Fix encoding: remove all non-ASCII chars from app.py

app.py

@@ -1,5 +1,5 @@
-"""
-ROCmPort AI ΓÇö Gradio Space entry point
+"""
+ROCmPort AI - Gradio Space entry point
 Calls the deployed FastAPI backend (Render) and streams agent events.
 """
 
@@ -9,20 +9,20 @@ import json
 
 BACKEND_URL = "https://rocmport-ai-q2b1.onrender.com"
 
-AGENT_ICONS = {
-    "analyzer": "🔍",
-    "translator": "🔄",
-    "optimizer": "ΓÜí",
-    "tester": "🧪",
-    "coordinator": "🎯",
+AGENT_LABELS = {
+    "analyzer":    "Analyzer",
+    "translator":  "Translator",
+    "optimizer":   "Optimizer",
+    "tester":      "Tester",
+    "coordinator": "Coordinator",
 }
 
-STATUS_ICONS = {
-    "waiting": "⏳",
-    "running": "🔄",
-    "done": "✅",
-    "failed": "❌",
-    "retrying": "🔁",
+STATUS_LABELS = {
+    "waiting":  "[waiting]",
+    "running":  "[running]",
+    "done":     "[done]",
+    "failed":   "[FAILED]",
+    "retrying": "[retrying]",
 }
 
 EXAMPLE_REDUCTION = """\
@@ -58,7 +58,7 @@ __global__ void vectorAdd(const float *A, const float *B, float *C, int n) {
         C[i] = A[i] + B[i];
         // Warp-size assumption: 32 threads per warp (wrong on AMD wavefront-64)
         if (threadIdx.x % 32 == 0) {
-            printf("Warp leader: %d\\n", threadIdx.x / 32);
+            printf("Warp leader: %d\n", threadIdx.x / 32);
         }
     }
 }"""
@@ -85,7 +85,7 @@ __global__ void matmul(float *A, float *B, float *C, int N) {
 def port_kernel(cuda_code: str, kernel_name: str, simple_mode: bool):
     """Generator: streams agent events and yields (log_markdown, hip_code)."""
     if not cuda_code or len(cuda_code.strip()) < 10:
-        yield "❌ Please provide CUDA kernel code (at least 10 characters).", ""
+        yield "Please provide CUDA kernel code (at least 10 characters).", ""
         return
 
     kernel_name = kernel_name.strip() or "custom"
@@ -98,7 +98,7 @@ def port_kernel(cuda_code: str, kernel_name: str, simple_mode: bool):
         "simple_mode": bool(simple_mode),
     }
 
-    log_lines.append("🚀 **Connecting to ROCmPort AI backend…**")
+    log_lines.append("**Connecting to ROCmPort AI backend...**")
     yield "\n\n".join(log_lines), hip_code
 
     try:
@@ -126,12 +126,11 @@ def port_kernel(cuda_code: str, kernel_name: str, simple_mode: bool):
                     message = event.get("message", "")
                     detail = event.get("detail") or ""
 
-                    icon = AGENT_ICONS.get(agent, "🤖")
-                    s_icon = STATUS_ICONS.get(status, "ΓÇó")
+                    label = AGENT_LABELS.get(agent, agent.capitalize())
+                    s_label = STATUS_LABELS.get(status, status)
 
-                    log_lines.append(f"{icon} **{agent.capitalize()}** {s_icon} ΓÇö {message}")
+                    log_lines.append(f"**{label}** {s_label} -- {message}")
 
-                    # Extract HIP code from coordinator or translator done events
                     if status == "done" and detail:
                         try:
                             detail_json = json.loads(detail)
@@ -150,58 +149,55 @@ def port_kernel(cuda_code: str, kernel_name: str, simple_mode: bool):
 
     except httpx.ConnectError:
         log_lines.append(
-            "❌ **Could not connect to backend.**\n\n"
-            "> The server may be in a cold-start state ΓÇö please wait ~30 s and retry."
+            "**Could not connect to backend.**\n\n"
+            "> The server may be cold-starting -- please wait ~30 s and retry."
         )
         yield "\n\n".join(log_lines), hip_code
         return
     except httpx.TimeoutException:
-        log_lines.append("⏱️ **Request timed out.** The pipeline may still be running — try again shortly.")
+        log_lines.append("**Request timed out.** The pipeline may still be running -- try again shortly.")
         yield "\n\n".join(log_lines), hip_code
         return
     except httpx.HTTPStatusError as exc:
-        log_lines.append(f"❌ **HTTP {exc.response.status_code}**: {exc.response.text[:300]}")
+        log_lines.append(f"**HTTP {exc.response.status_code}**: {exc.response.text[:300]}")
         yield "\n\n".join(log_lines), hip_code
         return
     except Exception as exc:  # noqa: BLE001
-        log_lines.append(f"❌ **Unexpected error**: {exc}")
+        log_lines.append(f"**Unexpected error**: {exc}")
         yield "\n\n".join(log_lines), hip_code
         return
 
     if not hip_code:
-        log_lines.append("\n⚠️ Pipeline finished but no HIP code was extracted. Check agent logs above.")
+        log_lines.append("\nPipeline finished but no HIP code was extracted. Check agent logs above.")
     else:
-        log_lines.append("\n✅ **Migration complete.** HIP code is shown on the right →")
+        log_lines.append("\n**Migration complete.** HIP code is shown on the right.")
 
     yield "\n\n".join(log_lines), hip_code
 
 
-# ΓöÇΓöÇ UI ΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓö��ΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇ
-
-CSS = """
-.panel-header { font-weight: 600; font-size: 1rem; margin-bottom: 4px; }
-footer { display: none !important; }
-"""
+CSS = (
+    ".panel-header { font-weight: 600; font-size: 1rem; margin-bottom: 4px; } "
+    "footer { display: none !important; }"
+)
 
-with gr.Blocks(title="ROCmPort AI — CUDA to ROCm Migration") as demo:
+with gr.Blocks(title="ROCmPort AI -- CUDA to ROCm Migration") as demo:
 
     gr.Markdown(
-        """# ΓÜí ROCmPort AI
-### Agentic CUDA → ROCm/HIP migration with wavefront-64 bug detection
+        """# ROCmPort AI
+### Agentic CUDA to ROCm/HIP migration with wavefront-64 bug detection
 
-> **Backend API**: [rocmport-ai-q2b1.onrender.com](https://rocmport-ai-q2b1.onrender.com) &nbsp;|&nbsp;
+> **Backend API**: [rocmport-ai-q2b1.onrender.com](https://rocmport-ai-q2b1.onrender.com) |
 > **GitHub**: [tazwaryayyyy/ROCmPort-AI](https://github.com/tazwaryayyyy/ROCmPort-AI)
 
-`hipify-clang` translates CUDA API calls mechanically ΓÇö it **cannot** detect that `if (tid < 32)` in a
-warp-level reduction silently skips lanes 32ΓÇô63 on AMD wavefront-64.
+`hipify-clang` translates CUDA API calls mechanically -- it **cannot** detect that `if (tid < 32)` in a
+warp-level reduction silently skips lanes 32-63 on AMD wavefront-64.
 The code compiles, the output is wrong, no errors. **ROCmPort AI catches this before execution.**
 """
     )
 
     with gr.Row():
-        # ΓöÇΓöÇ Left: input ΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇ
         with gr.Column(scale=1):
-            gr.Markdown("### 📥 Input", elem_classes="panel-header")
+            gr.Markdown("### Input", elem_classes="panel-header")
             cuda_input = gr.Code(
                 label="CUDA Kernel Code",
                 language="cpp",
@@ -216,13 +212,13 @@ The code compiles, the output is wrong, no errors. **ROCmPort AI catches this be
                     scale=2,
                 )
                 simple_mode = gr.Checkbox(
-                    label="Explain Like I'm 5",
+                    label="Explain Like I am 5",
                     value=False,
                     scale=1,
                 )
             with gr.Row():
-                port_btn = gr.Button("ΓÜí Port to ROCm", variant="primary", scale=3)
-                clear_btn = gr.Button("🗑 Clear", scale=1)
+                port_btn = gr.Button("Port to ROCm", variant="primary", scale=3)
+                clear_btn = gr.Button("Clear", scale=1)
 
             gr.Examples(
                 examples=[
@@ -234,15 +230,14 @@ The code compiles, the output is wrong, no errors. **ROCmPort AI catches this be
                 label="Demo Kernels (pre-loaded with intentional AMD bugs)",
             )
 
-        # ΓöÇΓöÇ Right: output ΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇ
         with gr.Column(scale=1):
-            gr.Markdown("### 📤 Output", elem_classes="panel-header")
+            gr.Markdown("### Output", elem_classes="panel-header")
             log_output = gr.Markdown(
                 value="*Agent steps will appear here once you click **Port to ROCm**.*",
                 label="Agent Pipeline Log",
             )
             hip_output = gr.Code(
-                label="Translated & Optimized HIP Code",
+                label="Translated and Optimized HIP Code",
                 language="cpp",
                 lines=18,
             )
@@ -254,16 +249,16 @@ The code compiles, the output is wrong, no errors. **ROCmPort AI catches this be
 
 | Agent | Role |
 |-------|------|
-| 🔍 **Analyzer** | Scans CUDA for AMD-specific risks: wavefront size, ballot/shuffle idioms, shared-memory layout |
-| 🔄 **Translator** | Runs `hipify` then applies LLM-guided fixes for bugs `hipify` cannot detect |
-| 🧪 **Tester** | Verifies compilation with `hipcc` and checks output correctness |
-| ΓÜí **Optimizer** | Proposes MI300X-specific optimisations; re-tested against baseline |
-| 🎯 **Coordinator** | Orchestrates the loop; retries up to 3× if the optimised output regresses |
+| **Analyzer** | Scans CUDA for AMD-specific risks: wavefront size, ballot/shuffle idioms, shared-memory layout |
+| **Translator** | Runs `hipify` then applies LLM-guided fixes for bugs `hipify` cannot detect |
+| **Tester** | Verifies compilation with `hipcc` and checks output correctness |
+| **Optimizer** | Proposes MI300X-specific optimisations; re-tested against baseline |
+| **Coordinator** | Orchestrates the loop; retries up to 3x if the optimised output regresses |
 
 ### The key bug: warp-size assumption
 
 ```c
-// NVIDIA (warpSize = 32) ΓÇö silently WRONG on AMD
+// NVIDIA (warpSize = 32) -- silently WRONG on AMD
 if (tid < 32) { vsmem[tid] += vsmem[tid + 32]; ... }
 
 // AMD-correct (wavefront = 64)
@@ -277,16 +272,15 @@ if (tid < 64) {
 
 | Kernel | Result |
 |--------|--------|
-| matrix_multiply 512×512 | 2.91× speedup over baseline HIP |
-| vector_add 32 M elements | ~3 918 GB/s (~74 % of MI300X peak) |
-| reduction 16 M elements | correctness PASS after wavefront-64 fix |
+| matrix_multiply 512x512 | 2.91x speedup over baseline HIP |
+| vector_add 32M elements | ~3918 GB/s (~74% of MI300X peak) |
+| reduction 16M elements | correctness PASS after wavefront-64 fix |
 
-> Source: `docs/benchmark_runs/` ΓÇö real `rocprof` CSV output, May 2026.
+> Source: `docs/benchmark_runs/` -- real `rocprof` CSV output, May 2026.
 > Results vary with kernel complexity; these figures are not guaranteed on every input.
 """
     )
 
-    # ΓöÇΓöÇ Event wiring ΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇΓöÇ
     port_btn.click(
         fn=port_kernel,
         inputs=[cuda_input, kernel_name, simple_mode],