add linker script parser

pyproject.toml

@@ -11,6 +11,7 @@ dependencies = [
     "angr",
     "pyelftools",
     "pypcode",
+    "lark",
     "lief",
     "keystone-engine",
     "intelhex",

src/patcherex2/components/utils/linker_script.lark

@@ -0,0 +1,41 @@
+start: (memoryblock | sectionblock | OHTERBLOCK)*
+
+memoryblock: "MEMORY" "{" memory_def* "}"
+memory_def: memory_name ["(" memory_attr ")"] ":" "ORIGIN" "=" memory_origin "," "LENGTH" "=" memory_length
+memory_attr: CNAME
+memory_name: CNAME
+memory_origin: HEX | INT
+memory_length: HEX | INT | SIZE
+
+sectionblock: "SECTIONS" "{" section_stmt* "}"
+section_stmt: section_def | IGNORE_ASSN
+section_def: section_name section_addr? ":" IGNORE_BRACE section_region? section_lma_region?
+section_addr: HEX | INT
+section_name: SECTION_NAME
+section_region: ">" CNAME
+section_lma_region: "AT>" CNAME
+
+OHTERBLOCK: IGNORE_STMT
+
+IGNORE_PAREN: "(" /[^\(\)]+/s ")"
+IGNORE_BRACE: "{" /[^\{\}]+/s "}"
+IGNORE_ASSN: /[^=;]+/s "=" /[^=;]+/s ";"
+IGNORE_STMT: ("_"|UCASE_LETTER) ("_"|UCASE_LETTER|DIGIT)* (IGNORE_BRACE | IGNORE_PAREN)
+
+SECTION_NAME: "." ("."|CNAME)* "*"?
+SIZE: INT ["K" | "M" | "G"]
+HEX: "0x" HEXDIGIT+
+
+%import common.LETTER
+%import common.UCASE_LETTER
+%import common.CNAME
+%import common.HEXDIGIT
+%import common.DIGIT
+%import common.WS
+%import common.INT
+%import common.CPP_COMMENT
+%import common.C_COMMENT
+
+%ignore CPP_COMMENT
+%ignore C_COMMENT
+%ignore WS

src/patcherex2/components/utils/linker_script_parser.py

@@ -0,0 +1,53 @@
+import os
+
+import lark
+
+
+class LinkerScriptParser:
+    def __init__(self, linker_script_content: str) -> None:
+        self.linker_script_content = linker_script_content
+        self.ast = self._parse()
+
+    @staticmethod
+    def from_file(linker_script_path: str):
+        with open(linker_script_path) as f:
+            return LinkerScriptParser(f.read())
+
+    @staticmethod
+    def from_string(linker_script_content: str):
+        return LinkerScriptParser(linker_script_content)
+
+    def _parse(self):
+        with open(os.path.join(os.path.dirname(__file__), "linker_script.lark")) as f:
+            parser = lark.Lark(f.read())
+            return parser.parse(self.linker_script_content)
+
+    def _get_ast_data(self, tree, name):
+        node = next(tree.find_data(name), None)
+        return node.children[0].value if node else None
+
+    def get_memory_regions(self):
+        memory_regions = []
+        for mem_def in self.ast.find_data("memory_def"):
+            memory_regions.append(
+                {
+                    "name": self._get_ast_data(mem_def, "memory_name"),
+                    "attr": self._get_ast_data(mem_def, "memory_attr"),
+                    "origin": self._get_ast_data(mem_def, "memory_origin"),
+                    "length": self._get_ast_data(mem_def, "memory_length"),
+                }
+            )
+        return memory_regions
+
+    def get_sections(self):
+        sections = []
+        for section_def in self.ast.find_data("section_def"):
+            sections.append(
+                {
+                    "name": self._get_ast_data(section_def, "section_name"),
+                    "addr": self._get_ast_data(section_def, "section_addr"),
+                    "region": self._get_ast_data(section_def, "section_region"),
+                    "lma_region": self._get_ast_data(section_def, "section_lma_region"),
+                }
+            )
+        return sections