add fuzzer input generator

4 files changed, 91 insertions, 59 deletions

diff --git a/bin/regen-fuzz b/bin/regen-fuzz
new file mode 100755
index 0000000..cda3cf8
--- /dev/null
+++ b/bin/regen-fuzz
@@ -0,0 +1,6 @@
+#!/bin/sh
+set -eu
+
+for file in tests/data/fixtures/*.in; do
+    cargo run --example generate_fuzz "$(basename "$file" .in)"
+done
diff --git a/examples/generate_fixture.rs b/examples/generate_fixture.rs
index 6e36a42..b4f372d 100644
--- a/examples/generate_fixture.rs
+++ b/examples/generate_fixture.rs
@@ -20,7 +20,7 @@ fn main() {
     for line in inputs.lines() {
         let line = line.unwrap();
 
-        let input = unhex(line.as_bytes());
+        let input = helpers::unhex(line.as_bytes());
         let mut input_file = std::fs::File::create(format!(
             "tests/data/fixtures/{}/{}.typescript",
             name, i
@@ -43,61 +43,3 @@ fn main() {
         i += 1;
     }
 }
-
-fn unhex(s: &[u8]) -> Vec<u8> {
-    let mut ret = vec![];
-    let mut i = 0;
-    while i < s.len() {
-        if s[i] == b'\\' {
-            match s[i + 1] {
-                b'\\' => {
-                    ret.push(b'\\');
-                    i += 2;
-                }
-                b'x' => {
-                    let upper = s[i + 2];
-                    let lower = s[i + 3];
-                    ret.push(helpers::hex(upper, lower).unwrap());
-                    i += 4;
-                }
-                b'u' => {
-                    assert_eq!(s[i + 2], b'{');
-                    let mut digits = vec![];
-                    let mut j = i + 3;
-                    while s[j] != b'}' {
-                        digits.push(s[j]);
-                        j += 1;
-                    }
-                    let digits: Vec<_> = digits
-                        .iter()
-                        .copied()
-                        .skip_while(|x| x == &b'0')
-                        .collect();
-                    let digits = String::from_utf8(digits).unwrap();
-                    let codepoint = u32::from_str_radix(&digits, 16).unwrap();
-                    let c = char::try_from(codepoint).unwrap();
-                    let mut bytes = [0; 4];
-                    ret.extend(c.encode_utf8(&mut bytes).bytes());
-                    i = j + 1;
-                }
-                b'r' => {
-                    ret.push(0x0d);
-                    i += 2;
-                }
-                b'n' => {
-                    ret.push(0x0a);
-                    i += 2;
-                }
-                b't' => {
-                    ret.push(0x09);
-                    i += 2;
-                }
-                _ => panic!("invalid escape"),
-            }
-        } else {
-            ret.push(s[i]);
-            i += 1;
-        }
-    }
-    ret
-}
diff --git a/examples/generate_fuzz.rs b/examples/generate_fuzz.rs
new file mode 100644
index 0000000..92c983b
--- /dev/null
+++ b/examples/generate_fuzz.rs
@@ -0,0 +1,25 @@
+use std::io::BufRead as _;
+use std::io::Write as _;
+
+#[path = "../tests/helpers/mod.rs"]
+mod helpers;
+
+fn main() {
+    let name = std::env::args().nth(1).unwrap();
+    let _ = std::fs::remove_file(format!("fuzz/in/{}", name));
+
+    let inputs =
+        std::fs::File::open(format!("tests/data/fixtures/{}.in", name))
+            .unwrap();
+    let inputs = std::io::BufReader::new(inputs);
+
+    let mut bytes = vec![];
+    for line in inputs.lines() {
+        let line = line.unwrap();
+        let input = helpers::unhex(line.as_bytes());
+        bytes.extend(input.iter());
+    }
+    let mut input_file =
+        std::fs::File::create(format!("fuzz/in/{}", name)).unwrap();
+    input_file.write_all(&bytes).unwrap();
+}
diff --git a/tests/helpers/mod.rs b/tests/helpers/mod.rs
index 20de2d4..d3dd05d 100644
--- a/tests/helpers/mod.rs
+++ b/tests/helpers/mod.rs
@@ -289,3 +289,62 @@ fn hex_char(c: u8) -> Result<u8, String> {
 pub fn hex(upper: u8, lower: u8) -> Result<u8, String> {
     Ok(hex_char(upper)? * 16 + hex_char(lower)?)
 }
+
+#[allow(dead_code)]
+pub fn unhex(s: &[u8]) -> Vec<u8> {
+    let mut ret = vec![];
+    let mut i = 0;
+    while i < s.len() {
+        if s[i] == b'\\' {
+            match s[i + 1] {
+                b'\\' => {
+                    ret.push(b'\\');
+                    i += 2;
+                }
+                b'x' => {
+                    let upper = s[i + 2];
+                    let lower = s[i + 3];
+                    ret.push(hex(upper, lower).unwrap());
+                    i += 4;
+                }
+                b'u' => {
+                    assert_eq!(s[i + 2], b'{');
+                    let mut digits = vec![];
+                    let mut j = i + 3;
+                    while s[j] != b'}' {
+                        digits.push(s[j]);
+                        j += 1;
+                    }
+                    let digits: Vec<_> = digits
+                        .iter()
+                        .copied()
+                        .skip_while(|x| x == &b'0')
+                        .collect();
+                    let digits = String::from_utf8(digits).unwrap();
+                    let codepoint = u32::from_str_radix(&digits, 16).unwrap();
+                    let c = char::try_from(codepoint).unwrap();
+                    let mut bytes = [0; 4];
+                    ret.extend(c.encode_utf8(&mut bytes).bytes());
+                    i = j + 1;
+                }
+                b'r' => {
+                    ret.push(0x0d);
+                    i += 2;
+                }
+                b'n' => {
+                    ret.push(0x0a);
+                    i += 2;
+                }
+                b't' => {
+                    ret.push(0x09);
+                    i += 2;
+                }
+                _ => panic!("invalid escape"),
+            }
+        } else {
+            ret.push(s[i]);
+            i += 1;
+        }
+    }
+    ret
+}