Switch to WASMI engine in icu_codepointtrie_builder (#4621)

Fixes #3241 

I switched our WASM engine to `wasmi` because it is faster to compile
and has fewer dependencies. I changed how we interact with C++, too, in
order to improve the performance.

components/collections/codepointtrie_builder/Cargo.toml

@@ -31,42 +31,18 @@ independent = true
 [features]
 # Use the wasm builder
 default = ["wasm"]
-wasm = ["wasmer", "wasmer-wasi", "once_cell"]
+wasm = ["dep:wasmi"]
 # Use the ICU4C builder
 # needs the ICU4C_LIB_PATH variable set and pointing to an ICU4C lib folder
 # containing dylibs. If you want to use staticlibs, set ICU4C_LINK_STATICALLY.
 # Will be silently disabled if the wasm feature is enabled
-icu4c = ["zerovec"]
+icu4c = []
 
 [dependencies]
 icu_collections = { workspace = true, features = ["serde"] }
-once_cell = { version = "1.18.0", optional = true }
 toml = "0.5"
-zerovec = { workspace = true, optional = true }
-
-[dependencies.wasmer]
-version = "2.2.1"
-default-features = false
-optional = true
-features = [
- # We are running on the local system (as opposed to JavaScript)
- "sys",
- # By default, use the Singlepass compiler for faster startup time
- "default-singlepass",
- # By default, use the Universal engine for in-memory compiled code
- "default-universal"
-]
-
-[dependencies.wasmer-wasi]
-version = "2.2.1"
-default-features = false
-optional = true
-features = [
- # We are running on the local system (as opposed to JavaScript)
- "sys",
- # We don't need real filesystem access, so use mem-fs
- "mem-fs"
-]
+zerovec = { workspace = true }
+wasmi = { version = "0.31.2", optional = true }
 
 [dev-dependencies]
 icu = { workspace = true }

components/collections/codepointtrie_builder/cpp/.gitignore

@@ -2,3 +2,4 @@ list_to_ucptrie
 list_to_ucptrie.wasm
 list_to_ucptrie.wat
 icu4c_obj/
+wasm_obj/

components/collections/codepointtrie_builder/cpp/Makefile

@@ -3,34 +3,79 @@
 # (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
 
 ICU4C_SOURCE ?= $(HOME)/icu/icu4c/source
-ICU4C_BUILD ?= $(ICU4C_SOURCE)
 CXX := clang++
-WASICXX := wasic++
 
 .PHONY: all clean
-all: list_to_ucptrie list_to_ucptrie.wasm
+all: ucptrie_wrap.wasm
 
-icu4c_obj/wasm/%.o: $(ICU4C_SOURCE)/%.cpp
- mkdir -p icu4c_obj/wasm/common
- mkdir -p icu4c_obj/wasm/tools/toolutil
- $(WASICXX) -DTRUE=1 -c -flto -I$(ICU4C_SOURCE)/common -I$(ICU4C_SOURCE)/tools/toolutil $< -o $@
+# The objects are built with the following flags:
+# --target=wasm32-unknown-wasi to build with WASI for system symbols
+# -DTRUE=1 to fix error use of undeclared identifier 'TRUE'
+# -DU_DISABLE_RENAMING so the same WASM module works across ICU versions
+# --compile to skip the linking step
+# -flto to include LTO (link-time optimization) metadata
 
-icu4c_obj_wasm_files = \
-icu4c_obj/wasm/common/cmemory.o \
-icu4c_obj/wasm/common/errorcode.o \
-icu4c_obj/wasm/common/ucptrie.o \
-icu4c_obj/wasm/common/umutablecptrie.o \
-icu4c_obj/wasm/common/uobject.o \
-icu4c_obj/wasm/common/utypes.o \
-icu4c_obj/wasm/tools/toolutil/writesrc.o
+wasm_obj/icu4c/%.o: $(ICU4C_SOURCE)/%.cpp
+ mkdir -p wasm_obj/icu4c/common
+ $(CXX) --target=wasm32-unknown-wasi \
+ -DTRUE=1 \
+ -DU_DISABLE_RENAMING=1 \
+ --compile \
+ -flto \
+ -I$(ICU4C_SOURCE)/common \
+ $< \
+ -o $@
 
-list_to_ucptrie: list_to_ucptrie.cpp $(icu4c_obj_native_files)
- $(CXX) -o list_to_ucptrie -ldl -I$(ICU4C_SOURCE)/common -I$(ICU4C_SOURCE)/tools/toolutil list_to_ucptrie.cpp $(ICU4C_BUILD)/lib/libicutu.a $(ICU4C_BUILD)/lib/libicuuc.a
+wasm_obj/ucptrie_wrap.o: ucptrie_wrap.cpp
+ mkdir -p wasm_obj
+ $(CXX) --target=wasm32-unknown-wasi \
+ -I/usr/include/wasm32-wasi \
+ --compile \
+ -flto \
+ -I$(ICU4C_SOURCE)/common \
+ $< \
+ -o $@
 
-list_to_ucptrie.wasm: list_to_ucptrie.cpp $(icu4c_obj_wasm_files)
- $(WASICXX) -DTRUE=1 -flto -Wl,--gc-sections -Wl,--strip-all -I$(ICU4C_SOURCE)/common -I$(ICU4C_SOURCE)/tools/toolutil $(icu4c_obj_wasm_files) list_to_ucptrie.cpp -o list_to_ucptrie.wasm
+wasm_obj_files = \
+ wasm_obj/icu4c/common/cmemory.o \
+ wasm_obj/icu4c/common/errorcode.o \
+ wasm_obj/icu4c/common/ucptrie.o \
+ wasm_obj/icu4c/common/umutablecptrie.o \
+ wasm_obj/icu4c/common/uobject.o \
+ wasm_obj/icu4c/common/utypes.o \
+ wasm_obj/ucptrie_wrap.o
+
+# Compile as a Reactor Module so we don't need a main function.
+# This also means we don't need WASI imports at runtime.
+# <https://dylibso.com/blog/wasi-command-reactor/>
+ucptrie_wrap.wasm: $(wasm_obj_files)
+ $(CXX) --target=wasm32-unknown-wasi \
+ -mexec-model=reactor \
+ -DTRUE=1 \
+ -flto \
+ -Wl,--gc-sections \
+ -Wl,--strip-all \
+ -Wl,--export=umutablecptrie_open \
+ -Wl,--export=umutablecptrie_set \
+ -Wl,--export=umutablecptrie_buildImmutable \
+ -Wl,--export=ucptrie_close \
+ -Wl,--export=umutablecptrie_close \
+ -Wl,--export=create_uerrorcode \
+ -Wl,--export=read_uerrorcode \
+ -Wl,--export=read_ucptrie_highStart \
+ -Wl,--export=read_ucptrie_shifted12HighStart \
+ -Wl,--export=read_ucptrie_index3NullOffset \
+ -Wl,--export=read_ucptrie_dataNullOffset \
+ -Wl,--export=read_ucptrie_nullValue \
+ -Wl,--export=get_index_ptr \
+ -Wl,--export=get_index_length \
+ -Wl,--export=get_data_ptr \
+ -Wl,--export=get_data_length \
+ -I$(ICU4C_SOURCE)/common \
+ -I$(ICU4C_SOURCE)/tools/toolutil \
+ $(wasm_obj_files) \
+ -o ucptrie_wrap.wasm
 
 clean:
- rm -rf icu4c_obj
- rm -f list_to_ucptrie
- rm -f list_to_ucptrie.wasm
+ rm -rf wasm_obj
+ rm -f ucptrie_wrap.wasm