git-mirrors/continuwuity
1
0
Fork 0
mirror of https://forgejo.ellis.link/continuwuation/continuwuity.git synced 2025-07-08 21:00:00 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

factor search tokenization out into a function

Browse source 
This ensures that the tokenization algorithm will remain in sync between
querying, indexing, and deindexing. The existing code had slightly
different behavior for querying, because it did not discard words with
>50 bytes. This was inconsequential, because >50 byte tokens are never
present in the index.

Signed-off-by: strawberry <strawberry@puppygock.gay>
This commit is contained in:
Benjamin Lee 2024-06-12 00:22:49 -04:00 • committed by strawberry
parent 58881100e7
commit de0ab64a60
1 changed files with 19 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

22
src/service/rooms/search/data.rs
View file

@ -10,14 +10,20 @@ pub trait Data: Send + Sync {
fn search_pdus<'a>(&'a self, room_id: &RoomId, search_string: &str) -> SearchPdusResult<'a>; fn search_pdus<'a>(&'a self, room_id: &RoomId, search_string: &str) -> SearchPdusResult<'a>;
} }
impl Data for KeyValueDatabase { /// Splits a string into tokens used as keys in the search inverted index
fn index_pdu(&self, shortroomid: u64, pdu_id: &[u8], message_body: &str) -> Result<()> { ///
let mut batch = message_body /// This may be used to tokenize both message bodies (for indexing) or search
.split_terminator(|c: char| !c.is_alphanumeric()) /// queries (for querying).
fn tokenize(body: &str) -> impl Iterator<Item = String> + '_ {
body.split_terminator(|c: char| !c.is_alphanumeric())
.filter(|s| !s.is_empty()) .filter(|s| !s.is_empty())
.filter(|word| word.len() <= 50) .filter(|word| word.len() <= 50)
.map(str::to_lowercase) .map(str::to_lowercase)
.map(|word| { }
impl Data for KeyValueDatabase {
fn index_pdu(&self, shortroomid: u64, pdu_id: &[u8], message_body: &str) -> Result<()> {
let mut batch = tokenize(message_body).map(|word| {
let mut key = shortroomid.to_be_bytes().to_vec(); let mut key = shortroomid.to_be_bytes().to_vec();
key.extend_from_slice(word.as_bytes()); key.extend_from_slice(word.as_bytes());
key.push(0xFF); key.push(0xFF);
@ -37,11 +43,7 @@ impl Data for KeyValueDatabase {
.to_be_bytes() .to_be_bytes()
.to_vec(); .to_vec();
let words: Vec<_> = search_string let words: Vec<_> = tokenize(search_string).collect();
.split_terminator(|c: char| !c.is_alphanumeric())
.filter(|s| !s.is_empty())
.map(str::to_lowercase)
.collect();
let iterators = words.clone().into_iter().map(move |word| { let iterators = words.clone().into_iter().map(move |word| {
let mut prefix2 = prefix.clone(); let mut prefix2 = prefix.clone();