1use std::fmt;
9
10use serde::de::{Deserialize, Deserializer, MapAccess, SeqAccess, Visitor};
11use serde_saphyr::Spanned;
12
13#[derive(Debug, Clone, PartialEq)]
15pub struct Node {
16 pub kind: Kind,
17 pub line: usize,
18 pub column: usize,
19}
20
21#[derive(Debug, Clone, PartialEq)]
22pub enum Kind {
23 Null,
24 Bool(bool),
25 Int(i64),
26 UInt(u64),
27 Float(f64),
28 Str(String),
29 Seq(Vec<Node>),
30 Map(Vec<(Key, Node)>),
32}
33
34#[derive(Debug, Clone, PartialEq, Eq)]
36pub struct Key {
37 pub name: String,
38 pub line: usize,
39}
40
41#[derive(Debug, Clone, PartialEq, Eq)]
43pub struct SyntaxError {
44 pub line: Option<usize>,
45 pub message: String,
46}
47
48impl Kind {
49 pub fn name(&self) -> &'static str {
51 match self {
52 Kind::Null => "nothing",
53 Kind::Bool(_) => "a boolean",
54 Kind::Int(_) | Kind::UInt(_) | Kind::Float(_) => "a number",
55 Kind::Str(_) => "a string",
56 Kind::Seq(_) => "a list",
57 Kind::Map(_) => "a map",
58 }
59 }
60}
61
62impl Node {
63 pub fn null() -> Node {
64 Node {
65 kind: Kind::Null,
66 line: 0,
67 column: 0,
68 }
69 }
70
71 pub fn as_str(&self) -> Option<&str> {
72 match &self.kind {
73 Kind::Str(s) => Some(s),
74 _ => None,
75 }
76 }
77
78 pub fn as_map(&self) -> Option<&[(Key, Node)]> {
79 match &self.kind {
80 Kind::Map(m) => Some(m),
81 _ => None,
82 }
83 }
84
85 pub fn get(&self, key: &str) -> Option<&Node> {
87 self.as_map()?.iter().find(|(k, _)| k.name == key).map(|(_, v)| v)
88 }
89
90 pub fn entry(&self, key: &str) -> Option<&(Key, Node)> {
92 self.as_map()?.iter().find(|(k, _)| k.name == key)
93 }
94
95 pub fn to_json(&self) -> serde_json::Value {
98 use serde_json::Value as J;
99 match &self.kind {
100 Kind::Null => J::Null,
101 Kind::Bool(b) => J::Bool(*b),
102 Kind::Int(i) => J::from(*i),
103 Kind::UInt(u) => J::from(*u),
104 Kind::Float(f) => {
105 serde_json::Number::from_f64(*f).map_or_else(|| J::String(f.to_string()), J::Number)
106 }
107 Kind::Str(s) => J::String(s.clone()),
108 Kind::Seq(s) => J::Array(s.iter().map(Node::to_json).collect()),
109 Kind::Map(m) => J::Object(m.iter().map(|(k, v)| (k.name.clone(), v.to_json())).collect()),
110 }
111 }
112}
113
114pub fn parse(text: &str) -> Result<Node, SyntaxError> {
116 let mut options = serde_saphyr::options::Options::default();
117 options.strict_booleans = true;
118 options.duplicate_keys = serde_saphyr::options::DuplicateKeyPolicy::Error;
119 options.with_snippet = false;
120 options.reject_non_finite_typeless_float = false;
122 if text.trim().is_empty() {
123 return Ok(Node::null());
124 }
125 match serde_saphyr::from_str_with_options::<Spanned<Raw>>(text, options) {
126 Ok(raw) => {
127 let mut node: Node = raw.into();
128 if let Kind::Map(m) = &mut node.kind {
131 m.retain(|(k, _)| !k.name.starts_with("x-"));
132 }
133 Ok(node)
134 }
135 Err(e) => Err(syntax_error(&e)),
136 }
137}
138
139fn syntax_error(e: &serde_saphyr::Error) -> SyntaxError {
140 static LOCATION: std::sync::LazyLock<regex::Regex> = std::sync::LazyLock::new(|| {
141 regex::Regex::new(r"^(?:error: )?(?:line \d+ column \d+: )?(.*?)(?: at line \d+, column \d+)?$")
142 .unwrap()
143 });
144 let line = e.location().map(|l| l.line() as usize).filter(|l| *l > 0);
145 let text = e.to_string();
146 let first = text.lines().next().unwrap_or_default();
147 let message = LOCATION
149 .captures(first)
150 .map_or(first, |c| c.get(1).map_or(first, |m| m.as_str()));
151 let message = message.replace(", set DuplicateKeyPolicy in Options if acceptable", "");
152 let message = if message.contains("merge value must be") {
153 "invalid YAML merge key (`<<`): `<<` must be a map or a list of maps".to_string()
154 } else {
155 message
156 };
157 SyntaxError { line, message }
158}
159
160enum Raw {
162 Null,
163 Bool(bool),
164 Int(i64),
165 UInt(u64),
166 Float(f64),
167 Str(String),
168 Seq(Vec<Spanned<Raw>>),
169 Map(Vec<(Spanned<KeyRaw>, Spanned<Raw>)>),
170}
171
172struct KeyRaw(String);
174
175impl From<Spanned<Raw>> for Node {
176 fn from(s: Spanned<Raw>) -> Node {
177 let kind = match s.value {
178 Raw::Null => Kind::Null,
179 Raw::Bool(b) => Kind::Bool(b),
180 Raw::Int(i) => Kind::Int(i),
181 Raw::UInt(u) => Kind::UInt(u),
182 Raw::Float(f) => Kind::Float(f),
183 Raw::Str(t) => Kind::Str(t),
184 Raw::Seq(items) => Kind::Seq(items.into_iter().map(Node::from).collect()),
185 Raw::Map(entries) => Kind::Map(
186 entries
187 .into_iter()
188 .map(|(k, v)| {
189 let key = Key {
190 name: k.value.0,
191 line: k.referenced.line() as usize,
192 };
193 (key, Node::from(v))
194 })
195 .collect(),
196 ),
197 };
198 Node {
199 kind,
200 line: s.referenced.line() as usize,
201 column: s.referenced.column() as usize,
202 }
203 }
204}
205
206impl<'de> Deserialize<'de> for Raw {
207 fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Self, D::Error> {
208 struct V;
209 impl<'de> Visitor<'de> for V {
210 type Value = Raw;
211 fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
212 f.write_str("a YAML value")
213 }
214 fn visit_unit<E>(self) -> Result<Raw, E> {
215 Ok(Raw::Null)
216 }
217 fn visit_none<E>(self) -> Result<Raw, E> {
218 Ok(Raw::Null)
219 }
220 fn visit_some<D: Deserializer<'de>>(self, d: D) -> Result<Raw, D::Error> {
221 Raw::deserialize(d)
222 }
223 fn visit_bool<E>(self, v: bool) -> Result<Raw, E> {
224 Ok(Raw::Bool(v))
225 }
226 fn visit_i64<E>(self, v: i64) -> Result<Raw, E> {
227 Ok(Raw::Int(v))
228 }
229 fn visit_u64<E>(self, v: u64) -> Result<Raw, E> {
230 Ok(i64::try_from(v).map_or(Raw::UInt(v), Raw::Int))
231 }
232 fn visit_f64<E>(self, v: f64) -> Result<Raw, E> {
233 Ok(Raw::Float(v))
234 }
235 fn visit_str<E>(self, v: &str) -> Result<Raw, E> {
236 Ok(Raw::Str(v.to_string()))
237 }
238 fn visit_string<E>(self, v: String) -> Result<Raw, E> {
239 Ok(Raw::Str(v))
240 }
241 fn visit_seq<A: SeqAccess<'de>>(self, mut a: A) -> Result<Raw, A::Error> {
242 let mut items = Vec::new();
243 while let Some(x) = a.next_element()? {
244 items.push(x);
245 }
246 Ok(Raw::Seq(items))
247 }
248 fn visit_map<A: MapAccess<'de>>(self, mut a: A) -> Result<Raw, A::Error> {
249 let mut entries = Vec::new();
250 while let Some(e) = a.next_entry()? {
251 entries.push(e);
252 }
253 Ok(Raw::Map(entries))
254 }
255 }
256 d.deserialize_any(V)
257 }
258}
259
260pub const COMPLEX_KEY: &str = "a map key is a list or a map; keys must be names";
262
263impl<'de> Deserialize<'de> for KeyRaw {
264 fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Self, D::Error> {
265 struct V;
266 impl<'de> Visitor<'de> for V {
267 type Value = KeyRaw;
268 fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
269 f.write_str("a key")
270 }
271 fn visit_unit<E>(self) -> Result<KeyRaw, E> {
272 Ok(KeyRaw("null".into()))
273 }
274 fn visit_none<E>(self) -> Result<KeyRaw, E> {
275 Ok(KeyRaw("null".into()))
276 }
277 fn visit_bool<E>(self, v: bool) -> Result<KeyRaw, E> {
278 Ok(KeyRaw(v.to_string()))
279 }
280 fn visit_i64<E>(self, v: i64) -> Result<KeyRaw, E> {
281 Ok(KeyRaw(v.to_string()))
282 }
283 fn visit_u64<E>(self, v: u64) -> Result<KeyRaw, E> {
284 Ok(KeyRaw(v.to_string()))
285 }
286 fn visit_f64<E>(self, v: f64) -> Result<KeyRaw, E> {
287 Ok(KeyRaw(v.to_string()))
288 }
289 fn visit_str<E>(self, v: &str) -> Result<KeyRaw, E> {
290 Ok(KeyRaw(v.to_string()))
291 }
292 fn visit_string<E>(self, v: String) -> Result<KeyRaw, E> {
293 Ok(KeyRaw(v))
294 }
295 fn visit_seq<A: SeqAccess<'de>>(self, _: A) -> Result<KeyRaw, A::Error> {
296 Err(serde::de::Error::custom(COMPLEX_KEY))
297 }
298 fn visit_map<A: MapAccess<'de>>(self, _: A) -> Result<KeyRaw, A::Error> {
299 Err(serde::de::Error::custom(COMPLEX_KEY))
300 }
301 }
302 d.deserialize_any(V)
303 }
304}
305
306#[cfg(test)]
307mod tests {
308 use super::*;
309
310 #[test]
311 fn values_keep_their_lines() {
312 let n = parse("a:\n b: 1\n c: [x, y]\n").unwrap();
313 let a = n.get("a").unwrap();
314 assert_eq!(n.entry("a").unwrap().0.line, 1);
315 assert_eq!(a.entry("c").unwrap().0.line, 3);
316 assert_eq!(a.get("b").unwrap().kind, Kind::Int(1));
317 assert_eq!(a.get("b").unwrap().line, 2);
318 }
319
320 #[test]
321 fn only_true_and_false_are_booleans() {
322 let n = parse("a: yes\nb: no\nc: on\nd: true\n").unwrap();
323 assert_eq!(n.get("a").unwrap().as_str(), Some("yes"));
324 assert_eq!(n.get("c").unwrap().as_str(), Some("on"));
325 assert_eq!(n.get("d").unwrap().kind, Kind::Bool(true));
326 }
327
328 #[test]
329 fn merge_keys_are_expanded_and_written_keys_win() {
330 let n = parse("base: &b {type: duckdb, path: x}\ndev: {<<: *b, path: y}\n").unwrap();
331 let dev = n.get("dev").unwrap();
332 assert_eq!(dev.get("type").unwrap().as_str(), Some("duckdb"));
333 assert_eq!(dev.get("path").unwrap().as_str(), Some("y"));
334 }
335
336 #[test]
337 fn duplicate_keys_are_errors_with_their_line() {
338 let e = parse("a: 1\na: 2\n").unwrap_err();
339 assert_eq!(e.line, Some(2));
340 assert_eq!(e.message, "duplicate mapping key: a");
341 }
342
343 #[test]
344 fn syntax_errors_have_a_line() {
345 let e = parse("a:\n b: [1\n").unwrap_err();
346 assert_eq!(e.line, Some(2));
347 assert!(!e.message.starts_with("line"), "{}", e.message);
348 }
349
350 #[test]
351 fn scalar_keys_are_strings_and_empty_files_are_null() {
352 assert_eq!(
353 parse("2024: x\n").unwrap().get("2024").unwrap().as_str(),
354 Some("x")
355 );
356 assert_eq!(parse("\n# nothing\n").unwrap().kind, Kind::Null);
357 }
358
359 #[test]
360 fn jinja_in_quoted_values_is_a_string() {
361 let n = parse("pw: \"{{ env_var('X') }}\"\n").unwrap();
362 assert_eq!(n.get("pw").unwrap().as_str(), Some("{{ env_var('X') }}"));
363 }
364}
365
366#[cfg(test)]
367mod x_key_tests {
368 use super::*;
369
370 #[test]
371 fn top_level_x_keys_hold_anchors_and_are_dropped() {
372 let n = parse("x-dest: &dest {profile: s3, path: a.csv}\noutput:\n destination:\n <<: *dest\n path: b.csv\n").unwrap();
373 let j = n.to_json();
374 assert!(j.get("x-dest").is_none(), "{j}");
375 assert_eq!(
376 j["output"]["destination"],
377 serde_json::json!({"profile": "s3", "path": "b.csv"})
378 );
379 }
380}