]> git.lizzy.rs Git - rust.git/blobdiff - src/string.rs
Use concat() instead of join("")
[rust.git] / src / string.rs
index 904260e81820462c987f9b27c07f7fd2c00d1b8a..d45f15a1c9bab59cb85710d453c096347cd6a748 100644 (file)
 
 // Format string literals.
 
-use unicode_segmentation::UnicodeSegmentation;
 use regex::Regex;
+use unicode_segmentation::UnicodeSegmentation;
 
-use Shape;
 use config::Config;
+use shape::Shape;
 use utils::wrap_str;
 
-use MIN_STRING;
+const MIN_STRING: usize = 10;
 
+/// Describes the layout of a piece of text.
 pub struct StringFormat<'a> {
+    /// The opening sequence of characters for the piece of text
     pub opener: &'a str,
+    /// The closing sequence of characters for the piece of text
     pub closer: &'a str,
+    /// The opening sequence of characters for a line
     pub line_start: &'a str,
+    /// The closing sequence of characters for a line
     pub line_end: &'a str,
+    /// The allocated box to fit the text into
     pub shape: Shape,
+    /// Trim trailing whitespaces
     pub trim_end: bool,
     pub config: &'a Config,
 }
 
-// FIXME: simplify this!
-pub fn rewrite_string<'a>(orig: &str, fmt: &StringFormat<'a>) -> Option<String> {
+impl<'a> StringFormat<'a> {
+    pub fn new(shape: Shape, config: &'a Config) -> StringFormat<'a> {
+        StringFormat {
+            opener: "\"",
+            closer: "\"",
+            line_start: " ",
+            line_end: "\\",
+            shape,
+            trim_end: false,
+            config,
+        }
+    }
+
+    /// Returns the maximum number of graphemes that is possible on a line while taking the
+    /// indentation into account.
+    ///
+    /// If we cannot put at least a single character per line, the rewrite won't succeed.
+    fn max_chars_with_indent(&self) -> Option<usize> {
+        Some(
+            self.shape
+                .width
+                .checked_sub(self.opener.len() + self.line_end.len() + 1)?
+                + 1,
+        )
+    }
+
+    /// Like max_chars_with_indent but the indentation is not substracted.
+    /// This allows to fit more graphemes from the string on a line when
+    /// SnippetState::EndWithLineFeed.
+    fn max_chars_without_indent(&self) -> Option<usize> {
+        Some(self.config.max_width().checked_sub(self.line_end.len())?)
+    }
+}
+
+pub fn rewrite_string<'a>(
+    orig: &str,
+    fmt: &StringFormat<'a>,
+    newline_max_chars: usize,
+) -> Option<String> {
+    let max_chars_with_indent = fmt.max_chars_with_indent()?;
+    let max_chars_without_indent = fmt.max_chars_without_indent()?;
+    let indent_with_newline = fmt.shape.indent.to_string_with_newline(fmt.config);
+    let indent_without_newline = fmt.shape.indent.to_string(fmt.config);
+
     // Strip line breaks.
-    let re = Regex::new(r"([^\\](\\\\)*)\\[\n\r][[:space:]]*").unwrap();
-    let stripped_str = re.replace_all(orig, "$1");
+    // With this regex applied, all remaining whitespaces are significant
+    let strip_line_breaks_re = Regex::new(r"([^\\](\\\\)*)\\[\n\r][[:space:]]*").unwrap();
+    let stripped_str = strip_line_breaks_re.replace_all(orig, "$1");
 
     let graphemes = UnicodeSegmentation::graphemes(&*stripped_str, false).collect::<Vec<&str>>();
-    let shape = fmt.shape.visual_indent(0);
-    let indent = shape.indent.to_string(fmt.config);
-    let punctuation = ":,;.";
 
     // `cur_start` is the position in `orig` of the start of the current line.
     let mut cur_start = 0;
@@ -50,98 +97,582 @@ pub fn rewrite_string<'a>(orig: &str, fmt: &StringFormat<'a>) -> Option<String>
     );
     result.push_str(fmt.opener);
 
-    let ender_length = fmt.line_end.len();
-    // If we cannot put at least a single character per line, the rewrite won't
-    // succeed.
-    let max_chars = try_opt!(shape.width.checked_sub(fmt.opener.len() + ender_length + 1)) + 1;
-
-    // Snip a line at a time from `orig` until it is used up. Push the snippet
+    // Snip a line at a time from `stripped_str` until it is used up. Push the snippet
     // onto result.
-    'outer: loop {
-        // `cur_end` will be where we break the line, as an offset into `orig`.
-        // Initialised to the maximum it could be (which may be beyond `orig`).
-        let mut cur_end = cur_start + max_chars;
-
-        // We can fit the rest of the string on this line, so we're done.
-        if cur_end >= graphemes.len() {
-            let line = &graphemes[cur_start..].join("");
-            result.push_str(line);
-            break 'outer;
+    let mut cur_max_chars = max_chars_with_indent;
+    let is_bareline_ok = fmt.line_start.is_empty() || is_whitespace(fmt.line_start);
+    loop {
+        // All the input starting at cur_start fits on the current line
+        if graphemes.len() - cur_start <= cur_max_chars {
+            for (i, grapheme) in graphemes[cur_start..].iter().enumerate() {
+                if is_line_feed(grapheme) {
+                    // take care of blank lines
+                    result = trim_right_but_line_feed(fmt.trim_end, result);
+                    result.push_str("\n");
+                    if !is_bareline_ok && cur_start + i + 1 < graphemes.len() {
+                        result.push_str(&indent_without_newline);
+                        result.push_str(fmt.line_start);
+                    }
+                } else {
+                    result.push_str(grapheme);
+                }
+            }
+            result = trim_right_but_line_feed(fmt.trim_end, result);
+            break;
+        }
+
+        // The input starting at cur_start needs to be broken
+        match break_string(
+            cur_max_chars,
+            fmt.trim_end,
+            fmt.line_end,
+            &graphemes[cur_start..],
+        ) {
+            SnippetState::LineEnd(line, len) => {
+                result.push_str(&line);
+                result.push_str(fmt.line_end);
+                result.push_str(&indent_with_newline);
+                result.push_str(fmt.line_start);
+                cur_max_chars = newline_max_chars;
+                cur_start += len;
+            }
+            SnippetState::EndWithLineFeed(line, len) => {
+                if line == "\n" && fmt.trim_end {
+                    result = result.trim_right().to_string();
+                }
+                result.push_str(&line);
+                if is_bareline_ok {
+                    // the next line can benefit from the full width
+                    cur_max_chars = max_chars_without_indent;
+                } else {
+                    result.push_str(&indent_without_newline);
+                    result.push_str(fmt.line_start);
+                    cur_max_chars = max_chars_with_indent;
+                }
+                cur_start += len;
+            }
+            SnippetState::EndOfInput(line) => {
+                result.push_str(&line);
+                break;
+            }
+        }
+    }
+
+    result.push_str(fmt.closer);
+    wrap_str(result, fmt.config.max_width(), fmt.shape)
+}
+
+/// Returns the index to the end of the url if the given string includes an
+/// URL or alike. Otherwise, returns None;
+fn detect_url(s: &[&str], index: usize) -> Option<usize> {
+    let start = match s[..=index].iter().rposition(|g| is_whitespace(g)) {
+        Some(pos) => pos + 1,
+        None => 0,
+    };
+    if s.len() < start + 8 {
+        return None;
+    }
+    let prefix = s[start..start + 8].concat();
+    if prefix.starts_with("https://")
+        || prefix.starts_with("http://")
+        || prefix.starts_with("ftp://")
+        || prefix.starts_with("file://")
+    {
+        match s[index..].iter().position(|g| is_whitespace(g)) {
+            Some(pos) => Some(index + pos - 1),
+            None => Some(s.len() - 1),
         }
+    } else {
+        None
+    }
+}
+
+/// Trims whitespaces to the right except for the line feed character.
+fn trim_right_but_line_feed(trim_end: bool, result: String) -> String {
+    let whitespace_except_line_feed = |c: char| c.is_whitespace() && c != '\n';
+    if trim_end && result.ends_with(whitespace_except_line_feed) {
+        result
+            .trim_right_matches(whitespace_except_line_feed)
+            .to_string()
+    } else {
+        result
+    }
+}
+
+/// Result of breaking a string so it fits in a line and the state it ended in.
+/// The state informs about what to do with the snippet and how to continue the breaking process.
+#[derive(Debug, PartialEq)]
+enum SnippetState {
+    /// The input could not be broken and so rewriting the string is finished.
+    EndOfInput(String),
+    /// The input could be broken and the returned snippet should be ended with a
+    /// `[StringFormat::line_end]`. The next snippet needs to be indented.
+    ///
+    /// The returned string is the line to print out and the number is the length that got read in
+    /// the text being rewritten. That length may be greater than the returned string if trailing
+    /// whitespaces got trimmed.
+    LineEnd(String, usize),
+    /// The input could be broken but a newline is present that cannot be trimmed. The next snippet
+    /// to be rewritten *could* use more width than what is specified by the given shape. For
+    /// example with a multiline string, the next snippet does not need to be indented, allowing
+    /// more characters to be fit within a line.
+    ///
+    /// The returned string is the line to print out and the number is the length that got read in
+    /// the text being rewritten.
+    EndWithLineFeed(String, usize),
+}
+
+fn not_whitespace_except_line_feed(g: &str) -> bool {
+    is_line_feed(g) || !is_whitespace(g)
+}
 
-        // Push cur_end left until we reach whitespace (or the line is too small).
-        while !graphemes[cur_end - 1].trim().is_empty() {
-            cur_end -= 1;
-            if cur_end < cur_start + MIN_STRING {
-                // We couldn't find whitespace before the string got too small.
-                // So start again at the max length and look for punctuation.
-                cur_end = cur_start + max_chars;
-                while !punctuation.contains(graphemes[cur_end - 1]) {
-                    cur_end -= 1;
-
-                    // If we can't break at whitespace or punctuation, grow the string instead.
-                    if cur_end < cur_start + MIN_STRING {
-                        cur_end = cur_start + max_chars;
-                        while !(punctuation.contains(graphemes[cur_end - 1]) ||
-                              graphemes[cur_end - 1].trim().is_empty())
-                        {
-                            if cur_end >= graphemes.len() {
-                                let line = &graphemes[cur_start..].join("");
-                                result.push_str(line);
-                                break 'outer;
-                            }
-                            cur_end += 1;
-                        }
-                        break;
+/// Break the input string at a boundary character around the offset `max_chars`. A boundary
+/// character is either a punctuation or a whitespace.
+fn break_string(max_chars: usize, trim_end: bool, line_end: &str, input: &[&str]) -> SnippetState {
+    let break_at = |index /* grapheme at index is included */| {
+        // Take in any whitespaces to the left/right of `input[index]` while
+        // preserving line feeds
+        let index_minus_ws = input[0..=index]
+            .iter()
+            .rposition(|grapheme| not_whitespace_except_line_feed(grapheme))
+            .unwrap_or(index);
+        // Take into account newlines occuring in input[0..=index], i.e., the possible next new
+        // line. If there is one, then text after it could be rewritten in a way that the available
+        // space is fully used.
+        for (i, grapheme) in input[0..=index].iter().enumerate() {
+            if is_line_feed(grapheme) {
+                if i <= index_minus_ws {
+                    let mut line = &input[0..i].concat()[..];
+                    if trim_end {
+                        line = line.trim_right();
                     }
+                    return SnippetState::EndWithLineFeed(format!("{}\n", line), i + 1);
                 }
                 break;
             }
         }
-        // Make sure there is no whitespace to the right of the break.
-        while cur_end < stripped_str.len() && graphemes[cur_end].trim().is_empty() {
-            cur_end += 1;
+
+        let mut index_plus_ws = index;
+        for (i, grapheme) in input[index + 1..].iter().enumerate() {
+            if !trim_end && is_line_feed(grapheme) {
+                return SnippetState::EndWithLineFeed(
+                    input[0..=index + 1 + i].concat(),
+                    index + 2 + i,
+                );
+            } else if not_whitespace_except_line_feed(grapheme) {
+                index_plus_ws = index + i;
+                break;
+            }
         }
 
-        // Make the current line and add it on to result.
-        let raw_line = graphemes[cur_start..cur_end].join("");
-        let line = if fmt.trim_end {
-            raw_line.trim()
+        if trim_end {
+            SnippetState::LineEnd(input[0..=index_minus_ws].concat(), index_plus_ws + 1)
         } else {
-            raw_line.as_str()
+            SnippetState::LineEnd(input[0..=index_plus_ws].concat(), index_plus_ws + 1)
+        }
+    };
+
+    // Find the position in input for breaking the string
+    if line_end.is_empty()
+        && trim_end
+        && !is_whitespace(input[max_chars - 1])
+        && is_whitespace(input[max_chars])
+    {
+        // At a breaking point already
+        // The line won't invalidate the rewriting because:
+        // - no extra space needed for the line_end character
+        // - extra whitespaces to the right can be trimmed
+        return break_at(max_chars - 1);
+    }
+    if let Some(url_index_end) = detect_url(input, max_chars) {
+        let index_plus_ws = url_index_end + input[url_index_end..]
+            .iter()
+            .skip(1)
+            .position(|grapheme| not_whitespace_except_line_feed(grapheme))
+            .unwrap_or(0);
+        return if trim_end {
+            SnippetState::LineEnd(input[..=url_index_end].concat(), index_plus_ws + 1)
+        } else {
+            return SnippetState::LineEnd(input[..=index_plus_ws].concat(), index_plus_ws + 1);
         };
+    }
+    match input[0..max_chars]
+        .iter()
+        .rposition(|grapheme| is_whitespace(grapheme))
+    {
+        // Found a whitespace and what is on its left side is big enough.
+        Some(index) if index >= MIN_STRING => break_at(index),
+        // No whitespace found, try looking for a punctuation instead
+        _ => match input[0..max_chars]
+            .iter()
+            .rposition(|grapheme| is_punctuation(grapheme))
+        {
+            // Found a punctuation and what is on its left side is big enough.
+            Some(index) if index >= MIN_STRING => break_at(index),
+            // Either no boundary character was found to the left of `input[max_chars]`, or the line
+            // got too small. We try searching for a boundary character to the right.
+            _ => match input[max_chars..]
+                .iter()
+                .position(|grapheme| is_whitespace(grapheme) || is_punctuation(grapheme))
+            {
+                // A boundary was found after the line limit
+                Some(index) => break_at(max_chars + index),
+                // No boundary to the right, the input cannot be broken
+                None => SnippetState::EndOfInput(input.concat()),
+            },
+        },
+    }
+}
 
-        result.push_str(line);
-        result.push_str(fmt.line_end);
-        result.push('\n');
-        result.push_str(&indent);
-        result.push_str(fmt.line_start);
+fn is_line_feed(grapheme: &str) -> bool {
+    grapheme.as_bytes()[0] == b'\n'
+}
 
-        // The next line starts where the current line ends.
-        cur_start = cur_end;
-    }
+fn is_whitespace(grapheme: &str) -> bool {
+    grapheme.chars().all(|c| c.is_whitespace())
+}
 
-    result.push_str(fmt.closer);
-    wrap_str(result, fmt.config.max_width(), fmt.shape)
+fn is_punctuation(grapheme: &str) -> bool {
+    match grapheme.as_bytes()[0] {
+        b':' | b',' | b';' | b'.' => true,
+        _ => false,
+    }
 }
 
 #[cfg(test)]
 mod test {
-    use super::{StringFormat, rewrite_string};
+    use super::{break_string, detect_url, rewrite_string, SnippetState, StringFormat};
+    use config::Config;
+    use shape::{Indent, Shape};
+    use unicode_segmentation::UnicodeSegmentation;
 
     #[test]
     fn issue343() {
         let config = Default::default();
+        let fmt = StringFormat::new(Shape::legacy(2, Indent::empty()), &config);
+        rewrite_string("eq_", &fmt, 2);
+    }
+
+    #[test]
+    fn should_break_on_whitespace() {
+        let string = "Placerat felis. Mauris porta ante sagittis purus.";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(20, false, "", &graphemes[..]),
+            SnippetState::LineEnd("Placerat felis. ".to_string(), 16)
+        );
+        assert_eq!(
+            break_string(20, true, "", &graphemes[..]),
+            SnippetState::LineEnd("Placerat felis.".to_string(), 16)
+        );
+    }
+
+    #[test]
+    fn should_break_on_punctuation() {
+        let string = "Placerat_felis._Mauris_porta_ante_sagittis_purus.";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(20, false, "", &graphemes[..]),
+            SnippetState::LineEnd("Placerat_felis.".to_string(), 15)
+        );
+    }
+
+    #[test]
+    fn should_break_forward() {
+        let string = "Venenatis_tellus_vel_tellus. Aliquam aliquam dolor at justo.";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(20, false, "", &graphemes[..]),
+            SnippetState::LineEnd("Venenatis_tellus_vel_tellus. ".to_string(), 29)
+        );
+        assert_eq!(
+            break_string(20, true, "", &graphemes[..]),
+            SnippetState::LineEnd("Venenatis_tellus_vel_tellus.".to_string(), 29)
+        );
+    }
+
+    #[test]
+    fn nothing_to_break() {
+        let string = "Venenatis_tellus_vel_tellus";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(20, false, "", &graphemes[..]),
+            SnippetState::EndOfInput("Venenatis_tellus_vel_tellus".to_string())
+        );
+    }
+
+    #[test]
+    fn significant_whitespaces() {
+        let string = "Neque in sem.      \n      Pellentesque tellus augue.";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(15, false, "", &graphemes[..]),
+            SnippetState::EndWithLineFeed("Neque in sem.      \n".to_string(), 20)
+        );
+        assert_eq!(
+            break_string(25, false, "", &graphemes[..]),
+            SnippetState::EndWithLineFeed("Neque in sem.      \n".to_string(), 20)
+        );
+
+        assert_eq!(
+            break_string(15, true, "", &graphemes[..]),
+            SnippetState::LineEnd("Neque in sem.".to_string(), 19)
+        );
+        assert_eq!(
+            break_string(25, true, "", &graphemes[..]),
+            SnippetState::EndWithLineFeed("Neque in sem.\n".to_string(), 20)
+        );
+    }
+
+    #[test]
+    fn big_whitespace() {
+        let string = "Neque in sem.            Pellentesque tellus augue.";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(20, false, "", &graphemes[..]),
+            SnippetState::LineEnd("Neque in sem.            ".to_string(), 25)
+        );
+        assert_eq!(
+            break_string(20, true, "", &graphemes[..]),
+            SnippetState::LineEnd("Neque in sem.".to_string(), 25)
+        );
+    }
+
+    #[test]
+    fn newline_in_candidate_line() {
+        let string = "Nulla\nconsequat erat at massa. Vivamus id mi.";
+
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(
+            break_string(25, false, "", &graphemes[..]),
+            SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
+        );
+        assert_eq!(
+            break_string(25, true, "", &graphemes[..]),
+            SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
+        );
+
+        let mut config: Config = Default::default();
+        config.set().max_width(27);
+        let fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
+        let rewritten_string = rewrite_string(string, &fmt, 27);
+        assert_eq!(
+            rewritten_string,
+            Some("\"Nulla\nconsequat erat at massa. \\\n Vivamus id mi.\"".to_string())
+        );
+    }
+
+    #[test]
+    fn last_line_fit_with_trailing_whitespaces() {
+        let string = "Vivamus id mi.  ";
+        let config: Config = Default::default();
+        let mut fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
+
+        fmt.trim_end = true;
+        let rewritten_string = rewrite_string(string, &fmt, 25);
+        assert_eq!(rewritten_string, Some("\"Vivamus id mi.\"".to_string()));
+
+        fmt.trim_end = false; // default value of trim_end
+        let rewritten_string = rewrite_string(string, &fmt, 25);
+        assert_eq!(rewritten_string, Some("\"Vivamus id mi.  \"".to_string()));
+    }
+
+    #[test]
+    fn last_line_fit_with_newline() {
+        let string = "Vivamus id mi.\nVivamus id mi.";
+        let config: Config = Default::default();
         let fmt = StringFormat {
-            opener: "\"",
-            closer: "\"",
-            line_start: " ",
-            line_end: "\\",
-            shape: ::Shape::legacy(2, ::Indent::empty()),
-            trim_end: false,
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "",
+            shape: Shape::legacy(100, Indent::from_width(&config, 4)),
+            trim_end: true,
+            config: &config,
+        };
+
+        let rewritten_string = rewrite_string(string, &fmt, 100);
+        assert_eq!(
+            rewritten_string,
+            Some("Vivamus id mi.\n    // Vivamus id mi.".to_string())
+        );
+    }
+
+    #[test]
+    fn overflow_in_non_string_content() {
+        let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
+        let config: Config = Default::default();
+        let fmt = StringFormat {
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "",
+            shape: Shape::legacy(30, Indent::from_width(&config, 8)),
+            trim_end: true,
+            config: &config,
+        };
+
+        assert_eq!(
+            rewrite_string(comment, &fmt, 30),
+            Some(
+                "Aenean metus.\n        // Vestibulum ac lacus. Vivamus\n        // porttitor"
+                    .to_string()
+            )
+        );
+    }
+
+    #[test]
+    fn overflow_in_non_string_content_with_line_end() {
+        let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
+        let config: Config = Default::default();
+        let fmt = StringFormat {
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "@",
+            shape: Shape::legacy(30, Indent::from_width(&config, 8)),
+            trim_end: true,
+            config: &config,
+        };
+
+        assert_eq!(
+            rewrite_string(comment, &fmt, 30),
+            Some(
+                "Aenean metus.\n        // Vestibulum ac lacus. Vivamus@\n        // porttitor"
+                    .to_string()
+            )
+        );
+    }
+
+    #[test]
+    fn blank_line_with_non_empty_line_start() {
+        let config: Config = Default::default();
+        let mut fmt = StringFormat {
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "",
+            shape: Shape::legacy(30, Indent::from_width(&config, 4)),
+            trim_end: true,
+            config: &config,
+        };
+
+        let comment = "Aenean metus. Vestibulum\n\nac lacus. Vivamus porttitor";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 30),
+            Some(
+                "Aenean metus. Vestibulum\n    //\n    // ac lacus. Vivamus porttitor".to_string()
+            )
+        );
+
+        fmt.shape = Shape::legacy(15, Indent::from_width(&config, 4));
+        let comment = "Aenean\n\nmetus. Vestibulum ac lacus. Vivamus porttitor";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 15),
+            Some(
+                r#"Aenean
+    //
+    // metus. Vestibulum
+    // ac lacus. Vivamus
+    // porttitor"#
+                    .to_string()
+            )
+        );
+    }
+
+    #[test]
+    fn retain_blank_lines() {
+        let config: Config = Default::default();
+        let fmt = StringFormat {
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "",
+            shape: Shape::legacy(20, Indent::from_width(&config, 4)),
+            trim_end: true,
             config: &config,
         };
 
-        rewrite_string("eq_", &fmt);
+        let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n\n";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 20),
+            Some(
+                "Aenean\n    //\n    // metus. Vestibulum ac\n    // lacus.\n    //\n".to_string()
+            )
+        );
+
+        let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 20),
+            Some("Aenean\n    //\n    // metus. Vestibulum ac\n    // lacus.\n".to_string())
+        );
+
+        let comment = "Aenean\n        \nmetus. Vestibulum ac lacus.";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 20),
+            Some("Aenean\n    //\n    // metus. Vestibulum ac\n    // lacus.".to_string())
+        );
+    }
+
+    #[test]
+    fn boundary_on_edge() {
+        let config: Config = Default::default();
+        let mut fmt = StringFormat {
+            opener: "",
+            closer: "",
+            line_start: "// ",
+            line_end: "",
+            shape: Shape::legacy(13, Indent::from_width(&config, 4)),
+            trim_end: true,
+            config: &config,
+        };
+
+        let comment = "Aenean metus. Vestibulum ac lacus.";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 13),
+            Some("Aenean metus.\n    // Vestibulum ac\n    // lacus.".to_string())
+        );
+
+        fmt.trim_end = false;
+        let comment = "Vestibulum ac lacus.";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 13),
+            Some("Vestibulum \n    // ac lacus.".to_string())
+        );
+
+        fmt.trim_end = true;
+        fmt.line_end = "\\";
+        let comment = "Vestibulum ac lacus.";
+        assert_eq!(
+            rewrite_string(comment, &fmt, 13),
+            Some("Vestibulum\\\n    // ac lacus.".to_string())
+        );
+    }
+
+    #[test]
+    fn detect_urls() {
+        let string = "aaa http://example.org something";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 8), Some(21));
+
+        let string = "https://example.org something";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 0), Some(18));
+
+        let string = "aaa ftp://example.org something";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 8), Some(20));
+
+        let string = "aaa file://example.org something";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 8), Some(21));
+
+        let string = "aaa http not an url";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 6), None);
+
+        let string = "aaa file://example.org";
+        let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
+        assert_eq!(detect_url(&graphemes, 8), Some(21));
     }
 }