Skip to content

Commit

Permalink
ParagraphStream iterator complete
Browse files Browse the repository at this point in the history
  • Loading branch information
@kwantam
kwantam committed Jun 18, 2014
1 parent eb71558 commit 2fa7c48
Show file tree
Hide file tree
Showing 4 changed files with 140 additions and 26 deletions.
15 changes: 9 additions & 6 deletions fmt/fmt.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -81,30 +81,33 @@ fn uumain(args: Vec<String>) -> int {
, mail : false
, uniform : false
, split_only : false
, use_prefix : false
, prefix : String::new()
, xprefix : false
, prefix_len : 0
, use_anti_prefix : false
, anti_prefix : String::new()
, xanti_prefix: false
, width : 78
, goal : 72
, tabwidth : 8
};

if matches.opt_present("c") { fmt_opts.crown = true; }
if matches.opt_present("t") { fmt_opts.tagged = true; fmt_opts.crown = false; }
if matches.opt_present("t") { fmt_opts.tagged = true; }
if matches.opt_present("c") { fmt_opts.crown = true; fmt_opts.tagged = false; }
if matches.opt_present("m") { fmt_opts.mail = true; }
if matches.opt_present("u") { fmt_opts.uniform = true; }
if matches.opt_present("s") { fmt_opts.split_only = true; fmt_opts.crown = false; fmt_opts.tagged = false; }
if matches.opt_present("x") { fmt_opts.xprefix = true; }
if matches.opt_present("X") { fmt_opts.xanti_prefix = true; }

match matches.opt_str("p") {
Some(s) => { fmt_opts.prefix = s; },
Some(s) => { fmt_opts.prefix = s; fmt_opts.use_prefix = true; fmt_opts.prefix_len = fmt_opts.prefix.as_slice().char_len() },
None => ()
};

match matches.opt_str("P") {
Some(s) => { fmt_opts.anti_prefix = s; },
Some(s) => { fmt_opts.anti_prefix = s; fmt_opts.use_anti_prefix = true; },
None => ()
};

Expand Down Expand Up @@ -156,8 +159,8 @@ fn uumain(args: Vec<String>) -> int {
}
Ok(f) => f
};
let mut fPara = parasplit::ParagraphStream::new(&fmt_opts, &mut fp);
fPara.count();
let fPara = parasplit::ParagraphStream::new(&fmt_opts, &mut fp);
fPara.map(|x| println!("{}",x)).count();
}

0
Expand Down
4 changes: 3 additions & 1 deletion fmt/fmtoptions.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,13 +13,15 @@ pub struct FmtOptions {
pub tagged : bool,
pub mail : bool,
pub split_only : bool,
pub use_prefix : bool,
pub prefix : String,
pub xprefix : bool,
pub prefix_len : uint,
pub use_anti_prefix : bool,
pub anti_prefix : String,
pub xanti_prefix: bool,
pub uniform : bool,
pub width : uint,
pub goal : uint,
pub tabwidth : uint,
}

141 changes: 122 additions & 19 deletions fmt/parasplit.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,21 +29,28 @@ pub fn open_file(filename : &str) -> IoResult<FileOrStdReader> {
}
}

pub struct FileLines<'a> {
struct FileLines<'a> {
opts : &'a FmtOptions,
lines : Lines<'a, FileOrStdReader>,
use_prefix : bool,
use_anti_prefix : bool,
}

#[deriving(Show)]
pub enum Line {
enum Line {
FormatLine(FileLine),
NoFormatLine(String, bool)
}

impl Line {
fn get_fileline(self) -> FileLine {
match self {
FormatLine(fl) => fl,
NoFormatLine(..) => fail!("Found NoFormatLine when expecting FormatLine")
}
}
}

#[deriving(Show)]
pub struct FileLine {
struct FileLine {
line : String,
indent_end : uint,
prefix_end : uint,
Expand All @@ -53,20 +60,20 @@ pub struct FileLine {
}

impl<'a> FileLines<'a> {
pub fn new<'a>(opts : &'a FmtOptions, lines : Lines<'a, FileOrStdReader>) -> FileLines<'a> {
FileLines { opts : opts, lines : lines, use_prefix : opts.prefix.len() > 0, use_anti_prefix : opts.anti_prefix.len() > 0 }
fn new<'a>(opts : &'a FmtOptions, lines : Lines<'a, FileOrStdReader>) -> FileLines<'a> {
FileLines { opts : opts, lines : lines }
}

// returns true if this line should be formatted
fn match_prefix(&self, line : &str) -> (bool, uint) {
if ! self.use_prefix { return (true, 0u); }
if ! self.opts.use_prefix { return (true, 0u); }

FileLines::match_prefix_generic(self.opts.prefix.as_slice(), line, self.opts.xprefix)
}

// returns true if this line should be formatted
fn match_anti_prefix(&self, line : &str) -> bool {
if ! self.use_anti_prefix { return true; }
if ! self.opts.use_anti_prefix { return true; }

match FileLines::match_prefix_generic(self.opts.anti_prefix.as_slice(), line, self.opts.xanti_prefix) {
(true, _) => false,
Expand Down Expand Up @@ -96,7 +103,7 @@ impl<'a> FileLines<'a> {

impl<'a> Iterator<Line> for FileLines<'a> {
fn next(&mut self) -> Option<Line> {
let n =
let mut n =
match self.lines.next() {
Some(t) => match t {
Ok(tt) => tt,
Expand Down Expand Up @@ -126,8 +133,25 @@ impl<'a> Iterator<Line> for FileLines<'a> {
return Some(NoFormatLine(n, false));
}

let nLen = n.len();
// replace trailing newline, if any, with space
{
let CharRange {ch, next : i} = n.as_slice().char_range_at_reverse(nLen);
if ch == '\n' {
unsafe {
let nmut = n.as_mut_bytes();
nmut[i] = 0x20;
}
let CharRange {ch, next : _} = n.as_slice().char_range_at_reverse(i);
if ch == '.' {
n.push_char(' ');
}
}
}

// figure out the indent, prefix, and prefixindent ending points
let (indEnd, pfxEnd, pfxIndEnd) =
if self.use_prefix {
if self.opts.use_prefix {
let pfxEnd = poffset + self.opts.prefix.len();
let nSlice = n.as_slice().slice_from(pfxEnd);
let nSlice2 = nSlice.trim_left();
Expand All @@ -136,9 +160,10 @@ impl<'a> Iterator<Line> for FileLines<'a> {
} else {
let nSlice = n.as_slice().trim_left();

(n.len() - nSlice.len(), 0, 0)
(nLen - nSlice.len(), 0, 0)
};

// indent length
let indLen =
if indEnd > 0 {
let nSlice = n.as_slice().slice(pfxEnd, indEnd);
Expand All @@ -147,6 +172,7 @@ impl<'a> Iterator<Line> for FileLines<'a> {
0
};

// prefix indent length
let pfxIndLen =
if pfxIndEnd > 0 {
let nSlice = n.as_slice().slice_to(pfxIndEnd);
Expand All @@ -161,9 +187,9 @@ impl<'a> Iterator<Line> for FileLines<'a> {
}
}

#[allow(dead_code)]
#[deriving(Show)]
pub struct Paragraph {
pub words : Vec<String>,
pub lines : Vec<FileLine>,
pub init_str : String,
pub init_len : uint,
pub indent_str : String,
Expand All @@ -172,14 +198,12 @@ pub struct Paragraph {
pub pfxind_len : uint
}

#[allow(dead_code)]
pub struct ParagraphStream<'a> {
lines : Peekable<Line,FileLines<'a>>,
next_mail : bool, // next line should be considered for a mail header
opts : &'a FmtOptions
}

#[allow(dead_code)]
impl<'a> ParagraphStream<'a> {
pub fn new<'a>(opts : &'a FmtOptions, reader : &'a mut FileOrStdReader) -> ParagraphStream<'a> {
let lines = FileLines::new(opts, reader.lines()).peekable();
Expand All @@ -204,6 +228,9 @@ impl<'a> ParagraphStream<'a> {
None => return false
};

// header field must be nonzero length
if colonPosn == 0 { return false; }

return lSlice.slice_to(colonPosn).chars()
.all(|x| match x as uint {
y if y < 33 || y > 126 => false,
Expand Down Expand Up @@ -243,11 +270,18 @@ impl<'a> Iterator<Paragraph> for ParagraphStream<'a> {
}

// now build a paragraph
let mut words : Vec<String> = vec!();
let mut init_str = String::new();
let mut init_len = 0;
let mut indent_str = String::new();
let mut indent_len = 0;
let mut pfxind_str = String::new();
let mut pfxind_len = 0;

let mut pLines : Vec<FileLine> = Vec::new();
let mut in_mail = false;
let mut second_done = false; // for when we use crown or tagged mode
loop {
// scope for fl - need to unborrow before we can take next()
// need to explicitly force fl out of scope before we can call self.lines.next()
{ // peek ahead
let fl =
match self.lines.peek() {
Expand All @@ -258,9 +292,78 @@ impl<'a> Iterator<Paragraph> for ParagraphStream<'a> {
&NoFormatLine(..) => break
}
};

if pLines.len() == 0 {
// first time through the loop, get things set up
// detect mail header
if self.opts.mail && self.next_mail && ParagraphStream::is_mail_header(fl) {
in_mail = true;
// there can't be any indent or pfxind because otherwise is_mail_header would fail
// since there cannot be any whitespace before the colon in a valid header field
indent_str.push_str(" ");
indent_len = 2;
} else {
if self.opts.crown || self.opts.tagged {
init_str.push_str(fl.line.as_slice().slice_to(fl.indent_end));
init_len = fl.indent_len + fl.pfxind_len + self.opts.prefix_len;
}

indent_str.push_str(fl.line.as_slice().slice(fl.prefix_end,fl.indent_end));
indent_len = fl.indent_len;

if self.opts.tagged {
indent_str.push_str(" ");
indent_len += 4;
}

if self.opts.use_prefix {
pfxind_str.push_str(fl.line.as_slice().slice_to(fl.pfxind_end));
pfxind_len = fl.pfxind_len;
}
}
} else if in_mail {
// lines following mail headers must begin with spaces
if (self.opts.use_prefix && fl.pfxind_end == 0) || (! self.opts.use_prefix && fl.indent_end == 0) {
break; // this line does not begin with spaces
}
second_done = true;
} else if ! second_done {
// now we have enough info to handle crown margin and tagged mode
if self.opts.crown {
// even in crown mode we require that pfxind is the same
if pfxind_len != fl.pfxind_len {
break;
} else {
// matching pfxind, so get indent_len from this line
indent_str.push_str(fl.line.as_slice().slice(fl.prefix_end,fl.indent_end));
indent_len = fl.indent_len;
}
} else if self.opts.tagged {
// in tagged mode, pfxind has to be the same and indent has to be *different*
if (pfxind_len != fl.pfxind_len) || (indent_len - 4 == fl.indent_len) {
break;
} else {
indent_str.push_str(fl.line.as_slice().slice(fl.prefix_end,fl.indent_end));
indent_len = fl.indent_len;
}
} else {
// detect mismatch
if (indent_len != fl.indent_len) || (pfxind_len != fl.pfxind_len) {
break;
}
}
second_done = true;
} else {
// detect mismatch
if (indent_len != fl.indent_len) || (pfxind_len != fl.pfxind_len) {
break;
}
}
}

pLines.push(self.lines.next().unwrap().get_fileline());
}

None
Some(Paragraph { lines : pLines, init_str : init_str, init_len : init_len, indent_str : indent_str, indent_len : indent_len, pfxind_str : pfxind_str, pfxind_len : pfxind_len })
}
}
6 changes: 6 additions & 0 deletions fmt/test.rs
View file
Original file line number Diff line number Diff line change
@@ -1,12 +1,14 @@

use std::io::{stdin};
use std::str::CharRange;

fn main() {
let mut s = stdin();

//let mut q = s.lines().scan(0u, |s, _| { *s = *s + 1; Some(*s) });
//q.count();

/*
let z = true ^ false;
let foo = s.lines().filter(|x| z && x.is_ok()).map(|x| x.unwrap());
Expand All @@ -16,4 +18,8 @@ fn main() {
println!("{}",bar);
println!("{} : {}","asdf".slice_to(2),"asdf".slice_from(2));
*/

let CharRange { ch, next } = "asdf\n".char_range_at_reverse("asdf\n".len());
println!("{} {}", ch, next);
}

0 comments on commit 2fa7c48

Please sign in to comment.