Support reading input from stdin (#3339)

5 files changed, 113 insertions, 27 deletions

diff --git a/crates/typst-cli/src/args.rs b/crates/typst-cli/src/args.rs
index c14f0277..973eea8b 100644
--- a/crates/typst-cli/src/args.rs
+++ b/crates/typst-cli/src/args.rs
@@ -65,6 +65,7 @@ pub struct CompileCommand {
     pub common: SharedArgs,
 
     /// Path to output file (PDF, PNG, or SVG)
+    #[clap(required_if_eq("input", "-"))]
     pub output: Option<PathBuf>,
 
     /// The format of the output file, inferred from the extension by default
@@ -121,8 +122,9 @@ pub enum SerializationFormat {
 /// Common arguments of compile, watch, and query.
 #[derive(Debug, Clone, Args)]
 pub struct SharedArgs {
-    /// Path to input Typst file
-    pub input: PathBuf,
+    /// Path to input Typst file, use `-` to read input from stdin
+    #[clap(value_parser = input_value_parser)]
+    pub input: Input,
 
     /// Configures the project root (for absolute paths)
     #[clap(long = "root", env = "TYPST_ROOT", value_name = "DIR")]
@@ -155,6 +157,26 @@ pub struct SharedArgs {
     pub diagnostic_format: DiagnosticFormat,
 }
 
+/// An input that is either stdin or a real path.
+#[derive(Debug, Clone)]
+pub enum Input {
+    /// Stdin, represented by `-`.
+    Stdin,
+    /// A non-empty path.
+    Path(PathBuf),
+}
+
+/// The clap value parser used by `SharedArgs.input`
+fn input_value_parser(value: &str) -> Result<Input, clap::error::Error> {
+    if value.is_empty() {
+        Err(clap::Error::new(clap::error::ErrorKind::InvalidValue))
+    } else if value == "-" {
+        Ok(Input::Stdin)
+    } else {
+        Ok(Input::Path(value.into()))
+    }
+}
+
 /// Parses key/value pairs split by the first equal sign.
 ///
 /// This function will return an error if the argument contains no equals sign
diff --git a/crates/typst-cli/src/compile.rs b/crates/typst-cli/src/compile.rs
index f66553f6..337ec966 100644
--- a/crates/typst-cli/src/compile.rs
+++ b/crates/typst-cli/src/compile.rs
@@ -16,7 +16,7 @@ use typst::syntax::{FileId, Source, Span};
 use typst::visualize::Color;
 use typst::{World, WorldExt};
 
-use crate::args::{CompileCommand, DiagnosticFormat, OutputFormat};
+use crate::args::{CompileCommand, DiagnosticFormat, Input, OutputFormat};
 use crate::timings::Timer;
 use crate::watch::Status;
 use crate::world::SystemWorld;
@@ -29,7 +29,10 @@ impl CompileCommand {
     /// The output path.
     pub fn output(&self) -> PathBuf {
         self.output.clone().unwrap_or_else(|| {
-            self.common.input.with_extension(
+            let Input::Path(path) = &self.common.input else {
+                panic!("output must be specified when input is from stdin, as guarded by the CLI");
+            };
+            path.with_extension(
                 match self.output_format().unwrap_or(OutputFormat::Pdf) {
                     OutputFormat::Pdf => "pdf",
                     OutputFormat::Png => "png",
@@ -163,8 +166,8 @@ fn export_pdf(
     command: &CompileCommand,
     world: &SystemWorld,
 ) -> StrResult<()> {
-    let ident = world.input().to_string_lossy();
-    let buffer = typst_pdf::pdf(document, Some(&ident), now());
+    let ident = world.input().map(|i| i.to_string_lossy());
+    let buffer = typst_pdf::pdf(document, ident.as_deref(), now());
     let output = command.output();
     fs::write(output, buffer)
         .map_err(|err| eco_format!("failed to write PDF file ({err})"))?;
diff --git a/crates/typst-cli/src/watch.rs b/crates/typst-cli/src/watch.rs
index 67276a3e..35861242 100644
--- a/crates/typst-cli/src/watch.rs
+++ b/crates/typst-cli/src/watch.rs
@@ -9,7 +9,7 @@ use notify::{RecommendedWatcher, RecursiveMode, Watcher};
 use same_file::is_same_file;
 use typst::diag::StrResult;
 
-use crate::args::CompileCommand;
+use crate::args::{CompileCommand, Input};
 use crate::compile::compile_once;
 use crate::terminal;
 use crate::timings::Timer;
@@ -168,7 +168,10 @@ impl Status {
         term_out.set_color(&color)?;
         write!(term_out, "watching")?;
         term_out.reset()?;
-        writeln!(term_out, " {}", command.common.input.display())?;
+        match &command.common.input {
+            Input::Stdin => writeln!(term_out, " <stdin>"),
+            Input::Path(path) => writeln!(term_out, " {}", path.display()),
+        }?;
 
         term_out.set_color(&color)?;
         write!(term_out, "writing to")?;
diff --git a/crates/typst-cli/src/world.rs b/crates/typst-cli/src/world.rs
index fee6b7db..72efa7fa 100644
--- a/crates/typst-cli/src/world.rs
+++ b/crates/typst-cli/src/world.rs
@@ -1,11 +1,13 @@
 use std::collections::HashMap;
+use std::io::Read;
 use std::path::{Path, PathBuf};
 use std::sync::OnceLock;
-use std::{fs, mem};
+use std::{fs, io, mem};
 
 use chrono::{DateTime, Datelike, Local};
 use comemo::Prehashed;
 use ecow::eco_format;
+use once_cell::sync::Lazy;
 use parking_lot::Mutex;
 use typst::diag::{FileError, FileResult, StrResult};
 use typst::foundations::{Bytes, Datetime, Dict, IntoValue};
@@ -14,17 +16,22 @@ use typst::text::{Font, FontBook};
 use typst::{Library, World};
 use typst_timing::{timed, TimingScope};
 
-use crate::args::SharedArgs;
+use crate::args::{Input, SharedArgs};
 use crate::compile::ExportCache;
 use crate::fonts::{FontSearcher, FontSlot};
 use crate::package::prepare_package;
 
+/// Static `FileId` allocated for stdin.
+/// This is to ensure that a file is read in the correct way.
+static STDIN_ID: Lazy<FileId> =
+    Lazy::new(|| FileId::new_fake(VirtualPath::new("<stdin>")));
+
 /// A world that provides access to the operating system.
 pub struct SystemWorld {
     /// The working directory.
     workdir: Option<PathBuf>,
     /// The canonical path to the input file.
-    input: PathBuf,
+    input: Option<PathBuf>,
     /// The root relative to which absolute paths are resolved.
     root: PathBuf,
     /// The input path.
@@ -52,25 +59,34 @@ impl SystemWorld {
         searcher.search(&command.font_paths);
 
         // Resolve the system-global input path.
-        let input = command.input.canonicalize().map_err(|_| {
-            eco_format!("input file not found (searched at {})", command.input.display())
-        })?;
+        let input = match &command.input {
+            Input::Stdin => None,
+            Input::Path(path) => Some(path.canonicalize().map_err(|_| {
+                eco_format!("input file not found (searched at {})", path.display())
+            })?),
+        };
 
         // Resolve the system-global root directory.
         let root = {
             let path = command
                 .root
                 .as_deref()
-                .or_else(|| input.parent())
+                .or_else(|| input.as_deref().and_then(|i| i.parent()))
                 .unwrap_or(Path::new("."));
             path.canonicalize().map_err(|_| {
                 eco_format!("root directory not found (searched at {})", path.display())
             })?
         };
 
-        // Resolve the virtual path of the main file within the project root.
-        let main_path = VirtualPath::within_root(&input, &root)
-            .ok_or("source file must be contained in project root")?;
+        let main = if let Some(path) = &input {
+            // Resolve the virtual path of the main file within the project root.
+            let main_path = VirtualPath::within_root(path, &root)
+                .ok_or("source file must be contained in project root")?;
+            FileId::new(None, main_path)
+        } else {
+            // Return the special id of STDIN otherwise
+            *STDIN_ID
+        };
 
         let library = {
             // Convert the input pairs to a dictionary.
@@ -87,7 +103,7 @@ impl SystemWorld {
             workdir: std::env::current_dir().ok(),
             input,
             root,
-            main: FileId::new(None, main_path),
+            main,
             library: Prehashed::new(library),
             book: Prehashed::new(searcher.book),
             fonts: searcher.fonts,
@@ -130,8 +146,8 @@ impl SystemWorld {
     }
 
     /// Return the canonical path to the input file.
-    pub fn input(&self) -> &PathBuf {
-        &self.input
+    pub fn input(&self) -> Option<&PathBuf> {
+        self.input.as_ref()
     }
 
     /// Lookup a source file by id.
@@ -231,7 +247,7 @@ impl FileSlot {
     /// Retrieve the source for this file.
     fn source(&mut self, project_root: &Path) -> FileResult<Source> {
         self.source.get_or_init(
-            || system_path(project_root, self.id),
+            || read(self.id, project_root),
             |data, prev| {
                 let name = if prev.is_some() { "reparsing file" } else { "parsing file" };
                 let _scope = TimingScope::new(name, None);
@@ -249,7 +265,7 @@ impl FileSlot {
     /// Retrieve the file's bytes.
     fn file(&mut self, project_root: &Path) -> FileResult<Bytes> {
         self.file
-            .get_or_init(|| system_path(project_root, self.id), |data, _| Ok(data.into()))
+            .get_or_init(|| read(self.id, project_root), |data, _| Ok(data.into()))
     }
 }
 
@@ -283,7 +299,7 @@ impl<T: Clone> SlotCell<T> {
     /// Gets the contents of the cell or initialize them.
     fn get_or_init(
         &mut self,
-        path: impl FnOnce() -> FileResult<PathBuf>,
+        load: impl FnOnce() -> FileResult<Vec<u8>>,
         f: impl FnOnce(Vec<u8>, Option<T>) -> FileResult<T>,
     ) -> FileResult<T> {
         // If we accessed the file already in this compilation, retrieve it.
@@ -294,7 +310,7 @@ impl<T: Clone> SlotCell<T> {
         }
 
         // Read and hash the file.
-        let result = timed!("loading file", path().and_then(|p| read(&p)));
+        let result = timed!("loading file", load());
         let fingerprint = timed!("hashing file", typst::util::hash128(&result));
 
         // If the file contents didn't change, yield the old processed data.
@@ -329,8 +345,20 @@ fn system_path(project_root: &Path, id: FileId) -> FileResult<PathBuf> {
     id.vpath().resolve(root).ok_or(FileError::AccessDenied)
 }
 
-/// Read a file.
-fn read(path: &Path) -> FileResult<Vec<u8>> {
+/// Reads a file from a `FileId`.
+///
+/// If the ID represents stdin it will read from standard input,
+/// otherwise it gets the file path of the ID and reads the file from disk.
+fn read(id: FileId, project_root: &Path) -> FileResult<Vec<u8>> {
+    if id == *STDIN_ID {
+        read_from_stdin()
+    } else {
+        read_from_disk(&system_path(project_root, id)?)
+    }
+}
+
+/// Read a file from disk.
+fn read_from_disk(path: &Path) -> FileResult<Vec<u8>> {
     let f = |e| FileError::from_io(e, path);
     if fs::metadata(path).map_err(f)?.is_dir() {
         Err(FileError::IsDirectory)
@@ -339,6 +367,18 @@ fn read(path: &Path) -> FileResult<Vec<u8>> {
     }
 }
 
+/// Read from stdin.
+fn read_from_stdin() -> FileResult<Vec<u8>> {
+    let mut buf = Vec::new();
+    let result = io::stdin().read_to_end(&mut buf);
+    match result {
+        Ok(_) => (),
+        Err(err) if err.kind() == io::ErrorKind::BrokenPipe => (),
+        Err(err) => return Err(FileError::from_io(err, Path::new("<stdin>"))),
+    }
+    Ok(buf)
+}
+
 /// Decode UTF-8 with an optional BOM.
 fn decode_utf8(buf: &[u8]) -> FileResult<&str> {
     // Remove UTF-8 BOM.
diff --git a/crates/typst-syntax/src/file.rs b/crates/typst-syntax/src/file.rs
index 40659c6a..6699f05d 100644
--- a/crates/typst-syntax/src/file.rs
+++ b/crates/typst-syntax/src/file.rs
@@ -57,6 +57,24 @@ impl FileId {
         id
     }
 
+    /// Create a new unique ("fake") file specification, which is not
+    /// accessible by path.
+    ///
+    /// Caution: the ID returned by this method is the *only* identifier of the
+    /// file, constructing a file ID with a path will *not* reuse the ID even
+    /// if the path is the same. This method should only be used for generating
+    /// "virtual" file ids such as content read from stdin.
+    #[track_caller]
+    pub fn new_fake(path: VirtualPath) -> Self {
+        let mut interner = INTERNER.write().unwrap();
+        let num = interner.from_id.len().try_into().expect("out of file ids");
+
+        let id = FileId(num);
+        let leaked = Box::leak(Box::new((None, path)));
+        interner.from_id.push(leaked);
+        id
+    }
+
     /// The package the file resides in, if any.
     pub fn package(&self) -> Option<&'static PackageSpec> {
         self.pair().0.as_ref()