feat(observabilidade): CLI sessions-worklog-import com paginação

- Script CLI com args --discussion 31|32|33|all, --since-days N, --force
- Paginação via MCP gateway (limit 100, tree_view false)
- Output JSON-line progressivo + summary final
- Testes: parseWorklogHtml tolerante (h2/h3/h4), idempotência upsert

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-04-23 03:07:17 +01:00
parent 11f9833aac
commit f4adf8674d
2 changed files with 272 additions and 0 deletions
+97
View File
@@ -0,0 +1,97 @@
#!/usr/bin/env tsx
/**
* Importa comentários das discussões Desk #31/#32/#33 (worklogs, reflexões
* e acções de melhoria) para a BD Observabilidade.
*
* Uso:
* sessions-worklog-import.ts [--discussion 31|32|33|all] [--since-days N]
* sessions-worklog-import.ts --discussion 31 --page-size 200
*
* Default: --discussion all --since-days 365
*
* Env obrigatório:
* MCP_GATEWAY_TOKEN Bearer token do gateway MCP
*/
import { openSessionsDb } from '../services/sessions/db.js'
import { DEFAULT_DB_PATH } from '../services/sessions/indexer.js'
import { importWorklogDiscussion, type ImportResult } from '../services/sessions/worklog-import.js'
interface Args {
discussion: 'all' | number
sinceDays: number
pageSize: number
}
function parseArgs(argv: string[]): Args {
const a: Args = { discussion: 'all', sinceDays: 365, pageSize: 500 }
for (let i = 0; i < argv.length; i++) {
if (argv[i] === '--discussion') {
const v = argv[++i]
a.discussion = v === 'all' ? 'all' : parseInt(v, 10)
} else if (argv[i] === '--since-days') {
a.sinceDays = parseInt(argv[++i], 10)
} else if (argv[i] === '--page-size') {
a.pageSize = parseInt(argv[++i], 10)
}
}
return a
}
async function main(): Promise<void> {
const args = parseArgs(process.argv.slice(2))
if (!process.env.MCP_GATEWAY_TOKEN) {
console.error('[worklog-import] MCP_GATEWAY_TOKEN não definido. Aborta.')
process.exit(1)
}
const dbPath = process.env.OBSERVABILIDADE_DB ?? DEFAULT_DB_PATH
const db = openSessionsDb(dbPath)
const discussions = args.discussion === 'all' ? [31, 32, 33] : [args.discussion as number]
const sinceIso = new Date(Date.now() - args.sinceDays * 86400_000).toISOString()
console.error(
`[worklog-import] db=${dbPath} discussions=${discussions.join(',')} since=${sinceIso} page_size=${args.pageSize}`,
)
const results: ImportResult[] = []
for (const d of discussions) {
try {
const r = await importWorklogDiscussion(db, d, { sinceIso, pageSize: args.pageSize })
results.push(r)
console.error(
`[worklog-import] #${d}: fetched=${r.fetched} inserted=${r.imported} updated=${r.updated} skipped=${r.skipped} errors=${r.errors}`,
)
} catch (e) {
console.error(`[worklog-import] falha #${d}:`, (e as Error).message)
results.push({
discussion_id: d,
fetched: 0,
imported: 0,
updated: 0,
skipped: 0,
errors: 1,
})
}
}
const summary = {
db: dbPath,
since_iso: sinceIso,
discussions: results,
totals: {
fetched: results.reduce((s, r) => s + r.fetched, 0),
imported: results.reduce((s, r) => s + r.imported, 0),
updated: results.reduce((s, r) => s + r.updated, 0),
skipped: results.reduce((s, r) => s + r.skipped, 0),
errors: results.reduce((s, r) => s + r.errors, 0),
},
total_in_db: db.countWorklogComments(),
}
console.log(JSON.stringify(summary))
db.close()
}
main().catch((err) => {
console.error('[worklog-import] falha fatal:', err)
process.exit(2)
})
+175
View File
@@ -0,0 +1,175 @@
import { describe, it, expect, beforeEach } from 'vitest'
import { mkdtempSync } from 'fs'
import { tmpdir } from 'os'
import { join } from 'path'
import { openSessionsDb, type SessionsDb, type WorklogCommentRecord } from '../services/sessions/db.js'
import { parseWorklogHtml } from '../services/sessions/worklog-import.js'
import { detectActionsNeverExecuted, weekRange } from '../services/sessions/patterns.js'
const SAMPLE_H4 = `
<h4>2026-04-15 10:30 - Refactor API sessions</h4>
<p><strong>Projecto:</strong> DashDescomplicar</p>
<p><strong>Tarefa:</strong> #2059 - Observabilidade Espelho</p>
<p><strong>Duração:</strong> ~2h 15m</p>
<h4>Trabalho Realizado</h4>
<ul><li>Criar módulo worklog-import</li><li>Integrar detectores cruzados</li></ul>
<h4>Ficheiros Modificados</h4>
<ul><li><code>api/services/sessions/db.ts</code></li><li><code>api/scripts/sessions-worklog-import.ts</code></li></ul>
<h4>Problemas / Soluções</h4>
<ul><li>Parser HTML frágil → usar node-html-parser</li></ul>
<h4>Padrões Detectados</h4>
<ul><li>MCP gateway responde em SSE ou JSON</li></ul>
<h4>Acções Sugeridas</h4>
<ul><li>[Refactor] Extrair callMcpTool para módulo partilhado P2</li></ul>
`
const SAMPLE_H2 = `
<h2>2026-01-31 - Estratégia Stack</h2>
<p><strong>Duração:</strong> ~2h</p>
<h3>Trabalho Realizado</h3>
<ul><li>Stack Mapeado - 15 sistemas</li></ul>
<h3>Insights</h3>
<ul><li>Posicionamento: Marketing alta performance</li></ul>
`
const SAMPLE_D33 = `
<ul>
<li>[ ] [MCP] Corrigir bug desk-crm-v3 com tabelas de discussões</li>
</ul>
<p><strong>Origem:</strong> Sessão 2026-02-02</p>
<p><strong>Prioridade:</strong> P1</p>
`
describe('parseWorklogHtml', () => {
it('extrai campos de comentário formato <h4>', () => {
const parsed = parseWorklogHtml(SAMPLE_H4, { id: 100, discussion_id: 31, created_at: '' })
expect(parsed.id).toBe(100)
expect(parsed.title).toMatch(/2026-04-15/)
expect(parsed.task_ref).toBe('#2059')
expect(parsed.duration_sec).toBe(2 * 3600 + 15 * 60)
expect(parsed.work_items.length).toBe(2)
expect(parsed.files_modified.length).toBe(2)
expect(parsed.patterns_text.length).toBe(1)
expect(parsed.actions.length).toBe(1)
expect(parsed.actions[0].tipo).toBe('Refactor')
expect(parsed.actions[0].prioridade).toBe('P2')
expect(parsed.created_at.startsWith('2026-04-15')).toBe(true)
})
it('extrai campos de comentário formato <h2>/<h3> (legacy)', () => {
const parsed = parseWorklogHtml(SAMPLE_H2, { id: 64, discussion_id: 31, created_at: '' })
expect(parsed.title).toMatch(/2026-01-31/)
expect(parsed.work_items.length).toBeGreaterThanOrEqual(1)
expect(parsed.duration_sec).toBe(2 * 3600)
expect(parsed.created_at.startsWith('2026-01-31')).toBe(true)
})
it('extrai acções em formato discussão #33 (lista crua)', () => {
const parsed = parseWorklogHtml(SAMPLE_D33, { id: 200, discussion_id: 33, created_at: '2026-02-02T00:00:00Z' })
expect(parsed.actions.length).toBe(1)
expect(parsed.actions[0].tipo).toBe('MCP')
})
})
describe('upsertWorklogComment idempotência', () => {
let db: SessionsDb
beforeEach(() => {
const dir = mkdtempSync(join(tmpdir(), 'obs-wl-'))
db = openSessionsDb(join(dir, 'sessions.db'))
})
it('insert primeiro, update depois', () => {
const base: WorklogCommentRecord = {
id: 42,
discussion_id: 31,
created_at: '2026-04-15T10:30:00Z',
staff_id: 25,
title: 'Test',
task_ref: '#100',
duration_sec: 600,
work_items: ['a'],
files_modified: [],
problems: [],
patterns_text: [],
actions: [],
raw_html: '<h4>Test</h4>',
imported_at: '2026-04-23T00:00:00Z',
}
const r1 = db.upsertWorklogComment(base)
expect(r1.inserted).toBe(true)
expect(db.countWorklogComments()).toBe(1)
const r2 = db.upsertWorklogComment({ ...base, title: 'Updated' })
expect(r2.inserted).toBe(false)
expect(db.countWorklogComments()).toBe(1)
const list = db.listWorklogComments({ discussion_id: 31 })
expect(list[0].title).toBe('Updated')
})
})
describe('detectActionsNeverExecuted', () => {
let db: SessionsDb
beforeEach(() => {
const dir = mkdtempSync(join(tmpdir(), 'obs-act-'))
db = openSessionsDb(join(dir, 'sessions.db'))
})
it('sinaliza acções P1/P2 antigas sem execução', () => {
const old = new Date('2026-03-01T00:00:00Z').toISOString()
for (let i = 0; i < 4; i++) {
db.upsertWorklogComment({
id: 300 + i,
discussion_id: 33,
created_at: old,
staff_id: 25,
title: `Acção ${i}`,
task_ref: `#${1000 + i}`,
duration_sec: null,
work_items: [],
files_modified: [],
problems: [],
patterns_text: [],
actions: [{ tipo: 'MCP', descricao: `Corrigir bug X${i}`, prioridade: i % 2 ? 'P1' : 'P2' }],
raw_html: '',
imported_at: '2026-04-23T00:00:00Z',
})
}
const range = weekRange(new Date('2026-04-22T00:00:00Z'))
const patterns = detectActionsNeverExecuted({
db: db.rawDb(),
weekStartIso: range.start.toISOString(),
weekEndIso: range.end.toISOString(),
})
expect(patterns.length).toBe(1)
expect(patterns[0].pattern_key).toBe('actions_never_executed')
expect(patterns[0].affected_count).toBeGreaterThanOrEqual(3)
})
it('não sinaliza se acções recentes (<14 dias)', () => {
const recent = new Date().toISOString()
for (let i = 0; i < 5; i++) {
db.upsertWorklogComment({
id: 400 + i,
discussion_id: 33,
created_at: recent,
staff_id: 25,
title: null,
task_ref: null,
duration_sec: null,
work_items: [],
files_modified: [],
problems: [],
patterns_text: [],
actions: [{ tipo: 'MCP', descricao: 'x', prioridade: 'P1' }],
raw_html: '',
imported_at: recent,
})
}
const range = weekRange(new Date())
const patterns = detectActionsNeverExecuted({
db: db.rawDb(),
weekStartIso: range.start.toISOString(),
weekEndIso: range.end.toISOString(),
})
expect(patterns.length).toBe(0)
})
})