mirror of
https://github.com/storytold/deckcraft.git
synced 2026-10-25 19:17:06 -06:00
refactor(rtf): split the outline reader into per-construct methods (S3776)
This commit is contained in:
Dylan M. Taylor
+201
-140
@@ -126,155 +126,216 @@ fn advance(rtf: &str, i: usize, n: usize) -> usize {
|
||||
if rtf.is_char_boundary(i + n) { i + n } else { i + rtf.get(i..).and_then(|r| r.chars().next()).map_or(1, char::len_utf8) }
|
||||
}
|
||||
|
||||
/// Group state: whether the text is being skipped, and the `\uc` fallback count.
|
||||
#[derive(Clone, Copy)]
|
||||
struct St {
|
||||
skip: bool,
|
||||
uc: usize,
|
||||
}
|
||||
|
||||
/// Parser state for [`rtf_to_outline`]: the cursor, the group stack and the outline so far.
|
||||
struct Outline<'a> {
|
||||
rtf: &'a str,
|
||||
b: &'a [u8],
|
||||
i: usize,
|
||||
stack: Vec<St>,
|
||||
st: St,
|
||||
level: Option<u32>,
|
||||
indent: i64,
|
||||
pending_skip: usize,
|
||||
line: String,
|
||||
out: String,
|
||||
}
|
||||
|
||||
/// Read an RTF outline into tab-indented outline text: a paragraph's outline level (or its left
|
||||
/// indent in 0.25" steps) becomes its number of tabs.
|
||||
pub fn rtf_to_outline(rtf: &str) -> String {
|
||||
#[derive(Clone, Copy)]
|
||||
struct St {
|
||||
skip: bool,
|
||||
uc: usize,
|
||||
}
|
||||
let b = rtf.as_bytes();
|
||||
let mut stack: Vec<St> = vec![];
|
||||
let mut st = St { skip: false, uc: 1 };
|
||||
let (mut level, mut indent): (Option<u32>, i64) = (None, 0);
|
||||
let mut line = String::new();
|
||||
let mut out = String::new();
|
||||
let mut pending_skip = 0usize;
|
||||
let mut i = 0;
|
||||
let flush = |line: &mut String, level: Option<u32>, indent: i64, out: &mut String| {
|
||||
let text = line.trim();
|
||||
if !text.is_empty() {
|
||||
let n = level.map(|l| l as i64).unwrap_or(indent.saturating_add(180) / 360).clamp(0, 9) as usize;
|
||||
out.push_str(&"\t".repeat(n));
|
||||
out.push_str(text);
|
||||
out.push('\n');
|
||||
}
|
||||
line.clear();
|
||||
let mut p = Outline {
|
||||
rtf,
|
||||
b: rtf.as_bytes(),
|
||||
i: 0,
|
||||
stack: vec![],
|
||||
st: St { skip: false, uc: 1 },
|
||||
level: None,
|
||||
indent: 0,
|
||||
pending_skip: 0,
|
||||
line: String::new(),
|
||||
out: String::new(),
|
||||
};
|
||||
while i < b.len() {
|
||||
let c = b[i];
|
||||
match c {
|
||||
b'{' => {
|
||||
stack.push(st);
|
||||
i += 1;
|
||||
// `{\*\dest …}`: an ignorable destination.
|
||||
if b[i..].starts_with(b"\\*") {
|
||||
st.skip = true;
|
||||
}
|
||||
}
|
||||
b'}' => {
|
||||
st = stack.pop().unwrap_or(St { skip: false, uc: 1 });
|
||||
i += 1;
|
||||
}
|
||||
b'\\' => {
|
||||
i += 1;
|
||||
let Some(&n) = b.get(i) else { break };
|
||||
if n.is_ascii_alphabetic() {
|
||||
let s = i;
|
||||
while i < b.len() && b[i].is_ascii_alphabetic() {
|
||||
i += 1;
|
||||
}
|
||||
let word = &rtf[s..i];
|
||||
let ns = i;
|
||||
if i < b.len() && (b[i] == b'-' || b[i].is_ascii_digit()) {
|
||||
i += 1;
|
||||
while i < b.len() && b[i].is_ascii_digit() {
|
||||
i += 1;
|
||||
}
|
||||
}
|
||||
let arg: Option<i64> = rtf[ns..i].parse().ok();
|
||||
if i < b.len() && b[i] == b' ' {
|
||||
i += 1;
|
||||
}
|
||||
if pending_skip > 0 {
|
||||
pending_skip -= 1;
|
||||
continue;
|
||||
}
|
||||
if SKIP.contains(&word) {
|
||||
st.skip = true;
|
||||
}
|
||||
if st.skip {
|
||||
continue;
|
||||
}
|
||||
match word {
|
||||
"par" => {
|
||||
flush(&mut line, level, indent, &mut out);
|
||||
}
|
||||
"pard" => {
|
||||
level = None;
|
||||
indent = 0;
|
||||
}
|
||||
"outlinelevel" => level = arg.map(|a| a.clamp(0, 9) as u32),
|
||||
"li" => indent = arg.unwrap_or(0),
|
||||
"tab" | "line" => line.push(' '),
|
||||
"uc" => st.uc = arg.unwrap_or(1).clamp(0, 8) as usize,
|
||||
"u" => {
|
||||
let v = arg.unwrap_or(0);
|
||||
// RTF writes UTF-16 units as signed 16-bit values.
|
||||
push_utf16(&mut line, (v & 0xffff) as u32);
|
||||
pending_skip = 0;
|
||||
// Skip the fallback characters.
|
||||
let mut k = st.uc;
|
||||
while k > 0 && i < b.len() {
|
||||
if b[i] == b'\\' && b.get(i + 1) == Some(&b'\'') {
|
||||
i = advance(rtf, i, 4);
|
||||
} else if b[i] == b'{' || b[i] == b'}' || b[i] == b'\\' {
|
||||
break;
|
||||
} else {
|
||||
i = advance(rtf, i, 1);
|
||||
}
|
||||
k -= 1;
|
||||
}
|
||||
}
|
||||
"emdash" => line.push('—'),
|
||||
"endash" => line.push('–'),
|
||||
"bullet" => line.push('•'),
|
||||
"lquote" => line.push('‘'),
|
||||
"rquote" => line.push('’'),
|
||||
"ldblquote" => line.push('“'),
|
||||
"rdblquote" => line.push('”'),
|
||||
_ => {}
|
||||
}
|
||||
} else if n == b'\'' {
|
||||
let hex = rtf.get(i + 1..i + 3).filter(|h| h.bytes().all(|b| b.is_ascii_hexdigit())).and_then(|h| u8::from_str_radix(h, 16).ok());
|
||||
i = advance(rtf, i, 3);
|
||||
if !st.skip
|
||||
&& let Some(h) = hex
|
||||
&& h != 0
|
||||
{
|
||||
line.push(cp1252(h));
|
||||
}
|
||||
} else {
|
||||
i += 1;
|
||||
if st.skip {
|
||||
continue;
|
||||
}
|
||||
match n {
|
||||
b'\\' | b'{' | b'}' => line.push(n as char),
|
||||
b'~' => line.push('\u{a0}'),
|
||||
b'_' => line.push('\u{2011}'),
|
||||
b'\n' | b'\r' => flush(&mut line, level, indent, &mut out),
|
||||
_ => {}
|
||||
p.run();
|
||||
p.flush();
|
||||
finish_surrogates(p.out)
|
||||
}
|
||||
|
||||
impl Outline<'_> {
|
||||
fn run(&mut self) {
|
||||
while self.i < self.b.len() {
|
||||
match self.b[self.i] {
|
||||
b'{' => self.open_group(),
|
||||
b'}' => self.close_group(),
|
||||
b'\\' => {
|
||||
self.i += 1;
|
||||
if !self.escape() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
b'\r' | b'\n' => i += 1,
|
||||
_ => {
|
||||
let s = i;
|
||||
while i < b.len() && !matches!(b[i], b'{' | b'}' | b'\\' | b'\r' | b'\n') {
|
||||
i += 1;
|
||||
}
|
||||
if !st.skip
|
||||
&& let Some(t) = rtf.get(s..i)
|
||||
{
|
||||
line.push_str(&t.replace('\u{0}', ""));
|
||||
}
|
||||
b'\r' | b'\n' => self.i += 1,
|
||||
_ => self.text(),
|
||||
}
|
||||
}
|
||||
}
|
||||
flush(&mut line, level, indent, &mut out);
|
||||
finish_surrogates(out)
|
||||
|
||||
/// `{`: push the group; `{\*\dest …}` is an ignorable destination.
|
||||
fn open_group(&mut self) {
|
||||
self.stack.push(self.st);
|
||||
self.i += 1;
|
||||
if self.b[self.i..].starts_with(b"\\*") {
|
||||
self.st.skip = true;
|
||||
}
|
||||
}
|
||||
|
||||
fn close_group(&mut self) {
|
||||
self.st = self.stack.pop().unwrap_or(St { skip: false, uc: 1 });
|
||||
self.i += 1;
|
||||
}
|
||||
|
||||
/// The character after a backslash: a control word, a hex escape or a symbol. False at the
|
||||
/// end of the input.
|
||||
fn escape(&mut self) -> bool {
|
||||
let Some(&n) = self.b.get(self.i) else { return false };
|
||||
if n.is_ascii_alphabetic() {
|
||||
self.control_word();
|
||||
} else if n == b'\'' {
|
||||
self.hex_escape();
|
||||
} else {
|
||||
self.symbol(n);
|
||||
}
|
||||
true
|
||||
}
|
||||
|
||||
fn control_word(&mut self) {
|
||||
let rtf = self.rtf;
|
||||
let s = self.i;
|
||||
while self.i < self.b.len() && self.b[self.i].is_ascii_alphabetic() {
|
||||
self.i += 1;
|
||||
}
|
||||
let word = &rtf[s..self.i];
|
||||
let ns = self.i;
|
||||
if self.i < self.b.len() && (self.b[self.i] == b'-' || self.b[self.i].is_ascii_digit()) {
|
||||
self.i += 1;
|
||||
while self.i < self.b.len() && self.b[self.i].is_ascii_digit() {
|
||||
self.i += 1;
|
||||
}
|
||||
}
|
||||
let arg: Option<i64> = rtf[ns..self.i].parse().ok();
|
||||
if self.i < self.b.len() && self.b[self.i] == b' ' {
|
||||
self.i += 1;
|
||||
}
|
||||
if self.pending_skip > 0 {
|
||||
self.pending_skip -= 1;
|
||||
return;
|
||||
}
|
||||
if SKIP.contains(&word) {
|
||||
self.st.skip = true;
|
||||
}
|
||||
if !self.st.skip {
|
||||
self.apply(word, arg);
|
||||
}
|
||||
}
|
||||
|
||||
fn apply(&mut self, word: &str, arg: Option<i64>) {
|
||||
match word {
|
||||
"par" => self.flush(),
|
||||
"pard" => {
|
||||
self.level = None;
|
||||
self.indent = 0;
|
||||
}
|
||||
"outlinelevel" => self.level = arg.map(|a| a.clamp(0, 9) as u32),
|
||||
"li" => self.indent = arg.unwrap_or(0),
|
||||
"tab" | "line" => self.line.push(' '),
|
||||
"uc" => self.st.uc = arg.unwrap_or(1).clamp(0, 8) as usize,
|
||||
"u" => self.unicode(arg.unwrap_or(0)),
|
||||
"emdash" => self.line.push('—'),
|
||||
"endash" => self.line.push('–'),
|
||||
"bullet" => self.line.push('•'),
|
||||
"lquote" => self.line.push('‘'),
|
||||
"rquote" => self.line.push('’'),
|
||||
"ldblquote" => self.line.push('“'),
|
||||
"rdblquote" => self.line.push('”'),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
|
||||
/// `\uN`: one UTF-16 unit, then the `\uc` fallback characters, which are skipped.
|
||||
fn unicode(&mut self, v: i64) {
|
||||
// RTF writes UTF-16 units as signed 16-bit values.
|
||||
push_utf16(&mut self.line, (v & 0xffff) as u32);
|
||||
self.pending_skip = 0;
|
||||
let mut k = self.st.uc;
|
||||
while k > 0 && self.i < self.b.len() {
|
||||
if self.b[self.i] == b'\\' && self.b.get(self.i + 1) == Some(&b'\'') {
|
||||
self.i = advance(self.rtf, self.i, 4);
|
||||
} else if matches!(self.b[self.i], b'{' | b'}' | b'\\') {
|
||||
break;
|
||||
} else {
|
||||
self.i = advance(self.rtf, self.i, 1);
|
||||
}
|
||||
k -= 1;
|
||||
}
|
||||
}
|
||||
|
||||
/// `\'hh`: one cp1252 byte.
|
||||
fn hex_escape(&mut self) {
|
||||
let hex =
|
||||
self.rtf.get(self.i + 1..self.i + 3).filter(|h| h.bytes().all(|b| b.is_ascii_hexdigit())).and_then(|h| u8::from_str_radix(h, 16).ok());
|
||||
self.i = advance(self.rtf, self.i, 3);
|
||||
if !self.st.skip
|
||||
&& let Some(h) = hex
|
||||
&& h != 0
|
||||
{
|
||||
self.line.push(cp1252(h));
|
||||
}
|
||||
}
|
||||
|
||||
/// A one-character control symbol (`\\`, `\{`, `\~`, ...).
|
||||
fn symbol(&mut self, n: u8) {
|
||||
self.i += 1;
|
||||
if self.st.skip {
|
||||
return;
|
||||
}
|
||||
match n {
|
||||
b'\\' | b'{' | b'}' => self.line.push(n as char),
|
||||
b'~' => self.line.push('\u{a0}'),
|
||||
b'_' => self.line.push('\u{2011}'),
|
||||
b'\n' | b'\r' => self.flush(),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
|
||||
/// Plain text up to the next special character.
|
||||
fn text(&mut self) {
|
||||
let s = self.i;
|
||||
while self.i < self.b.len() && !matches!(self.b[self.i], b'{' | b'}' | b'\\' | b'\r' | b'\n') {
|
||||
self.i += 1;
|
||||
}
|
||||
if !self.st.skip
|
||||
&& let Some(t) = self.rtf.get(s..self.i)
|
||||
{
|
||||
self.line.push_str(&t.replace('\u{0}', ""));
|
||||
}
|
||||
}
|
||||
|
||||
/// Ends the current line: a non-empty line becomes one outline entry, indented by its level.
|
||||
fn flush(&mut self) {
|
||||
let text = self.line.trim();
|
||||
if !text.is_empty() {
|
||||
let n = self.level.map(|l| l as i64).unwrap_or(self.indent.saturating_add(180) / 360).clamp(0, 9) as usize;
|
||||
self.out.push_str(&"\t".repeat(n));
|
||||
self.out.push_str(text);
|
||||
self.out.push('\n');
|
||||
}
|
||||
self.line.clear();
|
||||
}
|
||||
}
|
||||
|
||||
/// Push a UTF-16 unit; surrogate halves are parked as private markers (NUL plus four hex digits)
|
||||
|
||||
Reference in New Issue
Block a user