Spaces:
Runtime error
Runtime error
File size: 7,787 Bytes
cd8bd0a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 | import { csvToJsonlInputSchema, type CsvToJsonlInput } from "./schemas";
/**
* RFC 4180 minimal CSV parser.
*
* Supports: quoted fields, escaped double-quotes (""), CRLF/LF line endings,
* inline newlines inside quoted fields.
* Does NOT strip BOM β callers should strip before invoking if needed.
*/
function parseCsvRow(line: string): string[] {
const out: string[] = [];
let buf = "";
let inQuotes = false;
for (let i = 0; i < line.length; i++) {
const ch = line[i];
if (inQuotes) {
// Escaped quote: "" inside a quoted field
if (ch === '"' && line[i + 1] === '"') {
buf += '"';
i++;
continue;
}
if (ch === '"') {
inQuotes = false;
continue;
}
buf += ch;
} else {
if (ch === '"') {
inQuotes = true;
continue;
}
if (ch === ",") {
out.push(buf);
buf = "";
continue;
}
buf += ch;
}
}
out.push(buf);
return out;
}
/**
* Split CSV text into logical lines, preserving inline newlines inside quoted fields.
* Handles CRLF and LF; does not collapse multiple empty lines.
*/
function splitLines(s: string): string[] {
const out: string[] = [];
let buf = "";
let inQuotes = false;
for (let i = 0; i < s.length; i++) {
const ch = s[i];
if (ch === '"') {
if (inQuotes && s[i + 1] === '"') {
// Escaped quote inside quoted field: add both and advance
buf += '""';
i++;
continue;
}
inQuotes = !inQuotes;
buf += ch;
continue;
}
if ((ch === "\n" || ch === "\r") && !inQuotes) {
if (ch === "\r" && s[i + 1] === "\n") i++;
if (buf.length > 0) out.push(buf);
buf = "";
continue;
}
buf += ch;
}
if (buf.length > 0) out.push(buf);
return out;
}
// ββ Prototype-pollution guard βββββββββββββββββββββββββββββββββββββββββββββββββ
/**
* Keys that are forbidden as object property names (CWE-915 β prototype pollution).
* A legitimate CSV mapping path cannot contain these: the Zod schema
* `wizardCsvMappingSchema` validates paths come from a controlled UI dropdown;
* we add this explicit denylist as defence-in-depth.
*/
const FORBIDDEN_KEYS = new Set(["__proto__", "constructor", "prototype"]);
/**
* Write `value` to `obj[key]` using Object.defineProperty instead of bracket
* assignment. Object.defineProperty bypasses the prototype-chain write path,
* preventing prototype-pollution attacks (semgrep rule
* javascript.lang.security.audit.prototype-pollution-assignment).
*/
function safePropSet(obj: object, key: string | number, value: unknown): void {
Object.defineProperty(obj, key, {
value,
writable: true,
enumerable: true,
configurable: true,
});
}
/**
* Set a nested value in `target` using a dot/bracket path such as:
* "custom_id"
* "body.max_tokens"
* "body.messages[0].content"
*
* Silently ignores malformed path segments or forbidden key names.
*/
function setByPath(target: Record<string, unknown>, path: string, value: unknown): void {
const tokens: Array<string | number> = [];
for (const part of path.split(".")) {
const match = part.match(/^([^[]+)((?:\[\d+\])*)$/);
if (!match) continue;
const key = match[1];
if (FORBIDDEN_KEYS.has(key)) return; // reject forbidden key
tokens.push(key);
for (const idx of match[2].match(/\d+/g) ?? []) {
tokens.push(Number(idx));
}
}
if (tokens.length === 0) return;
let cur: Record<string, unknown> = target;
for (let i = 0; i < tokens.length - 1; i++) {
const k = tokens[i];
const next = tokens[i + 1];
if (!Object.prototype.hasOwnProperty.call(cur, k) || cur[k] == null) {
const child: unknown = typeof next === "number" ? [] : Object.create(null);
safePropSet(cur, k, child);
}
const nextCur = Object.prototype.hasOwnProperty.call(cur, k) ? cur[k] : undefined;
if (nextCur == null || typeof nextCur !== "object") return; // bail if tree navigation failed
cur = nextCur as Record<string, unknown>;
}
const lastKey = tokens.at(-1)!;
if (typeof lastKey === "string" && FORBIDDEN_KEYS.has(lastKey)) return;
safePropSet(cur, lastKey, value);
}
// ββ Public API ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
/**
* Convert a CSV string + column mapping into a JSONL batch request file.
*
* - Validates input via `csvToJsonlInputSchema` (Zod).
* - Parses CSV per RFC 4180 (quoted fields, escaped quotes, CRLF/LF).
* - Auto-fills `role: "user"` when mapping includes a content path but no role path.
* - Coerces max_tokens / temperature to numbers.
* - Skips rows missing `custom_id` or content; records them in `errors`.
*/
export function csvToJsonl(rawInput: CsvToJsonlInput): {
jsonl: string;
rowsParsed: number;
rowsSkipped: number;
errors: Array<{ row: number; reason: string }>;
} {
const input = csvToJsonlInputSchema.parse(rawInput);
// Strip UTF-8 BOM if present so the first header cell parses cleanly
const normalizedCsv = input.csv.replace(/^ο»Ώ/, "");
const lines = splitLines(normalizedCsv);
if (lines.length < 2) {
return {
jsonl: "",
rowsParsed: 0,
rowsSkipped: 0,
errors: [{ row: 0, reason: "CSV has no data rows" }],
};
}
const headers = parseCsvRow(lines[0]);
const out: string[] = [];
const errors: Array<{ row: number; reason: string }> = [];
let skipped = 0;
for (let r = 1; r < lines.length; r++) {
const cells = parseCsvRow(lines[r]);
// Skip blank rows
if (cells.length === 0 || cells.every((c) => c.trim() === "")) {
skipped++;
continue;
}
const request: Record<string, unknown> = {
method: input.defaults.method,
url: input.defaults.url,
body: Object.assign(Object.create(null) as Record<string, unknown>, {
model: input.defaults.model,
}),
};
let hasContent = false;
let hasCustomId = false;
for (let c = 0; c < headers.length; c++) {
const header = headers[c];
const path = input.mapping[header];
if (!path) continue;
const raw = cells[c] ?? "";
if (path === "custom_id") {
if (raw.trim().length === 0) {
errors.push({ row: r + 1, reason: "custom_id is empty" });
continue;
}
request.custom_id = raw;
hasCustomId = true;
} else if (path.startsWith("body.messages[") && path.endsWith(".content")) {
setByPath(request, path, raw);
// Auto-fill role: "user" unless the mapping already maps a role for this slot
const rolePath = path.replace(".content", ".role");
if (!Object.values(input.mapping).includes(rolePath)) {
setByPath(request, rolePath, "user");
}
hasContent = true;
} else if (path === "body.input" || path === "body.prompt") {
setByPath(request, path, raw);
hasContent = true;
} else if (path.startsWith("body.")) {
// Coerce numeric fields (max_tokens, temperature, top_p, etc.)
const num = Number(raw);
setByPath(request, path, Number.isFinite(num) && raw.trim() !== "" ? num : raw);
}
}
if (!hasContent || !hasCustomId) {
skipped++;
errors.push({ row: r + 1, reason: "missing content or custom_id" });
continue;
}
out.push(JSON.stringify(request));
}
return {
jsonl: out.join("\n") + (out.length > 0 ? "\n" : ""),
rowsParsed: out.length,
rowsSkipped: skipped,
errors,
};
}
|