fix: resolve ERB conditionals in C header exception causes (#917)

Use Ruby ERB processing instead of Python regex to handle templates like
'<%- if ext?(:H) -%>V<%- end -%>U-mode' in exception cause names.

- Ruby task processes ERB templates using cfg_arch.render_erb()
- Python script reads pre-resolved JSON data
- Generates valid C code: CAUSE_ENVIRONMENT_CALL_FROM_VU_MODE

Fixes #894

---------

Co-authored-by: 7908837174 <[email protected]>

Author: 7908837174

backends/generators/c_header/generate_encoding.py

@@ -8,6 +8,7 @@
 import logging
 import argparse
 import yaml
+import json
 
 # Add parent directory to path to import generator.py
 parent_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
@@ -29,15 +30,55 @@ def calculate_mask(match_str):
     return int("".join("0" if c == "-" else "1" for c in match_str), 2)
 
 
-def load_exception_codes(ext_dir, enabled_extensions=None, include_all=False):
-    """Load exception codes from extension YAML files."""
+def load_exception_codes(
+    ext_dir, enabled_extensions=None, include_all=False, resolved_codes_file=None
+):
+    """Load exception codes from extension YAML files or pre-resolved JSON file."""
     exception_codes = []
     found_extensions = 0
     found_files = 0
 
     if enabled_extensions is None:
         enabled_extensions = []
 
+    # If we have a resolved codes file, use it instead of processing YAML files
+    if resolved_codes_file and os.path.exists(resolved_codes_file):
+        try:
+            with open(resolved_codes_file, encoding="utf-8") as f:
+                resolved_codes = json.load(f)
+
+            for code in resolved_codes:
+                num = code.get("num")
+                name = code.get("name")
+                if num is not None and name is not None:
+                    sanitized_name = (
+                        name.lower()
+                        .replace(" ", "_")
+                        .replace("/", "_")
+                        .replace("-", "_")
+                    )
+                    exception_codes.append((num, sanitized_name))
+
+            logging.info(
+                f"Loaded {len(exception_codes)} pre-resolved exception codes from {resolved_codes_file}"
+            )
+
+            # Sort by exception code number and deduplicate
+            seen_nums = set()
+            unique_codes = []
+            for num, name in sorted(exception_codes, key=lambda x: x[0]):
+                if num not in seen_nums:
+                    seen_nums.add(num)
+                    unique_codes.append((num, name))
+
+            return unique_codes
+
+        except Exception as e:
+            logging.error(
+                f"Error loading resolved codes file {resolved_codes_file}: {e}"
+            )
+            # Fall back to processing YAML files
+
     for dirpath, _, filenames in os.walk(ext_dir):
         for fname in filenames:
             if not fname.endswith(".yaml"):
@@ -298,6 +339,10 @@ def main():
         action="store_true",
         help="Fallback to hardcoded exception causes if none are loaded from files",
     )
+    parser.add_argument(
+        "--resolved-codes",
+        help="JSON file containing pre-resolved exception codes",
+    )
 
     args = parser.parse_args()
 
@@ -321,7 +366,10 @@ def main():
     # Load exception codes
     logging.info(f"Loading exception codes from {args.ext_dir}")
     causes = load_exception_codes(
-        args.ext_dir, args.extensions, include_all=args.include_all
+        args.ext_dir,
+        args.extensions,
+        include_all=args.include_all,
+        resolved_codes_file=args.resolved_codes,
     )
 
     # Process instructions and calculate masks

backends/generators/tasks.rake

@@ -1,6 +1,8 @@
 # frozen_string_literal: true
 
 require "udb/resolver"
+require 'json'
+require 'tempfile'
 
 directory "#{$root}/gen/go"
 directory "#{$root}/gen/c_header"
@@ -53,8 +55,36 @@ namespace :gen do
     csr_dir = cfg_arch.path / "csr"
     ext_dir = cfg_arch.path / "ext"
 
-    # Run the C header generator script using the same Python environment
-    # The script generates encoding.h for inclusion in C programs
-    sh "#{$root}/.home/.venv/bin/python3 #{$root}/backends/generators/c_header/generate_encoding.py --inst-dir=#{inst_dir} --csr-dir=#{csr_dir} --ext-dir=#{ext_dir} --output=#{output_dir}encoding.out.h --include-all"
+    # Process ERB templates in exception codes using Ruby ERB processing
+    resolved_exception_codes = []
+
+    # Collect all exception codes from extensions and resolve ERB templates
+    cfg_arch.extensions.each do |ext|
+      ext.exception_codes.each do |ecode|
+        # Use Ruby's ERB processing to resolve templates in exception names
+        resolved_name = cfg_arch.render_erb(ecode.name, "exception code name: #{ecode.name}")
+
+        resolved_exception_codes << {
+          'num' => ecode.num,
+          'name' => resolved_name,
+          'var' => ecode.var,
+          'ext' => ext.name
+        }
+      end
+    end
+
+    # Write resolved exception codes to a temporary JSON file
+    resolved_codes_file = Tempfile.new(['resolved_exception_codes', '.json'])
+    resolved_codes_file.write(JSON.pretty_generate(resolved_exception_codes))
+    resolved_codes_file.flush
+
+    begin
+      # Run the C header generator script using the same Python environment
+      # The script generates encoding.h for inclusion in C programs
+      sh "#{$root}/.home/.venv/bin/python3 #{$root}/backends/generators/c_header/generate_encoding.py --inst-dir=#{inst_dir} --csr-dir=#{csr_dir} --ext-dir=#{ext_dir} --resolved-codes=#{resolved_codes_file.path} --output=#{output_dir}encoding.out.h --include-all"
+    ensure
+      resolved_codes_file.close
+      resolved_codes_file.unlink
+    end
   end
 end