updates

arch/Dockerfile

@@ -12,6 +12,9 @@ FROM envoyproxy/envoy:v1.31-latest as envoy
 
 #Build config generator, so that we have a single build image for both Rust and Python
 FROM python:3-slim as arch
+
+RUN apt-get update && apt-get install -y gettext-base && apt-get clean && rm -rf /var/lib/apt/lists/*
+
 COPY --from=builder /arch/target/wasm32-wasi/release/prompt_gateway.wasm /etc/envoy/proxy-wasm-plugins/prompt_gateway.wasm
 COPY --from=builder /arch/target/wasm32-wasi/release/llm_gateway.wasm /etc/envoy/proxy-wasm-plugins/llm_gateway.wasm
 COPY --from=envoy /usr/local/bin/envoy /usr/local/bin/envoy
@@ -22,4 +25,5 @@ COPY arch/tools/cli/config_generator.py .
 COPY arch/envoy.template.yaml .
 COPY arch/arch_config_schema.yaml .
 
-CMD ["sh", "-c", "python config_generator.py && envoy -c /etc/envoy/envoy.yaml --component-log-level wasm:debug"]
+
+ENTRYPOINT ["sh", "-c", "python config_generator.py && envsubst < /etc/envoy/envoy.yaml > /etc/envoy.env_sub.yaml && envoy -c /etc/envoy.env_sub.yaml --component-log-level wasm:debug"]

arch/build_filter_image.sh

@@ -1 +1 @@
-docker build -t archgw .. -f Dockerfile
+docker build  -f Dockerfile .. -t katanemo/archgw

arch/docker-compose.dev.yaml

@@ -1,6 +1,6 @@
 services:
   archgw:
-    image: archgw:latest
+    image: katanemo/archgw:latest
     ports:
       - "10000:10000"
       - "11000:11000"
@@ -10,9 +10,12 @@ services:
       - ${ARCH_CONFIG_FILE:-../demos/function_calling/arch_config.yaml}:/config/arch_config.yaml
       - /etc/ssl/cert.pem:/etc/ssl/cert.pem
       - ./envoy.template.yaml:/config/envoy.template.yaml
-      - ./target/wasm32-wasi/release/intelligent_prompt_gateway.wasm:/etc/envoy/proxy-wasm-plugins/intelligent_prompt_gateway.wasm
       - ./arch_config_schema.yaml:/config/arch_config_schema.yaml
       - ./tools/cli/config_generator.py:/config/config_generator.py
-      - ./arch_logs:/var/log/
-    env_file:
-      - stage.env
+      - ../crates/target/wasm32-wasi/release/llm_gateway.wasm:/etc/envoy/proxy-wasm-plugins/llm_gateway.wasm
+      - ../crates/target/wasm32-wasi/release/prompt_gateway.wasm:/etc/envoy/proxy-wasm-plugins/prompt_gateway.wasm
+      - ~/archgw_logs:/var/log/
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    environment:
+      - OPENAI_API_KEY=${OPENAI_API_KEY:?error}

arch/docker-compose.yaml

@@ -10,7 +10,7 @@ services:
       - ${ARCH_CONFIG_FILE:-./demos/function_calling/arch_confg.yaml}:/config/arch_config.yaml
       - /etc/ssl/cert.pem:/etc/ssl/cert.pem
       - ~/archgw_logs:/var/log/
-    env_file:
-      - stage.env
     extra_hosts:
       - "host.docker.internal:host-gateway"
+    env_file:
+      - stage.env

arch/envoy.template.yaml

@@ -206,6 +206,18 @@ static_resources:
                             body:
                               inline_string: "x-arch-llm-provider header not set, llm gateway cannot perform routing\n"
                 http_filters:
+
+
+                  - name: envoy.filters.http.compressor
+                    typed_config:
+                      "@type": type.googleapis.com/envoy.extensions.filters.http.compressor.v3.Compressor
+                      compressor_library:
+                        name: compress
+                        typed_config:
+                          "@type": type.googleapis.com/envoy.extensions.compression.gzip.compressor.v3.Gzip
+                          memory_level: 3
+                          window_bits: 10
+
                   - name: envoy.filters.http.wasm
                     typed_config:
                       "@type": type.googleapis.com/udpa.type.v1.TypedStruct
@@ -223,6 +235,22 @@ static_resources:
                             code:
                               local:
                                 filename: "/etc/envoy/proxy-wasm-plugins/llm_gateway.wasm"
+
+
+
+                  - name: envoy.filters.http.decompressor
+                    typed_config:
+                      "@type": type.googleapis.com/envoy.extensions.filters.http.decompressor.v3.Decompressor
+                      decompressor_library:
+                        name: decompress
+                        typed_config:
+                          "@type": "type.googleapis.com/envoy.extensions.compression.gzip.decompressor.v3.Gzip"
+                          window_bits: 9
+                          chunk_size: 8192
+                          # If this ratio is set too low, then body data will not be decompressed completely.
+                          max_inflate_ratio: 1000
+
+
                   - name: envoy.filters.http.router
                     typed_config:
                       "@type": type.googleapis.com/envoy.extensions.filters.http.router.v3.Router