Merge pull request #41 from WenqingZong/wenqing/fix_incorrect_width

[Bug Fix] Incorrect line width caused by whitespaces

Cargo.toml

@@ -21,6 +21,7 @@ license = "MIT OR Apache-2.0"
 clippy = []
 
 [dependencies]
+regex = "1.10.4"
 smallvec = "1.11.2"
 textwrap = "0.16.0"
 unicode-width = "0.1.11"

src/lib.rs

@@ -1,3 +1,4 @@
+use regex::Regex;
 use smallvec::*;
 use std::io::{Result, Write};
 use textwrap::fill;
@@ -84,8 +85,11 @@ where
     // Final output is stored here
     let mut write_buffer = SmallVec::<[u8; BUFSIZE]>::new();
 
+    // Pre process to merge continuous whitespaces into one space character
+    let input = merge_white_spaces(input);
+
     // Let textwrap work its magic
-    let wrapped = fill(input, max_width);
+    let wrapped = fill(input.as_str(), max_width);
 
     let lines: Vec<&str> = wrapped.lines().collect();
 
@@ -147,3 +151,9 @@ fn longest_line(lines: &[&str]) -> usize {
         .max()
         .unwrap_or(0)
 }
+
+/// Merge continues white spaces into one space character while preserving newline characters.
+fn merge_white_spaces(input: &str) -> String {
+    let re = Regex::new(r"([^\S\r\n])+").unwrap();
+    re.replace_all(input, " ").to_string()
+}

tests/integration_test.rs

@@ -253,3 +253,58 @@ fn multibyte_string() -> Result<(), ()> {
     assert_eq!(&expected[1..], actual);
     Ok(())
 }
+
+#[test]
+fn string_with_continues_white_spaces() -> Result<(), ()> {
+    #[cfg(not(feature = "clippy"))]
+    let expected = r#"
+ __________________________________
+/  A string with many white spaces \
+|                                  |
+|                                  |
+\  And many newline characters     /
+ ----------------------------------
+        \
+         \
+            _~^~^~_
+        \) /  o o  \ (/
+          '_   -   _'
+          / '-----' \
+"#;
+    #[cfg(feature = "clippy")]
+    let expected = r#"
+ __________________________________
+/  A string with many white spaces \
+|                                  |
+|                                  |
+\  And many newline characters     /
+ ----------------------------------
+        \
+         \
+            __
+           /  \
+           |  |
+           @  @
+           |  |
+           || |/
+           || ||
+           |\_/|
+           \___/
+"#;
+
+    let input = "			A string        with 					many white spaces
+					
+
+    And many newline characters";
+    let width = DEFAULT_WIDTH;
+
+    let mut vec = Vec::new();
+
+    say(input, width, &mut vec).unwrap();
+
+    let actual = std::str::from_utf8(&vec).unwrap();
+
+    assert_eq!(&expected[1..], actual);
+
+    Ok(())
+}