chore: Make the repo use examples and keep the library

examples/serve-console.rs

@@ -0,0 +1,90 @@
+use std::str::from_utf8_unchecked;
+use std::io::{self, Write};
+use structopt::StructOpt;
+use std::path::PathBuf;
+
+use fst::Streamer;
+use elapsed::measure_time;
+use rocksdb::{DB, DBOptions, IngestExternalFileOptions};
+use raptor::{automaton, Metadata, RankedStream, CommonWords};
+
+#[derive(Debug, StructOpt)]
+pub struct CommandConsole {
+    /// The stop word file, each word must be separated by a newline.
+    #[structopt(long = "stop-words", parse(from_os_str))]
+    pub stop_words: PathBuf,
+
+    /// Meta file name (e.g. relaxed-colden).
+    #[structopt(parse(from_os_str))]
+    pub meta_name: PathBuf,
+}
+
+pub struct ConsoleSearch {
+    common_words: CommonWords,
+    metadata: Metadata,
+    db: DB,
+}
+
+impl ConsoleSearch {
+    pub fn from_command(command: CommandConsole) -> io::Result<ConsoleSearch> {
+        let common_words = CommonWords::from_file(command.stop_words)?;
+
+        let map_file = command.meta_name.with_extension("map");
+        let idx_file = command.meta_name.with_extension("idx");
+        let sst_file = command.meta_name.with_extension("sst");
+
+        let metadata = unsafe { Metadata::from_paths(map_file, idx_file).unwrap() };
+
+        let rocksdb = "rocksdb/storage";
+        let db = DB::open_default(rocksdb).unwrap();
+        let sst_file = sst_file.to_str().unwrap();
+        db.ingest_external_file(&IngestExternalFileOptions::new(), &[sst_file]).unwrap();
+        drop(db);
+        let db = DB::open_for_read_only(DBOptions::default(), rocksdb, false).unwrap();
+
+        Ok(ConsoleSearch { common_words, metadata, db })
+    }
+
+    pub fn serve(self) {
+        loop {
+            print!("Searching for: ");
+            io::stdout().flush().unwrap();
+
+            let mut query = String::new();
+            io::stdin().read_line(&mut query).unwrap();
+            let query = query.trim().to_lowercase();
+
+            if query.is_empty() { break }
+
+            let (elapsed, _) = measure_time(|| search(&self.metadata, &self.db, &self.common_words, &query));
+            println!("Finished in {}", elapsed);
+        }
+    }
+}
+
+fn search(metadata: &Metadata, database: &DB, common_words: &CommonWords, query: &str) {
+    let mut automatons = Vec::new();
+    for query in query.split_whitespace().filter(|q| !common_words.contains(*q)) {
+        let lev = automaton::build(query);
+        automatons.push(lev);
+    }
+
+    let mut stream = RankedStream::new(&metadata, automatons, 20);
+    while let Some(document) = stream.next() {
+        let id_key = format!("{}-id", document.document_id);
+        let id = database.get(id_key.as_bytes()).unwrap().unwrap();
+        let id = unsafe { from_utf8_unchecked(&id) };
+        print!("{} ", id);
+
+        let title_key = format!("{}-title", document.document_id);
+        let title = database.get(title_key.as_bytes()).unwrap().unwrap();
+        let title = unsafe { from_utf8_unchecked(&title) };
+        println!("{:?}", title);
+    }
+}
+
+fn main() {
+    let command = CommandConsole::from_args();
+    let console = ConsoleSearch::from_command(command).unwrap();
+    console.serve()
+}