1 use std::collections::hash_map;
2 use std::collections::hash_map::DefaultHasher;
3 use std::collections::HashMap;
4 use std::default::Default;
5 use std::hash::BuildHasherDefault;
6 use std::hash::Hash;
7 use std::hash::Hasher;
8 use std::slice;
9 
10 use crate::reflect::ReflectValueRef;
11 use crate::rt;
12 use crate::wire_format::WireType;
13 use crate::zigzag::encode_zig_zag_32;
14 use crate::zigzag::encode_zig_zag_64;
15 use crate::CodedOutputStream;
16 
17 /// Unknown value.
18 ///
19 /// See [`UnknownFields`](crate::UnknownFields) for the explanations.
20 #[derive(Debug)]
21 pub enum UnknownValue {
22     /// 32-bit unknown (e. g. `fixed32` or `float`)
23     Fixed32(u32),
24     /// 64-bit unknown (e. g. `fixed64` or `double`)
25     Fixed64(u64),
26     /// Varint unknown (e. g. `int32` or `bool`)
27     Varint(u64),
28     /// Length-delimited unknown (e. g. `message` or `string`)
29     LengthDelimited(Vec<u8>),
30 }
31 
32 impl UnknownValue {
33     /// Wire type for this unknown
wire_type(&self) -> WireType34     pub fn wire_type(&self) -> WireType {
35         self.get_ref().wire_type()
36     }
37 
38     /// As ref
get_ref<'s>(&'s self) -> UnknownValueRef<'s>39     pub fn get_ref<'s>(&'s self) -> UnknownValueRef<'s> {
40         match *self {
41             UnknownValue::Fixed32(fixed32) => UnknownValueRef::Fixed32(fixed32),
42             UnknownValue::Fixed64(fixed64) => UnknownValueRef::Fixed64(fixed64),
43             UnknownValue::Varint(varint) => UnknownValueRef::Varint(varint),
44             UnknownValue::LengthDelimited(ref bytes) => UnknownValueRef::LengthDelimited(&bytes),
45         }
46     }
47 
48     /// Construct unknown value from `int64` value.
int32(i: i32) -> UnknownValue49     pub fn int32(i: i32) -> UnknownValue {
50         UnknownValue::int64(i as i64)
51     }
52 
53     /// Construct unknown value from `int64` value.
int64(i: i64) -> UnknownValue54     pub fn int64(i: i64) -> UnknownValue {
55         UnknownValue::Varint(i as u64)
56     }
57 
58     /// Construct unknown value from `sint32` value.
sint32(i: i32) -> UnknownValue59     pub fn sint32(i: i32) -> UnknownValue {
60         UnknownValue::Varint(encode_zig_zag_32(i) as u64)
61     }
62 
63     /// Construct unknown value from `sint64` value.
sint64(i: i64) -> UnknownValue64     pub fn sint64(i: i64) -> UnknownValue {
65         UnknownValue::Varint(encode_zig_zag_64(i))
66     }
67 
68     /// Construct unknown value from `float` value.
float(f: f32) -> UnknownValue69     pub fn float(f: f32) -> UnknownValue {
70         UnknownValue::Fixed32(f.to_bits())
71     }
72 
73     /// Construct unknown value from `double` value.
double(f: f64) -> UnknownValue74     pub fn double(f: f64) -> UnknownValue {
75         UnknownValue::Fixed64(f.to_bits())
76     }
77 
78     /// Construct unknown value from `sfixed32` value.
sfixed32(i: i32) -> UnknownValue79     pub fn sfixed32(i: i32) -> UnknownValue {
80         UnknownValue::Fixed32(i as u32)
81     }
82 
83     /// Construct unknown value from `sfixed64` value.
sfixed64(i: i64) -> UnknownValue84     pub fn sfixed64(i: i64) -> UnknownValue {
85         UnknownValue::Fixed64(i as u64)
86     }
87 }
88 
89 /// Reference to unknown value.
90 ///
91 /// See [`UnknownFields`](crate::UnknownFields) for explanations.
92 #[derive(Debug, PartialEq)]
93 pub enum UnknownValueRef<'o> {
94     /// 32-bit unknown
95     Fixed32(u32),
96     /// 64-bit unknown
97     Fixed64(u64),
98     /// Varint unknown
99     Varint(u64),
100     /// Length-delimited unknown
101     LengthDelimited(&'o [u8]),
102 }
103 
104 impl<'o> UnknownValueRef<'o> {
105     /// Wire-type to serialize this unknown
wire_type(&self) -> WireType106     pub fn wire_type(&self) -> WireType {
107         match *self {
108             UnknownValueRef::Fixed32(_) => WireType::Fixed32,
109             UnknownValueRef::Fixed64(_) => WireType::Fixed64,
110             UnknownValueRef::Varint(_) => WireType::Varint,
111             UnknownValueRef::LengthDelimited(_) => WireType::LengthDelimited,
112         }
113     }
114 
to_reflect_value_ref(&'o self) -> ReflectValueRef<'o>115     pub(crate) fn to_reflect_value_ref(&'o self) -> ReflectValueRef<'o> {
116         match self {
117             UnknownValueRef::Fixed32(v) => ReflectValueRef::U32(*v),
118             UnknownValueRef::Fixed64(v) => ReflectValueRef::U64(*v),
119             UnknownValueRef::Varint(v) => ReflectValueRef::U64(*v),
120             UnknownValueRef::LengthDelimited(v) => ReflectValueRef::Bytes(v),
121         }
122     }
123 }
124 
125 /// Field unknown values.
126 ///
127 /// See [`UnknownFields`](crate::UnknownFields) for explanations.
128 #[derive(Clone, PartialEq, Eq, Debug, Default, Hash)]
129 pub(crate) struct UnknownValues {
130     /// 32-bit unknowns
131     pub(crate) fixed32: Vec<u32>,
132     /// 64-bit unknowns
133     pub(crate) fixed64: Vec<u64>,
134     /// Varint unknowns
135     pub(crate) varint: Vec<u64>,
136     /// Length-delimited unknowns
137     pub(crate) length_delimited: Vec<Vec<u8>>,
138 }
139 
140 impl UnknownValues {
141     /// Add unknown value
add_value(&mut self, value: UnknownValue)142     pub fn add_value(&mut self, value: UnknownValue) {
143         match value {
144             UnknownValue::Fixed64(fixed64) => self.fixed64.push(fixed64),
145             UnknownValue::Fixed32(fixed32) => self.fixed32.push(fixed32),
146             UnknownValue::Varint(varint) => self.varint.push(varint),
147             UnknownValue::LengthDelimited(length_delimited) => {
148                 self.length_delimited.push(length_delimited)
149             }
150         };
151     }
152 
153     /// Iterate over unknown values
iter<'s>(&'s self) -> UnknownValuesIter<'s>154     pub fn iter<'s>(&'s self) -> UnknownValuesIter<'s> {
155         UnknownValuesIter {
156             fixed32: self.fixed32.iter(),
157             fixed64: self.fixed64.iter(),
158             varint: self.varint.iter(),
159             length_delimited: self.length_delimited.iter(),
160         }
161     }
162 
any(&self) -> Option<UnknownValueRef>163     pub(crate) fn any(&self) -> Option<UnknownValueRef> {
164         if let Some(last) = self.fixed32.last() {
165             Some(UnknownValueRef::Fixed32(*last))
166         } else if let Some(last) = self.fixed64.last() {
167             Some(UnknownValueRef::Fixed64(*last))
168         } else if let Some(last) = self.varint.last() {
169             Some(UnknownValueRef::Varint(*last))
170         } else if let Some(last) = self.length_delimited.last() {
171             Some(UnknownValueRef::LengthDelimited(last))
172         } else {
173             None
174         }
175     }
176 }
177 
178 impl<'a> IntoIterator for &'a UnknownValues {
179     type Item = UnknownValueRef<'a>;
180     type IntoIter = UnknownValuesIter<'a>;
181 
into_iter(self) -> UnknownValuesIter<'a>182     fn into_iter(self) -> UnknownValuesIter<'a> {
183         self.iter()
184     }
185 }
186 
187 /// Iterator over unknown values
188 pub(crate) struct UnknownValuesIter<'o> {
189     fixed32: slice::Iter<'o, u32>,
190     fixed64: slice::Iter<'o, u64>,
191     varint: slice::Iter<'o, u64>,
192     length_delimited: slice::Iter<'o, Vec<u8>>,
193 }
194 
195 impl<'o> Iterator for UnknownValuesIter<'o> {
196     type Item = UnknownValueRef<'o>;
197 
next(&mut self) -> Option<UnknownValueRef<'o>>198     fn next(&mut self) -> Option<UnknownValueRef<'o>> {
199         if let Some(fixed32) = self.fixed32.next() {
200             return Some(UnknownValueRef::Fixed32(*fixed32));
201         }
202         if let Some(fixed64) = self.fixed64.next() {
203             return Some(UnknownValueRef::Fixed64(*fixed64));
204         }
205         if let Some(varint) = self.varint.next() {
206             return Some(UnknownValueRef::Varint(*varint));
207         }
208         if let Some(length_delimited) = self.length_delimited.next() {
209             return Some(UnknownValueRef::LengthDelimited(&length_delimited));
210         }
211         None
212     }
213 }
214 
215 /// Hold "unknown" fields in parsed message.
216 ///
217 /// Field may be unknown if it they are added in newer version of `.proto`.
218 /// Unknown fields are stored in `UnknownFields` structure, so
219 /// protobuf message could process messages without losing data.
220 ///
221 /// For example, in this operation: load from DB, modify, store to DB,
222 /// even when working with older `.proto` file, new fields won't be lost.
223 #[derive(Clone, PartialEq, Eq, Debug, Default)]
224 pub struct UnknownFields {
225     /// The map.
226     //
227     // `Option` is needed, because HashMap constructor performs allocation,
228     // and very expensive.
229     //
230     // We use "default hasher" to make iteration order deterministic.
231     // Which is used to make codegen output deterministic in presence of unknown fields
232     // (e. g. file options are represented as unknown fields).
233     // Using default hasher is suboptimal, because it makes unknown fields less safe.
234     // Note, Google Protobuf C++ simply uses linear map (which can exploitable the same way),
235     // and Google Protobuf Java uses tree map to store unknown fields
236     // (which is more expensive than hashmap).
237     fields: Option<Box<HashMap<u32, UnknownValues, BuildHasherDefault<DefaultHasher>>>>,
238 }
239 
240 /// Very simple hash implementation of `Hash` for `UnknownFields`.
241 /// Since map is unordered, we cannot put entry hashes into hasher,
242 /// instead we summing hashes of entries.
243 impl Hash for UnknownFields {
hash<H: Hasher>(&self, state: &mut H)244     fn hash<H: Hasher>(&self, state: &mut H) {
245         if let Some(ref map) = self.fields {
246             if !map.is_empty() {
247                 let mut hash: u64 = 0;
248                 for (k, v) in &**map {
249                     let mut entry_hasher = DefaultHasher::new();
250                     Hash::hash(&(k, v), &mut entry_hasher);
251                     hash = hash.wrapping_add(entry_hasher.finish());
252                 }
253                 Hash::hash(&map.len(), state);
254                 Hash::hash(&hash, state);
255             }
256         }
257     }
258 }
259 
260 impl UnknownFields {
261     /// Empty unknown fields.
new() -> UnknownFields262     pub const fn new() -> UnknownFields {
263         UnknownFields { fields: None }
264     }
265 
266     /// Clear all unknown fields.
clear(&mut self)267     pub fn clear(&mut self) {
268         if let Some(ref mut fields) = self.fields {
269             fields.clear();
270         }
271     }
272 
init_map(&mut self)273     fn init_map(&mut self) {
274         if self.fields.is_none() {
275             self.fields = Some(Default::default());
276         }
277     }
278 
find_field<'a>(&'a mut self, number: &'a u32) -> &'a mut UnknownValues279     fn find_field<'a>(&'a mut self, number: &'a u32) -> &'a mut UnknownValues {
280         self.init_map();
281 
282         match self.fields.as_mut().unwrap().entry(*number) {
283             hash_map::Entry::Occupied(e) => e.into_mut(),
284             hash_map::Entry::Vacant(e) => e.insert(Default::default()),
285         }
286     }
287 
288     /// Add unknown fixed 32-bit
add_fixed32(&mut self, number: u32, fixed32: u32)289     pub fn add_fixed32(&mut self, number: u32, fixed32: u32) {
290         self.find_field(&number).fixed32.push(fixed32);
291     }
292 
293     /// Add unknown fixed 64-bit
add_fixed64(&mut self, number: u32, fixed64: u64)294     pub fn add_fixed64(&mut self, number: u32, fixed64: u64) {
295         self.find_field(&number).fixed64.push(fixed64);
296     }
297 
298     /// Add unknown varint
add_varint(&mut self, number: u32, varint: u64)299     pub fn add_varint(&mut self, number: u32, varint: u64) {
300         self.find_field(&number).varint.push(varint);
301     }
302 
303     /// Add unknown length delimited
add_length_delimited(&mut self, number: u32, length_delimited: Vec<u8>)304     pub fn add_length_delimited(&mut self, number: u32, length_delimited: Vec<u8>) {
305         self.find_field(&number)
306             .length_delimited
307             .push(length_delimited);
308     }
309 
310     /// Add unknown value
add_value(&mut self, number: u32, value: UnknownValue)311     pub fn add_value(&mut self, number: u32, value: UnknownValue) {
312         self.find_field(&number).add_value(value);
313     }
314 
315     /// Remove unknown field by number
remove(&mut self, field_number: u32)316     pub fn remove(&mut self, field_number: u32) {
317         if let Some(fields) = &mut self.fields {
318             fields.remove(&field_number);
319         }
320     }
321 
322     /// Iterate over all unknowns
iter<'s>(&'s self) -> UnknownFieldsIter<'s>323     pub fn iter<'s>(&'s self) -> UnknownFieldsIter<'s> {
324         UnknownFieldsIter {
325             entries: self.fields.as_ref().map(|m| UnknownFieldsNotEmptyIter {
326                 fields: m.iter(),
327                 current: None,
328             }),
329         }
330     }
331 
332     /// Get any value for unknown fields.
get(&self, field_number: u32) -> Option<UnknownValueRef>333     pub fn get(&self, field_number: u32) -> Option<UnknownValueRef> {
334         match &self.fields {
335             Some(map) => map.get(&field_number).and_then(|v| v.any()),
336             None => None,
337         }
338     }
339 
340     #[doc(hidden)]
write_to_bytes(&self) -> Vec<u8>341     pub fn write_to_bytes(&self) -> Vec<u8> {
342         let mut r = Vec::with_capacity(rt::unknown_fields_size(self) as usize);
343         let mut stream = CodedOutputStream::vec(&mut r);
344         // Do we need it stable everywhere?
345         stream.write_unknown_fields_sorted(self).unwrap();
346         stream.flush().unwrap();
347         drop(stream);
348         r
349     }
350 }
351 
352 impl<'a> IntoIterator for &'a UnknownFields {
353     type Item = (u32, UnknownValueRef<'a>);
354     type IntoIter = UnknownFieldsIter<'a>;
355 
into_iter(self) -> UnknownFieldsIter<'a>356     fn into_iter(self) -> UnknownFieldsIter<'a> {
357         self.iter()
358     }
359 }
360 
361 struct UnknownFieldsNotEmptyIter<'s> {
362     fields: hash_map::Iter<'s, u32, UnknownValues>,
363     current: Option<(u32, UnknownValuesIter<'s>)>,
364 }
365 
366 /// Iterator over [`UnknownFields`](crate::UnknownFields)
367 pub struct UnknownFieldsIter<'s> {
368     entries: Option<UnknownFieldsNotEmptyIter<'s>>,
369 }
370 
371 impl<'s> Iterator for UnknownFieldsNotEmptyIter<'s> {
372     type Item = (u32, UnknownValueRef<'s>);
373 
next(&mut self) -> Option<(u32, UnknownValueRef<'s>)>374     fn next(&mut self) -> Option<(u32, UnknownValueRef<'s>)> {
375         loop {
376             if let Some((field_number, values)) = &mut self.current {
377                 if let Some(value) = values.next() {
378                     return Some((*field_number, value));
379                 }
380             }
381             let (field_number, values) = self.fields.next()?;
382             self.current = Some((*field_number, values.iter()));
383         }
384     }
385 }
386 
387 impl<'s> Iterator for UnknownFieldsIter<'s> {
388     type Item = (u32, UnknownValueRef<'s>);
389 
next(&mut self) -> Option<(u32, UnknownValueRef<'s>)>390     fn next(&mut self) -> Option<(u32, UnknownValueRef<'s>)> {
391         self.entries.as_mut().and_then(|entries| entries.next())
392     }
393 }
394 
395 #[cfg(test)]
396 mod test {
397     use std::collections::hash_map::DefaultHasher;
398     use std::hash::Hash;
399     use std::hash::Hasher;
400 
401     use super::UnknownFields;
402 
403     #[test]
unknown_fields_hash()404     fn unknown_fields_hash() {
405         let mut unknown_fields_1 = UnknownFields::new();
406         let mut unknown_fields_2 = UnknownFields::new();
407 
408         // Check field order is not important
409 
410         unknown_fields_1.add_fixed32(10, 222);
411         unknown_fields_1.add_fixed32(10, 223);
412         unknown_fields_1.add_fixed64(14, 224);
413 
414         unknown_fields_2.add_fixed32(10, 222);
415         unknown_fields_2.add_fixed64(14, 224);
416         unknown_fields_2.add_fixed32(10, 223);
417 
418         fn hash(unknown_fields: &UnknownFields) -> u64 {
419             let mut hasher = DefaultHasher::new();
420             Hash::hash(unknown_fields, &mut hasher);
421             hasher.finish()
422         }
423 
424         assert_eq!(hash(&unknown_fields_1), hash(&unknown_fields_2));
425     }
426 
427     #[test]
unknown_fields_iteration_order_deterministic()428     fn unknown_fields_iteration_order_deterministic() {
429         let mut u_1 = UnknownFields::new();
430         let mut u_2 = UnknownFields::new();
431         for u in &mut [&mut u_1, &mut u_2] {
432             u.add_fixed32(10, 20);
433             u.add_varint(30, 40);
434             u.add_fixed64(50, 60);
435             u.add_length_delimited(70, Vec::new());
436             u.add_varint(80, 90);
437             u.add_fixed32(11, 22);
438             u.add_fixed64(33, 44);
439         }
440 
441         let items_1: Vec<_> = u_1.iter().collect();
442         let items_2: Vec<_> = u_2.iter().collect();
443         assert_eq!(items_1, items_2);
444     }
445 }
446