diff --git a/examples/herestring.ion b/examples/herestring.ion
index 55b988e2d7d96cf598082aa64d8993d24cf8d173..fe327822de4ddc499334daffe9d712a79cb1ff09 100644
--- a/examples/herestring.ion
+++ b/examples/herestring.ion
@@ -12,3 +12,9 @@ end
find "bar" "@output"
find "zar" "@output"
+
+tr '[a-z]' '[A-Z]' << EOF
+one two
+three four
+five six
+EOF
diff --git a/examples/herestring.out b/examples/herestring.out
index 46bae8fcf850d4c61e955279754ef3105e932720..e43349bccffab183736112be6e32141459664f8a 100644
--- a/examples/herestring.out
+++ b/examples/herestring.out
@@ -1,3 +1,6 @@
FOO
true
false
+ONE TWO
+THREE FOUR
+FIVE SIX
diff --git a/src/parser/peg.rs b/src/parser/peg.rs
index ce1db23927a4a9c66183cfea63a7dd313c1e73b5..405e0736d5b944198809eb9dea904367d62560ee 100644
--- a/src/parser/peg.rs
+++ b/src/parser/peg.rs
@@ -23,7 +23,8 @@ pub struct Redirection {
pub enum Input {
/// A file; the contents of said file will be written to the `stdin` of a process
File(String),
- /// A string literal that is written to the `stdin` of a process
+ /// A string literal that is written to the `stdin` of a process.
+ /// If there is a second string, that second string is the EOF phrase for the heredoc.
HereString(String),
}
diff --git a/src/parser/pipelines.rs b/src/parser/pipelines.rs
index fc2fe2a18f89c05ec2c52a4e8cd0f2de80166ed2..bc59a02372b2e6a5e5a8721618c3eeccc721c3c5 100644
--- a/src/parser/pipelines.rs
+++ b/src/parser/pipelines.rs
@@ -1,11 +1,5 @@
#![allow(eq_op)] // Required as a macro sets this clippy warning off.
-// TODO:
-// - Rewrite this module like the shell_expand::words module
-// - Implement Herestrings
-// - Implement Heredocs
-// - Fix the cyclomatic complexity issue
-
use std::collections::HashSet;
use std::iter::Peekable;
@@ -270,15 +264,33 @@ impl<'a> Collector<'a> {
},
b'<' => {
bytes.next();
- if Some(b'<') == self.peek(i + 1) && Some(b'<') == self.peek(i + 2) {
- // If the next two characters are arrows, then interpret
- // the next argument as a herestring
- bytes.next();
- bytes.next();
- if let Some(cmd) = self.arg(&mut bytes)? {
- input = Some(Input::HereString(cmd.into()));
+ if Some(b'<') == self.peek(i + 1) {
+ if Some(b'<') == self.peek(i + 2) {
+ // If the next two characters are arrows, then interpret
+ // the next argument as a herestring
+ bytes.next();
+ bytes.next();
+ if let Some(cmd) = self.arg(&mut bytes)? {
+ input = Some(Input::HereString(cmd.into()));
+ } else {
+ return Err("expected string argument after '<<<'");
+ }
} else {
- return Err("expected string argument after '<<<'");
+ // Otherwise, what we have is not a herestring, but a heredoc.
+ bytes.next();
+ // Collect the rest of the byte iterator and then trim the result
+ // in order to get the EOF phrase that will be used to terminate
+ // the heredoc.
+ let heredoc = {
+ let mut buffer = Vec::new();
+ while let Some((_, byte)) = bytes.next() {
+ buffer.push(byte);
+ }
+ unsafe { String::from_utf8_unchecked(buffer) }
+ };
+ let heredoc = heredoc.lines().collect::<Vec<&str>>();
+ // Then collect the heredoc from standard input.
+ input = Some(Input::HereString(heredoc[1..heredoc.len()-1].join("\n")));
}
} else if let Some(file) = self.arg(&mut bytes)? {
// Otherwise interpret it as stdin redirection
@@ -740,6 +752,17 @@ mod tests {
assert_eq!(Statement::Pipeline(expected), parse(input));
}
+ #[test]
+ fn heredoc() {
+ let input = "calc << EOF\n1 + 2\n3 + 4\nEOF";
+ let expected = Pipeline {
+ jobs: vec![Job::new(array!["calc"], JobKind::Last)],
+ stdin: Some(Input::HereString("1 + 2\n3 + 4".into())),
+ stdout: None,
+ };
+ assert_eq!(Statement::Pipeline(expected), parse(input));
+ }
+
#[test]
fn piped_herestring() {
let input = "cat | tr 'o' 'x' <<< $VAR > out.log";
diff --git a/src/parser/quotes.rs b/src/parser/quotes.rs
index 9c777a73fc42c8efa6810bc729d59d3508f9672c..193d04e927c381a021f6bd4bb92a9685fcefa8f2 100644
--- a/src/parser/quotes.rs
+++ b/src/parser/quotes.rs
@@ -7,54 +7,99 @@ bitflags! {
}
}
-
pub struct QuoteTerminator {
buffer: String,
+ eof: Option<String>,
+ eof_buffer: String,
read: usize,
flags: Flags,
}
impl QuoteTerminator {
pub fn new(input: String) -> QuoteTerminator {
- QuoteTerminator { buffer: input, read: 0, flags: Flags::empty() }
+ QuoteTerminator { buffer: input, eof: None, eof_buffer: String::new(), read: 0, flags: Flags::empty() }
}
pub fn append(&mut self, input: String) {
- self.buffer.push_str(if self.flags.contains(TRIM) { input.trim() } else { &input });
+ if self.eof.is_none() {
+ self.buffer.push_str(if self.flags.contains(TRIM) { input.trim() } else { &input });
+ } else {
+ self.eof_buffer.push_str(&input);
+ }
}
pub fn check_termination(&mut self) -> bool {
- for character in self.buffer.bytes().skip(self.read) {
- self.read += 1;
- match character {
- _ if self.flags.contains(BACKSL) => self.flags ^= BACKSL,
- b'\\' => self.flags ^= BACKSL,
- b'\'' if !self.flags.intersects(DQUOTE) => self.flags ^= SQUOTE,
- b'"' if !self.flags.intersects(SQUOTE) => self.flags ^= DQUOTE,
- _ => (),
+ let mut eof_line = None;
+ let eof = self.eof.clone();
+ let status = if let Some(ref eof) = eof {
+ let line = &self.eof_buffer;
+ eof_line = Some([&line, "\n"].concat());
+ line.trim() == eof
+ } else {
+ {
+ let mut eof_found = false;
+ {
+ let mut bytes = self.buffer.bytes().skip(self.read);
+ while let Some(character) = bytes.next() {
+ self.read += 1;
+ match character {
+ _ if self.flags.contains(BACKSL) => self.flags ^= BACKSL,
+ b'\\' => self.flags ^= BACKSL,
+ b'\'' if !self.flags.intersects(DQUOTE) => self.flags ^= SQUOTE,
+ b'"' if !self.flags.intersects(SQUOTE) => self.flags ^= DQUOTE,
+ b'<' if !self.flags.contains(SQUOTE | DQUOTE) => {
+ let as_bytes = self.buffer.as_bytes();
+ if Some(&b'<') == as_bytes.get(self.read) {
+ self.read += 1;
+ if Some(&b'<') != as_bytes.get(self.read) {
+ use std::str;
+ let eof_phrase = unsafe { str::from_utf8_unchecked(&as_bytes[self.read..]) };
+ self.eof = Some(eof_phrase.trim().to_owned());
+ eof_found = true;
+ break
+ }
+ }
+ }
+ _ => (),
+ }
+ }
+ }
+ if eof_found {
+ self.buffer.push('\n');
+ return false
+ }
}
- }
- if self.flags.intersects(SQUOTE | DQUOTE) {
- self.read += 1;
- self.buffer.push('\n');
- false
- } else {
- match self.buffer.bytes().last() {
- Some(b'\\') => {
- let _ = self.buffer.pop();
- self.read -= 1;
- self.flags |= TRIM;
- false
- },
- Some(b'|') | Some(b'&') => {
- // self.read -= 1;
- // self.flags |= TRIM;
- false
+ if self.flags.intersects(SQUOTE | DQUOTE) {
+ self.read += 1;
+ self.buffer.push('\n');
+ false
+ } else {
+ match self.buffer.bytes().last() {
+ Some(b'\\') => {
+ let _ = self.buffer.pop();
+ self.read -= 1;
+ self.flags |= TRIM;
+ false
+ },
+ Some(b'|') | Some(b'&') => {
+ false
+ }
+ _ => true
}
- _ => true
+ }
+ };
+
+ if let Some(line) = eof_line {
+ self.buffer.push_str(&line);
+ }
+ if self.eof.is_some() {
+ self.eof_buffer.clear();
+ if status {
+ self.eof = None;
}
}
+ status
}
pub fn consume(self) -> String { self.buffer }