| 1 | //! Just enough YAML for a gem's `metadata.gz`: the `Gem::Specification` |
| 2 | //! RubyGems writes with Psych. Block mappings and sequences (including a |
| 3 | //! sequence at its key's indent, and `- - ">="` nested ones), plain and |
| 4 | //! quoted scalars wrapped over lines, `|` and `>` block scalars, and empty |
| 5 | //! flow collections. Tags (`!ruby/object:Gem::Version`) are dropped: the |
| 6 | //! value is read as the plain mapping or scalar under them. |
| 7 | |
| 8 | use serde_json::{Map, Value}; |
| 9 | |
| 10 | #[derive(Clone, Debug)] |
| 11 | struct Line { |
| 12 | indent: usize, |
| 13 | text: String, |
| 14 | } |
| 15 | |
| 16 | /// Reads a document into JSON values: mappings as objects, sequences as |
| 17 | /// arrays, scalars as strings, and empty values as null. |
| 18 | pub fn parse(text: &str) -> Result<Value, String> { |
| 19 | let mut lines = Vec::new(); |
| 20 | for raw in text.lines() { |
| 21 | let trimmed = raw.trim_end(); |
| 22 | let indent = trimmed.len() - trimmed.trim_start().len(); |
| 23 | let content = trimmed.trim_start(); |
| 24 | if content.starts_with('#') || content == "..." { |
| 25 | continue; |
| 26 | } |
| 27 | if indent == 0 && content.starts_with("---") { |
| 28 | let after = strip_tag(content[3..].trim()); |
| 29 | if !after.is_empty() { |
| 30 | lines.push(Line { indent: 0, text: after.to_owned() }); |
| 31 | } |
| 32 | continue; |
| 33 | } |
| 34 | lines.push(Line { indent, text: content.to_owned() }); |
| 35 | } |
| 36 | let mut parser = Parser { lines, at: 0 }; |
| 37 | parser.skip_blank(); |
| 38 | if parser.at >= parser.lines.len() { |
| 39 | return Ok(Value::Null); |
| 40 | } |
| 41 | let indent = parser.lines[parser.at].indent; |
| 42 | parser.node(indent) |
| 43 | } |
| 44 | |
| 45 | /// The text after a leading tag: `!ruby/object:Gem::Version` alone is empty. |
| 46 | fn strip_tag(text: &str) -> &str { |
| 47 | if text.starts_with('!') { |
| 48 | text.split_once(' ').map_or("", |(_, rest)| rest.trim_start()) |
| 49 | } else { |
| 50 | text |
| 51 | } |
| 52 | } |
| 53 | |
| 54 | /// Where a mapping key ends: the `:` followed by a space or the end of the |
| 55 | /// line, outside quotes. |
| 56 | fn key_end(text: &str) -> Option<usize> { |
| 57 | if text.starts_with('"') || text.starts_with('\'') { |
| 58 | let quote = text.as_bytes()[0] as char; |
| 59 | let close = text[1..].find(quote)? + 1; |
| 60 | return text[close + 1..].starts_with(':').then_some(close + 1).filter(|at| { |
| 61 | let after = &text[at + 1..]; |
| 62 | after.is_empty() || after.starts_with(' ') |
| 63 | }); |
| 64 | } |
| 65 | if text.starts_with('-') && (text.len() == 1 || text.as_bytes()[1] == b' ') { |
| 66 | return None; |
| 67 | } |
| 68 | let bytes = text.as_bytes(); |
| 69 | (0..bytes.len()).find(|&i| bytes[i] == b':' && (i + 1 == bytes.len() || bytes[i + 1] == b' ')) |
| 70 | } |
| 71 | |
| 72 | fn unquote_key(key: &str) -> String { |
| 73 | let key = key.trim(); |
| 74 | match scalar(key) { |
| 75 | Value::String(s) => s, |
| 76 | _ => key.to_owned(), |
| 77 | } |
| 78 | } |
| 79 | |
| 80 | /// A scalar as written on one line (continuations already joined). |
| 81 | fn scalar(text: &str) -> Value { |
| 82 | let text = strip_tag(text.trim()); |
| 83 | if text.is_empty() || text == "~" || text == "null" { |
| 84 | return Value::Null; |
| 85 | } |
| 86 | if text == "[]" { |
| 87 | return Value::Array(Vec::new()); |
| 88 | } |
| 89 | if text == "{}" { |
| 90 | return Value::Object(Map::new()); |
| 91 | } |
| 92 | if let Some(inner) = text.strip_prefix('[').and_then(|t| t.strip_suffix(']')) { |
| 93 | return Value::Array(inner.split(',').map(|item| scalar(item.trim())).collect()); |
| 94 | } |
| 95 | if let Some(inner) = text.strip_prefix('\'').and_then(|t| t.strip_suffix('\'')) { |
| 96 | return Value::String(inner.replace("''", "'")); |
| 97 | } |
| 98 | if let Some(inner) = text.strip_prefix('"').and_then(|t| t.strip_suffix('"')) { |
| 99 | let mut out = String::new(); |
| 100 | let mut chars = inner.chars(); |
| 101 | while let Some(c) = chars.next() { |
| 102 | if c != '\\' { |
| 103 | out.push(c); |
| 104 | continue; |
| 105 | } |
| 106 | match chars.next() { |
| 107 | Some('n') => out.push('\n'), |
| 108 | Some('t') => out.push('\t'), |
| 109 | Some('"') => out.push('"'), |
| 110 | Some('\\') => out.push('\\'), |
| 111 | Some('/') => out.push('/'), |
| 112 | Some('0') => out.push('\0'), |
| 113 | Some(' ') => out.push(' '), |
| 114 | Some('x') => { |
| 115 | let hex: String = chars.by_ref().take(2).collect(); |
| 116 | out.extend(u32::from_str_radix(&hex, 16).ok().and_then(char::from_u32)); |
| 117 | } |
| 118 | Some('u') => { |
| 119 | let hex: String = chars.by_ref().take(4).collect(); |
| 120 | out.extend(u32::from_str_radix(&hex, 16).ok().and_then(char::from_u32)); |
| 121 | } |
| 122 | Some(other) => { |
| 123 | out.push('\\'); |
| 124 | out.push(other); |
| 125 | } |
| 126 | None => out.push('\\'), |
| 127 | } |
| 128 | } |
| 129 | return Value::String(out); |
| 130 | } |
| 131 | Value::String(text.to_owned()) |
| 132 | } |
| 133 | |
| 134 | struct Parser { |
| 135 | lines: Vec<Line>, |
| 136 | at: usize, |
| 137 | } |
| 138 | |
| 139 | impl Parser { |
| 140 | fn skip_blank(&mut self) { |
| 141 | while self.at < self.lines.len() && self.lines[self.at].text.is_empty() { |
| 142 | self.at += 1; |
| 143 | } |
| 144 | } |
| 145 | |
| 146 | fn peek(&mut self) -> Option<&Line> { |
| 147 | self.skip_blank(); |
| 148 | self.lines.get(self.at) |
| 149 | } |
| 150 | |
| 151 | /// The node starting at the current line, which is at `indent`. |
| 152 | fn node(&mut self, indent: usize) -> Result<Value, String> { |
| 153 | let Some(line) = self.peek().cloned() else { |
| 154 | return Ok(Value::Null); |
| 155 | }; |
| 156 | if line.text == "-" || line.text.starts_with("- ") { |
| 157 | return self.sequence(line.indent); |
| 158 | } |
| 159 | if key_end(&line.text).is_some() { |
| 160 | return self.mapping(line.indent); |
| 161 | } |
| 162 | self.at += 1; |
| 163 | let text = self.continued(line.text.clone(), indent.saturating_sub(1)); |
| 164 | Ok(scalar(&text)) |
| 165 | } |
| 166 | |
| 167 | /// A scalar with the lines that continue it: deeper than `parent`, and |
| 168 | /// not themselves a key or an item. |
| 169 | fn continued(&mut self, mut text: String, parent: usize) -> String { |
| 170 | let open_quote = |t: &str| { |
| 171 | let t = t.trim(); |
| 172 | (t.starts_with('"') && (t.len() == 1 || !t.ends_with('"') || t.ends_with("\\\""))) |
| 173 | || (t.starts_with('\'') && (t.len() == 1 || !t.ends_with('\''))) |
| 174 | }; |
| 175 | loop { |
| 176 | let quoted = open_quote(&text); |
| 177 | let Some(next) = self.lines.get(self.at) else { break }; |
| 178 | if next.text.is_empty() { |
| 179 | if quoted { |
| 180 | text.push('\n'); |
| 181 | self.at += 1; |
| 182 | continue; |
| 183 | } |
| 184 | break; |
| 185 | } |
| 186 | if next.indent <= parent || (!quoted && (key_end(&next.text).is_some() || next.text.starts_with("- "))) { |
| 187 | break; |
| 188 | } |
| 189 | text.push(' '); |
| 190 | text.push_str(&next.text); |
| 191 | self.at += 1; |
| 192 | } |
| 193 | text |
| 194 | } |
| 195 | |
| 196 | fn sequence(&mut self, indent: usize) -> Result<Value, String> { |
| 197 | let mut items = Vec::new(); |
| 198 | while let Some(line) = self.peek().cloned() { |
| 199 | if line.indent != indent || !(line.text == "-" || line.text.starts_with("- ")) { |
| 200 | break; |
| 201 | } |
| 202 | let rest = line.text[1..].trim_start(); |
| 203 | let rest = strip_tag(rest).to_owned(); |
| 204 | if rest.is_empty() { |
| 205 | self.at += 1; |
| 206 | match self.peek().cloned() { |
| 207 | Some(next) if next.indent > indent => items.push(self.node(next.indent)?), |
| 208 | _ => items.push(Value::Null), |
| 209 | } |
| 210 | continue; |
| 211 | } |
| 212 | // The item's content stands in for a line at its own column. |
| 213 | let column = indent + (line.text.len() - rest.len()); |
| 214 | self.lines[self.at] = Line { indent: column, text: rest }; |
| 215 | items.push(self.node(column)?); |
| 216 | } |
| 217 | Ok(Value::Array(items)) |
| 218 | } |
| 219 | |
| 220 | fn mapping(&mut self, indent: usize) -> Result<Value, String> { |
| 221 | let mut map = Map::new(); |
| 222 | while let Some(line) = self.peek().cloned() { |
| 223 | if line.indent != indent { |
| 224 | break; |
| 225 | } |
| 226 | let Some(end) = key_end(&line.text) else { break }; |
| 227 | let key = unquote_key(&line.text[..end]); |
| 228 | let rest = strip_tag(line.text[end + 1..].trim()).to_owned(); |
| 229 | self.at += 1; |
| 230 | let value = if rest.is_empty() { |
| 231 | match self.peek().cloned() { |
| 232 | Some(next) if next.indent > indent => self.node(next.indent)?, |
| 233 | Some(next) if next.indent == indent && (next.text == "-" || next.text.starts_with("- ")) => self.sequence(indent)?, |
| 234 | _ => Value::Null, |
| 235 | } |
| 236 | } else if let Some(style) = rest.strip_prefix('|').map(|s| (true, s)).or_else(|| rest.strip_prefix('>').map(|s| (false, s))) { |
| 237 | self.block(indent, style.0, style.1) |
| 238 | } else { |
| 239 | let text = self.continued(rest, indent); |
| 240 | scalar(&text) |
| 241 | }; |
| 242 | map.insert(key, value); |
| 243 | } |
| 244 | Ok(Value::Object(map)) |
| 245 | } |
| 246 | |
| 247 | /// A `|` (literal) or `>` (folded) block scalar under a key at `indent`. |
| 248 | fn block(&mut self, indent: usize, literal: bool, chomp: &str) -> Value { |
| 249 | let mut lines: Vec<(usize, String)> = Vec::new(); |
| 250 | while let Some(next) = self.lines.get(self.at) { |
| 251 | if !next.text.is_empty() && next.indent <= indent { |
| 252 | break; |
| 253 | } |
| 254 | lines.push((next.indent, next.text.clone())); |
| 255 | self.at += 1; |
| 256 | } |
| 257 | while lines.last().is_some_and(|(_, t)| t.is_empty()) { |
| 258 | lines.pop(); |
| 259 | } |
| 260 | let base = lines.iter().filter(|(_, t)| !t.is_empty()).map(|(i, _)| *i).min().unwrap_or(0); |
| 261 | let shown: Vec<String> = lines |
| 262 | .iter() |
| 263 | .map(|(i, t)| if t.is_empty() { String::new() } else { format!("{}{t}", " ".repeat(i - base)) }) |
| 264 | .collect(); |
| 265 | let mut text = if literal { |
| 266 | shown.join("\n") |
| 267 | } else { |
| 268 | let mut folded = String::new(); |
| 269 | for (n, line) in shown.iter().enumerate() { |
| 270 | if n > 0 { |
| 271 | folded.push(if line.is_empty() || shown[n - 1].is_empty() { '\n' } else { ' ' }); |
| 272 | } |
| 273 | folded.push_str(line); |
| 274 | } |
| 275 | folded |
| 276 | }; |
| 277 | if !chomp.contains('-') && !text.is_empty() { |
| 278 | text.push('\n'); |
| 279 | } |
| 280 | Value::String(text) |
| 281 | } |
| 282 | } |
| 283 | |
| 284 | #[cfg(test)] |
| 285 | mod tests { |
| 286 | use super::*; |
| 287 | use serde_json::json; |
| 288 | |
| 289 | const SPEC: &str = r#"--- !ruby/object:Gem::Specification |
| 290 | name: hello-world |
| 291 | version: !ruby/object:Gem::Version |
| 292 | version: 0.1.0 |
| 293 | platform: ruby |
| 294 | authors: |
| 295 | - Ada Lovelace |
| 296 | autorequire: |
| 297 | bindir: exe |
| 298 | cert_chain: [] |
| 299 | date: 2026-10-06 00:00:00.000000000 Z |
| 300 | dependencies: |
| 301 | - !ruby/object:Gem::Dependency |
| 302 | name: rack |
| 303 | requirement: !ruby/object:Gem::Requirement |
| 304 | requirements: |
| 305 | - - ">=" |
| 306 | - !ruby/object:Gem::Version |
| 307 | version: '2.0' |
| 308 | - - "<" |
| 309 | - !ruby/object:Gem::Version |
| 310 | version: '4' |
| 311 | type: :runtime |
| 312 | prerelease: false |
| 313 | - !ruby/object:Gem::Dependency |
| 314 | name: rspec |
| 315 | requirement: !ruby/object:Gem::Requirement |
| 316 | requirements: |
| 317 | - - "~>" |
| 318 | - !ruby/object:Gem::Version |
| 319 | version: '3.12' |
| 320 | type: :development |
| 321 | description: |- |
| 322 | Says hello. |
| 323 | |
| 324 | Then says it again. |
| 325 | email: |
| 326 | - ada@example.com |
| 327 | homepage: https://g1t.sh/acme/hello-world |
| 328 | licenses: |
| 329 | - MIT |
| 330 | metadata: |
| 331 | source_code_uri: https://g1t.sh/acme/hello-world |
| 332 | "quoted key": 'it''s' |
| 333 | required_ruby_version: !ruby/object:Gem::Requirement |
| 334 | requirements: |
| 335 | - - ">=" |
| 336 | - !ruby/object:Gem::Version |
| 337 | version: 3.0.0 |
| 338 | summary: A summary long enough that Psych wraps it onto a second line when it |
| 339 | writes the specification |
| 340 | test_files: [] |
| 341 | "#; |
| 342 | |
| 343 | #[test] |
| 344 | fn a_gemspec_reads_as_rubygems_wrote_it() { |
| 345 | let spec = parse(SPEC).unwrap(); |
| 346 | assert_eq!(spec["name"], "hello-world"); |
| 347 | assert_eq!(spec["version"]["version"], "0.1.0"); |
| 348 | assert_eq!(spec["platform"], "ruby"); |
| 349 | assert_eq!(spec["authors"], json!(["Ada Lovelace"])); |
| 350 | assert_eq!(spec["autorequire"], Value::Null); |
| 351 | assert_eq!(spec["cert_chain"], json!([])); |
| 352 | let deps = spec["dependencies"].as_array().unwrap(); |
| 353 | assert_eq!(deps.len(), 2); |
| 354 | assert_eq!(deps[0]["name"], "rack"); |
| 355 | assert_eq!(deps[0]["type"], ":runtime"); |
| 356 | assert_eq!(deps[0]["requirement"]["requirements"], json!([[">=", { "version": "2.0" }], ["<", { "version": "4" }]])); |
| 357 | assert_eq!(deps[1]["type"], ":development"); |
| 358 | assert_eq!(spec["description"], "Says hello.\n\nThen says it again."); |
| 359 | assert_eq!(spec["metadata"]["source_code_uri"], "https://g1t.sh/acme/hello-world"); |
| 360 | assert_eq!(spec["metadata"]["quoted key"], "it's"); |
| 361 | assert_eq!(spec["required_ruby_version"]["requirements"][0][1]["version"], "3.0.0"); |
| 362 | assert_eq!(spec["summary"], "A summary long enough that Psych wraps it onto a second line when it writes the specification"); |
| 363 | assert_eq!(spec["test_files"], json!([])); |
| 364 | } |
| 365 | |
| 366 | #[test] |
| 367 | fn scalars_and_blocks() { |
| 368 | let doc = parse("a: \"line\\nnext\"\nb: >\n folded\n text\nc: [x, 'y']\nd: ~\n").unwrap(); |
| 369 | assert_eq!(doc["a"], "line\nnext"); |
| 370 | assert_eq!(doc["b"], "folded text\n"); |
| 371 | assert_eq!(doc["c"], json!(["x", "y"])); |
| 372 | assert_eq!(doc["d"], Value::Null); |
| 373 | assert_eq!(parse("").unwrap(), Value::Null); |
| 374 | assert_eq!(parse("- a\n- b\n").unwrap(), json!(["a", "b"])); |
| 375 | } |
| 376 | } |