feat: switching index to tag based (#265)

* add tag based index

* update for python

* fix lint

Cargo.lock

@@ -2847,7 +2847,6 @@ dependencies = [
  "temp_testdir",
  "tracing",
  "tracing-test",
- "tree-sitter-java",
  "tree-sitter-javascript",
  "tree-sitter-python",
  "tree-sitter-tags",
@@ -3507,16 +3506,6 @@ dependencies = [
  "regex",
 ]
 
-[[package]]
-name = "tree-sitter-java"
-version = "0.20.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f0bf5d3f508cbffcbfe1805834101c0d24297a8b6c2184ad9c595556c46d2420"
-dependencies = [
- "cc",
- "tree-sitter",
-]
-
 [[package]]
 name = "tree-sitter-javascript"
 version = "0.20.0"

crates/tabby-scheduler/Cargo.toml

@@ -19,7 +19,6 @@ lazy_static = { workspace = true }
 serde = { workspace = true }
 serde-jsonlines = { workspace = true }
 file-rotate = "0.7.5"
-tree-sitter-java = "0.20.0"
 tree-sitter-python = "0.20.2"
 
 [dev-dependencies]

crates/tabby-scheduler/src/dataset.rs

@@ -207,7 +207,8 @@ lazy_static! {
             ("sql", vec!["sql"]),
             ("scala", vec!["scala"]),
             ("shellscript", vec!["sh", "bash", "command", "zsh"]),
-            ("typescript", vec!["ts", "tsx"]),
+            ("typescript", vec!["ts"]),
+            ("tsx", vec!["tsx"]),
             ("tex", vec!["tex"]),
             ("vb", vec!["vb"]),
         ])
@@ -223,19 +224,7 @@ lazy_static! {
         map
     };
     static ref LANGUAGE_TAGS: HashMap<&'static str, TagsConfigurationSync> = {
-        HashMap::from([
+        HashMap::from([(
-            (
-                "javascript",
-                TagsConfigurationSync(
-                    TagsConfiguration::new(
-                        tree_sitter_javascript::language(),
-                        tree_sitter_javascript::TAGGING_QUERY,
-                        tree_sitter_javascript::LOCALS_QUERY,
-                    )
-                    .unwrap(),
-                ),
-            ),
-            (
             "python",
             TagsConfigurationSync(
                 TagsConfiguration::new(
@@ -245,18 +234,6 @@ lazy_static! {
                 )
                 .unwrap(),
             ),
-            ),
+        )])
-            (
-                "java",
-                TagsConfigurationSync(
-                    TagsConfiguration::new(
-                        tree_sitter_java::language(),
-                        tree_sitter_java::TAGGING_QUERY,
-                        "",
-                    )
-                    .unwrap(),
-                ),
-            ),
-        ])
     };
 }

crates/tabby-scheduler/src/index.rs

@@ -1,6 +1,7 @@
-use std::fs;
+use std::{collections::HashMap, fs};
 
 use anyhow::Result;
+use lazy_static::lazy_static;
 use tabby_common::{config::Config, path::index_dir, Document};
 use tantivy::{
     directory::MmapDirectory,
@@ -11,8 +12,8 @@ use tantivy::{
 
 pub fn index_repositories(_config: &Config) -> Result<()> {
     let mut builder = Schema::builder();
-    let git_url = builder.add_text_field("git_url", STRING | STORED);
+    let git_url = builder.add_text_field("name", STRING | STORED);
-    let filepath = builder.add_text_field("filepath", STRING | STORED);
+    let filepath = builder.add_text_field("body", STRING | STORED);
     let content = builder.add_text_field("content", TEXT | STORED);
     let language = builder.add_text_field("language", TEXT | STORED);
     let schema = builder.build();
@@ -24,6 +25,24 @@ pub fn index_repositories(_config: &Config) -> Result<()> {
     writer.delete_all_documents()?;
 
     for doc in Document::all()? {
+        for tag in doc.tags {
+            let name = doc.content.get(tag.name_range).unwrap();
+            if name.len() < 5 {
+                continue;
+            }
+
+            let body = doc.content.get(tag.range).unwrap();
+            let count_body_lines = body.lines().count();
+            if !(3..=10).contains(&count_body_lines) {
+                continue;
+            }
+
+            if let Some(blacklist) = LANGUAGE_NAME_BLACKLIST.get(doc.language.as_str()) {
+                if blacklist.contains(&name) {
+                    continue;
+                }
+            }
+        }
         writer.add_document(doc!(
                 git_url => doc.git_url,
                 filepath => doc.filepath,
@@ -36,3 +55,8 @@ pub fn index_repositories(_config: &Config) -> Result<()> {
 
     Ok(())
 }
+
+lazy_static! {
+    static ref LANGUAGE_NAME_BLACKLIST: HashMap<&'static str, Vec<&'static str>> =
+        HashMap::from([("python", vec!["__init__"])]);
+}