searchengines-web/MeiliSearch
searchengines-web/MeiliSearch
1
0
Fork 0
mirror of https://github.com/meilisearch/MeiliSearch synced 2024-11-29 16:24:26 +01:00
Code Issues Releases Wiki Activity

feat: Use the new Tokenizer in the json-line-indexer

Browse Source
This commit is contained in:
Clément Renault 2018-09-27 17:01:16 +02:00
parent 8a0c82d51e
commit a066c084fe
1 changed files with 32 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

56
raptor-indexer/src/main.rs
View File

@ -7,9 +7,9 @@ use std::path::Path;
use std::collections::{HashSet, BTreeMap}; use std::collections::{HashSet, BTreeMap};
use std::io::{self, BufReader, BufRead}; use std::io::{self, BufReader, BufRead};
use std::fs::File; use std::fs::File;
use std::{env, iter}; use std::env;
use raptor::{MetadataBuilder, DocIndex}; use raptor::{MetadataBuilder, DocIndex, Tokenizer};
use rocksdb::{SstFileWriter, EnvOptions, ColumnFamilyOptions}; use rocksdb::{SstFileWriter, EnvOptions, ColumnFamilyOptions};
use serde_json::from_str; use serde_json::from_str;
use unidecode::unidecode; use unidecode::unidecode;
@ -37,6 +37,30 @@ where P: AsRef<Path>,
Ok(set) Ok(set)
} }
fn insert_document_words<'a, I, A, B>(builder: &mut MetadataBuilder<A, B>, doc_index: u64, attr: u8, words: I)
where A: io::Write,
B: io::Write,
I: IntoIterator<Item=(usize, &'a str)>,
{
for (index, word) in words {
let doc_index = DocIndex {
document: doc_index,
attribute: attr,
attribute_index: index as u32,
};
// insert the exact representation
let word_lower = word.to_lowercase();
// and the unidecoded lowercased version
let word_unidecoded = unidecode(word).to_lowercase();
if word_lower != word_unidecoded {
builder.insert(word_unidecoded, doc_index);
}
builder.insert(word_lower, doc_index);
}
}
fn main() { fn main() {
let data_path = env::args().nth(1).expect("Missing data json lines file (e.g. products.json_lines)"); let data_path = env::args().nth(1).expect("Missing data json lines file (e.g. products.json_lines)");
let data = File::open(data_path).unwrap(); let data = File::open(data_path).unwrap();
@ -69,29 +93,13 @@ fn main() {
let product: Product = from_str(&line).unwrap(); let product: Product = from_str(&line).unwrap();
{ let title = Tokenizer::new(&product.title);
let title = iter::repeat(0).zip(product.title.split_whitespace()).filter(|&(_, w)| !common_words.contains(w)).enumerate(); let title = title.iter().filter(|&(_, w)| !common_words.contains(w));
let description = iter::repeat(1).zip(product.ft.split_whitespace()).filter(|&(_, w)| !common_words.contains(w)).enumerate(); insert_document_words(&mut builder, product.group_id, 0, title);
let words = title.chain(description); let description = Tokenizer::new(&product.ft);
for (i, (attr, word)) in words { let description = description.iter().filter(|&(_, w)| !common_words.contains(w));
let doc_index = DocIndex { insert_document_words(&mut builder, product.group_id, 1, description);
document: product.group_id,
attribute: attr,
attribute_index: i as u32,
};
// insert the exact representation
let word_lower = word.to_lowercase();
// and the unidecoded lowercased version
let word_unidecoded = unidecode(word).to_lowercase();
if word_lower != word_unidecoded {
builder.insert(word_unidecoded, doc_index);
}
builder.insert(word_lower, doc_index);
}
}
// TODO simplify this by using functions and // TODO simplify this by using functions and
// use the MetadataBuilder internal BTreeMap ? // use the MetadataBuilder internal BTreeMap ?