skip binary and empty files (#15)

Cargo.toml

@@ -24,6 +24,7 @@ regex = "1.6.0"
 serde = { version = "1.0.145", features = ["derive"] }
 serde_json = "1.0.85"
 serde-sarif = "0.3.4"
+content_inspector = "0.2.4"
 
 [dev-dependencies]
 assert_cmd = "2.0"

src/rules/pls_no_land.rs

@@ -2,6 +2,7 @@ extern crate regex;
 
 use crate::diagnostic;
 use anyhow::Context;
+use content_inspector::inspect;
 use regex::Regex;
 use std::collections::HashSet;
 use std::fs::File;
@@ -25,25 +26,31 @@ pub fn pls_no_land(paths: &HashSet<PathBuf>) -> anyhow::Result<Vec<diagnostic::D
     Ok(ret)
 }
 
-type LinesView = Vec<String>;
+fn pls_no_land_impl(path: &PathBuf) -> anyhow::Result<Vec<diagnostic::Diagnostic>> {
+    let in_file = File::open(path).with_context(|| format!("failed to open: {:#?}", path))?;
+    let mut in_buf = BufReader::new(in_file);
 
-fn lines_view<R: BufRead>(reader: R) -> anyhow::Result<LinesView> {
-    let mut ret: LinesView = LinesView::default();
-    for line in reader.lines() {
-        let line = line?;
-        ret.push(line);
+    let mut first_line = vec![];
+    in_buf.read_until(b'\n', &mut first_line)?;
+
+    if first_line.is_empty() || inspect(&first_line[..]).is_binary() {
+        return Ok(vec![]);
     }
-    Ok(ret)
-}
 
-fn pls_no_land_impl(path: &PathBuf) -> anyhow::Result<Vec<diagnostic::Diagnostic>> {
-    let in_file = File::open(path).with_context(|| format!("failed to open: {:#?}", path))?;
-    let in_buf = BufReader::new(in_file);
-    let lines_view = lines_view(in_buf).context("failed to build lines view")?;
+    let first_line_view = String::from_utf8(first_line)
+        .with_context(|| format!("could not read first line of {:#?}", path))?;
+    let lines_view = in_buf
+        .lines()
+        .collect::<std::io::Result<Vec<String>>>()
+        .with_context(|| format!("failed to read lines of text from {:#?}", path))?;
 
     let mut ret = Vec::new();
 
-    for (i, line) in lines_view.iter().enumerate() {
+    for (i, line) in [first_line_view]
+        .iter()
+        .chain(lines_view.iter())
+        .enumerate()
+    {
         if line.contains("trunk-ignore(|-begin|-end|-all)\\(trunk-toolbox/do-not-land\\)") {
             continue;
         }

tests/do_not_land_test.rs

@@ -8,7 +8,10 @@ fn do_not_land() {
     let test_repo = TestRepo::make().unwrap();
 
     test_repo
-        .write("alpha.foo", "lorem ipsum dolor\ndo-NOT-lAnD\nsit amet\n")
+        .write(
+            "alpha.foo",
+            "lorem ipsum dolor\ndo-NOT-lAnD\nsit amet\n".as_bytes(),
+        )
         .unwrap();
     test_repo.git_add_all().unwrap();
     let horton = test_repo.run_horton().unwrap();
@@ -20,3 +23,34 @@ fn do_not_land() {
         predicates::str::contains("Found 'do-NOT-lAnD'").eval(&horton)
     );
 }
+
+#[test]
+fn do_not_land_ignores_binary_files() {
+    let test_repo = TestRepo::make().unwrap();
+
+    test_repo
+        .write("alpha.foo.binary", include_bytes!("trunk-logo.png"))
+        .unwrap();
+    test_repo.git_add_all().unwrap();
+    let horton = test_repo.run_horton().unwrap();
+    let result: serde_json::Value = serde_json::from_str(&horton).unwrap();
+    let runs = result
+        .as_object()
+        .unwrap()
+        .get("runs")
+        .unwrap()
+        .as_array()
+        .unwrap();
+
+    assert_eq!(runs.len(), 1);
+    assert_eq!(
+        runs.get(0)
+            .unwrap()
+            .get("results")
+            .unwrap()
+            .as_array()
+            .unwrap()
+            .is_empty(),
+        true
+    );
+}

tests/integration_testing.rs

@@ -49,7 +49,7 @@ impl TestRepo {
         Ok(TestRepo { dir })
     }
 
-    pub fn write(&self, relpath: &str, data: &str) -> anyhow::Result<()> {
+    pub fn write(&self, relpath: &str, data: &[u8]) -> anyhow::Result<()> {
         let path = {
             let mut path = self.dir.path().to_path_buf();
             path.push(relpath);