Modify the codemap code to use more slices and to information about

columns within a line, not just the line numbers. Also try to clarify
and use the term `line_index` when 0-based.
This commit is contained in:
Niko Matsakis 2015-04-09 14:46:03 -04:00
parent 16574e3fbd
commit 5156b3a6cd
2 changed files with 133 additions and 40 deletions

View File

@ -49,7 +49,7 @@ pub struct BytePos(pub u32);
/// A character offset. Because of multibyte utf8 characters, a byte offset
/// is not equivalent to a character offset. The CodeMap will convert BytePos
/// values to CharPos values as necessary.
#[derive(Copy, Clone, PartialEq, Hash, PartialOrd, Debug)]
#[derive(Copy, Clone, PartialEq, Eq, Hash, PartialOrd, Debug)]
pub struct CharPos(pub usize);
// FIXME: Lots of boilerplate in these impls, but so far my attempts to fix
@ -305,9 +305,21 @@ impl ExpnId {
pub type FileName = String;
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
pub struct LineInfo {
/// Index of line, starting from 0.
pub line_index: usize,
/// Column in line where span begins, starting from 0.
pub start_col: CharPos,
/// Column in line where span ends, starting from 0, exclusive.
pub end_col: CharPos,
}
pub struct FileLines {
pub file: Rc<FileMap>,
pub lines: Vec<usize>
pub lines: Vec<LineInfo>
}
/// Identifies an offset of a multi-byte character in a FileMap
@ -479,9 +491,9 @@ impl FileMap {
lines.push(pos);
}
/// get a line from the list of pre-computed line-beginnings
///
pub fn get_line(&self, line_number: usize) -> Option<String> {
/// get a line from the list of pre-computed line-beginnings.
/// line-number here is 0-based.
pub fn get_line(&self, line_number: usize) -> Option<&str> {
match self.src {
Some(ref src) => {
let lines = self.lines.borrow();
@ -492,7 +504,7 @@ impl FileMap {
match slice.find('\n') {
Some(e) => &slice[..e],
None => slice
}.to_string()
}
})
}
None => None
@ -661,10 +673,29 @@ impl CodeMap {
pub fn span_to_lines(&self, sp: Span) -> FileLines {
let lo = self.lookup_char_pos(sp.lo);
let hi = self.lookup_char_pos(sp.hi);
let mut lines = Vec::new();
for i in lo.line - 1..hi.line {
lines.push(i);
};
let mut lines = Vec::with_capacity(hi.line - lo.line + 1);
// The span starts partway through the first line,
// but after that it starts from offset 0.
let mut start_col = lo.col;
// For every line but the last, it extends from `start_col`
// and to the end of the line. Be careful because the line
// numbers in Loc are 1-based, so we subtract 1 to get 0-based
// lines.
for line_index in lo.line-1 .. hi.line-1 {
let line_len = lo.file.get_line(line_index).map(|s| s.len()).unwrap_or(0);
lines.push(LineInfo { line_index: line_index,
start_col: start_col,
end_col: CharPos::from_usize(line_len) });
start_col = CharPos::from_usize(0);
}
// For the last line, it extends from `start_col` to `hi.col`:
lines.push(LineInfo { line_index: hi.line - 1,
start_col: start_col,
end_col: hi.col });
FileLines {file: lo.file, lines: lines}
}
@ -919,6 +950,7 @@ pub struct MalformedCodemapPositions {
#[cfg(test)]
mod test {
use super::*;
use std::rc::Rc;
#[test]
fn t1 () {
@ -926,10 +958,10 @@ mod test {
let fm = cm.new_filemap("blork.rs".to_string(),
"first line.\nsecond line".to_string());
fm.next_line(BytePos(0));
assert_eq!(fm.get_line(0), Some("first line.".to_string()));
assert_eq!(fm.get_line(0), Some("first line."));
// TESTING BROKEN BEHAVIOR:
fm.next_line(BytePos(10));
assert_eq!(fm.get_line(1), Some(".".to_string()));
assert_eq!(fm.get_line(1), Some("."));
}
#[test]
@ -1057,7 +1089,54 @@ mod test {
assert_eq!(file_lines.file.name, "blork.rs");
assert_eq!(file_lines.lines.len(), 1);
assert_eq!(file_lines.lines[0], 1);
assert_eq!(file_lines.lines[0].line_index, 1);
}
/// Given a string like " ^~~~~~~~~~~~ ", produces a span
/// coverting that range. The idea is that the string has the same
/// length as the input, and we uncover the byte positions. Note
/// that this can span lines and so on.
fn span_from_selection(input: &str, selection: &str) -> Span {
assert_eq!(input.len(), selection.len());
let left_index = selection.find('^').unwrap() as u32;
let right_index = selection.rfind('~').unwrap() as u32;
Span { lo: BytePos(left_index), hi: BytePos(right_index + 1), expn_id: NO_EXPANSION }
}
fn new_filemap_and_lines(cm: &CodeMap, filename: &str, input: &str) -> Rc<FileMap> {
let fm = cm.new_filemap(filename.to_string(), input.to_string());
let mut byte_pos: u32 = 0;
for line in input.lines() {
// register the start of this line
fm.next_line(BytePos(byte_pos));
// update byte_pos to include this line and the \n at the end
byte_pos += line.len() as u32 + 1;
}
fm
}
/// Test span_to_snippet and span_to_lines for a span coverting 3
/// lines in the middle of a file.
#[test]
fn span_to_snippet_and_lines_spanning_multiple_lines() {
let cm = CodeMap::new();
let inputtext = "aaaaa\nbbbbBB\nCCC\nDDDDDddddd\neee\n";
let selection = " \n ^~\n~~~\n~~~~~ \n \n";
new_filemap_and_lines(&cm, "blork.rs", inputtext);
let span = span_from_selection(inputtext, selection);
// check that we are extracting the text we thought we were extracting
assert_eq!(&cm.span_to_snippet(span).unwrap(), "BB\nCCC\nDDDDD");
// check that span_to_lines gives us the complete result with the lines/cols we expected
let lines = cm.span_to_lines(span);
let expected = vec![
LineInfo { line_index: 1, start_col: CharPos(4), end_col: CharPos(6) },
LineInfo { line_index: 2, start_col: CharPos(0), end_col: CharPos(3) },
LineInfo { line_index: 3, start_col: CharPos(0), end_col: CharPos(5) }
];
assert_eq!(lines.lines, expected);
}
#[test]

View File

@ -483,25 +483,39 @@ fn highlight_lines(err: &mut EmitterWriter,
cm: &codemap::CodeMap,
sp: Span,
lvl: Level,
lines: codemap::FileLines) -> io::Result<()> {
lines: codemap::FileLines)
-> io::Result<()>
{
let fm = &*lines.file;
let mut elided = false;
let mut display_lines = &lines.lines[..];
if display_lines.len() > MAX_LINES {
display_lines = &display_lines[0..MAX_LINES];
elided = true;
}
let line_strings: Option<Vec<&str>> =
lines.lines.iter()
.map(|info| fm.get_line(info.line_index))
.collect();
let line_strings = match line_strings {
None => { return Ok(()); }
Some(line_strings) => line_strings
};
// Display only the first MAX_LINES lines.
let all_lines = lines.lines.len();
let display_lines = cmp::min(all_lines, MAX_LINES);
let display_line_infos = &lines.lines[..display_lines];
let display_line_strings = &line_strings[..display_lines];
// Print the offending lines
for &line_number in display_lines {
if let Some(line) = fm.get_line(line_number) {
try!(write!(&mut err.dst, "{}:{} {}\n", fm.name,
line_number + 1, line));
}
for (line_info, line) in display_line_infos.iter().zip(display_line_strings.iter()) {
try!(write!(&mut err.dst, "{}:{} {}\n",
fm.name,
line_info.line_index + 1,
line));
}
if elided {
let last_line = display_lines[display_lines.len() - 1];
let s = format!("{}:{} ", fm.name, last_line + 1);
// If we elided something, put an ellipsis.
if display_lines < all_lines {
let last_line_index = display_line_infos.last().unwrap().line_index;
let s = format!("{}:{} ", fm.name, last_line_index + 1);
try!(write!(&mut err.dst, "{0:1$}...\n", "", s.len()));
}
@ -510,7 +524,7 @@ fn highlight_lines(err: &mut EmitterWriter,
if lines.lines.len() == 1 {
let lo = cm.lookup_char_pos(sp.lo);
let mut digits = 0;
let mut num = (lines.lines[0] + 1) / 10;
let mut num = (lines.lines[0].line_index + 1) / 10;
// how many digits must be indent past?
while num > 0 { num /= 10; digits += 1; }
@ -522,7 +536,7 @@ fn highlight_lines(err: &mut EmitterWriter,
for _ in 0..skip {
s.push(' ');
}
if let Some(orig) = fm.get_line(lines.lines[0]) {
if let Some(orig) = fm.get_line(lines.lines[0].line_index) {
let mut col = skip;
let mut lastc = ' ';
let mut iter = orig.chars().enumerate();
@ -597,32 +611,32 @@ fn end_highlight_lines(w: &mut EmitterWriter,
let lines = &lines.lines[..];
if lines.len() > MAX_LINES {
if let Some(line) = fm.get_line(lines[0]) {
if let Some(line) = fm.get_line(lines[0].line_index) {
try!(write!(&mut w.dst, "{}:{} {}\n", fm.name,
lines[0] + 1, line));
lines[0].line_index + 1, line));
}
try!(write!(&mut w.dst, "...\n"));
let last_line_number = lines[lines.len() - 1];
if let Some(last_line) = fm.get_line(last_line_number) {
let last_line_index = lines[lines.len() - 1].line_index;
if let Some(last_line) = fm.get_line(last_line_index) {
try!(write!(&mut w.dst, "{}:{} {}\n", fm.name,
last_line_number + 1, last_line));
last_line_index + 1, last_line));
}
} else {
for &line_number in lines {
if let Some(line) = fm.get_line(line_number) {
for line_info in lines {
if let Some(line) = fm.get_line(line_info.line_index) {
try!(write!(&mut w.dst, "{}:{} {}\n", fm.name,
line_number + 1, line));
line_info.line_index + 1, line));
}
}
}
let last_line_start = format!("{}:{} ", fm.name, lines[lines.len()-1]+1);
let last_line_start = format!("{}:{} ", fm.name, lines[lines.len()-1].line_index + 1);
let hi = cm.lookup_char_pos(sp.hi);
let skip = last_line_start.width(false);
let mut s = String::new();
for _ in 0..skip {
s.push(' ');
}
if let Some(orig) = fm.get_line(lines[0]) {
if let Some(orig) = fm.get_line(lines[0].line_index) {
let iter = orig.chars().enumerate();
for (pos, ch) in iter {
// Span seems to use half-opened interval, so subtract 1