From a56d5f2ddfcfb290c308f2b1a3464c3bebc340de Mon Sep 17 00:00:00 2001 From: sigoden Date: Sun, 8 Sep 2024 19:50:02 +0800 Subject: feat: save rag in YAML instead of bin (#848) --- src/rag/serde_vectors.rs | 69 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 69 insertions(+) create mode 100644 src/rag/serde_vectors.rs (limited to 'src/rag/serde_vectors.rs') diff --git a/src/rag/serde_vectors.rs b/src/rag/serde_vectors.rs new file mode 100644 index 0000000..894c22c --- /dev/null +++ b/src/rag/serde_vectors.rs @@ -0,0 +1,69 @@ +use super::*; + +use base64::{engine::general_purpose::STANDARD, Engine}; +use serde::{de, Deserializer, Serializer}; + +pub fn serialize( + vectors: &IndexMap>, + serializer: S, +) -> Result +where + S: Serializer, +{ + let encoded_map: IndexMap = vectors + .iter() + .map(|(key, vec)| { + let (h, l) = split_document_id(*key); + let byte_slice = unsafe { + std::slice::from_raw_parts( + vec.as_ptr() as *const u8, + vec.len() * std::mem::size_of::(), + ) + }; + (format!("{h}-{l}"), STANDARD.encode(byte_slice)) + }) + .collect(); + + encoded_map.serialize(serializer) +} + +pub fn deserialize<'de, D>(deserializer: D) -> Result>, D::Error> +where + D: Deserializer<'de>, +{ + let encoded_map: IndexMap = + IndexMap::::deserialize(deserializer)?; + + let mut decoded_map = IndexMap::new(); + for (key, base64_str) in encoded_map { + let decoded_key: DocumentId = key + .split_once('-') + .and_then(|(h, l)| { + let h = h.parse::().ok()?; + let l = l.parse::().ok()?; + Some(combine_document_id(h, l)) + }) + .ok_or_else(|| de::Error::custom(format!("Invalid key '{key}'")))?; + + let decoded_data = STANDARD.decode(&base64_str).map_err(de::Error::custom)?; + + if decoded_data.len() % std::mem::size_of::() != 0 { + return Err(de::Error::custom(format!("Invalid vector at '{key}'"))); + } + + let num_f32s = decoded_data.len() / std::mem::size_of::(); + + let mut vec_f32 = vec![0.0f32; num_f32s]; + unsafe { + std::ptr::copy_nonoverlapping( + decoded_data.as_ptr(), + vec_f32.as_mut_ptr() as *mut u8, + decoded_data.len(), + ); + } + + decoded_map.insert(decoded_key, vec_f32); + } + + Ok(decoded_map) +} -- cgit v1.2.3