mariano/meetus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

merging

Browse Source
This commit is contained in:
Mariano Gabriel
2026-07-05 22:37:46 -03:00
parent b65d6075be
commit 8cfb4c2cbe
5 changed files with 289 additions and 186 deletions
Download Patch File Download Diff File Expand all files Collapse all files

369
ui/doocus-app/server/doocusApi.ts
View File

@@ -1,15 +1,21 @@
/**
* Vite dev middleware exposing the local doocus tree index to the app.
* Vite dev middleware exposing one or more local doocus output trees.
*
* Reads `docs-output/index.json` (produced by process_tree.py) — a replica of
* the whole local drive tree — plus the extracted `<file>.doocus/content.md`
* sidecars, and streams the ORIGINAL files by relative path (resolved against
* the index's recorded root). All reads are local; nothing leaves the machine.
* Each output dir is a "collection" (its own index.json + sidecars + source
* root). The client can turn collections on/off; file paths are composite —
* "<collectionId>/<relpath>" — so a single path identifies both the collection
* and the file within it. All reads are local; nothing leaves the machine.
*
* Routes:
* GET /api/tree the index.json tree
* GET /api/detail?path=<rel> node + extracted content + meta + inline text
* GET /api/original?path=<rel> stream the original file (native viewer / packaging)
* Routes (path params are composite, query is ?path=/?q=):
* GET /api/collections available collections
* GET /api/tree?collections=a,b merged tree of the active collections
* GET /api/detail?path=<c/rel> node + extracted content + meta + text
* GET /api/original?path=<c/rel> stream the original (Range-enabled)
* GET /api/meeting?path=<c/rel> meetus review data for a meeting
* GET /api/meeting/frame?path=&file= a meeting frame
* GET /api/runs/:cid (cid=c/rel) meetus-shaped run (for the embed)
* PUT /api/runs/:cid/review write the meeting's review.json
* GET /api/search?q=&collections=a,b search cached text across active cols
*/
import type { Plugin } from 'vite'
import type { IncomingMessage, ServerResponse } from 'node:http'
@@ -18,7 +24,14 @@ import fsp from 'node:fs/promises'
import path from 'node:path'
interface Options {
outputDir: string
outputDirs: string[]
}
interface Collection {
id: string
name: string
dir: string
indexPath: string
}
interface Node {
@@ -31,16 +44,16 @@ interface Node {
modified: string
url: string | null
out?: string
title?: string
warnings?: string[]
}
interface Index {
root: string
generatedAt: string
counts: Record<string, number>
files: Node[]
counts?: Record<string, number>
}
// Text-native files: readable as-is, so the "original" doubles as its own text.
const TEXT_EXTS = new Set(['md', 'markdown', 'txt', 'json', 'yaml', 'yml', 'csv', 'html', 'htm'])
const MIME: Record<string, string> = {
@@ -53,19 +66,36 @@ const MIME: Record<string, string> = {
'.mp4': 'video/mp4',
}
export function doocusApi(opts: Options): Plugin {
const outputDir = path.resolve(opts.outputDir)
const indexPath = path.join(outputDir, 'index.json')
function buildCollections(dirs: string[]): Collection[] {
const seen = new Map<string, number>()
return dirs.map((d) => {
const dir = path.resolve(d)
let id = path.basename(dir) || 'output'
const n = seen.get(id) ?? 0
seen.set(id, n + 1)
if (n > 0) id = `${id}-${n + 1}`
return { id, name: id, dir, indexPath: path.join(dir, 'index.json') }
})
}
function readIndex(): Index | null {
try {
return JSON.parse(fs.readFileSync(indexPath, 'utf-8'))
} catch {
return null
}
export function doocusApi(opts: Options): Plugin {
const collections = buildCollections(opts.outputDirs)
const byId = new Map(collections.map((c) => [c.id, c]))
function readIndexFor(c: Collection): Index | null {
try { return JSON.parse(fs.readFileSync(c.indexPath, 'utf-8')) } catch { return null }
}
/** Resolve a node's original file, rejecting traversal outside the index root. */
/** Split a composite "<collectionId>/<rel>" into its collection + rel path. */
function splitPath(p: string): { col: Collection; rel: string } | null {
const i = p.indexOf('/')
const id = i < 0 ? p : p.slice(0, i)
const rel = i < 0 ? '' : p.slice(i + 1)
const col = byId.get(id)
return col ? { col, rel } : null
}
/** Resolve a file within its collection's source root, rejecting traversal. */
function originalAbs(index: Index, rel: string): string | null {
const root = path.resolve(index.root)
const abs = path.resolve(root, rel)
@@ -73,57 +103,43 @@ export function doocusApi(opts: Options): Plugin {
return abs
}
// Lazy search index over the cached small text — rebuilt when index.json changes.
let searchCache: { mtime: number; entries: Array<{ path: string; hay: string; bytes: number }> } | null = null
function meetusDir(col: Collection, rel: string): { dir: string; stem: string } {
return { dir: path.join(col.dir, rel + '.meetus'), stem: path.parse(rel).name }
}
// Per-collection search index over cached text — rebuilt on index.json change.
const searchCache = new Map<string, { mtime: number; entries: Array<{ path: string; hay: string; bytes: number }> }>()
// Read full text for reliable search (local files; speed is secondary). The
// generous cap only guards against a pathologically huge file.
async function readCap(p: string | null, cap = 20_000_000): Promise<string> {
if (!p) return ''
try { return (await fsp.readFile(p, 'utf-8')).slice(0, cap) } catch { return '' }
}
async function buildSearchIndex(): Promise<void> {
const index = readIndex()
if (!index) { searchCache = { mtime: 0, entries: [] }; return }
async function buildSearchIndex(c: Collection): Promise<void> {
const index = readIndexFor(c)
let mtime = 0
try { mtime = fs.statSync(indexPath).mtimeMs } catch { /* */ }
try { mtime = fs.statSync(c.indexPath).mtimeMs } catch { /* */ }
const entries: Array<{ path: string; hay: string; bytes: number }> = []
for (const f of index.files) {
let text = `${f.path} ${(f as any).title ?? ''}`
for (const f of index?.files ?? []) {
let text = `${f.path} ${f.title ?? ''}`
if (f.mode === 'extracted' && f.out) {
text += ' ' + await readCap(path.join(outputDir, f.out, 'content.md'))
text += ' ' + await readCap(path.join(c.dir, f.out, 'content.md'))
} else if (f.mode === 'meeting') {
const { dir, stem } = meetusDirFor(f.path)
const { dir, stem } = meetusDir(c, f.path)
text += ' ' + await readCap(path.join(dir, `${stem}_enhanced.txt`))
} else if (TEXT_EXTS.has(f.ext)) {
} else if (TEXT_EXTS.has(f.ext) && index) {
text += ' ' + await readCap(originalAbs(index, f.path))
}
entries.push({ path: f.path, hay: text.toLowerCase(), bytes: f.bytes })
entries.push({ path: `${c.id}/${f.path}`, hay: text.toLowerCase(), bytes: f.bytes })
}
searchCache = { mtime, entries }
}
async function search(res: ServerResponse, query: string): Promise<void> {
const q = query.trim().toLowerCase()
let curMtime = 0
try { curMtime = fs.statSync(indexPath).mtimeMs } catch { /* */ }
if (!searchCache || searchCache.mtime !== curMtime) await buildSearchIndex()
const entries = searchCache?.entries ?? []
// Empty query matches nothing (the UI shows the full tree when not searching).
const hits = q ? entries.filter((e) => e.hay.includes(q)) : []
sendJson(res, 200, {
query: q,
matches: hits.map((e) => e.path),
count: hits.length,
bytes: hits.reduce((s, e) => s + e.bytes, 0),
})
searchCache.set(c.id, { mtime, entries })
}
return {
name: 'doocus-api',
configureServer(server) {
server.config.logger.info(`[doocus-api] serving tree from: ${indexPath}`)
server.config.logger.info(
`[doocus-api] ${collections.length} collection(s): ${collections.map((c) => `${c.id} → ${c.dir}`).join(', ')}`)
server.middlewares.use('/api', (req, res, next) => {
const url = new URL(req.url ?? '/', 'http://localhost')
const parts = url.pathname.split('/').filter(Boolean)
@@ -136,54 +152,100 @@ export function doocusApi(opts: Options): Plugin {
}
async function handle(req: IncomingMessage, res: ServerResponse, parts: string[], q: URLSearchParams): Promise<boolean> {
const range = req.headers.range
if (parts[0] === 'tree') {
const index = readIndex()
if (!index) { sendJson(res, 404, { error: `index.json not found in ${outputDir}` }); return true }
sendJson(res, 200, index)
return true
}
if (parts[0] === 'detail') {
await getDetail(res, q.get('path') ?? '')
return true
}
if (parts[0] === 'original') {
await serveOriginal(res, q.get('path') ?? '', range)
return true
}
// GET /api/meeting?path=<rel> meetus review data for a meeting
// GET /api/meeting/frame?path=<rel>&file=<f> stream a frame from its .meetus/
if (parts[0] === 'meeting' && parts[1] === 'frame') {
serveMeetingFrame(res, q.get('path') ?? '', q.get('file') ?? '')
return true
}
if (parts[0] === 'meeting') {
await getMeeting(res, q.get('path') ?? '')
return true
}
if (parts[0] === 'search') {
await search(res, q.get('q') ?? '')
return true
}
// meetus-shaped run API (id = meeting rel path), so the composed meetus
// review store works unchanged when embedded in doocus.
if (parts[0] === 'collections') { listCollections(res); return true }
if (parts[0] === 'tree') { getTree(res, q.get('collections')); return true }
if (parts[0] === 'detail') { await getDetail(res, q.get('path') ?? ''); return true }
if (parts[0] === 'original') { await serveOriginal(res, q.get('path') ?? '', req.headers.range); return true }
if (parts[0] === 'meeting' && parts[1] === 'frame') { serveMeetingFrame(res, q.get('path') ?? '', q.get('file') ?? ''); return true }
if (parts[0] === 'meeting') { await getMeeting(res, q.get('path') ?? ''); return true }
if (parts[0] === 'search') { await search(res, q.get('q') ?? '', q.get('collections')); return true }
if (parts[0] === 'runs' && parts.length >= 2) {
const id = decodeURIComponent(parts[1])
if (parts.length === 2 && req.method === 'GET') { await getRunShaped(res, id); return true }
if (parts.length === 3 && parts[2] === 'review' && req.method === 'PUT') { await saveRunReview(req, res, id); return true }
}
return false
}
/** The .meetus sidecar dir + the video stem for a meeting node's rel path. */
function meetusDirFor(rel: string): { dir: string; stem: string } {
return { dir: path.join(outputDir, rel + '.meetus'), stem: path.parse(rel).name }
function listCollections(res: ServerResponse): void {
const out = collections.map((c) => {
const idx = readIndexFor(c)
return { id: c.id, name: c.name, available: !!idx, count: idx?.files.length ?? 0 }
})
sendJson(res, 200, { collections: out })
}
function getTree(res: ServerResponse, activeCsv: string | null): void {
const active = activeCsv ? new Set(activeCsv.split(',').filter(Boolean)) : null
const cols = collections.filter((c) => !active || active.has(c.id))
const files: Array<Node & { collection: string }> = []
const counts: Record<string, number> = { extracted: 0, meeting: 0, link: 0, total: 0 }
for (const c of cols) {
const idx = readIndexFor(c)
if (!idx) continue
for (const f of idx.files) {
files.push({ ...f, collection: c.id, path: `${c.id}/${f.path}` })
counts[f.mode] = (counts[f.mode] ?? 0) + 1
counts.total += 1
}
}
sendJson(res, 200, {
collections: collections.map((c) => ({ id: c.id, name: c.name })),
files,
counts,
})
}
async function getDetail(res: ServerResponse, composite: string): Promise<void> {
const sp = splitPath(composite)
if (!sp) { sendJson(res, 404, { error: 'collection not found' }); return }
const idx = readIndexFor(sp.col)
const node = idx?.files.find((f) => f.path === sp.rel)
if (!idx || !node) { sendJson(res, 404, { error: 'node not found' }); return }
let content: string | null = null
let meta: unknown = null
let text: string | null = null
if (node.mode === 'extracted' && node.out) {
try { content = await fsp.readFile(path.join(sp.col.dir, node.out, 'content.md'), 'utf-8') } catch { /* */ }
try { meta = JSON.parse(await fsp.readFile(path.join(sp.col.dir, node.out, 'meta.json'), 'utf-8')) } catch { /* */ }
} else if (node.mode === 'link' && TEXT_EXTS.has(node.ext)) {
const abs = originalAbs(idx, sp.rel)
if (abs) { try { text = await fsp.readFile(abs, 'utf-8') } catch { /* */ } }
}
sendJson(res, 200, {
node: { ...node, collection: sp.col.id, path: composite },
content, meta, text,
mime: MIME[path.extname(node.name).toLowerCase()] ?? 'application/octet-stream',
originalUrl: `/api/original?path=${encodeURIComponent(composite)}`,
})
}
async function serveOriginal(res: ServerResponse, composite: string, range?: string): Promise<void> {
const sp = splitPath(composite)
const idx = sp ? readIndexFor(sp.col) : null
const abs = sp && idx ? originalAbs(idx, sp.rel) : null
if (!abs || !isFile(abs)) { sendJson(res, 404, { error: 'original not reachable' }); return }
const mime = MIME[path.extname(abs).toLowerCase()] ?? 'application/octet-stream'
const size = fs.statSync(abs).size
res.setHeader('Content-Type', mime)
res.setHeader('Accept-Ranges', 'bytes')
const m = range?.match(/bytes=(\d*)-(\d*)/)
if (m) {
const start = m[1] ? parseInt(m[1], 10) : 0
const end = m[2] ? parseInt(m[2], 10) : size - 1
if (start >= size || end >= size || start > end) {
res.statusCode = 416; res.setHeader('Content-Range', `bytes */${size}`); res.end(); return
}
res.statusCode = 206
res.setHeader('Content-Range', `bytes ${start}-${end}/${size}`)
res.setHeader('Content-Length', String(end - start + 1))
fs.createReadStream(abs, { start, end }).pipe(res)
return
}
res.setHeader('Content-Length', String(size))
fs.createReadStream(abs).pipe(res)
}
interface MeetingData {
@@ -192,9 +254,10 @@ export function doocusApi(opts: Options): Plugin {
enhancedAvailable: boolean
}
/** Read a meeting's .meetus output into the shared review shape. */
async function meetingData(rel: string): Promise<MeetingData> {
const { dir, stem } = meetusDirFor(rel)
async function meetingData(composite: string): Promise<MeetingData | null> {
const sp = splitPath(composite)
if (!sp) return null
const { dir, stem } = meetusDir(sp.col, sp.rel)
const segments: MeetingData['segments'] = []
const whisper = path.join(dir, `${stem}.json`)
if (fs.existsSync(whisper)) {
@@ -208,8 +271,6 @@ export function doocusApi(opts: Options): Plugin {
}
} catch { /* none */ }
}
// Frame → time from enhanced.txt ([MM:SS] before each `Frame: frames/…`).
const enhanced = path.join(dir, `${stem}_enhanced.txt`)
const times = new Map<string, number>()
let enhancedAvailable = false
@@ -225,7 +286,6 @@ export function doocusApi(opts: Options): Plugin {
}
} catch { /* none */ }
}
const frames: MeetingData['frames'] = []
try {
const framesDir = path.join(dir, 'frames')
@@ -235,36 +295,37 @@ export function doocusApi(opts: Options): Plugin {
try { size = (await fsp.stat(path.join(framesDir, file))).size } catch { /* */ }
frames.push({
file,
url: `/api/meeting/frame?path=${encodeURIComponent(rel)}&file=${encodeURIComponent(file)}`,
url: `/api/meeting/frame?path=${encodeURIComponent(composite)}&file=${encodeURIComponent(file)}`,
time: fny ? Number(fny[1]) : times.get(file) ?? null,
size,
})
}
} catch { /* no frames */ }
return { segments, frames, enhancedAvailable }
}
async function getMeeting(res: ServerResponse, rel: string): Promise<void> {
const { dir } = meetusDirFor(rel)
const data = await meetingData(rel)
async function getMeeting(res: ServerResponse, composite: string): Promise<void> {
const sp = splitPath(composite)
const data = await meetingData(composite)
if (!sp || !data) { sendJson(res, 404, { error: 'meeting not found' }); return }
sendJson(res, 200, {
path: rel,
hasOutput: fs.existsSync(dir),
path: composite,
hasOutput: fs.existsSync(meetusDir(sp.col, sp.rel).dir),
...data,
videoUrl: `/api/original?path=${encodeURIComponent(rel)}`,
videoUrl: `/api/original?path=${encodeURIComponent(composite)}`,
})
}
/** meetus store's /api/runs/:id shape (id = meeting rel path). */
async function getRunShaped(res: ServerResponse, id: string): Promise<void> {
const { dir } = meetusDirFor(id)
const sp = splitPath(id)
const data = await meetingData(id)
if (!sp || !data) { sendJson(res, 404, { error: 'run not found' }); return }
const { dir } = meetusDir(sp.col, sp.rel)
let review: unknown = null
try { review = JSON.parse(await fsp.readFile(path.join(dir, 'review.json'), 'utf-8')) } catch { /* none */ }
try { review = JSON.parse(await fsp.readFile(path.join(dir, 'review.json'), 'utf-8')) } catch { /* */ }
sendJson(res, 200, {
id,
manifest: { video: { name: path.basename(id) } },
manifest: { video: { name: path.basename(sp.rel) } },
segments: data.segments,
frames: data.frames,
enhancedAvailable: data.enhancedAvailable,
@@ -274,9 +335,10 @@ export function doocusApi(opts: Options): Plugin {
})
}
/** Persist the review sidecar next to the meeting's .meetus output (atomic). */
async function saveRunReview(req: IncomingMessage, res: ServerResponse, id: string): Promise<void> {
const { dir } = meetusDirFor(id)
const sp = splitPath(id)
if (!sp) { sendJson(res, 404, { error: 'run not found' }); return }
const { dir } = meetusDir(sp.col, sp.rel)
if (!fs.existsSync(dir)) { sendJson(res, 404, { error: 'no meetus output for this meeting' }); return }
let parsed: unknown
try { parsed = JSON.parse((await readBody(req)) || '{}') } catch { sendJson(res, 400, { error: 'invalid JSON' }); return }
@@ -286,8 +348,10 @@ export function doocusApi(opts: Options): Plugin {
sendJson(res, 200, { ok: true })
}
function serveMeetingFrame(res: ServerResponse, rel: string, file: string): void {
const { dir } = meetusDirFor(rel)
function serveMeetingFrame(res: ServerResponse, composite: string, file: string): void {
const sp = splitPath(composite)
if (!sp) { sendJson(res, 404, { error: 'not found' }); return }
const { dir } = meetusDir(sp.col, sp.rel)
const safe = path.basename(file)
const full = path.join(dir, 'frames', safe)
if (safe !== file || !fs.existsSync(full)) { sendJson(res, 404, { error: 'frame not found' }); return }
@@ -296,63 +360,24 @@ export function doocusApi(opts: Options): Plugin {
fs.createReadStream(full).pipe(res)
}
async function getDetail(res: ServerResponse, rel: string): Promise<void> {
const index = readIndex()
if (!index) { sendJson(res, 404, { error: 'no index' }); return }
const node = index.files.find((f) => f.path === rel)
if (!node) { sendJson(res, 404, { error: 'node not found' }); return }
let content: string | null = null // extracted search text (heavy formats)
let meta: unknown = null
let text: string | null = null // inline text for text-native originals
if (node.mode === 'extracted' && node.out) {
try { content = await fsp.readFile(path.join(outputDir, node.out, 'content.md'), 'utf-8') } catch { /* */ }
try { meta = JSON.parse(await fsp.readFile(path.join(outputDir, node.out, 'meta.json'), 'utf-8')) } catch { /* */ }
} else if (node.mode === 'link' && TEXT_EXTS.has(node.ext)) {
const abs = originalAbs(index, rel)
if (abs) { try { text = await fsp.readFile(abs, 'utf-8') } catch { /* */ } }
}
sendJson(res, 200, {
node,
content,
meta,
text,
mime: MIME[path.extname(node.name).toLowerCase()] ?? 'application/octet-stream',
originalUrl: `/api/original?path=${encodeURIComponent(rel)}`,
})
}
async function serveOriginal(res: ServerResponse, rel: string, range?: string): Promise<void> {
const index = readIndex()
if (!index) { sendJson(res, 404, { error: 'no index' }); return }
const abs = originalAbs(index, rel)
if (!abs || !isFile(abs)) { sendJson(res, 404, { error: 'original not reachable' }); return }
const mime = MIME[path.extname(abs).toLowerCase()] ?? 'application/octet-stream'
const size = fs.statSync(abs).size
res.setHeader('Content-Type', mime)
res.setHeader('Accept-Ranges', 'bytes')
// Range support — needed so large meeting videos can seek/stream.
const m = range?.match(/bytes=(\d*)-(\d*)/)
if (m) {
const start = m[1] ? parseInt(m[1], 10) : 0
const end = m[2] ? parseInt(m[2], 10) : size - 1
if (start >= size || end >= size || start > end) {
res.statusCode = 416
res.setHeader('Content-Range', `bytes */${size}`)
res.end()
return
async function search(res: ServerResponse, query: string, activeCsv: string | null): Promise<void> {
const qq = query.trim().toLowerCase()
const active = activeCsv ? new Set(activeCsv.split(',').filter(Boolean)) : null
const cols = collections.filter((c) => !active || active.has(c.id))
let matches: string[] = []
let bytes = 0
if (qq) {
for (const c of cols) {
let cur = 0
try { cur = fs.statSync(c.indexPath).mtimeMs } catch { /* */ }
const cached = searchCache.get(c.id)
if (!cached || cached.mtime !== cur) await buildSearchIndex(c)
for (const e of searchCache.get(c.id)?.entries ?? []) {
if (e.hay.includes(qq)) { matches.push(e.path); bytes += e.bytes }
}
}
res.statusCode = 206
res.setHeader('Content-Range', `bytes ${start}-${end}/${size}`)
res.setHeader('Content-Length', String(end - start + 1))
fs.createReadStream(abs, { start, end }).pipe(res)
return
}
res.setHeader('Content-Length', String(size))
fs.createReadStream(abs).pipe(res)
sendJson(res, 200, { query: qq, matches, count: matches.length, bytes })
}
}