feat: encode doc into binary (#441)

thorseraq · web-flow · commit 4faa5df9f17c · 2023-05-25T23:58:18.000+08:00
diff --git a/libs/jwst-codec/src/doc/codec/update.rs b/libs/jwst-codec/src/doc/codec/update.rs
@@ -393,6 +393,7 @@ mod tests {
     use super::*;
     use serde::Deserialize;
     use std::{num::ParseIntError, path::PathBuf};
+    use yrs::{Map, Transact};
 
     fn struct_item(id: (Client, Clock), len: usize) -> StructInfo {
         StructInfo::Item {
@@ -581,4 +582,23 @@ mod tests {
             &OrderRange::from(vec![(10..12), (13..15)])
         );
     }
+
+    #[test]
+    fn double_run_test_with_yrs_basic() {
+        let yrs_doc = yrs::Doc::new();
+
+        let map = yrs_doc.get_or_insert_map("abc");
+        let mut trx = yrs_doc.transact_mut();
+        map.insert(&mut trx, "a", 1).unwrap();
+
+        let binary_from_yrs = trx.encode_update_v1().unwrap();
+
+        let mut decoder = RawDecoder::new(binary_from_yrs.clone());
+        let update = Update::read(&mut decoder).unwrap();
+        let mut doc = Doc::default();
+        doc.apply_update(update).unwrap();
+        let binary = doc.encode_update_v1().unwrap();
+
+        assert_eq!(binary_from_yrs, binary);
+    }
 }
diff --git a/libs/jwst-codec/src/doc/document.rs b/libs/jwst-codec/src/doc/document.rs
@@ -416,4 +416,12 @@ impl Doc {
             Ok(())
         })
     }
+
+    pub fn encode_update_v1(self) -> JwstCodecResult<Vec<u8>> {
+        let store = self.store;
+        let mut encoder = RawEncoder::default();
+        store.encode_with_state_vector(&StateVector::default(), &mut encoder)?;
+
+        Ok(encoder.into_inner())
+    }
 }
diff --git a/libs/jwst-codec/src/doc/store.rs b/libs/jwst-codec/src/doc/store.rs
@@ -1,5 +1,6 @@
 use super::*;
 use crate::doc::StateVector;
+use std::collections::VecDeque;
 use std::{
     cell::{Ref, RefCell, RefMut},
     collections::{hash_map::Entry, HashMap},
@@ -10,6 +11,21 @@ use std::{
 #[derive(Clone, PartialEq, Debug)]
 pub struct StructRef(Arc<RefCell<StructInfo>>);
 
+impl StructRef {
+    pub fn read<R: CrdtReader>(decoder: &mut R, id: Id) -> JwstCodecResult<Self> {
+        match StructInfo::read(decoder, id) {
+            Ok(info) => Ok(info.into()),
+            Err(err) => Err(err),
+        }
+    }
+}
+
+impl<W: CrdtWriter> CrdtWrite<W> for StructRef {
+    fn write(&self, writer: &mut W) -> JwstCodecResult {
+        self.0.borrow().write(writer)
+    }
+}
+
 impl From<StructInfo> for StructRef {
     fn from(info: StructInfo) -> Self {
         Self(Arc::new(RefCell::new(info)))
@@ -91,8 +107,10 @@ impl StructRef {
 #[derive(Default)]
 pub struct DocStore {
     pub items: Arc<RwLock<HashMap<Client, Vec<StructRef>>>>,
-    pub types: Arc<RwLock<HashMap<String, TypeStoreRef>>>,
     pub delete_set: Arc<RwLock<DeleteSet>>,
+
+    // following fields are only used in memory
+    pub types: Arc<RwLock<HashMap<String, TypeStoreRef>>>,
     pub pending: Option<Update>,
 }
 
@@ -361,6 +379,77 @@ impl DocStore {
             delete_set.add(id.client, id.clock, len)
         }
     }
+
+    fn diff_state_vectors(
+        local_state_vector: &StateVector,
+        remote_state_vector: &StateVector,
+    ) -> Vec<(Client, Clock)> {
+        let mut diff = Vec::new();
+
+        for (client, &remote_clock) in remote_state_vector.iter() {
+            let local_clock = local_state_vector.get(client);
+            if local_clock > remote_clock {
+                diff.push((*client, remote_clock));
+            }
+        }
+
+        for (client, _) in local_state_vector.iter() {
+            if remote_state_vector.get(client) == 0 {
+                diff.push((*client, 0));
+            }
+        }
+
+        diff
+    }
+
+    pub fn encode_with_state_vector<W: CrdtWriter>(
+        &self,
+        sv: &StateVector,
+        encoder: &mut W,
+    ) -> JwstCodecResult {
+        let local_state_vector = self.get_state_vector();
+        let diff = Self::diff_state_vectors(&local_state_vector, sv);
+        let mut update_structs: HashMap<u64, VecDeque<StructInfo>> = HashMap::new();
+
+        for (client, clock) in diff {
+            // We have made sure that the client is in the local state vector in diff_state_vectors()
+            if let Some(items) = self.items.read().unwrap().get(&client) {
+                if items.is_empty() {
+                    continue;
+                }
+
+                update_structs.insert(client, VecDeque::new());
+                let vec_struct_info = update_structs.get_mut(&client).unwrap();
+
+                // the smallest clock in items may exceed the clock
+                let clock = items.first().unwrap().id().clock.max(clock);
+                if let Some(index) = Self::get_item_index(items, clock) {
+                    let first_block = items.get(index).unwrap();
+                    let offset = first_block.clock() - clock;
+                    if offset != 0 {
+                        // needs to implement Content split first
+                        unimplemented!()
+                    } else {
+                        vec_struct_info.push_back(first_block.borrow().clone());
+                    }
+
+                    for item in items.iter().skip(index + 1) {
+                        vec_struct_info.push_back(item.borrow().clone());
+                    }
+                }
+            }
+        }
+
+        let update = Update {
+            structs: update_structs,
+            delete_set: self.delete_set.read().unwrap().clone(),
+            ..Update::default()
+        };
+
+        update.write(encoder)?;
+
+        Ok(())
+    }
 }
 
 #[cfg(test)]

Original file line number	Diff line number	Diff line change
`@@ -416,4 +416,12 @@ impl Doc {`
`416`	`416`	`Ok(())`
`417`	`417`	`})`
`418`	`418`	`}`
	`419`	`+`
	`420`	`+ pub fn encode_update_v1(self) -> JwstCodecResult<Vec<u8>> {`
	`421`	`+ let store = self.store;`
	`422`	`+ let mut encoder = RawEncoder::default();`
	`423`	`+ store.encode_with_state_vector(&StateVector::default(), &mut encoder)?;`
	`424`	`+`
	`425`	`+ Ok(encoder.into_inner())`
	`426`	`+ }`
`419`	`427`	`}`