first pass of merging in warp (doesn't build)
This commit is contained in:
@@ -1,7 +1,9 @@
|
||||
use crate::index::full_source_code_embedding::chunker::{coalesce_fragments, Fragment};
|
||||
use std::path::Path;
|
||||
|
||||
use itertools::Itertools;
|
||||
use line_span::{LineSpan, LineSpans};
|
||||
use std::path::Path;
|
||||
|
||||
use crate::index::full_source_code_embedding::chunker::{coalesce_fragments, Fragment};
|
||||
|
||||
/// Chunks the given file into [`Fragment`]s. Each chunk is at most `num_lines_per_chunk` lines long, and contains at most `max_bytes_per_chunk` bytes.
|
||||
pub(super) fn chunk_code<'a>(
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
use super::*;
|
||||
use std::path::Path;
|
||||
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn test_chunker() {
|
||||
let code = "This is some text content\nthat should be chunked\nusing the naive chunker\nbecause the language isn't recognized.";
|
||||
|
||||
@@ -48,7 +48,11 @@ pub(super) fn chunk_code<'a>(
|
||||
// of the allocator).
|
||||
//
|
||||
// See: https://github.com/tree-sitter/tree-sitter/issues/3129
|
||||
#[cfg(all(target_os = "linux", target_env = "gnu", not(feature = "jemalloc")))]
|
||||
#[cfg(all(
|
||||
any(target_os = "linux", target_os = "freebsd"),
|
||||
target_env = "gnu",
|
||||
not(feature = "jemalloc")
|
||||
))]
|
||||
unsafe {
|
||||
nix::libc::malloc_trim(0);
|
||||
}
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
use std::path::Path;
|
||||
|
||||
use languages::language_by_filename;
|
||||
use warp_util::standardized_path::StandardizedPath;
|
||||
|
||||
use super::*;
|
||||
|
||||
@@ -33,12 +34,14 @@ fn main() {
|
||||
"#;
|
||||
|
||||
let max_chunk_size = 128;
|
||||
let language_path =
|
||||
StandardizedPath::try_new("/test.rs").expect("test path should be absolute");
|
||||
|
||||
let chunks = chunk_code(
|
||||
source_code,
|
||||
Path::new("test.rs"),
|
||||
max_chunk_size,
|
||||
&language_by_filename(Path::new("test.rs"))
|
||||
&language_by_filename(&language_path)
|
||||
.expect("Rust language must exist")
|
||||
.grammar,
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user