| 1 | use super::{JsonPath, JsonPathToken, ParseError}; |
| 2 | use std::str::FromStr; |
| 3 | |
| 4 | impl<'a> JsonPath<'a> { |
| 5 | pub fn parse(path: &'a str) -> Result<Self, ParseError> { |
| 6 | let mut tokens = Vec::new(); |
| 7 | let mut i = 0; |
| 8 | let mut iter = path.chars().peekable(); |
| 9 | while let Some(c) = iter.next() { |
| 10 | i += c.len_utf8(); |
| 11 | match c { |
| 12 | '$' => tokens.push(JsonPathToken::Root), |
| 13 | '@' => tokens.push(JsonPathToken::Current), |
| 14 | '.' => { |
| 15 | let c = iter.peek(); |
| 16 | match c { |
| 17 | Some('.') => { |
| 18 | // '..' => recursive descent |
| 19 | tokens.push(JsonPathToken::RecursiveDescend); |
| 20 | iter.next(); |
| 21 | i += 1; |
| 22 | } |
| 23 | Some('*') => { |
| 24 | // '.*' => wildcard operator |
| 25 | tokens.push(JsonPathToken::Wildcard); |
| 26 | iter.next(); |
| 27 | i += 1; |
| 28 | } |
| 29 | Some(a) if a.is_alphabetic() => { |
| 30 | // '.{name}' => field name |
| 31 | let start = i; |
| 32 | while let Some(a) = iter.peek() { |
| 33 | if a.is_alphanumeric() || a == &'_' { |
| 34 | i += a.len_utf8(); |
| 35 | iter.next(); |
| 36 | } else { |
| 37 | break; |
| 38 | } |
| 39 | } |
| 40 | let end = i; |
| 41 | let member = &path[start..end]; |
| 42 | tokens.push(JsonPathToken::Member(member)); |
| 43 | } |
| 44 | Some(a) => { |
| 45 | return Err(invalid_char(*a, path)); |
| 46 | } |
| 47 | None => { |
| 48 | return Err(ParseError::InvalidJsonPath(format!( |
| 49 | "Path cannot end with '.': `{}`", |
| 50 | path |
| 51 | ))) |
| 52 | } |
| 53 | } |
| 54 | } |
| 55 | '[' => { |
| 56 | // get all values between '[..]' braces |
| 57 | let start = i; |
| 58 | let mut slice = &path[i..]; |
| 59 | let mut quote_start = false; |
| 60 | let mut nesting = 1; |
| 61 | for c in iter.by_ref() { |
| 62 | i += c.len_utf8(); |
| 63 | if c == '[' { |
| 64 | nesting += 1; |
| 65 | } |
| 66 | if c == ']' { |
| 67 | nesting -= 1; |
| 68 | if nesting == 0 && !quote_start { |
| 69 | slice = &slice[..(i - start - 1)]; |
| 70 | break; |
| 71 | } |
| 72 | } |
| 73 | if c == '\'' { |
| 74 | quote_start = !quote_start; |
| 75 | } |
| 76 | } |
| 77 | if slice == "*" { |
| 78 | // '[*]' => wildcard operator |
| 79 | tokens.push(JsonPathToken::Wildcard); |
| 80 | } else if let Ok(index) = slice.parse::<i32>() { |
| 81 | // '[{number}]' => array index |
| 82 | tokens.push(JsonPathToken::Index(index)); |
| 83 | } else if slice.starts_with('?') { |
| 84 | return Err(ParseError::InvalidJsonPath(format!( |
| 85 | "JSON Path predicate expressions are not supported yet: `{}`", |
| 86 | slice |
| 87 | ))); |
| 88 | } else if slice.contains(':') { |
| 89 | // '[{?from}:{?to}:{?by}]' => slice operator |
| 90 | let mut split = slice.split(':'); |
| 91 | let start = split |
| 92 | .next() |
| 93 | .and_then(|s| u32::from_str(s).ok()) |
| 94 | .unwrap_or(0); |
| 95 | let end = split |
| 96 | .next() |
| 97 | .and_then(|s| u32::from_str(s).ok()) |
| 98 | .unwrap_or(u32::MAX); |
| 99 | let by = split |
| 100 | .next() |
| 101 | .and_then(|s| u32::from_str(s).ok()) |
| 102 | .unwrap_or(1); |
| 103 | tokens.push(JsonPathToken::Slice(start, end, by)); |
| 104 | } else if slice.contains(',') { |
| 105 | // either member union ['a','b','c'] or index union [1, 2, 3] |
| 106 | let subslices = slice.split(','); |
| 107 | let mut i = subslices.into_iter(); |
| 108 | if let Some(n) = i.next() { |
| 109 | let mut n = n.trim(); |
| 110 | if n.starts_with('\'') && n.ends_with('\'') { |
| 111 | // member union |
| 112 | let mut members = Vec::new(); |
| 113 | while { |
| 114 | members.push(&n[1..(n.len() - 1)]); // remove ' on both ends |
| 115 | match i.next() { |
| 116 | None => false, |
| 117 | Some(member) => { |
| 118 | n = member.trim(); |
| 119 | if n.starts_with('\'') && n.ends_with('\'') { |
| 120 | true |
| 121 | } else { |
| 122 | return Err(ParseError::InvalidJsonPath(format!( |
| 123 | "substring `{}` is missing pairing `'` symbol", |
| 124 | n |
| 125 | ))); |
| 126 | } |
| 127 | } |
| 128 | } |
| 129 | } {} |
| 130 | tokens.push(JsonPathToken::MemberUnion(members)) |
| 131 | } else { |
| 132 | // index union |
| 133 | let mut indices = Vec::new(); |
| 134 | while { |
| 135 | match n.parse::<i32>() { |
| 136 | Ok(i) => indices.push(i), |
| 137 | Err(err) => return Err(ParseError::InvalidJsonPath(format!("substring `{}` doesn't contain valid union of keys or indices: {}", slice, err))), |
| 138 | } |
| 139 | match i.next() { |
| 140 | None => false, |
| 141 | Some(index) => { |
| 142 | n = index.trim(); |
| 143 | true |
| 144 | } |
| 145 | } |
| 146 | } {} |
| 147 | tokens.push(JsonPathToken::IndexUnion(indices)) |
| 148 | } |
| 149 | } |
| 150 | } else if slice.starts_with('\'') && slice.ends_with('\'') { |
| 151 | // "['name with spaces']" => field name with spaces |
| 152 | let member = &slice[1..slice.len() - 1]; |
| 153 | tokens.push(JsonPathToken::Member(member)); |
| 154 | } else { |
| 155 | return Err(ParseError::InvalidJsonPath(format!( |
| 156 | "substring `{}` is not supported: `{}`", |
| 157 | slice, path |
| 158 | ))); |
| 159 | } |
| 160 | } |
| 161 | '*' => tokens.push(JsonPathToken::Wildcard), |
| 162 | c if c.is_alphabetic() => { |
| 163 | // handle cases like `..name`, `$name` or `@name` |
| 164 | let start = i - c.len_utf8(); |
| 165 | while let Some(a) = iter.peek() { |
| 166 | if a.is_alphanumeric() || a == &'_' { |
| 167 | i += a.len_utf8(); |
| 168 | iter.next(); |
| 169 | } else { |
| 170 | break; |
| 171 | } |
| 172 | } |
| 173 | let end = i; |
| 174 | let member = &path[start..end]; |
| 175 | tokens.push(JsonPathToken::Member(member)); |
| 176 | } |
| 177 | _ => { |
| 178 | return Err(invalid_char(c, path)); |
| 179 | } |
| 180 | } |
| 181 | } |
| 182 | Ok(JsonPath { tokens }) |
| 183 | } |
| 184 | } |
| 185 | |
| 186 | fn invalid_char(c: char, path: &str) -> ParseError { |
| 187 | ParseError::InvalidJsonPath(format!("Invalid character `{}` in path: `{}`", c, path)) |
| 188 | } |
| 189 | |
| 190 | #[cfg(test)] |
| 191 | mod test { |
| 192 | use crate::json_path::parse::{JsonPath, JsonPathToken}; |
| 193 | |
| 194 | #[test] |
| 195 | fn parse_root() { |
| 196 | let path = JsonPath::parse("$").unwrap(); |
| 197 | assert_eq!(path.tokens, vec![JsonPathToken::Root]); |
| 198 | } |
| 199 | |
| 200 | #[test] |
| 201 | fn parse_current() { |
| 202 | let path = JsonPath::parse("@").unwrap(); |
| 203 | assert_eq!(path.tokens, vec![JsonPathToken::Current]); |
| 204 | } |
| 205 | |
| 206 | #[test] |
| 207 | fn parse_wildcard() { |
| 208 | let path = JsonPath::parse("$.*").unwrap(); |
| 209 | assert_eq!( |
| 210 | path.tokens, |
| 211 | vec![JsonPathToken::Root, JsonPathToken::Wildcard] |
| 212 | ); |
| 213 | } |
| 214 | |
| 215 | #[test] |
| 216 | fn parse_wildcard_quoted() { |
| 217 | let path = JsonPath::parse("$[*]").unwrap(); |
| 218 | assert_eq!( |
| 219 | path.tokens, |
| 220 | vec![JsonPathToken::Root, JsonPathToken::Wildcard] |
| 221 | ); |
| 222 | } |
| 223 | |
| 224 | #[test] |
| 225 | fn parse_descend() { |
| 226 | let path = JsonPath::parse("$..").unwrap(); |
| 227 | assert_eq!( |
| 228 | path.tokens, |
| 229 | vec![JsonPathToken::Root, JsonPathToken::RecursiveDescend] |
| 230 | ); |
| 231 | } |
| 232 | |
| 233 | #[test] |
| 234 | fn parse_descend_gradual() { |
| 235 | let path = JsonPath::parse("$..name").unwrap(); |
| 236 | assert_eq!( |
| 237 | path.tokens, |
| 238 | vec![ |
| 239 | JsonPathToken::Root, |
| 240 | JsonPathToken::RecursiveDescend, |
| 241 | JsonPathToken::Member("name") |
| 242 | ] |
| 243 | ); |
| 244 | } |
| 245 | |
| 246 | #[test] |
| 247 | fn parse_dot_member() { |
| 248 | let path = JsonPath::parse("$.key").unwrap(); |
| 249 | assert_eq!( |
| 250 | path.tokens, |
| 251 | vec![JsonPathToken::Root, JsonPathToken::Member("key")] |
| 252 | ); |
| 253 | } |
| 254 | |
| 255 | #[test] |
| 256 | fn parse_quote_member() { |
| 257 | let path = JsonPath::parse("$['key']").unwrap(); |
| 258 | assert_eq!( |
| 259 | path.tokens, |
| 260 | vec![JsonPathToken::Root, JsonPathToken::Member("key")] |
| 261 | ); |
| 262 | } |
| 263 | |
| 264 | #[test] |
| 265 | fn parse_index() { |
| 266 | let path = JsonPath::parse("$[0]").unwrap(); |
| 267 | assert_eq!( |
| 268 | path.tokens, |
| 269 | vec![JsonPathToken::Root, JsonPathToken::Index(0)] |
| 270 | ); |
| 271 | } |
| 272 | |
| 273 | #[test] |
| 274 | fn parse_negative_index() { |
| 275 | let path = JsonPath::parse("$[-3]").unwrap(); |
| 276 | assert_eq!( |
| 277 | path.tokens, |
| 278 | vec![JsonPathToken::Root, JsonPathToken::Index(-3)] |
| 279 | ); |
| 280 | } |
| 281 | |
| 282 | #[test] |
| 283 | fn parse_slice_bounded() { |
| 284 | let path = JsonPath::parse("$[1:3]").unwrap(); |
| 285 | assert_eq!( |
| 286 | path.tokens, |
| 287 | vec![JsonPathToken::Root, JsonPathToken::Slice(1, 3, 1)] |
| 288 | ); |
| 289 | } |
| 290 | |
| 291 | #[test] |
| 292 | fn parse_slice_left_unbounded() { |
| 293 | let path = JsonPath::parse("$[:3]").unwrap(); |
| 294 | assert_eq!( |
| 295 | path.tokens, |
| 296 | vec![JsonPathToken::Root, JsonPathToken::Slice(0, 3, 1)] |
| 297 | ); |
| 298 | } |
| 299 | |
| 300 | #[test] |
| 301 | fn parse_slice_right_unbounded() { |
| 302 | let path = JsonPath::parse("$[3:]").unwrap(); |
| 303 | assert_eq!( |
| 304 | path.tokens, |
| 305 | vec![JsonPathToken::Root, JsonPathToken::Slice(3, u32::MAX, 1)] |
| 306 | ); |
| 307 | } |
| 308 | |
| 309 | #[test] |
| 310 | fn parse_slice_with_step() { |
| 311 | let path = JsonPath::parse("$[3::2]").unwrap(); |
| 312 | assert_eq!( |
| 313 | path.tokens, |
| 314 | vec![JsonPathToken::Root, JsonPathToken::Slice(3, u32::MAX, 2)] |
| 315 | ); |
| 316 | let path = JsonPath::parse("$[3:5:2]").unwrap(); |
| 317 | assert_eq!( |
| 318 | path.tokens, |
| 319 | vec![JsonPathToken::Root, JsonPathToken::Slice(3, 5, 2)] |
| 320 | ); |
| 321 | let path = JsonPath::parse("$[:3:2]").unwrap(); |
| 322 | assert_eq!( |
| 323 | path.tokens, |
| 324 | vec![JsonPathToken::Root, JsonPathToken::Slice(0, 3, 2)] |
| 325 | ); |
| 326 | } |
| 327 | |
| 328 | #[test] |
| 329 | fn parse_member_union() { |
| 330 | let path = JsonPath::parse("$.users['a', 'bc', '']").unwrap(); |
| 331 | assert_eq!( |
| 332 | path.tokens, |
| 333 | vec![ |
| 334 | JsonPathToken::Root, |
| 335 | JsonPathToken::Member("users"), |
| 336 | JsonPathToken::MemberUnion(vec!["a", "bc", ""]) |
| 337 | ] |
| 338 | ); |
| 339 | } |
| 340 | |
| 341 | #[test] |
| 342 | fn parse_index_union() { |
| 343 | let path = JsonPath::parse("$.users[0, 123, -1]").unwrap(); |
| 344 | assert_eq!( |
| 345 | path.tokens, |
| 346 | vec![ |
| 347 | JsonPathToken::Root, |
| 348 | JsonPathToken::Member("users"), |
| 349 | JsonPathToken::IndexUnion(vec![0, 123, -1]) |
| 350 | ] |
| 351 | ); |
| 352 | } |
| 353 | |
| 354 | #[test] |
| 355 | fn parse_complex() { |
| 356 | let path = JsonPath::parse("$.key_1[0].key2[*]").unwrap(); |
| 357 | assert_eq!( |
| 358 | path.tokens, |
| 359 | vec![ |
| 360 | JsonPathToken::Root, |
| 361 | JsonPathToken::Member("key_1"), |
| 362 | JsonPathToken::Index(0), |
| 363 | JsonPathToken::Member("key2"), |
| 364 | JsonPathToken::Wildcard |
| 365 | ] |
| 366 | ); |
| 367 | } |
| 368 | |
| 369 | #[test] |
| 370 | fn parse_predicate_error() { |
| 371 | let res = JsonPath::parse("$.users[?(@['name'] == 'Alice')].surname"); |
| 372 | assert!(res.is_err()); |
| 373 | let err = res.unwrap_err(); |
| 374 | assert!(err.to_string().contains("?(@['name'] == 'Alice')")); |
| 375 | } |
| 376 | } |
| 377 | |
amadan