Skip to main content

dre_core/
preflight.rs

1//! Jinja pre-flight: syntax checks and static scans of `var()`, `env_var()` and `run.*` call
2//! sites. Nothing is rendered with live values here.
3
4use std::sync::LazyLock;
5
6use regex::Regex;
7
8/// `run.*` attributes the runtime context provides.
9pub const RUN_ATTRS: &[&str] = &[
10    "report",
11    "set",
12    "target",
13    "profile",
14    "source_type",
15    "schedule",
16    "date",
17    "date_format",
18    "now",
19    "timezone",
20];
21/// Attributes and methods of `run.date`.
22pub use crate::dates::DATE_ATTRS;
23
24static SEGMENT: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"(?s)\{\{.*?\}\}|\{%.*?%\}").unwrap());
25static VAR: LazyLock<Regex> =
26    LazyLock::new(|| Regex::new(r#"(?:^|[^\w.])(env_var|var)\s*\(\s*(['"])([^'"]+)['"]\s*([,)])"#).unwrap());
27static RUN: LazyLock<Regex> =
28    LazyLock::new(|| Regex::new(r"(?:^|[^\w.])run\.([A-Za-z_]\w*)(?:\.([A-Za-z_]\w*))?").unwrap());
29
30static MACRO: LazyLock<Regex> = LazyLock::new(|| {
31    Regex::new(r"(?s)\{%-?\s*macro\s+([A-Za-z_]\w*)\s*\(.*?%\}(.*?)\{%-?\s*endmacro\s*-?%\}").unwrap()
32});
33static CALL: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"(?:^|[^\w.])([A-Za-z_]\w*)\s*\(").unwrap());
34
35static REF: LazyLock<Regex> =
36    LazyLock::new(|| Regex::new(r#"(?:^|[^\w.])ref\s*\(\s*['"]([^'"]+)['"]\s*\)"#).unwrap());
37
38/// Every `ref('name')` with a literal name inside Jinja blocks, with its 1-based line.
39pub fn refs(src: &str) -> Vec<(String, usize)> {
40    let mut out = Vec::new();
41    for seg in SEGMENT.find_iter(src) {
42        for c in REF.captures_iter(seg.as_str()) {
43            out.push((
44                c[1].to_string(),
45                line_at(src, seg.start() + c.get(1).unwrap().start()),
46            ));
47        }
48    }
49    out
50}
51
52/// A macro defined in a macro file.
53#[derive(Debug, Clone)]
54pub struct MacroDef {
55    pub name: String,
56    /// The macro's body.
57    pub body: String,
58    /// 0-based line offset of the body within its file.
59    pub line_offset: usize,
60}
61
62/// Every `{% macro name(...) %} ... {% endmacro %}` in a source.
63pub fn macro_defs(src: &str) -> Vec<MacroDef> {
64    MACRO
65        .captures_iter(src)
66        .map(|c| {
67            let body = c.get(2).unwrap();
68            MacroDef {
69                name: c[1].to_string(),
70                body: body.as_str().to_string(),
71                line_offset: line_at(src, body.start()) - 1,
72            }
73        })
74        .collect()
75}
76
77/// Names called like functions inside Jinja blocks (`name(`).
78pub fn called_names(src: &str) -> Vec<String> {
79    let mut out = Vec::new();
80    for seg in SEGMENT.find_iter(src) {
81        for c in CALL.captures_iter(seg.as_str()) {
82            out.push(c[1].to_string());
83        }
84    }
85    out
86}
87
88/// Whether a string contains Jinja at all.
89pub fn is_templated(s: &str) -> bool {
90    s.contains("{{") || s.contains("{%")
91}
92
93/// Compile-check a template; returns `(line, message)` on a syntax error.
94pub fn check_syntax(name: &str, src: &str) -> Result<(), (Option<usize>, String)> {
95    let env = minijinja::Environment::new();
96    match env.template_from_named_str(name, src) {
97        Ok(_) => Ok(()),
98        Err(e) => Err((e.line(), e.detail().unwrap_or("syntax error").to_string())),
99    }
100}
101
102#[derive(Debug, Clone, PartialEq, Eq)]
103pub struct Call {
104    /// `var` or `env_var`.
105    pub func: &'static str,
106    pub name: String,
107    pub has_default: bool,
108    /// 1-based line within the scanned source.
109    pub line: usize,
110}
111
112fn line_at(src: &str, offset: usize) -> usize {
113    src[..offset].matches('\n').count() + 1
114}
115
116/// Every `var('x')` / `env_var('X')` call inside Jinja blocks.
117pub fn calls(src: &str) -> Vec<Call> {
118    let mut out = Vec::new();
119    for seg in SEGMENT.find_iter(src) {
120        for c in VAR.captures_iter(seg.as_str()) {
121            let m = c.get(1).unwrap();
122            out.push(Call {
123                func: if m.as_str() == "var" { "var" } else { "env_var" },
124                name: c[3].to_string(),
125                has_default: &c[4] == ",",
126                line: line_at(src, seg.start() + m.start()),
127            });
128        }
129    }
130    out
131}
132
133/// `run.*` references inside Jinja blocks that aren't part of the runtime context.
134pub fn unknown_run_refs(src: &str) -> Vec<(String, usize)> {
135    let mut out = Vec::new();
136    for seg in SEGMENT.find_iter(src) {
137        for c in RUN.captures_iter(seg.as_str()) {
138            let attr = &c[1];
139            let line = line_at(src, seg.start() + c.get(0).unwrap().start());
140            if !RUN_ATTRS.contains(&attr) {
141                out.push((format!("run.{attr}"), line));
142            } else if attr == "date"
143                && let Some(sub) = c.get(2)
144                && !DATE_ATTRS.contains(&sub.as_str())
145            {
146                out.push((format!("run.date.{}", sub.as_str()), line));
147            }
148        }
149    }
150    out
151}
152
153#[cfg(test)]
154mod tests {
155    use super::*;
156
157    #[test]
158    fn finds_var_calls_with_and_without_defaults() {
159        let src = "select {{ var('a') }},\n {{ var(\"b\", 1) }}, {{ env_var('HOME') }} from t";
160        let c = calls(src);
161        assert_eq!(c.len(), 3);
162        assert_eq!((c[0].name.as_str(), c[0].has_default, c[0].line), ("a", false, 1));
163        assert_eq!((c[1].name.as_str(), c[1].has_default, c[1].line), ("b", true, 2));
164        assert_eq!((c[2].func, c[2].name.as_str()), ("env_var", "HOME"));
165    }
166
167    #[test]
168    fn ignores_sql_outside_jinja_and_method_like_names() {
169        assert!(calls("select var('x') from t").is_empty());
170        assert!(calls("{{ my.var('x') }} {{ env_var_x('y') }}").is_empty());
171        assert!(unknown_run_refs("select run.dat from runs run").is_empty());
172    }
173
174    #[test]
175    fn flags_unknown_run_attributes() {
176        let src = "{{ run.report }} {{ run.dat }}\n{{ run.date.yyyymmdd }} {{ run.date.yymm }}";
177        assert_eq!(
178            unknown_run_refs(src),
179            vec![("run.dat".to_string(), 1), ("run.date.yymm".to_string(), 2)]
180        );
181    }
182}