From 77458bbafa52bf3c9e9c5d3add6339d4281f1121 Mon Sep 17 00:00:00 2001
From: Elango Cheran <elango@unicode.org>
Date: Thu, 21 Oct 2021 16:17:00 -0700
Subject: [PATCH] Initial code for connecting properties provider to the
 icu4x_datagen exporter tool

---
 Cargo.lock                                |  2 +
 components/properties/src/provider.rs     | 10 ++--
 provider/testdata/src/paths.rs            |  5 ++
 provider/uprops/src/enum_codepointtrie.rs | 19 +++++++
 tools/datagen/Cargo.toml                  |  2 +
 tools/datagen/src/bin/datagen.rs          | 65 +++++++++++++++++++++++
 6 files changed, 98 insertions(+), 5 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index a6e2e5ca52c..28425d8f458 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1089,10 +1089,12 @@ dependencies = [
  "clap",
  "futures",
  "icu_locid",
+ "icu_properties",
  "icu_provider",
  "icu_provider_blob",
  "icu_provider_cldr",
  "icu_provider_fs",
+ "icu_provider_uprops",
  "icu_testdata",
  "log",
  "reqwest",
diff --git a/components/properties/src/provider.rs b/components/properties/src/provider.rs
index d4e2d5abaa8..4c5f8896bad 100644
--- a/components/properties/src/provider.rs
+++ b/components/properties/src/provider.rs
@@ -22,19 +22,18 @@ pub mod key {
 
     /// Macro to help define resource keys and store them in a list.
     macro_rules! define_resource_keys {
-        ($count:expr; $(($k:ident, $s:literal)),+,) => {
+        ($allkeys:ident; $count:expr; $(($k:ident, $s:literal)),+,) => {
             $(
                 #[allow(missing_docs)] // These constants don't need individual documentation.
                 pub const $k: ResourceKey = resource_key!(UnicodeSet, $s, 1);
             )+
 
             /// The set of all resource keys supported by [`icu_uniset`](crate).
-            pub const ALL_KEYS: [ResourceKey; $count] = [$($k,)+];
+            pub const $allkeys: [ResourceKey; $count] = [$($k,)+];
         };
     }
 
-    define_resource_keys!(267;
-
+    define_resource_keys!(ALL_SET_KEYS; 265;
         //
         // Binary property UnicodeSets
         //
@@ -313,7 +312,9 @@ pub mod key {
         (SCRIPT_YEZIDI_V1, "sc=Yezi"),
         (SCRIPT_YI_V1, "sc=Yiii"),
         (SCRIPT_ZANABAZAR_SQUARE_V1, "sc=Zanb"),
+    );
 
+    define_resource_keys!(ALL_MAP_KEYS; 2;
         //
         // Enumerated property CodePointMaps
         //
@@ -322,7 +323,6 @@ pub mod key {
 
         (GENERAL_CATEGORY_V1, "gc"),
         (SCRIPT_V1, "sc"),
-
     );
 }
 
diff --git a/provider/testdata/src/paths.rs b/provider/testdata/src/paths.rs
index ed3cc0ac372..fa9a60cd485 100644
--- a/provider/testdata/src/paths.rs
+++ b/provider/testdata/src/paths.rs
@@ -14,6 +14,11 @@ pub fn cldr_json_root() -> PathBuf {
     data_root().join("cldr")
 }
 
+/// Returns the absolute path to the UProps TOML root directory.
+pub fn uprops_toml_root() -> PathBuf {
+    data_root().join("uprops")
+}
+
 /// Returns the absolute path to the ICU4X JSON root directory.
 pub fn icu4x_json_root() -> PathBuf {
     data_root().join("json")
diff --git a/provider/uprops/src/enum_codepointtrie.rs b/provider/uprops/src/enum_codepointtrie.rs
index a8cf5f2a6ad..8b5a0e39c37 100644
--- a/provider/uprops/src/enum_codepointtrie.rs
+++ b/provider/uprops/src/enum_codepointtrie.rs
@@ -8,8 +8,11 @@ use crate::uprops_serde::enumerated::EnumeratedPropertyCodePointTrie;
 
 use icu_codepointtrie::codepointtrie::{CodePointTrie, CodePointTrieHeader, TrieType, TrieValue};
 use icu_properties::provider::{UnicodePropertyMapV1, UnicodePropertyMapV1Marker};
+use icu_provider::iter::IterableDataProviderCore;
 use icu_provider::prelude::*;
 use zerovec::ZeroVec;
+use icu_properties::provider::*;
+use icu_properties::{Script, GeneralSubcategory};
 
 use core::convert::TryFrom;
 
@@ -110,6 +113,22 @@ impl<'data, T: TrieValue> DataProvider<'data, UnicodePropertyMapV1Marker<T>>
     }
 }
 
+icu_provider::impl_dyn_provider!(EnumeratedPropertyCodePointTrieProvider, {
+    key::GENERAL_CATEGORY_V1 => UnicodePropertyMapV1Marker<GeneralSubcategory>,
+    key::SCRIPT_V1 => UnicodePropertyMapV1Marker<Script>,
+}, SERDE_SE, 'data);
+
+impl IterableDataProviderCore for EnumeratedPropertyCodePointTrieProvider {
+    fn supported_options_for_key(
+        &self,
+        _resc_key: &ResourceKey,
+    ) -> Result<Box<dyn Iterator<Item = ResourceOptions>>, DataError> {
+        let list: Vec<ResourceOptions> = vec![ResourceOptions::default()];
+        Ok(Box::new(list.into_iter()))
+    }
+}
+
+
 #[cfg(test)]
 mod tests {
     use super::*;
diff --git a/tools/datagen/Cargo.toml b/tools/datagen/Cargo.toml
index 2a86839f247..1cf5cc5f611 100644
--- a/tools/datagen/Cargo.toml
+++ b/tools/datagen/Cargo.toml
@@ -31,10 +31,12 @@ anyhow = "1.0"
 clap = "2.33"
 futures = "0.3"
 icu_locid = { version = "0.3", path = "../../components/locid", features = ["std"]}
+icu_properties = { version = "0.3", path = "../../components/properties", features = ["std"]}
 icu_provider = { version = "0.3", path = "../../provider/core", features = ["std"]}
 icu_provider_blob = { version = "0.3", path = "../../provider/blob", features = ["export"] }
 icu_provider_cldr = { version = "0.3", path = "../../provider/cldr", features = ["download"] }
 icu_provider_fs = { version = "0.3", path = "../../provider/fs", features = ["export", "provider_json", "provider_bincode"] }
+icu_provider_uprops = { version = "0.3", path = "../../provider/uprops" }
 icu_testdata = { version = "0.3", path = "../../provider/testdata", features = ["metadata"] }
 log = "0.4"
 reqwest = { version = "0.11", features = ["json", "stream", "gzip"] }
diff --git a/tools/datagen/src/bin/datagen.rs b/tools/datagen/src/bin/datagen.rs
index a2ddeada49e..6589d342ff2 100644
--- a/tools/datagen/src/bin/datagen.rs
+++ b/tools/datagen/src/bin/datagen.rs
@@ -5,6 +5,7 @@
 use anyhow::Context;
 use clap::{App, Arg, ArgGroup, ArgMatches};
 use icu_locid::LanguageIdentifier;
+use icu_provider::prelude::*;
 use icu_provider::export::DataExporter;
 use icu_provider::filter::Filterable;
 use icu_provider::hello_world::{self, HelloWorldProvider};
@@ -20,6 +21,8 @@ use icu_provider_fs::export::fs_exporter;
 use icu_provider_fs::export::serializers;
 use icu_provider_fs::export::FilesystemExporter;
 use icu_provider_fs::manifest;
+use icu_properties::provider::key::{ALL_SET_KEYS, ALL_MAP_KEYS};
+use icu_provider_uprops::{EnumeratedPropertyCodePointTrieProvider, PropertiesDataProvider};
 use simple_logger::SimpleLogger;
 use std::collections::HashSet;
 use std::path::PathBuf;
@@ -261,6 +264,8 @@ fn main() -> anyhow::Result<()> {
     if matches.is_present("ALL_KEYS") || matches.is_present("KEYS") {
         let keys = matches.values_of("KEYS").map(|values| values.collect());
         export_cldr(&matches, exporter, locales_vec.as_deref(), keys.as_ref())?;
+        export_set_props(&matches, exporter, keys.as_ref())?;
+        export_map_props(&matches, exporter, keys.as_ref())?;
     }
 
     if matches.is_present("HELLO_WORLD") {
@@ -402,6 +407,66 @@ fn export_cldr<'data>(
     Ok(())
 }
 
+fn export_set_props<'data>(
+    matches: &ArgMatches,
+    exporter: &mut (impl DataExporter<'data, SerdeSeDataStructMarker> + ?Sized),
+    allowed_keys: Option<&HashSet<&str>>,
+) -> anyhow::Result<()> {
+    let provider = 
+        if let Some(path) = matches.value_of("CLDR_ROOT") {
+            PropertiesDataProvider::new(PathBuf::from(path))
+        } else {
+            anyhow::bail!("Value for --cldr-root must be specified",)
+        };
+
+    let keys = ALL_SET_KEYS;
+    let keys: Vec<ResourceKey> = if let Some(allowed_keys) = allowed_keys {
+        keys.into_iter()
+            .filter(|k| allowed_keys.contains(&*k.writeable_to_string()))
+            .copied()
+            .collect()
+    } else {
+        keys.to_vec()
+    };
+
+    for key in keys.iter() {
+        log::info!("Writing key: {}", key);
+        icu_provider::export::export_from_iterable(key, &provider, exporter)?;
+    }
+
+    Ok(())
+}
+
+fn export_map_props<'data>(
+    matches: &ArgMatches,
+    exporter: &mut (impl DataExporter<'data, SerdeSeDataStructMarker> + ?Sized),
+    allowed_keys: Option<&HashSet<&str>>,
+) -> anyhow::Result<()> {
+    let provider = 
+        if let Some(path) = matches.value_of("CLDR_ROOT") {
+            EnumeratedPropertyCodePointTrieProvider::new(PathBuf::from(path))
+        } else {
+            anyhow::bail!("Value for --cldr-root must be specified",)
+        };
+
+    let keys = ALL_MAP_KEYS;
+    let keys: Vec<ResourceKey> = if let Some(allowed_keys) = allowed_keys {
+        keys.into_iter()
+            .filter(|k| allowed_keys.contains(&*k.writeable_to_string()))
+            .copied()
+            .collect()
+    } else {
+        keys.to_vec()
+    };
+
+    for key in keys.iter() {
+        log::info!("Writing key: {}", key);
+        icu_provider::export::export_from_iterable(key, &provider, exporter)?;
+    }
+
+    Ok(())
+}
+
 fn export_hello_world<'data>(
     _: &ArgMatches,
     exporter: &mut (impl DataExporter<'data, SerdeSeDataStructMarker> + ?Sized),