Move crates under a sub folder to clean up the code

crates/milli/examples/index.rs

@@ -0,0 +1,114 @@
+use std::error::Error;
+use std::fs::File;
+use std::io::{BufRead, BufReader, Cursor, Seek};
+use std::path::Path;
+
+use heed::EnvOpenOptions;
+use milli::documents::{DocumentsBatchBuilder, DocumentsBatchReader};
+use milli::update::{IndexDocuments, IndexDocumentsConfig, IndexerConfig, Settings};
+use milli::{Index, Object};
+
+fn usage(error: &str, program_name: &str) -> String {
+    format!(
+        "{}. Usage: {} <PATH-TO-INDEX> <PATH-TO-DATASET> [searchable_fields] [filterable_fields]",
+        error, program_name
+    )
+}
+
+fn main() -> Result<(), Box<dyn Error>> {
+    let mut args = std::env::args();
+    let program_name = args.next().expect("No program name");
+    let index_path =
+        args.next().unwrap_or_else(|| panic!("{}", usage("Missing path to index.", &program_name)));
+    let dataset_path = args
+        .next()
+        .unwrap_or_else(|| panic!("{}", usage("Missing path to source dataset.", &program_name)));
+    // let primary_key = args.next().unwrap_or_else(|| "id".into());
+    // "title overview"
+    let searchable_fields: Vec<String> = args
+        .next()
+        .map(|arg| arg.split_whitespace().map(ToString::to_string).collect())
+        .unwrap_or_default();
+
+    println!("{searchable_fields:?}");
+    // "release_date genres"
+    let filterable_fields: Vec<String> = args
+        .next()
+        .map(|arg| arg.split_whitespace().map(ToString::to_string).collect())
+        .unwrap_or_default();
+
+    let mut options = EnvOpenOptions::new();
+    options.map_size(100 * 1024 * 1024 * 1024); // 100 GB
+
+    std::fs::create_dir_all(&index_path).unwrap();
+    let index = Index::new(options, index_path).unwrap();
+    let mut wtxn = index.write_txn().unwrap();
+
+    let config = IndexerConfig::default();
+    let mut builder = Settings::new(&mut wtxn, &index, &config);
+    // builder.set_primary_key(primary_key);
+    let searchable_fields = searchable_fields.iter().map(|s| s.to_string()).collect();
+    builder.set_searchable_fields(searchable_fields);
+    let filterable_fields = filterable_fields.iter().map(|s| s.to_string()).collect();
+    builder.set_filterable_fields(filterable_fields);
+
+    builder.execute(|_| (), || false).unwrap();
+
+    let config = IndexerConfig::default();
+    let indexing_config = IndexDocumentsConfig::default();
+
+    let builder =
+        IndexDocuments::new(&mut wtxn, &index, &config, indexing_config, |_| (), || false).unwrap();
+
+    let documents = documents_from(
+        &dataset_path,
+        Path::new(&dataset_path).extension().unwrap_or_default().to_str().unwrap_or_default(),
+    );
+    let (builder, user_error) = builder.add_documents(documents).unwrap();
+    user_error.unwrap();
+    builder.execute().unwrap();
+    wtxn.commit().unwrap();
+
+    index.prepare_for_closing().wait();
+    Ok(())
+}
+fn documents_from(filename: &str, filetype: &str) -> DocumentsBatchReader<impl BufRead + Seek> {
+    let reader = File::open(filename)
+        .unwrap_or_else(|_| panic!("could not find the dataset in: {}", filename));
+    let reader = BufReader::new(reader);
+    let documents = match filetype {
+        "csv" => documents_from_csv(reader).unwrap(),
+        "json" => documents_from_json(reader).unwrap(),
+        "jsonl" => documents_from_jsonl(reader).unwrap(),
+        otherwise => panic!("invalid update format {:?}", otherwise),
+    };
+    DocumentsBatchReader::from_reader(Cursor::new(documents)).unwrap()
+}
+
+fn documents_from_jsonl(reader: impl BufRead) -> milli::Result<Vec<u8>> {
+    let mut documents = DocumentsBatchBuilder::new(Vec::new());
+
+    for result in serde_json::Deserializer::from_reader(reader).into_iter::<Object>() {
+        let object = result.unwrap();
+        documents.append_json_object(&object)?;
+    }
+
+    documents.into_inner().map_err(Into::into)
+}
+
+fn documents_from_json(reader: impl BufRead) -> milli::Result<Vec<u8>> {
+    let mut documents = DocumentsBatchBuilder::new(Vec::new());
+
+    documents.append_json_array(reader)?;
+
+    documents.into_inner().map_err(Into::into)
+}
+
+fn documents_from_csv(reader: impl BufRead) -> milli::Result<Vec<u8>> {
+    let csv = csv::Reader::from_reader(reader);
+
+    let mut documents = DocumentsBatchBuilder::new(Vec::new());
+    documents.append_csv(csv)?;
+
+    documents.into_inner().map_err(Into::into)
+}

crates/milli/examples/search.rs

@@ -0,0 +1,124 @@
+use std::error::Error;
+use std::io::stdin;
+use std::path::Path;
+use std::time::Instant;
+
+use heed::EnvOpenOptions;
+use milli::{
+    execute_search, filtered_universe, DefaultSearchLogger, GeoSortStrategy, Index, SearchContext,
+    SearchLogger, TermsMatchingStrategy, TimeBudget,
+};
+
+#[global_allocator]
+static ALLOC: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
+fn main() -> Result<(), Box<dyn Error>> {
+    let mut args = std::env::args();
+    let program_name = args.next().expect("No program name");
+    let dataset = args.next().unwrap_or_else(|| {
+        panic!(
+            "Missing path to index. Usage: {} <PATH-TO-INDEX> [<logger-dir>] [print-documents]",
+            program_name
+        )
+    });
+    let detailed_logger_dir = args.next();
+    let print_documents: bool =
+        if let Some(arg) = args.next() { arg == "print-documents" } else { false };
+
+    let mut options = EnvOpenOptions::new();
+    options.map_size(100 * 1024 * 1024 * 1024); // 100 GB
+
+    let index = Index::new(options, dataset)?;
+    let txn = index.read_txn()?;
+    let mut query = String::new();
+    while stdin().read_line(&mut query)? > 0 {
+        for _ in 0..2 {
+            let mut default_logger = DefaultSearchLogger;
+            // FIXME: consider resetting the state of the logger between search executions as otherwise panics are possible.
+            // Workaround'd here by recreating the logger on each iteration of the loop
+            let mut detailed_logger = detailed_logger_dir
+                .as_ref()
+                .map(|logger_dir| (milli::VisualSearchLogger::default(), logger_dir));
+            let logger: &mut dyn SearchLogger<_> =
+                if let Some((detailed_logger, _)) = detailed_logger.as_mut() {
+                    detailed_logger
+                } else {
+                    &mut default_logger
+                };
+
+            let start = Instant::now();
+
+            let mut ctx = SearchContext::new(&index, &txn)?;
+            let universe = filtered_universe(ctx.index, ctx.txn, &None)?;
+
+            let docs = execute_search(
+                &mut ctx,
+                (!query.trim().is_empty()).then(|| query.trim()),
+                TermsMatchingStrategy::Last,
+                milli::score_details::ScoringStrategy::Skip,
+                false,
+                universe,
+                &None,
+                &None,
+                GeoSortStrategy::default(),
+                0,
+                20,
+                None,
+                &mut DefaultSearchLogger,
+                logger,
+                TimeBudget::max(),
+                None,
+                None,
+            )?;
+            if let Some((logger, dir)) = detailed_logger {
+                logger.finish(&mut ctx, Path::new(dir))?;
+            }
+            let elapsed = start.elapsed();
+            println!("new: {}us, docids: {:?}", elapsed.as_micros(), docs.documents_ids);
+            if print_documents {
+                let documents = index
+                    .documents(&txn, docs.documents_ids.iter().copied())
+                    .unwrap()
+                    .into_iter()
+                    .map(|(id, obkv)| {
+                        let mut object = serde_json::Map::default();
+                        for (fid, fid_name) in index.fields_ids_map(&txn).unwrap().iter() {
+                            let value = obkv.get(fid).unwrap();
+                            let value: serde_json::Value = serde_json::from_slice(value).unwrap();
+                            object.insert(fid_name.to_owned(), value);
+                        }
+                        (id, serde_json::to_string_pretty(&object).unwrap())
+                    })
+                    .collect::<Vec<_>>();
+
+                for (id, document) in documents {
+                    println!("{id}:");
+                    println!("{document}");
+                }
+
+                let documents = index
+                    .documents(&txn, docs.documents_ids.iter().copied())
+                    .unwrap()
+                    .into_iter()
+                    .map(|(id, obkv)| {
+                        let mut object = serde_json::Map::default();
+                        for (fid, fid_name) in index.fields_ids_map(&txn).unwrap().iter() {
+                            let value = obkv.get(fid).unwrap();
+                            let value: serde_json::Value = serde_json::from_slice(value).unwrap();
+                            object.insert(fid_name.to_owned(), value);
+                        }
+                        (id, serde_json::to_string_pretty(&object).unwrap())
+                    })
+                    .collect::<Vec<_>>();
+                println!("{}us: {:?}", elapsed.as_micros(), docs.documents_ids);
+                for (id, document) in documents {
+                    println!("{id}:");
+                    println!("{document}");
+                }
+            }
+        }
+        query.clear();
+    }
+
+    Ok(())
+}

crates/milli/examples/settings.rs

@@ -0,0 +1,33 @@
+// use big_s::S;
+use heed::EnvOpenOptions;
+// use maplit::hashset;
+use milli::{
+    update::{IndexerConfig, Settings},
+    Criterion, Index,
+};
+
+fn main() {
+    let mut options = EnvOpenOptions::new();
+    options.map_size(100 * 1024 * 1024 * 1024); // 100 GB
+
+    let index = Index::new(options, "data_movies.ms").unwrap();
+    let mut wtxn = index.write_txn().unwrap();
+
+    let config = IndexerConfig::default();
+    let mut builder = Settings::new(&mut wtxn, &index, &config);
+
+    // builder.set_min_word_len_one_typo(5);
+    // builder.set_min_word_len_two_typos(7);
+    // builder.set_sortable_fields(hashset! { S("release_date") });
+    builder.set_criteria(vec![
+        Criterion::Words,
+        Criterion::Typo,
+        Criterion::Proximity,
+        Criterion::Attribute,
+        Criterion::Sort,
+        Criterion::Exactness,
+    ]);
+
+    builder.execute(|_| (), || false).unwrap();
+    wtxn.commit().unwrap();
+}