From db4f985e2314f8caecb030851104d66561ed1e3c Mon Sep 17 00:00:00 2001
From: Tom Caruso <carusot42@gmail.com>
Date: Tue, 12 Apr 2022 16:11:02 -0400
Subject: [PATCH 1/2] initial commit: option to interpret null byte as an empty
 cell

---
 src/main.rs    | 28 ++++++++++++++++++++++------
 tests/tests.rs | 42 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+), 6 deletions(-)

diff --git a/src/main.rs b/src/main.rs
index defa4be..d09360b 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -99,6 +99,13 @@ struct Opt {
     /// quoting.
     #[structopt(value_name = "CHAR", long = "quote", default_value = "\"")]
     quote: CharSpecifier,
+
+    /// Should a null-byte (\x0) be interpreted as an empty cell
+    /// this option exists because it is not always possible to pass
+    /// a null byte into --null on the command line.
+    /// https://github.com/faradayio/scrubcsv/issues/17
+    #[structopt(long = "null-byte-as-empty")]
+    null_byte_as_empty: bool
 }
 
 lazy_static! {
@@ -122,16 +129,25 @@ fn run() -> Result<()> {
     // Remember the time we started.
     let start_time = now();
 
-    // Build a regex containing our `--null` value.
-    let null_re = if let Some(null_re_str) = opt.null.as_ref() {
-        // Always match the full CSV value.
-        let s = format!("^{}$", null_re_str);
-        let re = Regex::new(&s).context("can't compile regular expression")?;
-        Some(re)
+    let mut pattern= if opt.null_byte_as_empty == true {
+        Some(String::from(r"\x00"))
     } else {
         None
     };
 
+    if let Some(null_re_str) = opt.null.as_ref() {
+        pattern = match pattern {
+            Some(p) => Some(format!(r"{}|^{}$", p, null_re_str)),
+            None => Some(format!("^{}$", null_re_str)),
+        }
+    }
+
+    // Build a regex containing our `--null` value(s).
+    let null_re = match pattern {
+        Some(p) => Some(Regex::new(&p).context("can't build regular expression")?),
+        None => None,
+    };
+
     // Fetch our input from either standard input or a file.  The only tricky
     // detail here is that we use a `Box<dyn Read>` to represent "some object
     // implementing `Read`, stored on the heap."  This allows us to do runtime
diff --git a/tests/tests.rs b/tests/tests.rs
index b47dd37..83787a2 100644
--- a/tests/tests.rs
+++ b/tests/tests.rs
@@ -200,3 +200,45 @@ a,b,c
 "#
     );
 }
+
+#[test]
+fn test_null_byte_as_empty_cell() {
+    let testdir = TestDir::new("scrubcsv", "null_byte_as_empty");
+    let output = testdir
+        .cmd()
+        .arg("--null-byte-as-empty")
+        .output_with_stdin(
+            br#"c1,c2,c3
+1,2,3
+\x00,\x00,1
+"#,
+        ).expect("error running scrubcsv test");
+    assert_eq!(
+        output.stdout_str(),
+        r#"c1,c2,c3
+1,2,3
+,,1
+"#
+    );
+
+}#[test]
+fn test_null_byte_as_empty_cell_with_null_re() {
+    let testdir = TestDir::new("scrubcsv", "null_byte_as_empty_with_null_re");
+    let output = testdir
+        .cmd()
+        .arg("--null-byte-as-empty")
+        .args(&["--null", "NULL"])
+        .output_with_stdin(
+            br#"c1,c2,c3
+1,2,NULL
+\x00,\x00,1
+"#,
+        ).expect("error running scrubcsv test");
+    assert_eq!(
+        output.stdout_str(),
+        r#"c1,c2,c3
+1,2,
+,,1
+"#
+    );
+}

From beef518d40dbba5f90a3918cf9d5e5463258e3d2 Mon Sep 17 00:00:00 2001
From: Tom Caruso <carusot42@gmail.com>
Date: Tue, 12 Apr 2022 16:13:25 -0400
Subject: [PATCH 2/2] cargo fmt

---
 src/main.rs    |  4 ++--
 tests/tests.rs | 10 ++++++----
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/src/main.rs b/src/main.rs
index d09360b..d74ce90 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -105,7 +105,7 @@ struct Opt {
     /// a null byte into --null on the command line.
     /// https://github.com/faradayio/scrubcsv/issues/17
     #[structopt(long = "null-byte-as-empty")]
-    null_byte_as_empty: bool
+    null_byte_as_empty: bool,
 }
 
 lazy_static! {
@@ -129,7 +129,7 @@ fn run() -> Result<()> {
     // Remember the time we started.
     let start_time = now();
 
-    let mut pattern= if opt.null_byte_as_empty == true {
+    let mut pattern = if opt.null_byte_as_empty == true {
         Some(String::from(r"\x00"))
     } else {
         None
diff --git a/tests/tests.rs b/tests/tests.rs
index 83787a2..f86f782 100644
--- a/tests/tests.rs
+++ b/tests/tests.rs
@@ -212,7 +212,8 @@ fn test_null_byte_as_empty_cell() {
 1,2,3
 \x00,\x00,1
 "#,
-        ).expect("error running scrubcsv test");
+        )
+        .expect("error running scrubcsv test");
     assert_eq!(
         output.stdout_str(),
         r#"c1,c2,c3
@@ -220,8 +221,8 @@ fn test_null_byte_as_empty_cell() {
 ,,1
 "#
     );
-
-}#[test]
+}
+#[test]
 fn test_null_byte_as_empty_cell_with_null_re() {
     let testdir = TestDir::new("scrubcsv", "null_byte_as_empty_with_null_re");
     let output = testdir
@@ -233,7 +234,8 @@ fn test_null_byte_as_empty_cell_with_null_re() {
 1,2,NULL
 \x00,\x00,1
 "#,
-        ).expect("error running scrubcsv test");
+        )
+        .expect("error running scrubcsv test");
     assert_eq!(
         output.stdout_str(),
         r#"c1,c2,c3