Skip to main content

trace_spacex/
trace_spacex.rs

1use datacortex_core::dcx::Mode;
2use datacortex_core::format::{detect_format, preprocess, reverse_preprocess};
3
4fn main() {
5    let data = std::fs::read("/tmp/spacex.json").expect("read spacex.json");
6    let format = detect_format(&data);
7    eprintln!("Format: {:?}", format);
8    eprintln!("Original: {} bytes", data.len());
9
10    // Test Fast mode (where the bug is)
11    let (preprocessed, chain) = preprocess(&data, format, Mode::Fast);
12    eprintln!("Preprocessed: {} bytes", preprocessed.len());
13    eprintln!("Transforms: {}", chain.records.len());
14    for (i, rec) in chain.records.iter().enumerate() {
15        let name = match rec.id {
16            7 => "JSON_ARRAY_COLUMNAR",
17            15 => "NESTED_FLATTEN",
18            14 => "TYPED_ENCODING",
19            13 => "VALUE_DICT",
20            _ => "?",
21        };
22        eprintln!(
23            "  [{}] ID={} ({}) metadata={} bytes",
24            i,
25            rec.id,
26            name,
27            rec.metadata.len()
28        );
29    }
30
31    let restored = reverse_preprocess(&preprocessed, &chain);
32    if restored == data {
33        eprintln!("Full pipeline roundtrip: OK");
34    } else {
35        eprintln!(
36            "Full pipeline roundtrip: FAILED ({} vs {} bytes)",
37            restored.len(),
38            data.len()
39        );
40        // Find first difference
41        for (i, (a, b)) in restored.iter().zip(data.iter()).enumerate() {
42            if a != b {
43                let start = i.saturating_sub(20);
44                let end = (i + 20).min(restored.len()).min(data.len());
45                eprintln!("  First diff at byte {}", i);
46                eprintln!(
47                    "  Restored: {:?}",
48                    String::from_utf8_lossy(&restored[start..end])
49                );
50                eprintln!(
51                    "  Original: {:?}",
52                    String::from_utf8_lossy(&data[start..end])
53                );
54                break;
55            }
56        }
57        if restored.len() != data.len() {
58            eprintln!("  Length diff: {} vs {}", restored.len(), data.len());
59        }
60    }
61
62    // Test Balanced mode (should pass)
63    let (preprocessed_bal, chain_bal) = preprocess(&data, format, Mode::Balanced);
64    let restored_bal = reverse_preprocess(&preprocessed_bal, &chain_bal);
65    if restored_bal == data {
66        eprintln!("Balanced pipeline roundtrip: OK");
67    } else {
68        eprintln!("Balanced pipeline roundtrip: FAILED");
69    }
70}