Add CI and some test

Author: friendlymatthew

.github/workflows/tests.yml

@@ -0,0 +1,52 @@
+name: CI
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+
+jobs:
+  tests:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@master
+
+      - uses: dtolnay/rust-toolchain@master
+        with:
+          toolchain: "nightly"
+
+      - name: version info
+        run: rustc --version; cargo --version;
+
+      - name: Run all tests
+        run: cargo test
+
+  fmt:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@master
+
+      - uses: dtolnay/rust-toolchain@master
+        with:
+          toolchain: nightly
+
+      - run: |
+          rustup component add rustfmt
+          cargo fmt --all -- --check
+
+  clippy:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@master
+
+      - uses: dtolnay/rust-toolchain@master
+        with:
+          toolchain: "nightly"
+
+      - run: |
+          rustup component add clippy
+          cargo clippy -- -D warnings
+

src/reader.rs

@@ -71,10 +71,13 @@ impl CsvReader {
             let mut valid_commas = self.comma_bitsets[i] & !inside_quotations;
             let mut valid_whitespace = self.whitespace_bitsets[i] & !inside_quotations;
 
+            dbg!(valid_commas, valid_whitespace);
+
             let mut first_comma = valid_commas.leading_zeros() as usize;
             let mut first_lf = valid_whitespace.leading_zeros() as usize;
 
             if first_comma == 0 && first_lf == 0 {
+                // congratulations, we just skipped 64 bytes.
                 continue;
             }
 
@@ -168,6 +171,22 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn read_basic2() -> Result<()> {
+        let data = b"aaa,\"bbb\",ccc\n";
+
+        check_line(
+            data,
+            vec![vec![
+                "aaa".to_string(),
+                "\"bbb\"".to_string(),
+                "ccc".to_string(),
+            ]],
+        )?;
+
+        Ok(())
+    }
+
     #[test]
     fn read_nested() -> Result<()> {
         let data = b"\"aaa,howdy\",\"b\"\"bb\",\"ccc\"\n";
@@ -184,6 +203,15 @@ mod tests {
         Ok(())
     }
 
+    // #[test]
+    // fn read_nested2() -> Result<()> {
+    //     let data = b"aaa,\"b\"\"\"\n";
+    // 
+    //     check_line(data, vec![vec!["aaa".to_string(), "\"b\"\"\"".to_string()]])?;
+    // 
+    //     Ok(())
+    // }
+
     #[test]
     fn read_newline_field() -> Result<()> {
         let data = b"\"aaa,ho\nwdy\",\"b\"\"bb\",\"ccc\"\n";
@@ -216,10 +244,34 @@ mod tests {
         Ok(())
     }
 
-    /*
+    #[test]
+    fn read_taxi_zone_lookup_header() -> Result<()> {
+        let data = b"\"LocationID\",\"Borough\",\"Zone\",\"service_zone\"\r\n";
+
+        check_line(
+            data,
+            vec![vec![
+                r#""LocationID""#.to_string(),
+                r#""Borough""#.to_string(),
+                r#""Zone""#.to_string(),
+                r#""service_zone""#.to_string(),
+            ]],
+        )?;
+
+        Ok(())
+    }
+
     #[test]
     fn read_taxi_zone_lookup() -> Result<()> {
         let data = std::fs::read("taxi_zone_lookup.csv")?;
+        assert_eq!(
+            b"\"LocationID\",\"Borough\",\"Zone\",\"service_zone\"\r\n",
+            &data[..46]
+        );
+
+        let rows = CsvReader::new(&data).read()?;
+        assert_eq!(rows.len(), 1);
+        dbg!(&rows);
 
         for row in CsvReader::new(&data).read()? {
             let fields = row
@@ -231,9 +283,6 @@ mod tests {
             dbg!(fields);
         }
 
-        assert!(false, "Just to see the dbg");
-
         Ok(())
     }
-     */
 }