[feat] add --merge CLI flag to control output behavior; update integration tests accordingly
This commit is contained in:
263
src/main.rs
263
src/main.rs
@@ -40,6 +40,10 @@ struct Args {
|
||||
#[arg(short, long, value_name = "FILE")]
|
||||
output: Option<String>,
|
||||
|
||||
/// Merge all inputs into a single output; if not set, each input is written as a separate output
|
||||
#[arg(short = 'm', long = "merge")]
|
||||
merge: bool,
|
||||
|
||||
/// Language code to use for transcription (e.g., en, de). No auto-detection.
|
||||
#[arg(short, long, value_name = "LANG")]
|
||||
language: Option<String>,
|
||||
@@ -426,119 +430,174 @@ fn main() -> Result<()> {
|
||||
return Err(anyhow!("Please specify --language (e.g., --language en). Language detection was removed."));
|
||||
}
|
||||
|
||||
let mut entries: Vec<OutputEntry> = Vec::new();
|
||||
if args.merge {
|
||||
// MERGED MODE (previous default)
|
||||
let mut entries: Vec<OutputEntry> = Vec::new();
|
||||
for input_path in &inputs {
|
||||
let path = Path::new(input_path);
|
||||
let speaker = sanitize_speaker_name(
|
||||
path.file_stem()
|
||||
.and_then(|s| s.to_str())
|
||||
.unwrap_or("speaker")
|
||||
);
|
||||
|
||||
for input_path in &inputs {
|
||||
let path = Path::new(input_path);
|
||||
let speaker = sanitize_speaker_name(
|
||||
path.file_stem()
|
||||
.and_then(|s| s.to_str())
|
||||
.unwrap_or("speaker")
|
||||
);
|
||||
|
||||
let mut buf = String::new();
|
||||
if is_audio_file(path) {
|
||||
let items = transcribe_native(path, &speaker, lang_hint.as_deref())?;
|
||||
for e in items {
|
||||
entries.push(e);
|
||||
let mut buf = String::new();
|
||||
if is_audio_file(path) {
|
||||
let items = transcribe_native(path, &speaker, lang_hint.as_deref())?;
|
||||
for e in items { entries.push(e); }
|
||||
continue;
|
||||
} else if is_json_file(path) {
|
||||
File::open(path)
|
||||
.with_context(|| format!("Failed to open: {}", input_path))?
|
||||
.read_to_string(&mut buf)
|
||||
.with_context(|| format!("Failed to read: {}", input_path))?;
|
||||
} else {
|
||||
return Err(anyhow!(format!("Unsupported input type (expected .json or audio media): {}", input_path)));
|
||||
}
|
||||
continue;
|
||||
} else if is_json_file(path) {
|
||||
File::open(path)
|
||||
.with_context(|| format!("Failed to open: {}", input_path))?
|
||||
.read_to_string(&mut buf)
|
||||
.with_context(|| format!("Failed to read: {}", input_path))?;
|
||||
|
||||
let root: InputRoot = serde_json::from_str(&buf)
|
||||
.with_context(|| format!("Invalid JSON transcript parsed from {}", input_path))?;
|
||||
|
||||
for seg in root.segments {
|
||||
entries.push(OutputEntry {
|
||||
id: 0,
|
||||
speaker: speaker.clone(),
|
||||
start: seg.start,
|
||||
end: seg.end,
|
||||
text: seg.text,
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
// Sort globally by (start, end)
|
||||
entries.sort_by(|a, b| {
|
||||
match a.start.partial_cmp(&b.start) {
|
||||
Some(std::cmp::Ordering::Equal) | None => {}
|
||||
Some(o) => return o,
|
||||
}
|
||||
a.end
|
||||
.partial_cmp(&b.end)
|
||||
.unwrap_or(std::cmp::Ordering::Equal)
|
||||
});
|
||||
for (i, e) in entries.iter_mut().enumerate() { e.id = i as u64; }
|
||||
let out = OutputRoot { items: entries };
|
||||
|
||||
if let Some(path) = output_path {
|
||||
let base_path = Path::new(&path);
|
||||
let parent_opt = base_path.parent();
|
||||
if let Some(parent) = parent_opt {
|
||||
if !parent.as_os_str().is_empty() {
|
||||
create_dir_all(parent).with_context(|| {
|
||||
format!("Failed to create parent directory for output: {}", parent.display())
|
||||
})?;
|
||||
}
|
||||
}
|
||||
let stem = base_path.file_stem().and_then(|s| s.to_str()).unwrap_or("output");
|
||||
let date = date_prefix();
|
||||
let base_name = format!("{}_{}", date, stem);
|
||||
let dir = parent_opt.unwrap_or(Path::new(""));
|
||||
let json_path = dir.join(format!("{}.json", &base_name));
|
||||
let toml_path = dir.join(format!("{}.toml", &base_name));
|
||||
let srt_path = dir.join(format!("{}.srt", &base_name));
|
||||
|
||||
let mut json_file = File::create(&json_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", json_path.display()))?;
|
||||
serde_json::to_writer_pretty(&mut json_file, &out)?; writeln!(&mut json_file)?;
|
||||
|
||||
let toml_str = toml::to_string_pretty(&out)?;
|
||||
let mut toml_file = File::create(&toml_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", toml_path.display()))?;
|
||||
toml_file.write_all(toml_str.as_bytes())?; if !toml_str.ends_with('\n') { writeln!(&mut toml_file)?; }
|
||||
|
||||
let srt_str = render_srt(&out.items);
|
||||
let mut srt_file = File::create(&srt_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", srt_path.display()))?;
|
||||
srt_file.write_all(srt_str.as_bytes())?;
|
||||
} else {
|
||||
return Err(anyhow!(format!("Unsupported input type (expected .json or audio media): {}", input_path)));
|
||||
let stdout = io::stdout();
|
||||
let mut handle = stdout.lock();
|
||||
serde_json::to_writer_pretty(&mut handle, &out)?; writeln!(&mut handle)?;
|
||||
}
|
||||
} else {
|
||||
// SEPARATE MODE (default now)
|
||||
// If writing to stdout with multiple inputs, not supported
|
||||
if output_path.is_none() && inputs.len() > 1 {
|
||||
return Err(anyhow!("Multiple inputs without --merge require -o OUTPUT_DIR to write separate files"));
|
||||
}
|
||||
|
||||
let root: InputRoot = serde_json::from_str(&buf)
|
||||
.with_context(|| format!("Invalid JSON transcript parsed from {}", input_path))?;
|
||||
|
||||
for seg in root.segments {
|
||||
entries.push(OutputEntry {
|
||||
id: 0, // will be reassigned after sorting
|
||||
speaker: speaker.clone(),
|
||||
start: seg.start,
|
||||
end: seg.end,
|
||||
text: seg.text,
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
// Sort globally by (start, end)
|
||||
entries.sort_by(|a, b| {
|
||||
match a.start.partial_cmp(&b.start) {
|
||||
Some(std::cmp::Ordering::Equal) | None => {}
|
||||
Some(o) => return o,
|
||||
}
|
||||
a.end
|
||||
.partial_cmp(&b.end)
|
||||
.unwrap_or(std::cmp::Ordering::Equal)
|
||||
});
|
||||
|
||||
// Reassign contiguous ids by chronological order
|
||||
for (i, e) in entries.iter_mut().enumerate() {
|
||||
e.id = i as u64;
|
||||
}
|
||||
|
||||
// Output as an object containing the array (valid JSON)
|
||||
// Schema equivalent to:
|
||||
// {
|
||||
// [
|
||||
// id: number,
|
||||
// speaker: string,
|
||||
// start: number,
|
||||
// end: number,
|
||||
// text: string
|
||||
// ], ...
|
||||
// }
|
||||
let out = OutputRoot { items: entries };
|
||||
|
||||
if let Some(path) = output_path {
|
||||
let base_path = Path::new(&path);
|
||||
let parent_opt = base_path.parent();
|
||||
if let Some(parent) = parent_opt {
|
||||
if !parent.as_os_str().is_empty() {
|
||||
create_dir_all(parent).with_context(|| {
|
||||
format!("Failed to create parent directory for output: {}", parent.display())
|
||||
})?;
|
||||
// If output_path is provided, treat it as a directory. Create it.
|
||||
let out_dir: Option<PathBuf> = output_path.as_ref().map(|p| PathBuf::from(p));
|
||||
if let Some(dir) = &out_dir {
|
||||
if !dir.as_os_str().is_empty() {
|
||||
create_dir_all(dir).with_context(|| format!("Failed to create output directory: {}", dir.display()))?;
|
||||
}
|
||||
}
|
||||
let stem = base_path.file_stem().and_then(|s| s.to_str()).unwrap_or("output");
|
||||
let date = date_prefix();
|
||||
let base_name = format!("{}_{}", date, stem);
|
||||
let dir = parent_opt.unwrap_or(Path::new(""));
|
||||
let json_path = dir.join(format!("{}.json", &base_name));
|
||||
let toml_path = dir.join(format!("{}.toml", &base_name));
|
||||
let srt_path = dir.join(format!("{}.srt", &base_name));
|
||||
|
||||
// JSON
|
||||
let mut json_file = File::create(&json_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", json_path.display()))?;
|
||||
serde_json::to_writer_pretty(&mut json_file, &out)?;
|
||||
writeln!(&mut json_file)?;
|
||||
for input_path in &inputs {
|
||||
let path = Path::new(input_path);
|
||||
let speaker = sanitize_speaker_name(
|
||||
path.file_stem().and_then(|s| s.to_str()).unwrap_or("speaker")
|
||||
);
|
||||
|
||||
// TOML
|
||||
let toml_str = toml::to_string_pretty(&out)?;
|
||||
let mut toml_file = File::create(&toml_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", toml_path.display()))?;
|
||||
toml_file.write_all(toml_str.as_bytes())?;
|
||||
if !toml_str.ends_with('\n') {
|
||||
writeln!(&mut toml_file)?;
|
||||
// Collect entries per file
|
||||
let mut entries: Vec<OutputEntry> = Vec::new();
|
||||
if is_audio_file(path) {
|
||||
let items = transcribe_native(path, &speaker, lang_hint.as_deref())?;
|
||||
entries.extend(items);
|
||||
} else if is_json_file(path) {
|
||||
let mut buf = String::new();
|
||||
File::open(path)
|
||||
.with_context(|| format!("Failed to open: {}", input_path))?
|
||||
.read_to_string(&mut buf)
|
||||
.with_context(|| format!("Failed to read: {}", input_path))?;
|
||||
let root: InputRoot = serde_json::from_str(&buf)
|
||||
.with_context(|| format!("Invalid JSON transcript parsed from {}", input_path))?;
|
||||
for seg in root.segments {
|
||||
entries.push(OutputEntry { id: 0, speaker: speaker.clone(), start: seg.start, end: seg.end, text: seg.text });
|
||||
}
|
||||
} else {
|
||||
return Err(anyhow!(format!("Unsupported input type (expected .json or audio media): {}", input_path)));
|
||||
}
|
||||
|
||||
// Sort and reassign ids per file
|
||||
entries.sort_by(|a, b| {
|
||||
match a.start.partial_cmp(&b.start) { Some(std::cmp::Ordering::Equal) | None => {} Some(o) => return o }
|
||||
a.end.partial_cmp(&b.end).unwrap_or(std::cmp::Ordering::Equal)
|
||||
});
|
||||
for (i, e) in entries.iter_mut().enumerate() { e.id = i as u64; }
|
||||
let out = OutputRoot { items: entries };
|
||||
|
||||
if let Some(dir) = &out_dir {
|
||||
// Build file names using input stem
|
||||
let stem = path.file_stem().and_then(|s| s.to_str()).unwrap_or("output");
|
||||
let date = date_prefix();
|
||||
let base_name = format!("{}_{}", date, stem);
|
||||
let json_path = dir.join(format!("{}.json", &base_name));
|
||||
let toml_path = dir.join(format!("{}.toml", &base_name));
|
||||
let srt_path = dir.join(format!("{}.srt", &base_name));
|
||||
|
||||
let mut json_file = File::create(&json_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", json_path.display()))?;
|
||||
serde_json::to_writer_pretty(&mut json_file, &out)?; writeln!(&mut json_file)?;
|
||||
|
||||
let toml_str = toml::to_string_pretty(&out)?;
|
||||
let mut toml_file = File::create(&toml_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", toml_path.display()))?;
|
||||
toml_file.write_all(toml_str.as_bytes())?; if !toml_str.ends_with('\n') { writeln!(&mut toml_file)?; }
|
||||
|
||||
let srt_str = render_srt(&out.items);
|
||||
let mut srt_file = File::create(&srt_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", srt_path.display()))?;
|
||||
srt_file.write_all(srt_str.as_bytes())?;
|
||||
} else {
|
||||
// stdout (only single input reaches here)
|
||||
let stdout = io::stdout();
|
||||
let mut handle = stdout.lock();
|
||||
serde_json::to_writer_pretty(&mut handle, &out)?; writeln!(&mut handle)?;
|
||||
}
|
||||
}
|
||||
|
||||
// SRT
|
||||
let srt_str = render_srt(&out.items);
|
||||
let mut srt_file = File::create(&srt_path)
|
||||
.with_context(|| format!("Failed to create output file: {}", srt_path.display()))?;
|
||||
srt_file.write_all(srt_str.as_bytes())?;
|
||||
} else {
|
||||
let stdout = io::stdout();
|
||||
let mut handle = stdout.lock();
|
||||
serde_json::to_writer_pretty(&mut handle, &out)?;
|
||||
writeln!(&mut handle)?;
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user