Fixing icu_datagen no-download path (#3446)

unicode-org · May 23, 2023 · d8cc2f3 · d8cc2f3
1 parent 5436b31
commit d8cc2f3
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 5 deletions.
diff --git a/provider/datagen/src/bin/datagen/args.rs b/provider/datagen/src/bin/datagen/args.rs
@@ -121,21 +121,19 @@ pub struct Cli {
     icuexport_tag: String,
 
     #[arg(long, value_name = "PATH")]
-    #[cfg_attr(not(feature = "networking"), arg(default_value = "builtin"))]
     #[arg(
         help = "Path to a local icuexport directory (see https://github.com/unicode-org/icu/releases).\n\
                   Note that some keys do not support versions before release-71-1."
     )]
     icuexport_root: Option<PathBuf>,
 
-    #[arg(long, value_name = "TAG")]
+    #[arg(long, value_name = "TAG", default_value = "latest")]
     #[arg(
         help = "Download segmentation LSTM models from this GitHub tag (https://github.com/unicode-org/lstm_word_segmentation/tags)\n\
                   Use 'latest' for the latest version verified to work with this version of the binary.\n\
                   Ignored if '--segmenter-lstm-root' is present. Requires binary to be built with `networking` Cargo feature (enabled by default)."
     )]
     #[cfg_attr(not(feature = "networking"), arg(hide = true))]
-    #[cfg_attr(feature = "networking", arg(default_value = "latest"))]
     segmenter_lstm_tag: String,
 
     #[arg(long, value_name = "PATH")]
@@ -333,7 +331,7 @@ impl Cli {
             #[cfg(feature = "networking")]
             (_, tag) => config::PathOrTag::Tag(String::from(tag)),
             #[cfg(not(feature = "networking"))]
-            _ => eyre::bail!("--{root_arg} flag is mandatory unless datagen is built with the `\"networking\"` Cargo feature"),
+            _ => config::PathOrTag::None,
         })
     }
 

diff --git a/provider/datagen/src/bin/datagen/config.rs b/provider/datagen/src/bin/datagen/config.rs
@@ -77,6 +77,8 @@ pub enum PathOrTag {
     Tag(String),
     #[cfg(feature = "networking")]
     Latest,
+    #[cfg(not(feature = "networking"))]
+    None,
 }
 
 #[derive(Debug, serde::Serialize, serde::Deserialize, PartialEq)]

diff --git a/provider/datagen/src/bin/datagen/mod.rs b/provider/datagen/src/bin/datagen/mod.rs
@@ -52,6 +52,8 @@ fn main() -> eyre::Result<()> {
         }
         #[cfg(feature = "networking")]
         config::PathOrTag::Tag(tag) => source_data.with_cldr_for_tag(&tag, Default::default())?,
+        #[cfg(not(feature = "networking"))]
+        config::PathOrTag::None => source_data,
     };
 
     source_data = match config.icu_export {
@@ -62,17 +64,20 @@ fn main() -> eyre::Result<()> {
         }
         #[cfg(feature = "networking")]
         config::PathOrTag::Tag(tag) => source_data.with_icuexport_for_tag(&tag)?,
+        #[cfg(not(feature = "networking"))]
+        config::PathOrTag::None => source_data,
     };
 
     source_data = match config.segmenter_lstm {
-        config::PathOrTag::Path(path) if path.as_os_str() == "builtin" => source_data,
         config::PathOrTag::Path(path) => source_data.with_icuexport(path)?,
         #[cfg(feature = "networking")]
         config::PathOrTag::Latest => {
             source_data.with_segmenter_lstm_for_tag(SourceData::LATEST_TESTED_SEGMENTER_LSTM_TAG)?
         }
         #[cfg(feature = "networking")]
         config::PathOrTag::Tag(tag) => source_data.with_segmenter_lstm_for_tag(&tag)?,
+        #[cfg(not(feature = "networking"))]
+        config::PathOrTag::None => source_data,
     };
 
     let provider = DatagenProvider::try_new(options, source_data)?;