| 66 |
const fam = meta.family;
|
| 67 |
const mod = r.model_name;
|
| 68 |
|
| 69 |
families.add(fam);
|
| 70 |
models.add(mod);
|
| 71 |
|
|
|
|
| 72 |
if (!familyData[fam]) familyData[fam] = {};
|
| 73 |
if (!familyData[fam][mod]) familyData[fam][mod] = {};
|
| 74 |
|
| 75 |
+
// Store by Language Name
|
| 76 |
familyData[fam][mod][meta.name] = {
|
| 77 |
wer: (r.datasets_avg_wer || 0) * 100,
|
| 78 |
+
cer: (r.datasets_avg_cer || 0) * 100,
|
| 79 |
+
iso: meta.lang_code_key // Store ISO here
|
| 80 |
};
|
| 81 |
});
|
| 82 |
|
|
|
|
| 85 |
Array.from(families).sort().forEach(fam => {
|
| 86 |
const modData = familyData[fam];
|
| 87 |
|
| 88 |
+
// Collect Languages with their ISO codes
|
| 89 |
+
const langMapForFamily = new Map();
|
| 90 |
+
Object.values(modData).forEach(m => {
|
| 91 |
+
Object.keys(m).forEach(langName => {
|
| 92 |
+
// Get ISO from the stored data object
|
| 93 |
+
const iso = m[langName].iso;
|
| 94 |
+
langMapForFamily.set(langName, iso);
|
| 95 |
+
});
|
| 96 |
+
});
|
| 97 |
+
|
| 98 |
+
// Create array of objects: { name: "Amharic", iso: "amh" }
|
| 99 |
+
const languages = Array.from(langMapForFamily.keys()).sort().map(name => ({
|
| 100 |
+
name: name,
|
| 101 |
+
iso: langMapForFamily.get(name)
|
| 102 |
+
}));
|
| 103 |
|
|
|
|
| 104 |
const rows = Object.keys(modData).map(modName => {
|
| 105 |
const scores = modData[modName];
|
| 106 |
let sW=0, sC=0, n=0;
|
| 107 |
|
| 108 |
+
languages.forEach(l => {
|
| 109 |
+
if(scores[l.name]) { sW += scores[l.name].wer; sC += scores[l.name].cer; n++; }
|
| 110 |
});
|
| 111 |
|
| 112 |
const row = {
|
|
|
|
| 115 |
Avg_CER: n ? sC/n : 0
|
| 116 |
};
|
| 117 |
|
| 118 |
+
languages.forEach(l => {
|
| 119 |
+
if(scores[l.name]) {
|
| 120 |
+
row[`WER_${l.name}`] = scores[l.name].wer;
|
| 121 |
+
row[`CER_${l.name}`] = scores[l.name].cer;
|
|
|
|
| 122 |
}
|
| 123 |
});
|
| 124 |
return row;
|
| 125 |
});
|
| 126 |
|
|
|
|
| 127 |
rows.sort((a,b) => a.Avg_WER - b.Avg_WER);
|
| 128 |
+
asrByFamily[fam] = { data: rows, languages: languages };
|
| 129 |
});
|
| 130 |
|
| 131 |
// 6. Format ASR By Model
|
|
|
|
| 134 |
const rows = [];
|
| 135 |
asrRecords.forEach(r => {
|
| 136 |
if(r.model_name === mod && r.task === 'asr') {
|
| 137 |
+
const meta = langMap[r.lang_code] || { name: r.lang_code, family: 'Unknown', lang_code_key: r.lang_code };
|
| 138 |
rows.push({
|
| 139 |
Language: meta.name,
|
| 140 |
+
ISO: meta.lang_code_key, // Include ISO
|
| 141 |
Family: meta.family,
|
| 142 |
WER: (r.datasets_avg_wer || 0) * 100,
|
| 143 |
CER: (r.datasets_avg_cer || 0) * 100
|
|
|
|
| 154 |
ttsData.forEach(r => {
|
| 155 |
ttsModels.add(r.model);
|
| 156 |
if(!ttsGrouped[r.model]) ttsGrouped[r.model] = [];
|
| 157 |
+
// Add ISO if available in metadata, else guess or use code
|
| 158 |
+
const meta = langMap[r.lang_code] || { lang_code_key: r.lang_code };
|
| 159 |
+
r.iso = meta.lang_code_key;
|
| 160 |
ttsGrouped[r.model].push(r);
|
| 161 |
});
|
| 162 |
|
|
|
|
| 170 |
tts: ttsGrouped,
|
| 171 |
slid: slidData
|
| 172 |
};
|
| 173 |
+
console.log("Data Ready.");
|
| 174 |
}
|
| 175 |
|
|
|
|
| 176 |
app.get('/api/data', async (req, res) => {
|
| 177 |
if (!CACHED_DATA) {
|
| 178 |
try { await loadAndProcessData(); }
|
|
|
|
| 181 |
res.json(CACHED_DATA);
|
| 182 |
});
|
| 183 |
|
|
|
|
| 184 |
app.listen(PORT, '0.0.0.0', () => {
|
| 185 |
+
console.log(`Server running on ${PORT}`);
|
| 186 |
+
loadAndProcessData();
|
| 187 |
});
|