@@ -5,8 +5,34 @@ export function normalizeOptionalString(value: unknown): string | undefined {
|
5 | 5 | return typeof value === "string" && value.trim().length > 0 ? value : undefined; |
6 | 6 | } |
7 | 7 | |
| 8 | +// Transcript readers repeatedly extract a fixed set of metadata fields from |
| 9 | +// oversized JSONL prefixes. Keep the compiled regexes process-local instead of |
| 10 | +// rebuilding them for every field on every oversized record. |
| 11 | +const TRANSCRIPT_FIELD_REGEX_CACHE = new Map< |
| 12 | +string, |
| 13 | +{ stringRe: RegExp; nullRe: RegExp; numberRe: RegExp } |
| 14 | +>(); |
| 15 | + |
| 16 | +function getTranscriptFieldRegexes(field: string): { |
| 17 | +stringRe: RegExp; |
| 18 | +nullRe: RegExp; |
| 19 | +numberRe: RegExp; |
| 20 | +} { |
| 21 | +let cached = TRANSCRIPT_FIELD_REGEX_CACHE.get(field); |
| 22 | +if (!cached) { |
| 23 | +const escapedField = escapeRegExp(field); |
| 24 | +cached = { |
| 25 | +stringRe: new RegExp(`"${escapedField}"\\s*:\\s*"((?:\\\\.|[^"\\\\])*)"`), |
| 26 | +nullRe: new RegExp(`"${escapedField}"\\s*:\\s*null`), |
| 27 | +numberRe: new RegExp(`"${escapedField}"\\s*:\\s*(-?\\d+(?:\\.\\d+)?(?:[eE][+-]?\\d+)?)`), |
| 28 | +}; |
| 29 | +TRANSCRIPT_FIELD_REGEX_CACHE.set(field, cached); |
| 30 | +} |
| 31 | +return cached; |
| 32 | +} |
| 33 | + |
8 | 34 | export function extractJsonStringFieldPrefix(prefix: string, field: string): string | undefined { |
9 | | -const match = new RegExp(`"${escapeRegExp(field)}"\\s*:\\s*"((?:\\\\.|[^"\\\\])*)"`).exec(prefix); |
| 35 | +const match = getTranscriptFieldRegexes(field).stringRe.exec(prefix); |
10 | 36 | if (!match) { |
11 | 37 | return undefined; |
12 | 38 | } |
@@ -22,16 +48,14 @@ export function extractJsonNullableStringFieldPrefix(
|
22 | 48 | prefix: string, |
23 | 49 | field: string, |
24 | 50 | ): string | null | undefined { |
25 | | -if (new RegExp(`"${escapeRegExp(field)}"\\s*:\\s*null`).test(prefix)) { |
| 51 | +if (getTranscriptFieldRegexes(field).nullRe.test(prefix)) { |
26 | 52 | return null; |
27 | 53 | } |
28 | 54 | return extractJsonStringFieldPrefix(prefix, field); |
29 | 55 | } |
30 | 56 | |
31 | 57 | export function extractJsonNumberFieldPrefix(prefix: string, field: string): number | undefined { |
32 | | -const match = new RegExp( |
33 | | -`"${escapeRegExp(field)}"\\s*:\\s*(-?\\d+(?:\\.\\d+)?(?:[eE][+-]?\\d+)?)`, |
34 | | -).exec(prefix); |
| 58 | +const match = getTranscriptFieldRegexes(field).numberRe.exec(prefix); |
35 | 59 | if (!match) { |
36 | 60 | return undefined; |
37 | 61 | } |
|