build: generate hex dumps of server assets on the fly

Author: ochafik

.gitignore

@@ -34,6 +34,7 @@ lcov-report/
 gcovr-report/
 
 build*
+!build.zig
 cmake-build-*
 out/
 tmp/
@@ -100,6 +101,9 @@ qnt-*.txt
 perf-*.txt
 
 examples/jeopardy/results.txt
+examples/server/*.html.hpp
+examples/server/*.js.hpp
+examples/server/*.mjs.hpp
 
 poetry.lock
 poetry.toml

Makefile

@@ -2,7 +2,8 @@
 BUILD_TARGETS = \
 	main quantize quantize-stats perplexity imatrix embedding vdot q8dot train-text-from-scratch convert-llama2c-to-ggml \
 	simple batched batched-bench save-load-state server gguf gguf-split eval-callback llama-bench libllava.a llava-cli baby-llama beam-search  \
-	retrieval speculative infill tokenize benchmark-matmult parallel finetune export-lora lookahead lookup passkey gritlm tests/test-c.o
+	retrieval speculative infill tokenize benchmark-matmult parallel finetune export-lora lookahead lookup passkey gritlm tests/test-c.o \
+	examples/server/index.html.hpp examples/server/index.js.hpp examples/server/completion.js.hpp examples/server/json-schema-to-grammar.mjs.hpp
 
 # Binaries only useful for tests
 TEST_TARGETS = \
@@ -788,10 +789,13 @@ save-load-state: examples/save-load-state/save-load-state.cpp ggml.o llama.o $(C
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
 
-server: examples/server/server.cpp examples/server/utils.hpp examples/server/httplib.h common/json.hpp examples/server/index.html.hpp examples/server/index.js.hpp examples/server/completion.js.hpp json-schema-to-grammar.o common/stb_image.h ggml.o llama.o $(COMMON_DEPS) grammar-parser.o $(OBJS)
+server: examples/server/server.cpp examples/server/utils.hpp examples/server/httplib.h common/json.hpp examples/server/index.html.hpp examples/server/index.js.hpp examples/server/completion.js.hpp examples/server/json-schema-to-grammar.mjs.hpp json-schema-to-grammar.o common/stb_image.h ggml.o llama.o $(COMMON_DEPS) grammar-parser.o $(OBJS)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h %.hpp $<,$^) -Iexamples/server $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS) $(LWINSOCK2)
 
+examples/server/%.hpp: examples/server/public/%
+	xxd -n $(notdir $<) -i $< $@
+
 gguf: examples/gguf/gguf.cpp ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)

build.zig

@@ -139,4 +139,13 @@ pub fn build(b: *std.build.Builder) !void {
     if (server.target.isWindows()) {
         server.linkSystemLibrary("ws2_32");
     }
+
+    const server_assets = [_][]const u8{ "index.html", "index.js", "completion.js", "json-schema-to-grammar.mjs" };
+    for (server_assets) |asset| {
+        // Note: code can be simplified once setCwd is standard.
+        const gen_hpp = b.addSystemCommand(
+            &.{ "xxd", "-n", asset, "-i", b.fmt("examples/server/public/{s}", .{asset}), b.fmt("examples/server/{s}.hpp", .{asset}) },
+        );
+        server.step.dependOn(&gen_hpp.step);
+    }
 }

examples/server/CMakeLists.txt

@@ -1,12 +1,34 @@
+set(SERVER_ASSETS 
+    index.html
+    index.js
+    completion.js
+    json-schema-to-grammar.mjs
+)
+set(SERVER_ASSETS_HPP "")
+foreach(asset ${SERVER_ASSETS})
+  list(APPEND SERVER_ASSETS_HPP "${CMAKE_CURRENT_BINARY_DIR}/${asset}.hpp")
+endforeach()
+foreach(asset ${SERVER_ASSETS})
+  add_custom_command(
+    OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/${asset}.hpp"
+    WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}/public"
+    COMMAND     xxd -i "${asset}" "${CMAKE_CURRENT_BINARY_DIR}/${asset}.hpp"
+    DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/public/${asset}"
+    VERBATIM
+  )
+endforeach()
+add_custom_target(generate_server_assets ALL DEPENDS ${SERVER_ASSETS_HPP})
+
 set(TARGET server)
 option(LLAMA_SERVER_VERBOSE "Build verbose logging option for Server" ON)
 option(LLAMA_SERVER_SSL "Build SSL support for the server" OFF)
-include_directories(${CMAKE_CURRENT_SOURCE_DIR})
+include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_BINARY_DIR})
 add_executable(${TARGET}
     server.cpp
     utils.hpp
     httplib.h
 )
+add_dependencies(${TARGET} generate_server_assets)
 install(TARGETS ${TARGET} RUNTIME)
 target_compile_definitions(${TARGET} PRIVATE
     SERVER_VERBOSE=$<BOOL:${LLAMA_SERVER_VERBOSE}>

examples/server/completion.js.hpp

@@ -8,13 +8,3 @@ PUBLIC=$DIR/public
 echo "download js bundle files"
 curl https://npm.reversehttp.com/@preact/signals-core,@preact/signals,htm/preact,preact,preact/hooks > $PUBLIC/index.js
 echo >> $PUBLIC/index.js # add newline
-
-FILES=$(ls $PUBLIC)
-
-cd $PUBLIC
-for FILE in $FILES; do
-  echo "generate $FILE.hpp"
-
-  # use simple flag for old version of xxd
-  xxd -i $FILE > $DIR/$FILE.hpp
-done