fix_bp_stack.py

#!/usr/bin/env python
import os, sys, bisect, re

class Module:
    def __init__(self, name):
        self.name = name
        self.files = []
        self.funcs = []
        self.lines = []

    def parse_line(self, line):
        head, tail = line.split(" ", 1)
        if head == "FILE":
            filenum, name = tail.split(" ", 1)
            filenum = int(filenum, 10)
            if name.startswith("hg:"):
                name = name.split(":")[2]
            while len(self.files) <= filenum:
                self.files.append(None)
            self.files[filenum] = name
        elif head == "FUNC":
            addr, length, params, name = tail.split(" ", 3)
            addr = int(addr, 16)
            length = int(length, 16)
            self.funcs.append((addr, length, name))
        elif head == "PUBLIC" or head == "STACK":
            pass
        else:
            addr = int(head, 16)
            length, line, filenum = tail.split(" ", 2)
            length = int(length, 16)
            line = int(line, 10)
            filenum = int(filenum, 10)
            self.lines.append((addr, length, line, filenum))

    def lookup(self, addr):
        bf = bisect.bisect(self.funcs, (addr,))
        if bf == 0:
            return None
        faddr, flen, fname = self.funcs[bf - 1]
        if addr > faddr + flen:
            return None
        bl = bisect.bisect(self.lines, (addr,))
        if bl == 0:
            return fname
        laddr, llen, lline, lfile = self.lines[bl - 1]
        if addr > laddr + llen:
            return fname
        return ("%s %s:%d" % (fname, self.files[lfile], lline))

def read_symbols(modules, path):
    this_module = None
    with open(path) as file:
        for line in file:
            line = line.rstrip()
            if line.startswith("MODULE "):
                kwd, os, arch, blob, name = line.split(" ", 4)
                if name not in modules:
                    modules[name] = Module(name)
                this_module = modules[name]
            else:
                this_module.parse_line(line)
    
def main(argv):
    modules = {}
    for arg in argv:
        if os.path.isfile(arg):
            read_symbols(modules, arg)
        else:
            for root, dirnames, filenames in os.walk(arg):
                for filename in filenames:
                    if filename.endswith(".sym"): #???
                        read_symbols(modules, os.path.join(root, filename))

    FIXABLE = re.compile('\?\?\?\[(.*/)?(?P<file>[^/]*)\+(?P<offset>0x[0-9a-f]+)\]')
    def fix(m):
        filename = m.group("file").rstrip()
        offset = int(m.group("offset"), 16)
        if filename not in modules:
            return m.group(0)
        xlation = modules[filename].lookup(offset)
        if not xlation:
            return m.group(0)
        return "%s (%s+0x%x)" % (xlation, filename, offset)

    for line in sys.stdin:
        sys.stdout.write(re.sub(FIXABLE, fix, line))

if __name__ == '__main__':
    main(sys.argv[1:])