@@ -17,6 +17,105 @@ def check_requirement(req, exts):
1717 return False
1818
1919
20+ def build_match_from_format (format_field ):
21+ """
22+ Build a match string from the format field in the new schema.
23+ """
24+ if not format_field or "opcodes" not in format_field :
25+ return None
26+
27+ # Determine instruction width by finding maximum bit position
28+ valid_locations = []
29+
30+ opcodes = format_field ["opcodes" ]
31+ # Check opcodes
32+ for field_data in opcodes .values ():
33+ if isinstance (field_data , dict ) and "location" in field_data :
34+ if isinstance (field_data ["location" ], str ):
35+ try :
36+ location = field_data ["location" ]
37+ split_location = location .split ("|" )
38+ high = max (
39+ (
40+ int (location .split ("-" )[0 ])
41+ if "-" in location
42+ else int (location )
43+ )
44+ for location in split_location
45+ )
46+ valid_locations .append (high )
47+ except (ValueError , IndexError ):
48+ raise ValueError (
49+ f"Invalid location format: { field_data ['location' ]}
50+ )
51+ elif isinstance (field_data ["location" ], int ):
52+ try :
53+ valid_locations .append (field_data ["location" ])
54+ except (ValueError , IndexError ):
55+ raise ValueError (
56+ f"Invalid location format: { field_data ['location' ]}
57+ )
58+ else :
59+ raise ValueError (f"Unknown location format: { field_data ['location' ]} )
60+
61+ if "variables" in format_field :
62+ variables = format_field ["variables" ]
63+ # Check variables
64+ for var_data in variables .values ():
65+ if isinstance (var_data , dict ) and "location" in var_data :
66+ if isinstance (var_data ["location" ], str ):
67+ try :
68+ location = var_data ["location" ]
69+ if "-" in location :
70+ high = int (location .split ("-" )[0 ])
71+ else :
72+ high = int (location )
73+ valid_locations .append (high )
74+ except (ValueError , IndexError ):
75+ raise ValueError (
76+ f"Invalid location format: { var_data ['location' ]}
77+ )
78+ elif isinstance (var_data ["location" ], int ):
79+ try :
80+ valid_locations .append (var_data ["location" ])
81+ except (ValueError , IndexError ):
82+ raise ValueError (
83+ f"Invalid location format: { var_data ['location' ]}
84+ )
85+ else :
86+ raise ValueError (f"Invalid location format: { var_data ['location' ]} )
87+
88+ if not valid_locations :
89+ raise ValueError ("No valid bit locations found in format field" )
90+
91+ max_bit = max (valid_locations )
92+
93+ # Set instruction width based on maximum bit position
94+ width = max_bit + 1
95+ match_bits = ["-" ] * width
96+
97+ # Populate match string with opcode bits
98+ for field_data in opcodes .values ():
99+ if isinstance (field_data , dict ):
100+ try :
101+ location = field_data ["location" ]
102+ if isinstance (location , str ) and "-" in location :
103+ high , low = map (int , location .split ("-" ))
104+ else :
105+ high = low = int (location )
106+
107+ if high < low or high >= width :
108+ logging .warning (f"Invalid bit range: { location } )
109+ continue # Skip invalid bit ranges
110+
111+ binary_value = format (field_data ["value" ], f"0{ high - low + 1 } )
112+ match_bits [width - high - 1 : width - low ] = binary_value
113+ except (ValueError , IndexError ):
114+ raise ValueError (f"Error processing opcode field: { field_data } )
115+
116+ return "" .join (match_bits )
117+
118+
20119def parse_extension_requirements (extensions_spec ):
21120 """
22121 Parse the extension requirements from the definedBy field.
@@ -177,11 +276,27 @@ def load_instructions(
177276
178277 encoding = data .get ("encoding" , {})
179278 if not encoding :
180- logging .error (
181- f"Missing 'encoding' field in instruction { name } { path }
182- )
183- encoding_filtered += 1
184- continue
279+ # Check if this instruction uses the new schema with a 'format' field
280+ format_field = data .get ("format" )
281+ if not format_field :
282+ logging .error (
283+ f"Missing 'encoding' field in instruction { name } { path }
284+ )
285+ encoding_filtered += 1
286+ continue
287+
288+ # Try to build a match string from the format field
289+ match_string = build_match_from_format (format_field )
290+ if not match_string :
291+ logging .error (
292+ f"Could not build encoding from format field in instruction { name } { path }
293+ )
294+ encoding_filtered += 1
295+ continue
296+
297+ # Create a synthetic encoding compatible with existing logic
298+ encoding = {"match" : match_string , "variables" : []}
299+ logging .debug (f"Built encoding from format field for { name } )
185300
186301 # Check if the instruction specifies a base architecture constraint
187302 base = data .get ("base" )
0 commit comments