search.rs
| 1 | //! Search API: one walk serves both scopes, streamed as SSE. |
| 2 | |
| 3 | use crate::common::*; |
| 4 | use axum::http::StatusCode; |
| 5 | |
| 6 | /// Root id for the whole-root (".") user root is 1 (first row inserted). |
| 7 | const ROOT: i64 = 1; |
| 8 | |
| 9 | /// The `data:` payloads of an SSE body, in order. |
| 10 | fn events(body: &str) -> Vec<serde_json::Value> { |
| 11 | body.lines() |
| 12 | .filter_map(|l| l.strip_prefix("data:")) |
| 13 | .map(|d| serde_json::from_str(d.trim()).expect("event is JSON")) |
| 14 | .collect() |
| 15 | } |
| 16 | |
| 17 | #[tokio::test] |
| 18 | async fn both_scopes_stream_from_one_walk() { |
| 19 | let env = Env::new().await; |
| 20 | let admin = env.admin().await; |
| 21 | let r = admin |
| 22 | .get(&format!("/api/search?q=hello&scope=both&root={ROOT}")) |
| 23 | .await; |
| 24 | assert_eq!(r.status, StatusCode::OK); |
| 25 | assert_eq!( |
| 26 | r.header("content-type").as_deref(), |
| 27 | Some("text/event-stream") |
| 28 | ); |
| 29 | assert_eq!(r.header("x-accel-buffering").as_deref(), Some("no")); |
| 30 | |
| 31 | let evs = events(&r.text()); |
| 32 | // The name hit: matched on the entry's own name, with the server's |
| 33 | // sniffed kind. |
| 34 | let file = evs |
| 35 | .iter() |
| 36 | .find(|e| e["type"] == "file") |
| 37 | .expect("a name hit"); |
| 38 | assert_eq!(file["path"], "docs/inner/hello.txt"); |
| 39 | assert_eq!(file["kind"], "text"); |
| 40 | assert_eq!(file["is_dir"], false); |
| 41 | // The content hit, from the same walk. |
| 42 | let m = evs |
| 43 | .iter() |
| 44 | .find(|e| e["type"] == "match") |
| 45 | .expect("a content hit"); |
| 46 | assert_eq!(m["path"], "docs/inner/hello.txt"); |
| 47 | assert_eq!(m["line"], 1); |
| 48 | assert_eq!(m["text"], "hello world"); |
| 49 | // The stream always ends with the summary. |
| 50 | let done = evs.last().expect("a done event"); |
| 51 | assert_eq!(done["type"], "done"); |
| 52 | assert_eq!(done["stopped"], false); |
| 53 | assert!(done["scanned"].as_u64().unwrap() >= 8); |
| 54 | } |
| 55 | |
| 56 | #[tokio::test] |
| 57 | async fn path_narrows_the_walk_to_a_subfolder() { |
| 58 | let env = Env::new().await; |
| 59 | let admin = env.admin().await; |
| 60 | let paths = |body: String| -> Vec<String> { |
| 61 | events(&body) |
| 62 | .iter() |
| 63 | .filter(|e| e["type"] == "file") |
| 64 | .map(|e| e["path"].as_str().unwrap().to_string()) |
| 65 | .collect() |
| 66 | }; |
| 67 | // Inside `docs`: the hit is found and its path stays root-relative. |
| 68 | let r = admin |
| 69 | .get(&format!( |
| 70 | "/api/search?q=hello&scope=name&root={ROOT}&path=docs" |
| 71 | )) |
| 72 | .await; |
| 73 | assert_eq!(r.status, StatusCode::OK); |
| 74 | assert_eq!(paths(r.text()), vec!["docs/inner/hello.txt".to_string()]); |
| 75 | // The start folder itself is not a result. |
| 76 | let r = admin |
| 77 | .get(&format!( |
| 78 | "/api/search?q=docs&scope=name&root={ROOT}&path=docs" |
| 79 | )) |
| 80 | .await; |
| 81 | assert!(paths(r.text()).is_empty()); |
| 82 | // Outside `docs`: nothing. |
| 83 | let r = admin |
| 84 | .get(&format!( |
| 85 | "/api/search?q=hello&scope=name&root={ROOT}&path=src" |
| 86 | )) |
| 87 | .await; |
| 88 | assert!(paths(r.text()).is_empty()); |
| 89 | // A missing or escaping start folder is rejected. |
| 90 | let r = admin |
| 91 | .get(&format!("/api/search?q=hello&root={ROOT}&path=nope")) |
| 92 | .await; |
| 93 | assert!(r.status.is_client_error()); |
| 94 | let r = admin |
| 95 | .get(&format!("/api/search?q=hello&root={ROOT}&path=../")) |
| 96 | .await; |
| 97 | assert!(r.status.is_client_error()); |
| 98 | } |
| 99 | |
| 100 | #[tokio::test] |
| 101 | async fn name_scope_ignores_the_parent_path() { |
| 102 | let env = Env::new().await; |
| 103 | let admin = env.admin().await; |
| 104 | let r = admin |
| 105 | .get(&format!("/api/search?q=docs&scope=name&root={ROOT}")) |
| 106 | .await; |
| 107 | let paths: Vec<String> = events(&r.text()) |
| 108 | .iter() |
| 109 | .filter(|e| e["type"] == "file") |
| 110 | .map(|e| e["path"].as_str().unwrap().to_string()) |
| 111 | .collect(); |
| 112 | // The folder itself, never the files inside it. |
| 113 | assert_eq!(paths, vec!["docs".to_string()]); |
| 114 | } |
| 115 | |
| 116 | #[tokio::test] |
| 117 | async fn search_rejects_bad_input_and_anonymous_callers() { |
| 118 | let env = Env::new().await; |
| 119 | let admin = env.admin().await; |
| 120 | let anon = Client::new(env.app.clone()); |
| 121 | |
| 122 | assert_eq!( |
| 123 | anon.get("/api/search?q=hello").await.status, |
| 124 | StatusCode::UNAUTHORIZED |
| 125 | ); |
| 126 | assert_eq!( |
| 127 | admin.get("/api/search?q=%20").await.status, |
| 128 | StatusCode::BAD_REQUEST |
| 129 | ); |
| 130 | assert_eq!( |
| 131 | admin.get("/api/search?q=hello&scope=nope").await.status, |
| 132 | StatusCode::BAD_REQUEST |
| 133 | ); |
| 134 | assert_eq!( |
| 135 | admin.get("/api/search?q=hello&root=999").await.status, |
| 136 | StatusCode::FORBIDDEN |
| 137 | ); |
| 138 | } |
| 139 | |
| 140 | #[tokio::test] |
| 141 | async fn hidden_and_gitignored_entries_are_searched() { |
| 142 | let env = Env::new().await; |
| 143 | let p = env.root.path(); |
| 144 | std::fs::create_dir_all(p.join(".hidden")).unwrap(); |
| 145 | std::fs::write(p.join(".hidden/secretnote.txt"), "needle here").unwrap(); |
| 146 | std::fs::write(p.join("ignoredfile.log"), "needle here").unwrap(); |
| 147 | std::fs::write(p.join(".gitignore"), "*.log\n").unwrap(); |
| 148 | let admin = env.admin().await; |
| 149 | |
| 150 | let r = admin |
| 151 | .get(&format!("/api/search?q=needle&scope=content&root={ROOT}")) |
| 152 | .await; |
| 153 | let evs = events(&r.text()); |
| 154 | let paths: Vec<String> = evs |
| 155 | .iter() |
| 156 | .filter_map(|e| e["path"].as_str().map(str::to_string)) |
| 157 | .collect(); |
| 158 | assert!( |
| 159 | paths.contains(&".hidden/secretnote.txt".to_string()), |
| 160 | "{paths:?}" |
| 161 | ); |
| 162 | assert!(paths.contains(&"ignoredfile.log".to_string()), "{paths:?}"); |
| 163 | } |
| 164 | |
| 165 | /// An excluded folder is invisible to search: not as a name hit, and not as |
| 166 | /// a source of content hits from inside it. |
| 167 | #[tokio::test] |
| 168 | async fn excluded_folders_never_appear_in_results() { |
| 169 | let env = Env::new().await; |
| 170 | let admin = env.admin().await; |
| 171 | |
| 172 | // Baseline: "docs" and the file under it are both findable. |
| 173 | let r = admin |
| 174 | .get(&format!("/api/search?q=hello&scope=both&root={ROOT}")) |
| 175 | .await; |
| 176 | let paths: Vec<String> = events(&r.text()) |
| 177 | .iter() |
| 178 | .filter(|e| e["type"] == "file") |
| 179 | .map(|e| e["path"].as_str().unwrap_or_default().to_string()) |
| 180 | .collect(); |
| 181 | assert!(paths.contains(&"docs/inner/hello.txt".to_string())); |
| 182 | |
| 183 | let r = admin |
| 184 | .put_json( |
| 185 | "/api/admin/settings", |
| 186 | &serde_json::json!({ |
| 187 | "allow_writable_shares": false, |
| 188 | // Normalisation: the stored form has no surrounding slashes. |
| 189 | "search_excludes": ["/docs/"], |
| 190 | }), |
| 191 | ) |
| 192 | .await; |
| 193 | assert_eq!(r.status, StatusCode::OK, "settings: {}", r.text()); |
| 194 | assert_eq!(r.json()["search_excludes"][0], "docs"); |
| 195 | |
| 196 | let r = admin |
| 197 | .get(&format!("/api/search?q=hello&scope=both&root={ROOT}")) |
| 198 | .await; |
| 199 | let evs = events(&r.text()); |
| 200 | for e in &evs { |
| 201 | let p = e["path"].as_str().unwrap_or_default(); |
| 202 | assert!( |
| 203 | !p.starts_with("docs"), |
| 204 | "excluded folder leaked into results: {e}" |
| 205 | ); |
| 206 | } |
| 207 | // A search for the folder's own name finds nothing either. |
| 208 | let r = admin |
| 209 | .get(&format!("/api/search?q=docs&scope=name&root={ROOT}")) |
| 210 | .await; |
| 211 | assert!( |
| 212 | !events(&r.text()).iter().any(|e| e["type"] == "file"), |
| 213 | "the excluded folder itself was still listed" |
| 214 | ); |
| 215 | |
| 216 | // Everything outside it is untouched. |
| 217 | let r = admin |
| 218 | .get(&format!("/api/search?q=main&scope=name&root={ROOT}")) |
| 219 | .await; |
| 220 | assert!( |
| 221 | events(&r.text()).iter().any(|e| e["path"] == "src/main.rs"), |
| 222 | "an unrelated folder was excluded too" |
| 223 | ); |
| 224 | } |
| 225 | |
| 226 | /// "." would exclude the whole root, which turns search off rather than |
| 227 | /// narrowing it. Blank and duplicate entries are dropped the same way. |
| 228 | #[tokio::test] |
| 229 | async fn exclude_list_is_normalised() { |
| 230 | let env = Env::new().await; |
| 231 | let admin = env.admin().await; |
| 232 | let r = admin |
| 233 | .put_json( |
| 234 | "/api/admin/settings", |
| 235 | &serde_json::json!({ |
| 236 | "allow_writable_shares": false, |
| 237 | "search_excludes": [".", "", " ", "docs", "docs/", "/src"], |
| 238 | }), |
| 239 | ) |
| 240 | .await; |
| 241 | assert_eq!(r.status, StatusCode::OK); |
| 242 | let got = r.json(); |
| 243 | let list: Vec<String> = got["search_excludes"] |
| 244 | .as_array() |
| 245 | .unwrap() |
| 246 | .iter() |
| 247 | .map(|v| v.as_str().unwrap().to_string()) |
| 248 | .collect(); |
| 249 | assert_eq!(list, vec!["docs".to_string(), "src".to_string()]); |
| 250 | |
| 251 | // And it survives a round trip. |
| 252 | let r = admin.get("/api/admin/settings").await; |
| 253 | assert_eq!(r.json()["search_excludes"], got["search_excludes"]); |
| 254 | } |
| 255 | |
| 256 | /// Windows-style separators must survive normalisation. Trimming the |
| 257 | /// slashes before converting the backslashes left `\docs\` stored as |
| 258 | /// `/docs/`, which then matched nothing. |
| 259 | #[tokio::test] |
| 260 | async fn backslash_paths_are_normalised_before_trimming() { |
| 261 | let env = Env::new().await; |
| 262 | let admin = env.admin().await; |
| 263 | let r = admin |
| 264 | .put_json( |
| 265 | "/api/admin/settings", |
| 266 | &serde_json::json!({ |
| 267 | "allow_writable_shares": false, |
| 268 | "search_excludes": ["\\docs\\"], |
| 269 | }), |
| 270 | ) |
| 271 | .await; |
| 272 | assert_eq!(r.status, StatusCode::OK); |
| 273 | assert_eq!(r.json()["search_excludes"][0], "docs"); |
| 274 | |
| 275 | // And it actually excludes. |
| 276 | let r = admin |
| 277 | .get(&format!("/api/search?q=hello&scope=both&root={ROOT}")) |
| 278 | .await; |
| 279 | for e in events(&r.text()) { |
| 280 | let p = e["path"].as_str().unwrap_or_default(); |
| 281 | assert!(!p.starts_with("docs"), "not excluded: {e}"); |
| 282 | } |
| 283 | } |
| 284 |