sheeter_engine/
inspect.rs

1use crate::{CanonicalValue, Format, SqlDialect};
2use std::error::Error;
3
4/// Prints a structural summary of an input document.
5///
6/// Reports the format, the number of top-level documents (sheets), and for
7/// each sheet the number of rows and columns with inferred types.
8///
9/// # Examples
10///
11/// ```
12/// use sheeter_engine::{inspect, Format};
13///
14/// let report = inspect(b"name,age\nAlice,30\nBob,25\n", Format::Csv).unwrap();
15/// assert!(report.contains("rows: 2"));
16/// assert!(report.contains("name"));
17/// ```
18pub fn inspect(input: &[u8], format: Format) -> Result<String, Box<dyn Error>> {
19    let sheets = load_sheets(input, format)?;
20    let mut out = String::new();
21    out.push_str(&format!("format: {}\n", format_name(format)));
22    out.push_str(&format!("documents: {}\n", sheets.len()));
23    for (name, value) in &sheets {
24        describe_sheet(&mut out, name, value);
25    }
26    Ok(out)
27}
28
29fn format_name(f: Format) -> &'static str {
30    match f {
31        Format::Csv => "csv",
32        Format::Tsv => "tsv",
33        Format::Json => "json",
34        Format::Yaml => "yaml",
35        Format::Xml => "xml",
36        Format::Toml => "toml",
37        Format::Xlsx => "xlsx",
38        Format::Sql => "sql",
39        Format::Ndjson => "ndjson",
40        Format::Ods => "ods",
41    }
42}
43
44fn load_sheets(
45    input: &[u8],
46    format: Format,
47) -> Result<Vec<(String, CanonicalValue)>, Box<dyn Error>> {
48    match format {
49        Format::Xlsx => CanonicalValue::from_xlsx_workbook(input),
50        Format::Sql => {
51            let text = std::str::from_utf8(input)?;
52            CanonicalValue::from_sql_workbook(text, SqlDialect::Generic)
53        }
54        Format::Ndjson => {
55            let text = std::str::from_utf8(input)?;
56            Ok(vec![(
57                "Sheet1".to_string(),
58                CanonicalValue::from_ndjson(text)?,
59            )])
60        }
61        _ => {
62            let text = std::str::from_utf8(input)?;
63            let hub = match format {
64                Format::Csv => CanonicalValue::from_csv(text)?,
65                Format::Tsv => CanonicalValue::from_tsv(text)?,
66                Format::Json => CanonicalValue::from_json(text)?,
67                Format::Yaml => CanonicalValue::from_yaml(text)?,
68                Format::Xml => CanonicalValue::from_xml(text)?,
69                Format::Toml => CanonicalValue::from_toml(text)?,
70                Format::Sql | Format::Ndjson | Format::Xlsx | Format::Ods => unreachable!(),
71            };
72            Ok(vec![("Sheet1".to_string(), hub)])
73        }
74    }
75}
76
77fn describe_sheet(out: &mut String, name: &str, value: &CanonicalValue) {
78    out.push_str(&format!("  {name}:\n"));
79    match value {
80        CanonicalValue::Array(rows) => {
81            let cols = collect_headers(rows);
82            out.push_str(&format!("    rows: {}\n", rows.len()));
83            out.push_str(&format!("    columns: {}\n", cols.len()));
84            for (i, col) in cols.iter().enumerate() {
85                let ty = infer_type(rows, col);
86                out.push_str(&format!("      {}: {} ({})\n", i + 1, col, ty));
87            }
88        }
89        CanonicalValue::Object(o) => {
90            out.push_str(&format!("    fields: {}\n", o.len()));
91            for (k, v) in o {
92                out.push_str(&format!("      {k}: {}\n", infer_scalar(v)));
93            }
94        }
95        CanonicalValue::String(s) => {
96            out.push_str(&format!("    scalar: string ({} bytes)\n", s.len()));
97        }
98        CanonicalValue::Bytes(b) => {
99            out.push_str(&format!("    scalar: binary ({} bytes)\n", b.len()));
100        }
101        CanonicalValue::Number(n) => out.push_str(&format!("    scalar: number ({n})\n")),
102        CanonicalValue::Bool(b) => out.push_str(&format!("    scalar: bool ({b})\n")),
103        CanonicalValue::Null => out.push_str("    scalar: null\n"),
104    }
105}
106
107fn collect_headers(rows: &[CanonicalValue]) -> Vec<String> {
108    let mut out: Vec<String> = Vec::new();
109    for row in rows {
110        if let CanonicalValue::Object(o) = row {
111            for (k, _) in o {
112                if !out.contains(k) {
113                    out.push(k.clone());
114                }
115            }
116        }
117    }
118    out
119}
120
121fn infer_type(rows: &[CanonicalValue], col: &str) -> &'static str {
122    let mut has_int = false;
123    let mut has_float = false;
124    let mut has_str = false;
125    let mut has_bool = false;
126    let mut has_null = false;
127    for row in rows {
128        if let CanonicalValue::Object(o) = row {
129            match o.iter().find(|(k, _)| k == col).map(|(_, v)| v) {
130                Some(CanonicalValue::Number(n)) => {
131                    if n.is_i64() || n.is_u64() {
132                        has_int = true;
133                    } else {
134                        has_float = true;
135                    }
136                }
137                Some(CanonicalValue::String(_)) => has_str = true,
138                Some(CanonicalValue::Bool(_)) => has_bool = true,
139                Some(CanonicalValue::Null) | None => has_null = true,
140                _ => has_str = true,
141            }
142        }
143    }
144    match (has_int, has_float, has_str, has_bool, has_null) {
145        (true, false, false, false, _) => "integer",
146        (_, true, false, false, _) => "number",
147        (false, false, false, true, _) => "boolean",
148        (false, false, true, false, _) => "string",
149        (false, false, false, false, true) => "null",
150        _ => "mixed",
151    }
152}
153
154fn infer_scalar(v: &CanonicalValue) -> &'static str {
155    match v {
156        CanonicalValue::Null => "null",
157        CanonicalValue::Bool(_) => "boolean",
158        CanonicalValue::Number(n) if n.is_i64() || n.is_u64() => "integer",
159        CanonicalValue::Number(_) => "number",
160        CanonicalValue::String(_) => "string",
161        CanonicalValue::Bytes(_) => "binary",
162        CanonicalValue::Array(_) => "array",
163        CanonicalValue::Object(_) => "object",
164    }
165}
sheeter_engine/inspect.rs

sheeter_engine/
inspect.rs
