flaskpaste/documentation/api.md

# FlaskPaste API Reference

## Overview

FlaskPaste provides a RESTful API for creating, retrieving, and deleting text and binary pastes.

**Base URL:** `http://your-server:5000/`

**Content Types:**
- Requests: `application/json`, `text/plain`, `application/octet-stream`, or any binary type
- Responses: `application/json` for metadata, original MIME type for raw content

## Authentication

Authentication is optional and uses client certificate fingerprints passed via the `X-SSL-Client-SHA1` header.

```http
X-SSL-Client-SHA1: a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2
```

The fingerprint must be exactly 40 lowercase hexadecimal characters (SHA1).

**Benefits of authentication:**
- Larger upload limit (50 MiB vs 3 MiB)
- Ability to delete owned pastes

---

## Endpoints

### GET /challenge

Get a proof-of-work challenge for paste creation. Required when PoW is enabled.

**Request:**
```http
GET /challenge HTTP/1.1
Host: localhost:5000
```

**Response (PoW disabled):**
```json
{
  "enabled": false,
  "difficulty": 0
}
```

**Response (PoW enabled):**
```json
{
  "enabled": true,
  "nonce": "a1b2c3d4e5f6a7b8a1b2c3d4e5f6a7b8",
  "difficulty": 20,
  "expires": 1700000300,
  "token": "a1b2c3d4...:1700000300:20:signature"
}
```

The client must find a number `N` such that `SHA256(nonce + ":" + N)` has at least `difficulty` leading zero bits.

---

### GET /health

Health check endpoint for load balancers and monitoring.

**Request:**
```http
GET /health HTTP/1.1
Host: localhost:5000
```

**Response (200 OK):**
```json
{
  "status": "healthy",
  "database": "ok"
}
```

**Response (503 Service Unavailable):**
```json
{
  "status": "unhealthy",
  "database": "error"
}
```

---

### GET /

Returns API information and usage instructions.

**Request:**
```http
GET / HTTP/1.1
Host: localhost:5000
```

**Response:**
```json
{
  "name": "FlaskPaste",
  "version": "1.2.0",
  "endpoints": {
    "GET /": "API information",
    "GET /health": "Health check",
    "POST /": "Create paste",
    "GET /<id>": "Retrieve paste metadata",
    "GET /<id>/raw": "Retrieve raw paste content",
    "DELETE /<id>": "Delete paste"
  },
  "usage": {
    "raw": "curl --data-binary @file.txt http://host/",
    "pipe": "cat file.txt | curl --data-binary @- http://host/",
    "json": "curl -H 'Content-Type: application/json' -d '{\"content\":\"...\"}' http://host/"
  },
  "note": "Use --data-binary (not -d) to preserve newlines"
}
```

---

### POST /

Create a new paste.

**Request (Raw Binary):**
```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: application/octet-stream

```

**Request (JSON):**
```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: application/json

```

**Request (with Proof-of-Work):**

When PoW is enabled, include the challenge token and solution:

```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: text/plain
X-PoW-Token: a1b2c3d4...:1700000300:20:signature
X-PoW-Solution: 12345678

```

**Request (Burn-After-Read):**

Create a paste that deletes itself after first retrieval:

```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: text/plain
X-Burn-After-Read: true

```

**Request (Custom Expiry):**

Create a paste with custom expiry time (in seconds):

```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: text/plain
X-Expiry: 3600

```

**Request (Password Protected):**

Create a paste that requires a password to access:

```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: text/plain
X-Paste-Password: secretpassword

```

**Response (201 Created):**
```json
{
  "id": "abc12345",
  "url": "/abc12345",
  "raw": "/abc12345/raw",
  "mime_type": "text/plain",
  "created_at": 1700000000,
  "owner": "a1b2c3...",        // Only present if authenticated
  "burn_after_read": true,     // Only present if enabled
  "expires_at": 1700003600,    // Only present if custom expiry set
  "password_protected": true   // Only present if password set
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 400 | No content provided |
| 400 | Password too long (max 1024 chars) |
| 400 | Proof-of-work required (when PoW enabled) |
| 400 | Proof-of-work failed (invalid/expired challenge) |
| 400 | Paste too small (below minimum size) |
| 413 | Paste too large |
| 429 | Duplicate content rate limit exceeded |

**Size Limits:**
- Minimum: disabled by default (`FLASKPASTE_MIN_SIZE`, e.g. 64 bytes for encryption enforcement)
- Anonymous: 3 MiB (configurable via `FLASKPASTE_MAX_ANON`)
- Authenticated: 50 MiB (configurable via `FLASKPASTE_MAX_AUTH`)

---

### GET /{id}

### HEAD /{id}

Retrieve paste metadata. HEAD returns headers only (no body).

**Request:**
```http
GET /abc12345 HTTP/1.1
Host: localhost:5000
```

**Request (Password Protected):**
```http
GET /abc12345 HTTP/1.1
Host: localhost:5000
X-Paste-Password: secretpassword
```

**Response (200 OK):**
```json
{
  "id": "abc12345",
  "mime_type": "text/plain",
  "size": 1234,
  "created_at": 1700000000,
  "raw": "/abc12345/raw",
  "password_protected": true  // Only present if protected
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 400 | Invalid paste ID format |
| 401 | Password required |
| 403 | Invalid password |
| 404 | Paste not found |

---

### GET /{id}/raw

### HEAD /{id}/raw

Retrieve raw paste content with correct MIME type. HEAD returns headers only (useful for checking MIME type and size without downloading content).

**Request:**
```http
GET /abc12345/raw HTTP/1.1
Host: localhost:5000
```

**Request (Password Protected):**
```http
GET /abc12345/raw HTTP/1.1
Host: localhost:5000
X-Paste-Password: secretpassword
```

**Response (200 OK):**
```http
HTTP/1.1 200 OK
Content-Type: image/png
Content-Disposition: inline

<binary content>
```

- `Content-Disposition: inline` is set for `image/*` and `text/*` types
- Content-Type matches the detected/stored MIME type

**Errors:**
| Code | Description |
|------|-------------|
| 400 | Invalid paste ID format |
| 401 | Password required |
| 403 | Invalid password |
| 404 | Paste not found |

---

### DELETE /{id}

Delete a paste. Requires authentication and ownership.

**Request:**
```http
DELETE /abc12345 HTTP/1.1
Host: localhost:5000
X-SSL-Client-SHA1: a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2
```

**Response (200 OK):**
```json
{
  "message": "Paste deleted"
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 400 | Invalid paste ID format |
| 401 | Authentication required |
| 403 | Permission denied (not owner) |
| 404 | Paste not found |

---

## MIME Type Detection

FlaskPaste automatically detects MIME types using:

1. **Magic byte signatures** (highest priority)
   - PNG: `\x89PNG\r\n\x1a\n`
   - JPEG: `\xff\xd8\xff`
   - GIF: `GIF87a` or `GIF89a`
   - WebP: `RIFF....WEBP`
   - ZIP: `PK\x03\x04`
   - PDF: `%PDF`
   - GZIP: `\x1f\x8b`

2. **Explicit Content-Type header** (if not generic)

3. **UTF-8 detection** (falls back to `text/plain`)

4. **Binary fallback** (`application/octet-stream`)

---

## Paste Expiry

Pastes expire based on last access time (default: 5 days).

- Every `GET /{id}` or `GET /{id}/raw` updates the last access timestamp
- Cleanup runs automatically (hourly, throttled)
- Configurable via `FLASKPASTE_EXPIRY` environment variable

**Custom Expiry:**

Pastes can have custom expiry times using the `X-Expiry` header:

```bash
# Paste expires in 1 hour
curl -H "X-Expiry: 3600" --data-binary @file.txt http://host/
```

**Configuration:**
```bash
export FLASKPASTE_EXPIRY=432000       # Default expiry: 5 days
export FLASKPASTE_MAX_EXPIRY=2592000  # Max custom expiry: 30 days
```

**Notes:**
- Custom expiry is capped at `FLASKPASTE_MAX_EXPIRY`
- Invalid or negative values are ignored (uses default)
- Response includes `expires_at` timestamp when custom expiry is set

---

## Burn-After-Read

Single-access pastes that delete themselves after first retrieval.

**How it works:**
- Set `X-Burn-After-Read: true` header on creation
- First `GET /{id}/raw` returns content and deletes paste
- Subsequent requests return 404
- Metadata `GET /{id}` does not trigger burn
- `HEAD` requests do not trigger burn

**Usage:**
```bash
# Create burn-after-read paste
curl -H "X-Burn-After-Read: true" --data-binary @secret.txt http://host/

# Response indicates burn is enabled
{
  "id": "abc12345",
  "burn_after_read": true,
  ...
}
```

**Notes:**
- Response includes `X-Burn-After-Read: true` header when content is retrieved
- Can be combined with custom expiry (paste expires OR burns, whichever first)
- Accepted values: `true`, `1`, `yes` (case-insensitive)

---

## Password Protection

Pastes can be protected with a password using PBKDF2-HMAC-SHA256 hashing.

**Creating a protected paste:**
```http
POST / HTTP/1.1
Host: localhost:5000
Content-Type: text/plain
X-Paste-Password: mysecretpassword

```

**Response (201 Created):**
```json
{
  "id": "abc12345",
  "url": "/abc12345",
  "raw": "/abc12345/raw",
  "mime_type": "text/plain",
  "created_at": 1700000000,
  "password_protected": true
}
```

**Accessing protected paste:**
```http
GET /abc12345 HTTP/1.1
Host: localhost:5000
X-Paste-Password: mysecretpassword
```

**Errors:**
| Code | Description |
|------|-------------|
| 400 | Password too long (max 1024 chars) |
| 401 | Password required |
| 403 | Invalid password |

**Response (401 Unauthorized):**
```json
{
  "error": "Password required",
  "password_protected": true
}
```

**Response (403 Forbidden):**
```json
{
  "error": "Invalid password"
}
```

**Security Implementation:**
- PBKDF2-HMAC-SHA256 with 600,000 iterations (OWASP 2023)
- 32-byte random salt per password
- Constant-time comparison prevents timing attacks
- Passwords never logged or stored in plaintext
- Maximum 1024 characters to prevent DoS via expensive hashing

**CLI Usage:**
```bash
# Create protected paste
./fpaste create -p "mypassword" secret.txt

# Retrieve protected paste
./fpaste get -p "mypassword" abc12345
```

**Notes:**
- Password protection can be combined with burn-after-read and custom expiry
- Unicode passwords are supported
- Special characters are allowed

---

## Abuse Prevention

FlaskPaste includes content-hash based deduplication to prevent spam and abuse.

**How it works:**
- Each paste's SHA256 content hash is tracked
- Repeated submissions of identical content are throttled
- After exceeding the threshold, further duplicates are rejected with 429

**Default limits:**
- Window: 1 hour (`FLASKPASTE_DEDUP_WINDOW`)
- Maximum: 3 identical submissions per window (`FLASKPASTE_DEDUP_MAX`)

**Response (429 Too Many Requests):**
```json
{
  "error": "Duplicate content rate limit exceeded",
  "count": 3,
  "window_seconds": 3600
}
```

**Configuration:**
```bash
export FLASKPASTE_DEDUP_WINDOW=3600  # Window in seconds (default: 1 hour)
export FLASKPASTE_DEDUP_MAX=3        # Max duplicates per window (default: 3)
```

**Notes:**
- Different content is not affected by other content's limits
- Counter resets after the window expires
- Hash records are cleaned up periodically

---

## Entropy Enforcement

FlaskPaste can require minimum content entropy to enforce client-side encryption.

**How it works:**
- Shannon entropy is calculated for submitted content (bits per byte)
- Encrypted/random data: ~7.5-8.0 bits/byte
- English text: ~4.0-5.0 bits/byte
- Content below threshold is rejected with 400

**Configuration:**
```bash
export FLASKPASTE_MIN_ENTROPY=6.0      # Require encryption-level entropy (0=disabled)
export FLASKPASTE_MIN_ENTROPY_SIZE=256 # Only check content >= this size (default: 256)
```

**Response (400 Bad Request):**
```json
{
  "error": "Content entropy too low",
  "entropy": 4.12,
  "min_entropy": 7.0,
  "hint": "Encrypt content before uploading (fpaste encrypts by default)"
}
```

**Caveats:**
- Small data is exempt (configurable via `MIN_ENTROPY_SIZE`, default 256 bytes)
- Compressed data (gzip, zip) also has high entropy — not distinguishable from encrypted
- This is a heuristic, not cryptographic proof of encryption

**Recommended thresholds:**
| Threshold | Effect |
|-----------|--------|
| 0 | Disabled (default) |
| 5.0 | Blocks most plaintext |
| 6.0 | Requires encryption or compression |
| 7.0 | Requires encryption + sufficient size |

---

### Binary Content Requirement

FlaskPaste can require unrecognizable binary content (MIME-based encryption enforcement).

**How it works:**
- Content is checked for magic bytes (PNG, JPEG, PDF, ZIP, etc.)
- Content is checked for valid UTF-8 text
- Recognized formats are rejected with 400
- Only application/octet-stream (unrecognizable binary) is allowed

**Configuration:**
```bash
export FLASKPASTE_REQUIRE_BINARY=1  # Reject recognizable formats (0=disabled)
```

**Response (400 Bad Request):**
```json
{
  "error": "Recognizable format not allowed",
  "detected": "text/plain",
  "hint": "Encrypt content before uploading (fpaste encrypts by default)"
}
```

**Detected formats:**
- `text/plain` (valid UTF-8 text)
- `image/png`, `image/jpeg`, `image/gif`, `image/webp`
- `application/pdf`, `application/zip`, `application/gzip`

**vs Entropy enforcement:**
| Method | Detects | False positives |
|--------|---------|-----------------|
| Entropy | Random-looking data | Compressed files pass |
| Binary | No magic bytes + invalid UTF-8 | Minimal |

Use both together for maximum encryption enforcement:
```bash
export FLASKPASTE_REQUIRE_BINARY=1
export FLASKPASTE_MIN_ENTROPY=6.0
export FLASKPASTE_MIN_SIZE=64
```

---

## Proof-of-Work

FlaskPaste includes an optional proof-of-work system to prevent automated spam.

**How it works:**
1. Client requests a challenge via `GET /challenge`
2. Server returns a nonce, difficulty, expiry time, and signed token
3. Client computes SHA256 hashes until finding one with enough leading zero bits
4. Client submits paste with `X-PoW-Token` and `X-PoW-Solution` headers

**Algorithm:**
```
For N = 0, 1, 2, ...:
  hash = SHA256(nonce + ":" + N)
  if leading_zero_bits(hash) >= difficulty:
    return N as solution
```

**Configuration:**
```bash
export FLASKPASTE_POW_DIFFICULTY=20   # Leading zero bits required (0=disabled)
export FLASKPASTE_POW_TTL=300         # Challenge validity in seconds
export FLASKPASTE_POW_SECRET="..."    # Optional signing key (auto-generated if empty)
```

**Headers:**
| Header | Description |
|--------|-------------|
| `X-PoW-Token` | The full token from `/challenge` response |
| `X-PoW-Solution` | The computed solution number |

**Error responses:**
```json
{
  "error": "Proof-of-work required",
  "hint": "GET /challenge for a new challenge"
}
```

```json
{
  "error": "Proof-of-work failed: Challenge expired"
}
```

**Notes:**
- Difficulty 20 requires approximately 1 million hash attempts on average
- Challenges are signed to prevent tampering
- Each challenge can only be used once (nonce uniqueness)

---

## Error Response Format

All errors return JSON:

```json
{
  "error": "Description of the error"
}
```

For size limit errors (413):
```json
{
  "error": "Paste too large",
  "size": 5000000,
  "max_size": 3145728,
  "authenticated": false
}
```

---

## Security Headers

All responses include the following security headers:

| Header | Value |
|--------|-------|
| `X-Content-Type-Options` | `nosniff` |
| `X-Frame-Options` | `DENY` |
| `X-XSS-Protection` | `1; mode=block` |
| `Referrer-Policy` | `strict-origin-when-cross-origin` |
| `Content-Security-Policy` | `default-src 'none'; frame-ancestors 'none'` |
| `Permissions-Policy` | `geolocation=(), microphone=(), camera=()` |
| `Strict-Transport-Security` | `max-age=31536000; includeSubDomains` |
| `Cache-Control` | `no-store, no-cache, must-revalidate, private` |
| `Pragma` | `no-cache` |

---

## Request Tracing

All requests include an `X-Request-ID` header for log correlation:

- If the client provides `X-Request-ID`, it is passed through
- If not provided, a UUID is generated
- The ID is echoed back in the response
- All log entries include `[rid=<request-id>]`

**Example:**
```bash
# Client-provided ID
curl -H "X-Request-ID: my-trace-123" https://paste.example.com/health

# Response includes:
# X-Request-ID: my-trace-123
```

---

## Proxy Trust Validation

When `FLASKPASTE_PROXY_SECRET` is configured, the application validates that requests come from a trusted reverse proxy by checking the `X-Proxy-Secret` header.

This provides defense-in-depth against header spoofing if an attacker bypasses the reverse proxy.

**Configuration:**
```bash
export FLASKPASTE_PROXY_SECRET="your-secret-value"
```

**Proxy Configuration (HAProxy):**
```
http-request set-header X-Proxy-Secret your-secret-value
```

If the secret doesn't match, authentication headers (`X-SSL-Client-SHA1`) are ignored and the request is treated as anonymous.

---

## PKI (Certificate Authority)

FlaskPaste includes an optional minimal PKI for issuing client certificates.

### Configuration

Enable PKI via environment variables:

```bash
export FLASKPASTE_PKI_ENABLED=1              # Enable PKI endpoints
export FLASKPASTE_PKI_CA_PASSWORD="secret"   # Required: CA password
export FLASKPASTE_PKI_CERT_DAYS=365          # Client certificate validity (days)
export FLASKPASTE_PKI_CA_DAYS=3650           # CA certificate validity (days)
```

### Certificate Revocation

When PKI is enabled, certificates issued by the CA are tracked. Revoked certificates are rejected during authentication (treated as anonymous).

---

### GET /pki

Get PKI status and CA information.

**Request:**
```http
GET /pki HTTP/1.1
Host: localhost:5000
```

**Response (PKI disabled):**
```json
{
  "enabled": false
}
```

**Response (PKI enabled, no CA):**
```json
{
  "enabled": true,
  "ca_exists": false,
  "hint": "POST /pki/ca to generate CA"
}
```

**Response (PKI enabled with CA):**
```json
{
  "enabled": true,
  "ca_exists": true,
  "common_name": "FlaskPaste CA",
  "fingerprint_sha1": "a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2",
  "created_at": 1700000000,
  "expires_at": 2015000000,
  "key_algorithm": "ec:secp384r1"
}
```

---

### POST /pki/ca

Generate a new Certificate Authority. Only works once (first-run bootstrap).

**Request:**
```http
POST /pki/ca HTTP/1.1
Host: localhost:5000
Content-Type: application/json

```

**Response (201 Created):**
```json
{
  "message": "CA generated",
  "common_name": "My CA",
  "fingerprint_sha1": "a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2",
  "created_at": 1700000000,
  "expires_at": 2015000000,
  "download": "/pki/ca.crt"
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 404 | PKI not enabled |
| 409 | CA already exists |
| 500 | PKI_CA_PASSWORD not configured |

---

### GET /pki/ca.crt

Download CA certificate in PEM format (for trust store).

**Request:**
```http
GET /pki/ca.crt HTTP/1.1
Host: localhost:5000
```

**Response (200 OK):**
```
-----BEGIN CERTIFICATE-----
MIICxDCCAaygAwIBAgIUY...
-----END CERTIFICATE-----
```

Content-Type: `application/x-pem-file`

**Errors:**
| Code | Description |
|------|-------------|
| 404 | PKI not enabled or CA not initialized |

---

### POST /pki/issue

Issue a new client certificate (open registration).

**Request:**
```http
POST /pki/issue HTTP/1.1
Host: localhost:5000
Content-Type: application/json

```

**Response (201 Created):**
```json
{
  "message": "Certificate issued",
  "serial": "00000000000000000000000000000001",
  "common_name": "alice",
  "fingerprint_sha1": "b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3",
  "created_at": 1700000000,
  "expires_at": 1731536000,
  "certificate_pem": "-----BEGIN CERTIFICATE-----\n...\n-----END CERTIFICATE-----\n",
  "private_key_pem": "-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n"
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 400 | common_name required |
| 404 | PKI not enabled or CA not initialized |
| 500 | Certificate issuance failed |

**Security Notes:**
- Private key is generated server-side and returned in response
- Store the private key securely; it is not recoverable
- The certificate can be used with nginx, HAProxy, or curl for mTLS

---

### GET /pki/certs

List certificates (own certificates only).

**Request (authenticated):**
```http
GET /pki/certs HTTP/1.1
Host: localhost:5000
X-SSL-Client-SHA1: a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2
```

**Response (200 OK):**
```json
{
  "certificates": [
    {
      "serial": "00000000000000000000000000000001",
      "common_name": "alice",
      "fingerprint_sha1": "b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3",
      "created_at": 1700000000,
      "expires_at": 1731536000,
      "status": "valid",
      "issued_to": "a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2"
    }
  ],
  "count": 1
}
```

**Notes:**
- Anonymous users receive an empty list
- Users see certificates they issued or certificates that match their fingerprint

---

### POST /pki/revoke/{serial}

Revoke a certificate by serial number. Requires authentication and ownership.

**Request:**
```http
POST /pki/revoke/00000000000000000000000000000001 HTTP/1.1
Host: localhost:5000
X-SSL-Client-SHA1: a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2
```

**Response (200 OK):**
```json
{
  "message": "Certificate revoked",
  "serial": "00000000000000000000000000000001"
}
```

**Errors:**
| Code | Description |
|------|-------------|
| 401 | Authentication required |
| 403 | Permission denied (not owner) |
| 404 | Certificate not found or PKI not enabled |
| 409 | Certificate already revoked |

**Authorization:**
- Must be authenticated
- Can revoke certificates you issued (issued_to matches your fingerprint)
- Can revoke your own certificate (fingerprint matches)