refactor: surface extraction cleanup failures as visible diagnostics
- Add IMH_EXTRACTION_CLEANUP_FAILED and IMH_EXTRACTION_RESIDUAL_ATTRIBUTES diagnostic codes with severity warning and extraction-error category - Check for residual data-imhotep-runtime-id attributes before extraction (both fast-geometry and CDP paths), emit diagnostic if prior run leaked them - Surface cleanup failures (attribute removal, CDP session detach) as returned diagnostics instead of silent console.warn - Wrap CDP sessionManager.detach() in try/catch to prevent finally-block unhandled throws on closed/navigated pages - Count injected vs cleaned runtime-id attributes; report mismatch as IMH_EXTRACTION_CLEANUP_FAILED with metrics - Move errors array declaration before try block in fast-geometry path so finally can append cleanup diagnostics
This commit is contained in:
John Dvorak
@@ -418,12 +418,35 @@ export async function extractWorldFastGeometry(
|
||||
selectorToIds: Array<[string, number[]]>
|
||||
}
|
||||
|
||||
const errors: ImhotepDiagnostic[] = []
|
||||
|
||||
const selectorPlans: SelectorPlan[] = []
|
||||
for (let i = 0; i < selectors.length; i++) {
|
||||
const queries = await materializeSemanticSelector(playwrightPage, selectors[i], i)
|
||||
selectorPlans.push({ key: selectors[i], queries })
|
||||
}
|
||||
|
||||
try {
|
||||
const residualBefore = await playwrightPage.evaluate(() =>
|
||||
document.querySelectorAll('[data-imhotep-runtime-id]').length,
|
||||
)
|
||||
if (residualBefore > 0) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_RESIDUAL_ATTRIBUTES',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `Found ${residualBefore} residual data-imhotep-runtime-id attribute(s) from a prior extraction that did not clean up.`,
|
||||
source: 'imhotep-playwright',
|
||||
related: [],
|
||||
fixHints: ['Leftover attributes indicate a prior extraction did not clean up. A page reload or navigating away and back may clear residual attributes.'],
|
||||
metrics: { residualCount: residualBefore },
|
||||
sourceRef: {},
|
||||
})
|
||||
}
|
||||
} catch {
|
||||
// Best-effort pre-check; proceed with extraction.
|
||||
}
|
||||
|
||||
try {
|
||||
const extracted = await playwrightPage.evaluate(({ plans, needs }: any) => {
|
||||
const elements: FastExtractedElement[] = []
|
||||
@@ -701,7 +724,6 @@ export async function extractWorldFastGeometry(
|
||||
}
|
||||
|
||||
const selectorToIds = new Map<string, number[]>(extracted.selectorToIds)
|
||||
const errors: ImhotepDiagnostic[] = []
|
||||
|
||||
for (const [selector, ids] of selectorToIds) {
|
||||
if (ids.length === 0 && !selector.startsWith('$')) {
|
||||
@@ -732,15 +754,45 @@ export async function extractWorldFastGeometry(
|
||||
}
|
||||
return { world, selectorToIds, errors }
|
||||
} finally {
|
||||
await playwrightPage.evaluate(() => {
|
||||
const nodes = Array.from(document.querySelectorAll('[data-imhotep-runtime-id]'))
|
||||
for (const el of nodes) {
|
||||
el.removeAttribute('data-imhotep-runtime-id')
|
||||
try {
|
||||
const result = await playwrightPage.evaluate(() => {
|
||||
const nodes = Array.from(document.querySelectorAll<HTMLElement>('[data-imhotep-runtime-id]'))
|
||||
const cleaned = nodes.length
|
||||
for (const el of nodes) {
|
||||
el.removeAttribute('data-imhotep-runtime-id')
|
||||
}
|
||||
// Verify all are gone
|
||||
const remaining = document.querySelectorAll('[data-imhotep-runtime-id]').length
|
||||
return { cleaned, remaining }
|
||||
})
|
||||
if (result.remaining > 0) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `Cleanup incomplete: ${result.remaining} data-imhotep-runtime-id attribute(s) could not be removed (injected ${result.cleaned}).`,
|
||||
source: 'imhotep-playwright',
|
||||
related: [],
|
||||
fixHints: ['Verify the page is still interactive (not closed or navigated away).'],
|
||||
metrics: { injected: result.cleaned, remaining: result.remaining },
|
||||
sourceRef: {},
|
||||
})
|
||||
}
|
||||
}).catch((err) => {
|
||||
} catch (err) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `Fast-geometry cleanup failed: ${err instanceof Error ? err.message : String(err)}`,
|
||||
source: 'imhotep-playwright',
|
||||
related: [],
|
||||
fixHints: ['Verify the page is still interactive (not closed or navigated away).'],
|
||||
metrics: {},
|
||||
sourceRef: {},
|
||||
})
|
||||
// eslint-disable-next-line no-console
|
||||
console.warn('[imhotep-playwright] fast-geometry cleanup failed:', err instanceof Error ? err.message : err)
|
||||
})
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -818,6 +870,28 @@ export async function extractWorldCdp(
|
||||
}
|
||||
|
||||
const sessionManager = createSessionManager(playwrightPage)
|
||||
|
||||
try {
|
||||
const residualBefore = await playwrightPage.evaluate(() =>
|
||||
document.querySelectorAll('[data-imhotep-runtime-id]').length,
|
||||
)
|
||||
if (residualBefore > 0) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_RESIDUAL_ATTRIBUTES',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `Found ${residualBefore} residual data-imhotep-runtime-id attribute(s) from a prior extraction that did not clean up.`,
|
||||
source: 'imhotep-cdp',
|
||||
related: [],
|
||||
fixHints: ['Leftover attributes indicate a prior extraction did not clean up. A page reload or navigating away and back may clear residual attributes.'],
|
||||
metrics: { residualCount: residualBefore },
|
||||
sourceRef: {},
|
||||
})
|
||||
}
|
||||
} catch {
|
||||
// Best-effort pre-check; proceed with extraction.
|
||||
}
|
||||
|
||||
try {
|
||||
await sessionManager.enableDomain('DOM')
|
||||
|
||||
@@ -925,16 +999,78 @@ export async function extractWorldCdp(
|
||||
}
|
||||
return { world, selectorToIds, errors }
|
||||
} finally {
|
||||
await playwrightPage.evaluate(() => {
|
||||
const nodes = Array.from(document.querySelectorAll('[data-imhotep-runtime-id]'))
|
||||
for (const el of nodes) {
|
||||
el.removeAttribute('data-imhotep-runtime-id')
|
||||
try {
|
||||
const result = await playwrightPage.evaluate(() => {
|
||||
const nodes = Array.from(document.querySelectorAll<HTMLElement>('[data-imhotep-runtime-id]'))
|
||||
const cleaned = nodes.length
|
||||
for (const el of nodes) {
|
||||
el.removeAttribute('data-imhotep-runtime-id')
|
||||
}
|
||||
const remaining = document.querySelectorAll('[data-imhotep-runtime-id]').length
|
||||
return { cleaned, remaining }
|
||||
})
|
||||
if (result.remaining > 0) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `CDP cleanup incomplete: ${result.remaining} data-imhotep-runtime-id attribute(s) could not be removed (injected ${result.cleaned}).`,
|
||||
source: 'imhotep-cdp',
|
||||
related: [],
|
||||
fixHints: ['Verify the page is still interactive (not closed or navigated away).'],
|
||||
metrics: { injected: result.cleaned, remaining: result.remaining },
|
||||
sourceRef: {},
|
||||
})
|
||||
}
|
||||
}).catch((err) => {
|
||||
} catch (err) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `CDP attribute cleanup failed: ${err instanceof Error ? err.message : String(err)}`,
|
||||
source: 'imhotep-cdp',
|
||||
related: [],
|
||||
fixHints: ['Verify the page is still interactive (not closed or navigated away).'],
|
||||
metrics: {},
|
||||
sourceRef: {},
|
||||
})
|
||||
// eslint-disable-next-line no-console
|
||||
console.warn('[imhotep-playwright] CDP cleanup failed:', err instanceof Error ? err.message : err)
|
||||
})
|
||||
await sessionManager.detach()
|
||||
}
|
||||
try {
|
||||
await sessionManager.detach()
|
||||
} catch (err) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `CDP session detach failed: ${err instanceof Error ? err.message : String(err)}`,
|
||||
source: 'imhotep-cdp',
|
||||
related: [],
|
||||
fixHints: ['The CDP session may already be closed. This is typically non-fatal.'],
|
||||
metrics: {},
|
||||
sourceRef: {},
|
||||
})
|
||||
// eslint-disable-next-line no-console
|
||||
console.warn('[imhotep-playwright] CDP cleanup failed:', err instanceof Error ? err.message : err)
|
||||
}
|
||||
try {
|
||||
await sessionManager.detach()
|
||||
} catch (err) {
|
||||
errors.push({
|
||||
code: 'IMH_EXTRACTION_CLEANUP_FAILED',
|
||||
severity: 'warning',
|
||||
category: 'extraction-error',
|
||||
message: `CDP session detach failed: ${err instanceof Error ? err.message : String(err)}`,
|
||||
source: 'imhotep-cdp',
|
||||
related: [],
|
||||
fixHints: ['The CDP session may already be closed. This is typically non-fatal.'],
|
||||
metrics: {},
|
||||
sourceRef: {},
|
||||
})
|
||||
// eslint-disable-next-line no-console
|
||||
console.warn('[imhotep-playwright] CDP session detach failed:', err instanceof Error ? err.message : err)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user