| 1 | // text-content extraction processors. `text-contents` fills |
| 2 | // `media_files.contents` for plain text files; `highlight-code` fills it |
| 3 | // with pre-rendered syntax highlighting html for source code. |
| 4 | const procLoadTextContents: Processor = { |
| 5 | name: "text-contents", |
| 6 | version: 2, |
| 7 | title: "load text content", |
| 8 | include: rules.extsReadContents, |
| 9 | cores: 1, |
| 10 | async run({ path, mediaFile, stat }) { |
| 11 | if (stat.size > 1_000_000) return; |
| 12 | const text = await path.read("utf-8"); |
| 13 | mediaFile.setContents(text); |
| 14 | }, |
| 15 | }; |
| 16 | |
| 17 | const procHighlightCode: Processor = { |
| 18 | name: "highlight-code", |
| 19 | version: 2, |
| 20 | title: "highlight source code", |
| 21 | include: new Set(rules.extsCode.keys()), |
| 22 | cores: 1, |
| 23 | async run({ path, mediaFile, stat }) { |
| 24 | const language = UNWRAP( |
| 25 | rules.extsCode.get(path.ext.toLowerCase()), |
| 26 | ); |
| 27 | // An issue is that .ts is an overloaded extension, shared between |
| 28 | // 'transport stream' and 'typescript'. |
| 29 | // |
| 30 | // Filter used here is: |
| 31 | // - more than 1mb |
| 32 | // - invalid UTF-8 |
| 33 | if (stat.size > 1_000_000) return; |
| 34 | let code; |
| 35 | const buf = await path.read(); |
| 36 | try { |
| 37 | code = new TextDecoder("utf-8", { fatal: true }).decode(buf); |
| 38 | } catch (error) { |
| 39 | mediaFile.setContents(""); |
| 40 | return; |
| 41 | } |
| 42 | const content = await highlight.highlightCode(code, language); |
| 43 | mediaFile.setContents(content); |
| 44 | }, |
| 45 | }; |
| 46 | |
| 47 | export const processors: Processor[] = [ |
| 48 | procLoadTextContents, |
| 49 | procHighlightCode, |
| 50 | ]; |
| 51 | |
| 52 | import { UNWRAP } from "@clo/lib/assert"; |
| 53 | |
| 54 | import * as highlight from "#src/file-viewer/highlight.ts"; |
| 55 | import * as rules from "#src/file-viewer/rules.ts"; |
| 56 | import type { Processor } from "../registry.ts"; |