feat: Batch requests and extract fn/paths from addr2line

Author: nebkat

esp_idf_panic_decoder/pc_address_decoder.py

@@ -1,7 +1,7 @@
 # SPDX-FileCopyrightText: 2023 Espressif Systems (Shanghai) CO LTD
 # SPDX-License-Identifier: Apache-2.0
-
-from typing import List, Optional, Union
+from itertools import groupby
+from typing import List, Optional, Union, Dict
 import re
 import subprocess
 
@@ -23,50 +23,105 @@ def __init__(
         self.toolchain_prefix = toolchain_prefix
         self.elf_files = elf_file if isinstance(elf_file, list) else [elf_file]
         self.rom_elf_file = rom_elf_file
-        self.pc_address_buffer = b''
         self.pc_address_matcher = [PcAddressMatcher(file) for file in self.elf_files]
-        if rom_elf_file is not None:
-            self.rom_pc_address_matcher = PcAddressMatcher(rom_elf_file)
-
-    def decode_address(self, line: bytes) -> str:
-        """Decoded possible addresses in line"""
-        line = self.pc_address_buffer + line
-        self.pc_address_buffer = b''
-        out = ''
-        for match in re.finditer(ADDRESS_RE, line.decode(errors='ignore')):
-            num = match.group()
-            address_int = int(num, 16)
-            translation = None
-
-            # Try looking for the address in the app ELF files
-            for matcher in self.pc_address_matcher:
-                if matcher.is_executable_address(address_int):
-                    translation = self.lookup_pc_address(num, elf_file=matcher.elf_path)
-                    if translation is not None:
-                        break
-            # Not found in app ELF file, check ROM ELF file (if it is available)
-            if translation is None and self.rom_elf_file is not None and \
-            self.rom_pc_address_matcher.is_executable_address(address_int):
-                translation = self.lookup_pc_address(num, is_rom=True, elf_file=self.rom_elf_file)
-
-            # Translation found either in the app or ROM ELF file
-            if translation is not None:
-                out += translation
-        return out
-
-    def lookup_pc_address(self, pc_addr: str, is_rom: bool = False, elf_file: str = '') -> Optional[str]:
-        """Decode address using addr2line tool"""
-        elf_file: str = elf_file if elf_file else self.rom_elf_file if is_rom else self.elf_files[0]  # type: ignore
-        cmd = [f'{self.toolchain_prefix}addr2line', '-pfiaC', '-e', elf_file, pc_addr]
+        if self.rom_elf_file:
+            self.pc_address_matcher.append(PcAddressMatcher(self.rom_elf_file))
+
+
+    def decode_addresses(self, line: bytes) -> Dict[str, List[dict]]:
+        """Decode possible addresses in a line, batching addr2line calls per ELF."""
+
+        # Find all hex addresses
+        addresses = re.findall(ADDRESS_RE, line.decode(errors='ignore'))
+
+        # Prepare result (to preserve overall order)
+        result: Dict[str, Optional[List[dict]]] = {addr: None for addr in addresses}
+
+        # Check each elf file for matches
+        for matcher in self.pc_address_matcher:
+            elf_path = matcher.elf_path
+            elf_addresses = [addr for addr in addresses if matcher.is_executable_address(int(addr, 16))]
+            if not elf_addresses:
+                continue
+
+            # Lookup addresses using addr2line
+            mapped_addresses = self.lookup_pc_address(elf_addresses, is_rom=(elf_path == self.rom_elf_file), elf_file=elf_path)
+
+            # Update result with the mapped addresses
+            result.update(mapped_addresses)
+
+            # Stop searching for addresses that have been found
+            addresses = [addr for addr in addresses if addr not in mapped_addresses.keys()]
+
+        # Filter out addresses that could not be resolved
+        result = {key: value for key, value in result.items() if value is not None}
+
+        return result
+
+
+    def lookup_pc_address(
+        self,
+        pc_addr: List[str],
+        is_rom: bool = False,
+        elf_file: str = ''
+    ) -> Dict[str, List[dict]]:
+        """
+        Decode a list of addresses using addr2line, returning a map from each address string
+        to a tuple (function_name, path:line).
+        """
+        elf_file = elf_file if elf_file else (self.rom_elf_file if is_rom else self.elf_files[0])  # type: ignore
+        cmd = [f'{self.toolchain_prefix}addr2line', '-fiaC', '-e', elf_file, *pc_addr]
 
         try:
-            translation = subprocess.check_output(cmd, cwd='.')
-            if b'?? ??:0' not in translation:
-                decoded = translation.decode()
-                return decoded if not is_rom else decoded.replace('at ??:?', 'in ROM')
+            batch_output = subprocess.check_output(cmd, cwd='.')
         except OSError as err:
             red_print(f'{" ".join(cmd)}: {err}')
+            return {}
         except subprocess.CalledProcessError as err:
             red_print(f'{" ".join(cmd)}: {err}')
             red_print('ELF file is missing or has changed, the build folder was probably modified.')
-        return None
+            return {}
+
+        decoded_output = batch_output.decode(errors='ignore')
+
+        # Step 1: Split into sections where each section starts with an 8-hex-digit address
+        sections = re.split(r'(?=0x[0-9A-Fa-f]{8}\r?\n)', decoded_output)
+
+        result: Dict[str, List[dict]] = {}
+        for section in sections:
+            section = section.strip()
+            if not section:
+                continue
+
+            # Step 2: Split the section by newlines
+            lines = section.split('\n')
+
+            # Step 3: First line is the address
+            address = lines[0].strip()
+
+            # Step 4: Build trace by consuming lines in pairs (function, path:line)
+            trace: List[dict] = []
+            for i in range(1, len(lines) - 1, 2):
+                fn = lines[i].strip()
+                path_line = lines[i + 1].strip()
+
+                # Remove any " (discriminator N)" suffix
+                path_line = re.sub(r' \(discriminator \d+\)$', '', path_line)
+
+                # Split on the last colon before digits to separate path and line number
+                parts = re.split(r':(?=\d+|\?$)', path_line, maxsplit=1)
+                if len(parts) == 2:
+                    path, line_str = parts
+                    line_num = int(line_str) if line_str != '?' else '?'
+                else:
+                    path = parts[0]
+                    line_num = 0
+
+                if path == '??' and is_rom:
+                    path = 'ROM'
+
+                trace.append({'fn': fn, 'path': path, 'line': line_num})
+
+            result[address] = trace
+
+        return result