[feat] add --merge-and-separate CLI flag; implement combined mode logic and tests

This commit is contained in:
2025-08-08 12:57:12 +02:00
parent 4d3016200e
commit 1d23be8066
3 changed files with 170 additions and 35 deletions

View File

@@ -61,6 +61,10 @@ struct Args {
#[arg(short = 'm', long = "merge")]
merge: bool,
/// Merge and also write separate outputs per input; requires -o OUTPUT_DIR
#[arg(long = "merge-and-separate")]
merge_and_separate: bool,
/// Language code to use for transcription (e.g., en, de). No auto-detection.
#[arg(short, long, value_name = "LANG")]
language: Option<String>,
@@ -88,7 +92,7 @@ struct InputSegment {
// other fields are ignored
}
#[derive(Debug, Serialize)]
#[derive(Debug, Serialize, Clone)]
struct OutputEntry {
id: u64,
speaker: String,
@@ -447,7 +451,108 @@ fn main() -> Result<()> {
return Err(anyhow!("Please specify --language (e.g., --language en). Language detection was removed."));
}
if args.merge {
if args.merge_and_separate {
// Combined mode: write separate outputs per input and also a merged output set
// Require an output directory
let out_dir = match output_path.as_ref() {
Some(p) => PathBuf::from(p),
None => return Err(anyhow!("--merge-and-separate requires -o OUTPUT_DIR")),
};
if !out_dir.as_os_str().is_empty() {
create_dir_all(&out_dir)
.with_context(|| format!("Failed to create output directory: {}", out_dir.display()))?;
}
let mut merged_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")
);
// Collect entries per file and extend merged
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.into_iter());
} 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; }
// Write separate outputs to out_dir
let out = OutputRoot { items: entries.clone() };
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 = out_dir.join(format!("{}.json", &base_name));
let toml_path = out_dir.join(format!("{}.toml", &base_name));
let srt_path = out_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())?;
// Extend merged with per-file entries
merged_entries.extend(out.items.into_iter());
}
// Now write merged output set into out_dir
merged_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 merged_entries.iter_mut().enumerate() { e.id = i as u64; }
let merged_out = OutputRoot { items: merged_entries };
let date = date_prefix();
let merged_base = format!("{}_merged", date);
let m_json = out_dir.join(format!("{}.json", &merged_base));
let m_toml = out_dir.join(format!("{}.toml", &merged_base));
let m_srt = out_dir.join(format!("{}.srt", &merged_base));
let mut mj = File::create(&m_json)
.with_context(|| format!("Failed to create output file: {}", m_json.display()))?;
serde_json::to_writer_pretty(&mut mj, &merged_out)?; writeln!(&mut mj)?;
let m_toml_str = toml::to_string_pretty(&merged_out)?;
let mut mt = File::create(&m_toml)
.with_context(|| format!("Failed to create output file: {}", m_toml.display()))?;
mt.write_all(m_toml_str.as_bytes())?; if !m_toml_str.ends_with('\n') { writeln!(&mut mt)?; }
let m_srt_str = render_srt(&merged_out.items);
let mut ms = File::create(&m_srt)
.with_context(|| format!("Failed to create output file: {}", m_srt.display()))?;
ms.write_all(m_srt_str.as_bytes())?;
} else if args.merge {
// MERGED MODE (previous default)
let mut entries: Vec<OutputEntry> = Vec::new();
for input_path in &inputs {