datalab-cli 0.1.0

A powerful CLI for converting, extracting, and processing documents using the Datalab API
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139

# extract-score

Score extraction results with confidence ratings.

## Synopsis

```
datalab extract-score [OPTIONS] --checkpoint-id <ID>
```

## Description

Score the results of a previous extraction using a checkpoint. Returns confidence ratings for the extracted data, helping you assess extraction quality.

!!! note
    This command requires a checkpoint ID from a previous extraction that was run with `--save-checkpoint`.

---

## Options

### Required Options

| Option | Description |
|--------|-------------|
| `--checkpoint-id <ID>` | Checkpoint ID from extraction with `--save-checkpoint` |

### Output Options

| Option | Description |
|--------|-------------|
| `-o, --output <FILE>` | Write result to file |

### Cache Options

| Option | Description |
|--------|-------------|
| `--skip-cache` | Skip local cache lookup |

### Advanced Options

| Option | Description | Default |
|--------|-------------|---------|
| `--timeout <SECS>` | Request timeout in seconds | `300` |

---

## Examples

### Basic Usage

First, run an extraction with checkpoint:

```bash
datalab extract invoice.pdf \
  --schema '{"fields": [{"name": "total", "type": "number"}]}' \
  --save-checkpoint

# Output includes: "checkpoint_id": "abc123"
```

Then score the extraction:

```bash
datalab extract-score --checkpoint-id abc123
```

### Save Score Results

```bash
datalab extract-score --checkpoint-id abc123 --output scores.json
```

---

## Output Schema

```json
{
  "scores": {
    "total": {
      "confidence": 0.95,
      "evidence": "Found clear total amount on page 1"
    },
    "invoice_number": {
      "confidence": 0.88,
      "evidence": "Invoice number found in header"
    }
  },
  "overall_confidence": 0.92,
  "metadata": {
    "checkpoint_id": "abc123",
    "processing_time": 0.5
  }
}
```

---

## Understanding Scores

| Confidence | Interpretation |
|------------|----------------|
| `0.9 - 1.0` | High confidence, likely accurate |
| `0.7 - 0.9` | Good confidence, probably accurate |
| `0.5 - 0.7` | Medium confidence, review recommended |
| `0.0 - 0.5` | Low confidence, manual verification needed |

---

## Workflow

1. **Extract with checkpoint**:
   ```bash
   datalab extract document.pdf --schema schema.json --save-checkpoint
   ```

2. **Review extraction results**

3. **Score if needed**:
   ```bash
   datalab extract-score --checkpoint-id <id>
   ```

4. **Use scores to filter or flag results**

---

## Related Commands

- [`extract`](extract.md) - Extract data from documents
- [`convert`](convert.md) - Convert documents (can save checkpoints)

---

## See Also

- [Checkpoints](../concepts/checkpoints.md)
- [Extracting Data Tutorial](../tutorials/extract-data.md)