From 79d8555620c9c47e20a9cb12794d341b869e0835 Mon Sep 17 00:00:00 2001
From: Kerollmops <renault.cle@gmail.com>
Date: Sun, 24 Jun 2018 19:27:24 +0200
Subject: [PATCH] feat: Make a more debug oriented search tool

---
 raptor-search/Cargo.lock  |  7 ++++
 raptor-search/Cargo.toml  |  1 +
 raptor-search/src/main.rs | 75 +++++++++++++++++++++++++++++++--------
 3 files changed, 69 insertions(+), 14 deletions(-)

diff --git a/raptor-search/Cargo.lock b/raptor-search/Cargo.lock
index 8a94a05a5..ddd348fd8 100644
--- a/raptor-search/Cargo.lock
+++ b/raptor-search/Cargo.lock
@@ -17,6 +17,11 @@ name = "cfg-if"
 version = "0.1.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 
+[[package]]
+name = "elapsed"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+
 [[package]]
 name = "env_logger"
 version = "0.3.5"
@@ -109,6 +114,7 @@ dependencies = [
 name = "raptor-search"
 version = "0.1.0"
 dependencies = [
+ "elapsed 0.1.2 (registry+https://github.com/rust-lang/crates.io-index)",
  "env_logger 0.3.5 (registry+https://github.com/rust-lang/crates.io-index)",
  "fst 0.3.0 (git+https://github.com/Kerollmops/fst.git?branch=op-builder-with-state)",
  "raptor 0.1.0",
@@ -169,6 +175,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 "checksum bincode 1.0.0 (registry+https://github.com/rust-lang/crates.io-index)" = "bda13183df33055cbb84b847becce220d392df502ebe7a4a78d7021771ed94d0"
 "checksum byteorder 1.2.3 (registry+https://github.com/rust-lang/crates.io-index)" = "74c0b906e9446b0a2e4f760cdb3fa4b2c48cdc6db8766a845c54b6ff063fd2e9"
 "checksum cfg-if 0.1.3 (registry+https://github.com/rust-lang/crates.io-index)" = "405216fd8fe65f718daa7102ea808a946b6ce40c742998fbfd3463645552de18"
+"checksum elapsed 0.1.2 (registry+https://github.com/rust-lang/crates.io-index)" = "6f4e5af126dafd0741c2ad62d47f68b28602550102e5f0dd45c8a97fc8b49c29"
 "checksum env_logger 0.3.5 (registry+https://github.com/rust-lang/crates.io-index)" = "15abd780e45b3ea4f76b4e9a26ff4843258dd8a3eed2775a0e7368c2e7936c2f"
 "checksum fst 0.3.0 (git+https://github.com/Kerollmops/fst.git?branch=op-builder-with-state)" = "<none>"
 "checksum group-by 0.1.0 (git+https://github.com/Kerollmops/group-by.git)" = "<none>"
diff --git a/raptor-search/Cargo.toml b/raptor-search/Cargo.toml
index 7b6e6d68d..983d95321 100644
--- a/raptor-search/Cargo.toml
+++ b/raptor-search/Cargo.toml
@@ -8,6 +8,7 @@ env_logger = { version = "0.3", default-features = false }
 raptor = { path = ".." }
 serde = "1.0"
 serde_derive = "1.0"
+elapsed = "0.1"
 
 [dependencies.fst]
 git = "https://github.com/Kerollmops/fst.git"
diff --git a/raptor-search/src/main.rs b/raptor-search/src/main.rs
index ad3cfad8f..5bd8be798 100644
--- a/raptor-search/src/main.rs
+++ b/raptor-search/src/main.rs
@@ -1,22 +1,16 @@
 extern crate env_logger;
 extern crate fst;
 extern crate raptor;
+extern crate elapsed;
 
-use std::env;
+use std::{env, fs};
+use std::process::Command;
+use std::io::{self, Write};
+use elapsed::measure_time;
 use fst::Streamer;
-use raptor::{load_map, RankedStream, LevBuilder};
-
-fn main() {
-    drop(env_logger::init());
-
-    let lev_builder = LevBuilder::new();
-    let map = load_map("map.fst", "values.vecs").unwrap();
-
-    let query = env::args().nth(1).expect("Please enter query words!");
-    let query = query.to_lowercase();
-
-    println!("Searching for: {:?}", query);
+use raptor::{load_map, DocIndexMap, RankedStream, LevBuilder};
 
+fn search(map: &DocIndexMap, lev_builder: &LevBuilder, query: &str) {
     let mut automatons = Vec::new();
     for query in query.split_whitespace() {
         let lev = lev_builder.build_automaton(query);
@@ -26,6 +20,59 @@ fn main() {
     let limit: Option<usize> = env::var("RAPTOR_OUTPUT_LIMIT").ok().and_then(|x| x.parse().ok());
     let mut stream = RankedStream::new(&map, map.values(), automatons, limit.unwrap_or(20));
     while let Some(document_id) = stream.next() {
-        println!("{:?}", document_id);
+        print!("{:?}", document_id);
+
+        /* only here to debug !
+        if let Ok(_) = fs::File::open("products.json_lines") {
+            let output = Command::new("rg")
+                                .arg(document_id.to_string())
+                                .arg("products.json_lines")
+                                .output();
+            if let Ok(Ok(output)) = output.map(|o| String::from_utf8(o.stdout)) {
+                if let Some(line) = output.lines().next() {
+                    let pattern = "\"title\":";
+                    if let Some(index) = line.find(pattern) {
+                        let line: String = line[index..].chars().skip(pattern.len()).take(100).collect();
+                        print!(" => {}", line);
+                    }
+                }
+            }
+        }
+        // */
+
+        println!();
+    }
+}
+
+fn main() {
+    drop(env_logger::init());
+
+    let (elapsed, (lev_builder, map)) = measure_time(|| {
+        let lev_builder = LevBuilder::new();
+        let map = load_map("map.fst", "values.vecs").unwrap();
+        (lev_builder, map)
+    });
+    println!("Loaded in {}", elapsed);
+
+    match env::args().nth(1) {
+        Some(query) => {
+            println!("Searching for: {:?}", query);
+            let query = query.to_lowercase();
+            let (elapsed, _) = measure_time(|| search(&map, &lev_builder, &query));
+            println!("Finished in {}", elapsed);
+        },
+        None => 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(&map, &lev_builder, &query));
+            println!("Finished in {}", elapsed);
+        },
     }
 }