Add flags to control generated files

Make dictionary processing async.

Signed-off-by: Tin Švagelj <tin.svagelj@live.com>

Cargo.toml

@@ -5,7 +5,6 @@ edition = "2021"
 description = "Wikipedia dump download and cleanup tool"
 authors = ["Tin Švagelj <tin.svagelj@live.com>"]
 license = "GPLv3"
-license-file = "LICENSE"
 
 [dependencies]
 tokio = { version = "1.38", features = ["full"] }

README.md

@@ -48,7 +48,7 @@ manager of choice) to compile the program.
 All configuration is done through CLI arguments.
 
 Print arguments with:
-```
+```sh
 wiki-extractor --help
 wiki-extractor local --help
 wiki-extractor remote --help
@@ -58,9 +58,9 @@ wiki-extractor remote --help
 
 - Find a closest [mirror](https://dumps.wikimedia.org/mirrors.html) to make the download faster.
   - Used URL should be the part before `/enwiki/latest` part once you locate the dump files.
-- Replace the command below with
+- Replace the URL in command below with your mirror of choice:
 
-```
+```sh
 wiki-extractor remote https://dumps.wikimedia.org/ -L en -w latest -o 
 ```
 

src/main.rs

@@ -47,9 +47,12 @@ pub struct Args {
     #[arg(short = 'o', long = "output", default_value = "./dump")]
     pub output: std::path::PathBuf,
 
+    /// Selection of generated files.
+    #[clap(flatten)]
+    pub generator: output::options::GeneratorOptions,
     /// Options for text dump generation.
     #[clap(flatten)]
-    pub text_options: output::options::TextOptions,
+    pub text: output::options::TextOptions,
 }
 
 fn main() -> anyhow::Result<()> {
@@ -60,9 +63,15 @@ fn main() -> anyhow::Result<()> {
     let Args {
         input,
         output,
-        text_options,
+        generator: generator_options,
+        text: text_options,
     } = Args::parse();
 
+    if !generator_options.any() {
+        log::info!("Nothing to do. See `--help` for list of generators.");
+        std::process::exit(0);
+    }
+
     let rt = tokio::runtime::Builder::new_multi_thread()
         .enable_all()
         .build()
@@ -75,7 +84,7 @@ fn main() -> anyhow::Result<()> {
         std::process::exit(1);
     }
 
-    let mut gen = DataGenerator::new(output, text_options)?;
+    let mut gen = DataGenerator::new(output, generator_options, text_options)?;
 
     if let Some(updated) = dump.updated {
         log::info!("Dump creation date: {updated}");