Delete Documents

Delete RFP Document

proposal_id

string

required

Unique identifier for the proposal

filename

string

required

Name of the file to delete

success

boolean

Indicates if deletion was successful

message

string

Confirmation message

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/{filename}" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Document deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/rfp/{filename}
DynamoDB Metadata:
- Removes file from uploaded_files.rfp-document array
- Clears RFP analysis data:
  - rfp_analysis
  - analysis_status_rfp
  - rfp_analysis_started_at
  - rfp_analysis_completed_at
  - rfp_analysis_error
Updates: Sets updated_at timestamp

Deleting an RFP document also removes all associated analysis results. You’ll need to re-upload and re-analyze the RFP to restore this data.

Delete Concept Document

proposal_id

string

required

Unique identifier for the proposal

filename

string

required

Name of the concept file to delete

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/concept/{filename}" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Concept file deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/initial_concept/{filename}
No vector cleanup needed (concept files are not vectorized)

Delete Concept Text

proposal_id

string

required

Unique identifier for the proposal

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/concept-text" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Concept text deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/initial_concept/concept_text.txt
No DynamoDB update (text is stored in S3 only)

Delete Reference Proposal

proposal_id

string

required

Unique identifier for the proposal

filename

string

required

Name of the reference file to delete

success

boolean

Deletion success indicator

message

string

Confirmation message

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/reference/{filename}" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Reference file deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/references/{filename}
Vector Embeddings:
- Deletes all vectors from reference-proposals-index
- Searches by document_name in vector keys
- Key format: *|*|*|*|{filename}|*|*
DynamoDB:
- Removes filename from uploaded_files.reference-proposals array
- Updates updated_at timestamp

Deleting a reference proposal removes all its vector embeddings. This affects AI retrieval for future proposal generation. Make sure you want to permanently remove this reference before deleting.

Vector Deletion Process

Deletion Flow

# 1. List all vectors in index
vectors = s3vectors.list_vectors(
    vectorBucketName=bucket,
    indexName="reference-proposals-index"
)

# 2. Filter by document_name (position 4 in key)
keys_to_delete = []
for vector in vectors:
    parts = vector['key'].split('|')
    if parts[4] == filename:  # document_name position
        keys_to_delete.append(vector['key'])

# 3. Delete matching vectors
s3vectors.delete_vectors(
    vectorBucketName=bucket,
    indexName="reference-proposals-index",
    keys=keys_to_delete
)

Delete Supporting Document

proposal_id

string

required

Unique identifier for the proposal

filename

string

required

Name of the supporting file to delete

success

boolean

Deletion success indicator

message

string

Confirmation message

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/supporting/{filename}" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Supporting file and vectors deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/supporting/{filename}
Vector Embeddings:
- Deletes all vectors from existing-work-index
- Searches by document_name in vector keys
DynamoDB:
- Removes filename from uploaded_files.supporting-docs array
- Updates updated_at timestamp

Delete Work Text

proposal_id

string

required

Unique identifier for the proposal

success

boolean

Deletion success indicator

message

string

Confirmation message

curl -X DELETE "https://api.igadinnovationhub.org/api/proposals/{proposal_id}/documents/work-text" \
  -H "Authorization: Bearer YOUR_TOKEN"

{
  "success": true,
  "message": "Work text deleted successfully"
}

What Gets Deleted

S3 File: {proposal_code}/documents/supporting/existing_work_text.txt
Vector Embeddings:
- Deletes all vectors from existing-work-index
- Searches for document_name = "existing_work_text"
- Removes all chunks (may be multiple vectors)
DynamoDB:
- Removes text_inputs.existing-work field
- Updates updated_at timestamp

Work text is chunked during upload, so deleting it removes multiple vector entries from the index.

Error Handling

Common Error Codes

Code	Error	Reason
`404`	Proposal not found	Invalid proposal ID or user doesn’t own it
`500`	S3 Delete Error	Failed to delete file from S3
`500`	Failed to delete vectors	Vector service error (non-critical)
`500`	DynamoDB update failed	Database update error

Non-Critical Failures

Vector deletion failures are logged but don’t fail the request:

Example

try:
    vector_service.delete_vectors_by_document_name(filename, index)
except Exception as e:
    print(f"⚠️  Vector deletion error (non-critical): {e}")
    # Continue with S3 and DynamoDB cleanup

Error Response Examples

{
  "detail": "Proposal not found"
}

Best Practices

1. Confirm Before Deleting

Always show a confirmation dialog before deleting documents, especially reference proposals:

Confirmation Example

const deleteDocument = async (filename: string) => {
  const confirmed = window.confirm(
    `Delete "${filename}"? This will remove the file and all associated vectors. This action cannot be undone.`
  )

  if (!confirmed) return

  try {
    const response = await fetch(
      `${API_URL}/api/proposals/${proposalId}/documents/reference/${filename}`,
      {
        method: 'DELETE',
        headers: { 'Authorization': `Bearer ${token}` },
      }
    )

    if (response.ok) {
      showSuccess('Document deleted successfully')
      refreshDocumentList()
    } else {
      const error = await response.json()
      showError('Deletion failed', error.detail)
    }
  } catch (err) {
    showError('Network error', 'Failed to delete document')
  }
}

2. Handle Partial Failures

Vector deletion may fail silently. Check logs if you suspect incomplete cleanup:

Check Logs

# Backend logs will show:
⚠️  No vectors found for document: filename.pdf
# or
⚠️  Vector deletion error (non-critical): ...

3. Refresh UI After Deletion

Refresh Pattern

const handleDelete = async (filename: string) => {
  await deleteDocument(filename)
  
  // Refresh document list
  await queryClient.invalidateQueries(['documents', proposalId])
  
  // Update vectorization status
  await queryClient.invalidateQueries(['vectorization-status', proposalId])
}

4. Bulk Deletion

For deleting multiple files, use Promise.all but handle individual failures:

Bulk Delete

const deleteMultiple = async (filenames: string[]) => {
  const results = await Promise.allSettled(
    filenames.map(filename =>
      fetch(`${API_URL}/api/proposals/${proposalId}/documents/reference/${filename}`, {
        method: 'DELETE',
        headers: { 'Authorization': `Bearer ${token}` },
      })
    )
  )

  const succeeded = results.filter(r => r.status === 'fulfilled').length
  const failed = results.filter(r => r.status === 'rejected').length

  showSuccess(`Deleted ${succeeded} files`)
  if (failed > 0) {
    showWarning(`Failed to delete ${failed} files`)
  }
}

5. Vector Index Cleanup

If vectors aren’t deleted properly (e.g., network error), you may need to manually clean up:

Manual Cleanup

from app.shared.vectors.service import VectorEmbeddingsService

vector_service = VectorEmbeddingsService()

# Delete by document name
vector_service.delete_vectors_by_document_name(
    document_name="orphaned_file.pdf",
    index_name="reference-proposals-index"
)

# Delete all vectors for a proposal
vector_service.delete_proposal_vectors(proposal_id="PROP-2024-001")

Cascade Deletion Summary

Document Type	S3	Vectors	DynamoDB	Analysis Data
RFP Document	✅	❌	✅	✅ Clears analysis
Concept File	✅	❌	❌	❌
Concept Text	✅	❌	❌	❌
Reference Proposal	✅	✅ `reference-proposals-index`	✅	❌
Supporting File	✅	✅ `existing-work-index`	✅	❌
Work Text	✅	✅ `existing-work-index` (all chunks)	✅	❌

Proposals

Proposal Workflow

Documents

Admin

Authentication

Delete RFP Document

What Gets Deleted

Delete Concept Document

What Gets Deleted

Delete Concept Text

What Gets Deleted

Delete Reference Proposal

What Gets Deleted

Vector Deletion Process

Delete Supporting Document

What Gets Deleted

Delete Work Text

What Gets Deleted

Error Handling

Common Error Codes

Non-Critical Failures

Error Response Examples

Best Practices

1. Confirm Before Deleting

2. Handle Partial Failures

3. Refresh UI After Deletion

4. Bulk Deletion

5. Vector Index Cleanup

Cascade Deletion Summary

Build docs developers (and LLMs) love

Proposals

Proposal Workflow

Documents

Admin

Authentication

​Delete RFP Document

​What Gets Deleted

​Delete Concept Document

​What Gets Deleted

​Delete Concept Text

​What Gets Deleted

​Delete Reference Proposal

​What Gets Deleted

​Vector Deletion Process

​Delete Supporting Document

​What Gets Deleted

​Delete Work Text

​What Gets Deleted

​Error Handling

​Common Error Codes

​Non-Critical Failures

​Error Response Examples

​Best Practices

​1. Confirm Before Deleting

​2. Handle Partial Failures

​3. Refresh UI After Deletion

​4. Bulk Deletion

​5. Vector Index Cleanup

​Cascade Deletion Summary

​Related Endpoints

Build docs developers (and LLMs) love

Delete RFP Document

What Gets Deleted

Delete Concept Document

What Gets Deleted

Delete Concept Text

What Gets Deleted

Delete Reference Proposal

What Gets Deleted

Vector Deletion Process

Delete Supporting Document

What Gets Deleted

Delete Work Text

What Gets Deleted

Error Handling

Common Error Codes

Non-Critical Failures

Error Response Examples

Best Practices

1. Confirm Before Deleting

2. Handle Partial Failures

3. Refresh UI After Deletion

4. Bulk Deletion

5. Vector Index Cleanup

Cascade Deletion Summary

Related Endpoints