Customized generate_sre_constants.py + ran to update constants.rs

RustPython · youknowone · Jan 5, 2026 · Jan 3, 2026 · Jan 3, 2026 · Jan 3, 2026
commit 468d314c5277c8499fad48c32bd809a8a364dc85
diff --git a/crates/sre_engine/src/constants.rs b/crates/sre_engine/src/constants.rs
@@ -3,17 +3,17 @@
  *
  * regular expression matching engine
  *
- * NOTE: This file is generated by sre_constants.py.  If you need
- * to change anything in here, edit sre_constants.py and run it.
+ * Auto-generated by scripts/generate_sre_constants.py from
+ * Lib/re/_constants.py.
  *
  * Copyright (c) 1997-2001 by Secret Labs AB.  All rights reserved.
  *
- * See the _sre.c file for information on usage and redistribution.
+ * See the sre.c file for information on usage and redistribution.
  */
 
 use bitflags::bitflags;
 
-pub const SRE_MAGIC: usize = 20221023;
+pub const SRE_MAGIC: usize = 20230612;
 #[derive(num_enum::TryFromPrimitive, Debug, PartialEq, Eq)]
 #[repr(u32)]
 #[allow(non_camel_case_types, clippy::upper_case_acronyms)]
@@ -62,6 +62,7 @@ pub enum SreOpcode {
     NOT_LITERAL_UNI_IGNORE = 41,
     RANGE_UNI_IGNORE = 42,
 }
+
 #[derive(num_enum::TryFromPrimitive, Debug, PartialEq, Eq)]
 #[repr(u32)]
 #[allow(non_camel_case_types, clippy::upper_case_acronyms)]
@@ -79,6 +80,7 @@ pub enum SreAtCode {
     UNI_BOUNDARY = 10,
     UNI_NON_BOUNDARY = 11,
 }
+
 #[derive(num_enum::TryFromPrimitive, Debug)]
 #[repr(u32)]
 #[allow(non_camel_case_types, clippy::upper_case_acronyms)]
@@ -102,10 +104,10 @@ pub enum SreCatCode {
     UNI_LINEBREAK = 16,
     UNI_NOT_LINEBREAK = 17,
 }
+
 bitflags! {
 #[derive(Debug, PartialEq, Eq, Clone, Copy)]
     pub struct SreFlag: u16 {
-        const TEMPLATE = 1;
         const IGNORECASE = 2;
         const LOCALE = 4;
         const MULTILINE = 8;
@@ -116,10 +118,12 @@ bitflags! {
         const ASCII = 256;
     }
 }
+
 bitflags! {
     pub struct SreInfo: u32 {
         const PREFIX = 1;
         const LITERAL = 2;
         const CHARSET = 4;
     }
 }
+
diff --git a/scripts/generate_sre_constants.py b/scripts/generate_sre_constants.py
@@ -1,5 +1,7 @@
 #! /usr/bin/env python3
-# This script generates Modules/_sre/sre_constants.h from Lib/re/_constants.py.
+# This script generates crates/sre_engine/src/constants.rs from Lib/re/_constants.py.
+
+SCRIPT_NAME = 'scripts/generate_sre_constants.py'
 
 
 def update_file(file, content):
@@ -13,13 +15,13 @@ def update_file(file, content):
         fobj.write(content)
     return True
 
-sre_constants_header = """\
+sre_constants_header = f"""\
 /*
  * Secret Labs' Regular Expression Engine
  *
  * regular expression matching engine
  *
- * Auto-generated by Tools/scripts/generate_sre_constants.py from
+ * Auto-generated by {SCRIPT_NAME} from
  * Lib/re/_constants.py.
  *
  * Copyright (c) 1997-2001 by Secret Labs AB.  All rights reserved.
@@ -29,50 +31,128 @@ def update_file(file, content):
 
 """
 
+def dump_enum(d, enum_name, derives, strip_prefix=""):
+    """Generate Rust enum definitions from a Python dictionary.
+
+    Args:
+        d (dict): The dictionary containing the enum variants.
+        enum_name (str): The name of the enum to generate.
+        derives (str): The derive attributes to include.
+        strip_prefix (str, optional): A prefix to strip from the variant names. Defaults to "".
+
+    Returns:
+        list: A list of strings representing the enum definition.
+    """
+    items = sorted(d)
+    content = [f"{derives}\n"]
+    content.append("#[repr(u32)]\n")
+    content.append("#[allow(non_camel_case_types, clippy::upper_case_acronyms)]\n")
+    content.append(f"pub enum {enum_name} {{\n")
+    for i, item in enumerate(items):
+        name = str(item).removeprefix(strip_prefix)
+        content.append(f"    {name} = {i},\n")
+    content.append("}\n\n")
+    return content
+
+
+def dump_bitflags(d, prefix, derives, struct_name, int_t):
+    """Generate Rust bitflags definitions from a Python dictionary.
+
+    Args:
+        d (dict): The dictionary containing the bitflag variants.
+        prefix (str): The prefix to strip from the variant names.
+        derives (str): The derive attributes to include.
+        struct_name (str): The name of the struct to generate.
+        int_t (str): The integer type to use for the bitflags.
+
+    Returns:
+        list: A list of strings representing the bitflags definition.
+    """
+    items = [(value, name) for name, value in d.items()
+                 if name.startswith(prefix)]
+    content = ["bitflags! {\n"]
+    content.append(f"{derives}\n") if derives else None
+    content.append(f"    pub struct {struct_name}: {int_t} {{\n")
+    for value, name in sorted(items):
+        name = str(name).removeprefix(prefix)
+        content.append(f"        const {name} = {value};\n")
+    content.append("    }\n")
+    content.append("}\n\n")
+    return content
+
 def main(
     infile="Lib/re/_constants.py",
-    outfile_constants="Modules/_sre/sre_constants.h",
-    outfile_targets="Modules/_sre/sre_targets.h",
+    outfile_constants="crates/sre_engine/src/constants.rs",
 ):
     ns = {}
     with open(infile) as fp:
         code = fp.read()
     exec(code, ns)
 
-    def dump(d, prefix):
-        items = sorted(d)
-        for item in items:
-            yield "#define %s_%s %d\n" % (prefix, item, item)
-
-    def dump2(d, prefix):
-        items = [(value, name) for name, value in d.items()
-                 if name.startswith(prefix)]
-        for value, name in sorted(items):
-            yield "#define %s %d\n" % (name, value)
-
-    def dump_gotos(d, prefix):
-        for i, item in enumerate(sorted(d)):
-            assert i == item
-            yield f"    &&{prefix}_{item},\n"
-
     content = [sre_constants_header]
-    content.append("#define SRE_MAGIC %d\n" % ns["MAGIC"])
-    content.extend(dump(ns["OPCODES"], "SRE_OP"))
-    content.extend(dump(ns["ATCODES"], "SRE"))
-    content.extend(dump(ns["CHCODES"], "SRE"))
-    content.extend(dump2(ns, "SRE_FLAG_"))
-    content.extend(dump2(ns, "SRE_INFO_"))
-
-    update_file(outfile_constants, ''.join(content))
+    content.append("use bitflags::bitflags;\n\n")
+    content.append(f"pub const SRE_MAGIC: usize = {ns['MAGIC']};\n")
+    content.extend(dump_enum(ns["OPCODES"], "SreOpcode", "#[derive(num_enum::TryFromPrimitive, Debug, PartialEq, Eq)]"))
+    content.extend(dump_enum(ns["ATCODES"], "SreAtCode", "#[derive(num_enum::TryFromPrimitive, Debug, PartialEq, Eq)]", "AT_"))
+    content.extend(dump_enum(ns["CHCODES"], "SreCatCode", "#[derive(num_enum::TryFromPrimitive, Debug)]", "CATEGORY_"))
 
-    content = [sre_constants_header]
-    content.append(f"static void *sre_targets[{len(ns['OPCODES'])}] = {{\n")
-    content.extend(dump_gotos(ns["OPCODES"], "TARGET_SRE_OP"))
-    content.append("};\n")
+    content.extend(dump_bitflags(ns, "SRE_FLAG_", "#[derive(Debug, PartialEq, Eq, Clone, Copy)]", "SreFlag", "u16"))
+    content.extend(dump_bitflags(ns, "SRE_INFO_",  "", "SreInfo", "u32"))
 
-    update_file(outfile_targets, ''.join(content))
+    update_file(outfile_constants, ''.join(content))
 
 
 if __name__ == '__main__':
     import sys
     main(*sys.argv[1:])
+
+
+
+
+
+#         dump(f, OPCODES, "SreOpcode", "u32", "")
+#         dump(f, ATCODES, "SreAtCode", "u32", "AT_")
+#         dump(f, CHCODES, "SreCatCode", "u32", "CATEGORY_")
+
+#         def bitflags(typ, int_t, prefix, flags):
+#             f.write(f"""\
+# bitflags! {{
+#     pub struct {typ}: {int_t} {{
+# """)
+#             for name in flags:
+#                 val = globals()[prefix + name]
+#                 f.write(f"        const {name} = {val};\n")
+#             f.write("""\
+#     }
+# }
+# """)
+
+#         bitflags("SreFlag", "u16", "SRE_FLAG_", [
+#             "TEMPLATE",
+#             "IGNORECASE",
+#             "LOCALE",
+#             "MULTILINE",
+#             "DOTALL",
+#             "UNICODE",
+#             "VERBOSE",
+#             "DEBUG",
+#             "ASCII",
+#         ])
+
+#         bitflags("SreInfo", "u32", "SRE_INFO_", [
+#             "PREFIX", "LITERAL", "CHARSET",
+#         ])
+
+#     print("done")
+
+
+
+
+# if __name__ == "__main__":
+#     import sys
+#     if len(sys.argv) > 1:
+#         constants_file = sys.argv[1]
+#     else:
+#         import os
+#         constants_file = os.path.join(os.path.dirname(__file__), "../../sre-engine/src/constants.rs")
+#     with open(constants_file, "w") as f: