From e83d1fc9fcfc6c80fb14b55e38fe7c1a7a4b7235 Mon Sep 17 00:00:00 2001
From: Antonio Scandurra <me@as-cii.com>
Date: Thu, 24 Feb 2022 16:33:31 +0100
Subject: [PATCH] Start on a regex implementation of `SearchQuery`

---
 crates/project/src/project.rs |  15 +++--
 crates/project/src/search.rs  | 108 ++++++++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+), 8 deletions(-)
 create mode 100644 crates/project/src/search.rs

diff --git a/crates/project/src/project.rs b/crates/project/src/project.rs
index 74890a2f51..aa9e47fcb7 100644
--- a/crates/project/src/project.rs
+++ b/crates/project/src/project.rs
@@ -2108,10 +2108,7 @@ impl Project {
                                             let matches = if let Some(file) =
                                                 fs.open_sync(&path).await.log_err()
                                             {
-                                                query
-                                                    .search(file)
-                                                    .next()
-                                                    .map_or(false, |range| range.is_ok())
+                                                query.is_contained_in_stream(file).unwrap_or(false)
                                             } else {
                                                 false
                                             };
@@ -2176,10 +2173,12 @@ impl Project {
                             let mut buffers_rx = buffers_rx.clone();
                             scope.spawn(async move {
                                 while let Some((buffer, snapshot)) = buffers_rx.next().await {
-                                    for range in query.search(
-                                        snapshot.as_rope().bytes_in_range(0..snapshot.len()),
-                                    ) {
-                                        let range = range.unwrap();
+                                    for range in query
+                                        .search(
+                                            snapshot.as_rope().bytes_in_range(0..snapshot.len()),
+                                        )
+                                        .unwrap()
+                                    {
                                         let range = snapshot.anchor_before(range.start)
                                             ..snapshot.anchor_after(range.end);
                                         worker_matched_buffers
diff --git a/crates/project/src/search.rs b/crates/project/src/search.rs
new file mode 100644
index 0000000000..69be605c93
--- /dev/null
+++ b/crates/project/src/search.rs
@@ -0,0 +1,108 @@
+use aho_corasick::{AhoCorasick, AhoCorasickBuilder};
+use anyhow::Result;
+use regex::{Regex, RegexBuilder};
+use std::{
+    borrow::Cow,
+    io::{BufRead, BufReader, Read},
+    ops::Range,
+    sync::Arc,
+};
+
+#[derive(Clone)]
+pub enum SearchQuery {
+    Text { search: Arc<AhoCorasick<usize>> },
+    Regex { multiline: bool, regex: Regex },
+}
+
+impl SearchQuery {
+    pub fn text(query: &str) -> Self {
+        let search = AhoCorasickBuilder::new()
+            .auto_configure(&[query])
+            .build(&[query]);
+        Self::Text {
+            search: Arc::new(search),
+        }
+    }
+
+    pub fn regex(query: &str, whole_word: bool, case_sensitive: bool) -> Result<Self> {
+        let mut query = Cow::Borrowed(query);
+        if whole_word {
+            let mut word_query = String::new();
+            word_query.push_str("\\b");
+            word_query.push_str(&query);
+            word_query.push_str("\\b");
+            query = Cow::Owned(word_query);
+        }
+
+        let multiline = query.contains("\n") || query.contains("\\n");
+        let regex = RegexBuilder::new(&query)
+            .case_insensitive(!case_sensitive)
+            .multi_line(multiline)
+            .build()?;
+        Ok(Self::Regex { multiline, regex })
+    }
+
+    pub fn is_contained_in_stream<T: Read>(&self, stream: T) -> Result<bool> {
+        match self {
+            SearchQuery::Text { search } => {
+                let mat = search.stream_find_iter(stream).next();
+                match mat {
+                    Some(Ok(_)) => Ok(true),
+                    Some(Err(err)) => Err(err.into()),
+                    None => Ok(false),
+                }
+            }
+            SearchQuery::Regex { multiline, regex } => {
+                let mut reader = BufReader::new(stream);
+                if *multiline {
+                    let mut text = String::new();
+                    if let Err(err) = reader.read_to_string(&mut text) {
+                        Err(err.into())
+                    } else {
+                        Ok(regex.find(&text).is_some())
+                    }
+                } else {
+                    for line in reader.lines() {
+                        let line = line?;
+                        if regex.find(&line).is_some() {
+                            return Ok(true);
+                        }
+                    }
+                    Ok(false)
+                }
+            }
+        }
+    }
+
+    pub fn search<'a, T: 'a + Read>(&'a self, stream: T) -> Result<Vec<Range<usize>>> {
+        let mut matches = Vec::new();
+        match self {
+            SearchQuery::Text { search } => {
+                for mat in search.stream_find_iter(stream) {
+                    let mat = mat?;
+                    matches.push(mat.start()..mat.end())
+                }
+            }
+            SearchQuery::Regex { multiline, regex } => {
+                let mut reader = BufReader::new(stream);
+                if *multiline {
+                    let mut text = String::new();
+                    reader.read_to_string(&mut text)?;
+                    matches.extend(regex.find_iter(&text).map(|mat| mat.start()..mat.end()));
+                } else {
+                    let mut line_ix = 0;
+                    for line in reader.lines() {
+                        let line = line?;
+                        matches.extend(
+                            regex
+                                .find_iter(&line)
+                                .map(|mat| (line_ix + mat.start())..(line_ix + mat.end())),
+                        );
+                        line_ix += line.len();
+                    }
+                }
+            }
+        }
+        Ok(matches)
+    }
+}