dataprof_core/
partial.rs

1use crate::{classification::DataType, source::FileFormat};
2
3/// Result of fast schema inference — column names paired with inferred data types.
4#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
5pub struct SchemaResult {
6    /// Columns with their inferred types. For CSV/Parquet the order matches
7    /// the source; for JSON/JSONL columns are sorted alphabetically.
8    pub columns: Vec<ColumnSchema>,
9    /// How many rows were sampled to infer the schema (0 for Parquet metadata).
10    pub rows_sampled: usize,
11    /// Time taken for inference in milliseconds.
12    pub inference_time_ms: u128,
13    /// `true` when the entire file was consumed or schema was read from
14    /// metadata; `false` when inference stopped at the sample-size cap and
15    /// the schema may not have fully stabilized.
16    pub schema_stable: bool,
17}
18
19/// A single column's name and inferred data type.
20#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
21pub struct ColumnSchema {
22    pub name: String,
23    pub data_type: DataType,
24}
25
26/// Lightweight structural summary for a file.
27///
28/// This intentionally sits between `quick_row_count()` / `infer_schema()` and a
29/// full profile: it reports shape and sampled per-column counters, not quality
30/// scores, pattern detection, recommendations, or raw values.
31#[derive(Debug, Clone, PartialEq, serde::Serialize, serde::Deserialize)]
32pub struct StructureReport {
33    /// Source identifier, currently the file path passed by the caller.
34    pub source: String,
35    /// Detected or explicitly provided file format.
36    pub format: FileFormat,
37    /// Exact or estimated row count for the source.
38    pub row_count: RowCountEstimate,
39    /// Number of rows consumed for the structural column summaries.
40    pub rows_sampled: usize,
41    /// True when the structural pass consumed the whole source.
42    pub source_exhausted: bool,
43    /// True when the structural pass stopped at the row sample cap.
44    pub truncated: bool,
45    /// Human-readable reason for truncation, if any.
46    pub truncation_reason: Option<String>,
47    /// CSV delimiter used for parsing, when applicable.
48    pub delimiter: Option<String>,
49    /// Per-column structural summaries.
50    pub columns: Vec<StructureColumnSummary>,
51    /// Non-fatal caveats, such as estimated row counts.
52    pub warnings: Vec<String>,
53}
54
55/// Sample-derived or metadata-derived structural summary for one column.
56#[derive(Debug, Clone, PartialEq, serde::Serialize, serde::Deserialize)]
57pub struct StructureColumnSummary {
58    pub name: String,
59    pub data_type: DataType,
60    pub total_count: Option<usize>,
61    pub null_count: Option<usize>,
62    pub null_ratio: Option<f64>,
63    pub unique_count: Option<usize>,
64    pub uniqueness_ratio: Option<f64>,
65    pub distinct_count_approximate: Option<bool>,
66    /// `"sample"` for CSV/JSON/JSONL, `"metadata"` for Parquet.
67    pub provenance: String,
68}
69
70/// Result of a quick row count operation.
71#[derive(Debug, Clone, PartialEq, serde::Serialize, serde::Deserialize)]
72pub struct RowCountEstimate {
73    /// The estimated or exact row count.
74    pub count: u64,
75    /// Whether the count is exact or an estimate.
76    pub exact: bool,
77    /// How the count was obtained.
78    pub method: CountMethod,
79    /// Time taken in milliseconds.
80    pub count_time_ms: u128,
81}
82
83/// Method used to obtain the row count.
84#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
85pub enum CountMethod {
86    /// Read from Parquet file footer metadata (exact, zero row reading).
87    ParquetMetadata,
88    /// Full scan of the file (exact).
89    FullScan,
90    /// Sample-based estimation (approximate).
91    Sampling,
92    /// Full scan of a streaming source (no file metadata available).
93    StreamFullScan,
94}
dataprof_core/partial.rs

dataprof_core/
partial.rs
