|
| 1 | +// SPDX-License-Identifier: GPL-2.0 |
| 2 | + |
| 3 | +//! JSON parser used to parse rustdoc output when retrieving doctests. |
| 4 | +
|
| 5 | +use std::collections::HashMap; |
| 6 | +use std::iter::Peekable; |
| 7 | +use std::str::FromStr; |
| 8 | + |
| 9 | +#[derive(Debug, PartialEq, Eq)] |
| 10 | +pub(crate) enum JsonValue { |
| 11 | + Object(HashMap<String, JsonValue>), |
| 12 | + String(String), |
| 13 | + Number(i32), |
| 14 | + Bool(bool), |
| 15 | + Array(Vec<JsonValue>), |
| 16 | + Null, |
| 17 | +} |
| 18 | + |
| 19 | +fn parse_ident<I: Iterator<Item = char>>( |
| 20 | + iter: &mut I, |
| 21 | + output: JsonValue, |
| 22 | + ident: &str, |
| 23 | +) -> Result<JsonValue, String> { |
| 24 | + let mut ident_iter = ident.chars().skip(1); |
| 25 | + |
| 26 | + loop { |
| 27 | + let i = ident_iter.next(); |
| 28 | + if i.is_none() { |
| 29 | + return Ok(output); |
| 30 | + } |
| 31 | + let c = iter.next(); |
| 32 | + if i != c { |
| 33 | + if let Some(c) = c { |
| 34 | + return Err(format!("Unexpected character `{c}` when parsing `{ident}`")); |
| 35 | + } |
| 36 | + return Err(format!("Missing character when parsing `{ident}`")); |
| 37 | + } |
| 38 | + } |
| 39 | +} |
| 40 | + |
| 41 | +fn parse_string<I: Iterator<Item = char>>(iter: &mut I) -> Result<JsonValue, String> { |
| 42 | + let mut out = String::new(); |
| 43 | + |
| 44 | + while let Some(c) = iter.next() { |
| 45 | + match c { |
| 46 | + '\\' => { |
| 47 | + let Some(c) = iter.next() else { break }; |
| 48 | + match c { |
| 49 | + '"' | '\\' | '/' => out.push(c), |
| 50 | + 'b' => out.push(char::from(0x8u8)), |
| 51 | + 'f' => out.push(char::from(0xCu8)), |
| 52 | + 't' => out.push('\t'), |
| 53 | + 'r' => out.push('\r'), |
| 54 | + 'n' => out.push('\n'), |
| 55 | + _ => { |
| 56 | + // This code doesn't handle codepoints so we put the string content as is. |
| 57 | + out.push('\\'); |
| 58 | + out.push(c); |
| 59 | + } |
| 60 | + } |
| 61 | + } |
| 62 | + '"' => { |
| 63 | + return Ok(JsonValue::String(out)); |
| 64 | + } |
| 65 | + _ => out.push(c), |
| 66 | + } |
| 67 | + } |
| 68 | + Err(format!("Unclosed JSON string `{out}`")) |
| 69 | +} |
| 70 | + |
| 71 | +fn parse_number<I: Iterator<Item = char>>( |
| 72 | + iter: &mut Peekable<I>, |
| 73 | + digit: char, |
| 74 | +) -> Result<JsonValue, String> { |
| 75 | + let mut nb = String::new(); |
| 76 | + |
| 77 | + nb.push(digit); |
| 78 | + loop { |
| 79 | + // We peek next character to prevent taking it from the iterator in case it's a comma. |
| 80 | + if matches!(iter.peek(), Some(',' | '}' | ']')) { |
| 81 | + break; |
| 82 | + } |
| 83 | + let Some(c) = iter.next() else { break }; |
| 84 | + if c.is_whitespace() { |
| 85 | + break; |
| 86 | + } else if !c.is_ascii_digit() { |
| 87 | + return Err(format!("Error when parsing number `{nb}`: found `{c}`")); |
| 88 | + } |
| 89 | + nb.push(c); |
| 90 | + } |
| 91 | + i32::from_str(&nb) |
| 92 | + .map(|nb| JsonValue::Number(nb)) |
| 93 | + .map_err(|error| format!("Invalid number: `{error}`")) |
| 94 | +} |
| 95 | + |
| 96 | +fn parse_array<I: Iterator<Item = char>>(iter: &mut Peekable<I>) -> Result<JsonValue, String> { |
| 97 | + let mut values = Vec::new(); |
| 98 | + |
| 99 | + 'main: loop { |
| 100 | + let Some(c) = iter.next() else { |
| 101 | + return Err("Unclosed array".to_string()); |
| 102 | + }; |
| 103 | + if c.is_whitespace() { |
| 104 | + continue; |
| 105 | + } else if c == ']' { |
| 106 | + break; |
| 107 | + } |
| 108 | + values.push(parse(iter, c)?); |
| 109 | + while let Some(c) = iter.next() { |
| 110 | + if c.is_whitespace() { |
| 111 | + continue; |
| 112 | + } else if c == ',' { |
| 113 | + break; |
| 114 | + } else if c == ']' { |
| 115 | + break 'main; |
| 116 | + } else { |
| 117 | + return Err(format!("Unexpected `{c}` when parsing array")); |
| 118 | + } |
| 119 | + } |
| 120 | + } |
| 121 | + Ok(JsonValue::Array(values)) |
| 122 | +} |
| 123 | + |
| 124 | +fn parse_object<I: Iterator<Item = char>>(iter: &mut Peekable<I>) -> Result<JsonValue, String> { |
| 125 | + let mut values = HashMap::new(); |
| 126 | + |
| 127 | + 'main: loop { |
| 128 | + let Some(c) = iter.next() else { |
| 129 | + return Err("Unclosed object".to_string()); |
| 130 | + }; |
| 131 | + let key; |
| 132 | + if c.is_whitespace() { |
| 133 | + continue; |
| 134 | + } else if c == '"' { |
| 135 | + let JsonValue::String(k) = parse_string(iter)? else { |
| 136 | + unreachable!() |
| 137 | + }; |
| 138 | + key = k; |
| 139 | + } else if c == '}' { |
| 140 | + break; |
| 141 | + } else { |
| 142 | + return Err(format!("Expected `\"` when parsing Object, found `{c}`")); |
| 143 | + } |
| 144 | + |
| 145 | + // We then get the `:` separator. |
| 146 | + loop { |
| 147 | + let Some(c) = iter.next() else { |
| 148 | + return Err(format!("Missing value after key `{key}`")); |
| 149 | + }; |
| 150 | + if c.is_whitespace() { |
| 151 | + continue; |
| 152 | + } else if c == ':' { |
| 153 | + break; |
| 154 | + } else { |
| 155 | + return Err(format!( |
| 156 | + "Expected `:` after key, found `{c}` when parsing object" |
| 157 | + )); |
| 158 | + } |
| 159 | + } |
| 160 | + // Then the value. |
| 161 | + let value = loop { |
| 162 | + let Some(c) = iter.next() else { |
| 163 | + return Err(format!("Missing value after key `{key}`")); |
| 164 | + }; |
| 165 | + if c.is_whitespace() { |
| 166 | + continue; |
| 167 | + } else { |
| 168 | + break parse(iter, c)?; |
| 169 | + } |
| 170 | + }; |
| 171 | + |
| 172 | + if values.contains_key(&key) { |
| 173 | + return Err(format!("Duplicated key `{key}`")); |
| 174 | + } |
| 175 | + values.insert(key, value); |
| 176 | + |
| 177 | + while let Some(c) = iter.next() { |
| 178 | + if c.is_whitespace() { |
| 179 | + continue; |
| 180 | + } else if c == ',' { |
| 181 | + break; |
| 182 | + } else if c == '}' { |
| 183 | + break 'main; |
| 184 | + } else { |
| 185 | + return Err(format!("Unexpected `{c}` when parsing array")); |
| 186 | + } |
| 187 | + } |
| 188 | + } |
| 189 | + Ok(JsonValue::Object(values)) |
| 190 | +} |
| 191 | + |
| 192 | +fn parse<I: Iterator<Item = char>>(iter: &mut Peekable<I>, c: char) -> Result<JsonValue, String> { |
| 193 | + match c { |
| 194 | + '{' => parse_object(iter), |
| 195 | + '"' => parse_string(iter), |
| 196 | + '[' => parse_array(iter), |
| 197 | + 't' => parse_ident(iter, JsonValue::Bool(true), "true"), |
| 198 | + 'f' => parse_ident(iter, JsonValue::Bool(false), "false"), |
| 199 | + 'n' => parse_ident(iter, JsonValue::Null, "null"), |
| 200 | + c => { |
| 201 | + if c.is_ascii_digit() || c == '-' { |
| 202 | + parse_number(iter, c) |
| 203 | + } else { |
| 204 | + Err(format!("Unexpected `{c}` character")) |
| 205 | + } |
| 206 | + } |
| 207 | + } |
| 208 | +} |
| 209 | + |
| 210 | +impl JsonValue { |
| 211 | + pub(crate) fn parse(input: &str) -> Result<Self, String> { |
| 212 | + let mut iter = input.chars().peekable(); |
| 213 | + let mut value = None; |
| 214 | + |
| 215 | + while let Some(c) = iter.next() { |
| 216 | + if c.is_whitespace() { |
| 217 | + continue; |
| 218 | + } |
| 219 | + value = Some(parse(&mut iter, c)?); |
| 220 | + break; |
| 221 | + } |
| 222 | + while let Some(c) = iter.next() { |
| 223 | + if c.is_whitespace() { |
| 224 | + continue; |
| 225 | + } else { |
| 226 | + return Err(format!("Unexpected character `{c}` after content")); |
| 227 | + } |
| 228 | + } |
| 229 | + if let Some(value) = value { |
| 230 | + Ok(value) |
| 231 | + } else { |
| 232 | + Err("Empty content".to_string()) |
| 233 | + } |
| 234 | + } |
| 235 | +} |
0 commit comments