bin_language.c File Reference

#include <rz_bin.h>

Go to the source code of this file.

Macros
#define	LANGUAGE_WITH_BLOCKS   " with blocks"
#define	language_apply_blocks_mask(x, b)   (b ? (RZ_BIN_LANGUAGE_BLOCKS | (x)) : (x))
#define	language_apply_blocks_string(x, b)   (RZ_BIN_LANGUAGE_HAS_BLOCKS(x) ? (b LANGUAGE_WITH_BLOCKS) : (b))

Functions
static bool	check_rust (RzBinSymbol *sym)
static bool	check_objc (RzBinSymbol *sym)
static bool	check_dlang (RzBinSymbol *sym)
static bool	check_swift (RzBinSymbol *sym)
static bool	check_golang (RzBinSymbol *sym)
static bool	check_cxx (RzBinSymbol *sym)
static bool	check_msvc (RzBinSymbol *sym)
static bool	check_kotlin (RzBinSymbol *sym)
static bool	check_groovy (RzBinSymbol *sym)
static bool	check_dart (RzBinSymbol *sym)
RZ_API RzBinLanguage	rz_bin_language_detect (RzBinFile *binfile)
	Tries to detect which language is used in the binary based on symbols and libraries. More...
RZ_API RzBinLanguage	rz_bin_language_to_id (const char *language)
	returns the language identifier based on the given lang name More...
RZ_API const char *	rz_bin_language_to_string (RzBinLanguage language)
	returns the language name based on the given language identifier More...

Macro Definition Documentation

language_apply_blocks_mask

#define language_apply_blocks_mask	(		x,
			b
	)		(b ? (RZ_BIN_LANGUAGE_BLOCKS | (x)) : (x))

Definition at line 9 of file bin_language.c.

language_apply_blocks_string

#define language_apply_blocks_string	(		x,
			b
	)		(RZ_BIN_LANGUAGE_HAS_BLOCKS(x) ? (b LANGUAGE_WITH_BLOCKS) : (b))

Definition at line 10 of file bin_language.c.

LANGUAGE_WITH_BLOCKS

#define LANGUAGE_WITH_BLOCKS   " with blocks"

Definition at line 8 of file bin_language.c.

Function Documentation

check_cxx()

static bool check_cxx ( RzBinSymbol *  sym )

inlinestatic

Definition at line 37 of file bin_language.c.

                                               {
    if (!strncmp(sym->name, "_Z", 2)) {
        return true;
    }
    return !strncmp(sym->name, "__Z", 3);
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_dart()

static bool check_dart ( RzBinSymbol *  sym )

inlinestatic

Definition at line 56 of file bin_language.c.

                                                {
    return strstr(sym->name, "io_flutter_");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_dlang()

static bool check_dlang ( RzBinSymbol *  sym )

inlinestatic

Definition at line 21 of file bin_language.c.

                                                 {
    if (!strncmp(sym->name, "_D2", 3)) {
        return true;
    }
    return !strncmp(sym->name, "_D4", 3);
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_golang()

static bool check_golang ( RzBinSymbol *  sym )

inlinestatic

Definition at line 32 of file bin_language.c.

                                                  {
    return !strncmp(sym->name, "go.", 3) ||
        strstr(sym->name, "gopclntab");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_groovy()

static bool check_groovy ( RzBinSymbol *  sym )

inlinestatic

Definition at line 52 of file bin_language.c.

                                                  {
    return strstr(sym->name, "_groovy");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_kotlin()

static bool check_kotlin ( RzBinSymbol *  sym )

inlinestatic

Definition at line 48 of file bin_language.c.

                                                  {
    return strstr(sym->name, "kotlin_");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_msvc()

static bool check_msvc ( RzBinSymbol *  sym )

inlinestatic

Definition at line 44 of file bin_language.c.

                                                {
    return *sym->name == '?';
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_objc()

static bool check_objc ( RzBinSymbol *  sym )

inlinestatic

Definition at line 17 of file bin_language.c.

                                                {
    return !strncmp(sym->name, "_OBJC_", 6);
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_rust()

static bool check_rust ( RzBinSymbol *  sym )

inlinestatic

Definition at line 12 of file bin_language.c.

                                                {
    return strstr(sym->name, "_$LT$") ||
        strstr(sym->name, "_rust_oom");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

check_swift()

static bool check_swift ( RzBinSymbol *  sym )

inlinestatic

Definition at line 28 of file bin_language.c.

                                                 {
    return strstr(sym->name, "swift_once");
}

References rz_bin_symbol_t::name.

Referenced by rz_bin_language_detect().

rz_bin_language_detect()

RZ_API RzBinLanguage rz_bin_language_detect

(

RzBinFile *

binfile

)

Tries to detect which language is used in the binary based on symbols and libraries.

Currently this method can detect the language only from bins that are either ELF, PE, Mach-O, Java Class and Dex.

The current supported languages are: c, cxx, dart, dlang, go, groovy, java, kotlin, msvc, objc, rust, swift.

Definition at line 69 of file bin_language.c.

                                                                {
    rz_return_val_if_fail(binfile && binfile->o, RZ_BIN_LANGUAGE_UNKNOWN);
    RzBinObject *o = binfile->o;
    RzBinInfo *info = o->info;
    RzBinSymbol *sym;
    RzBinSection *section;
    RzListIter *iter;
 
    if (!info) {
        return RZ_BIN_LANGUAGE_UNKNOWN;
    }
    RzBinLanguage lang = rz_bin_language_to_id(info->lang);
    if (lang != RZ_BIN_LANGUAGE_UNKNOWN &&
        lang != RZ_BIN_LANGUAGE_C &&
        lang != RZ_BIN_LANGUAGE_OBJC) {
        // avoid detecting a language if was already specified.
        return lang;
    }
 
    bool is_macho = info->rclass ? strstr(info->rclass, "mach") : false;
    bool is_elf = info->rclass ? strstr(info->rclass, "elf") : false;
    bool is_pe = info->rclass ? strstr(info->rclass, "pe") : false;
    bool is_class = info->rclass ? strstr(info->rclass, "class") : false;
    bool is_blocks = false;
    bool is_objc = false;
    bool is_cpp = false;
    char *lib = NULL;
 
    if (!is_macho && !is_elf && !is_pe && !is_class) {
        return RZ_BIN_LANGUAGE_UNKNOWN;
    }
 
    if (is_macho || is_elf) {
        rz_list_foreach (o->imports, iter, sym) {
            const char *name = sym->name;
            if (!strcmp(name, "_NSConcreteGlobalBlock")) {
                is_blocks = true;
            } else if (!strncmp(name, "objc_", 5)) {
                is_objc = true;
            }
        }
    }
    rz_list_foreach (o->libs, iter, lib) {
        if (is_macho && strstr(lib, "swift")) {
            info->lang = "swift";
            return language_apply_blocks_mask(RZ_BIN_LANGUAGE_SWIFT, is_blocks);
        } else if (strstr(lib, "stdc++") || strstr(lib, "c++")) {
            is_cpp = true;
        } else if (strstr(lib, "msvcp")) {
            info->lang = "msvc";
            return RZ_BIN_LANGUAGE_MSVC;
        } else if (strstr(lib, "phobos")) {
            info->lang = "dlang";
            return RZ_BIN_LANGUAGE_DLANG;
        }
    }
    if (is_objc) {
        info->lang = "objc";
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_OBJC, is_blocks);
    }
 
    rz_list_foreach (o->symbols, iter, sym) {
        if (!sym->name) {
            continue;
        }
        if (check_rust(sym)) {
            info->lang = "rust";
            return RZ_BIN_LANGUAGE_RUST;
        } else if (check_golang(sym)) {
            info->lang = "go";
            return RZ_BIN_LANGUAGE_GO;
        } else if (check_swift(sym)) {
            info->lang = "swift";
            return language_apply_blocks_mask(RZ_BIN_LANGUAGE_SWIFT, is_blocks);
        } else if (check_cxx(sym)) {
            is_cpp = true;
        } else if (check_objc(sym)) {
            info->lang = "objc";
            return language_apply_blocks_mask(RZ_BIN_LANGUAGE_OBJC, is_blocks);
        } else if (check_dlang(sym)) {
            info->lang = "dlang";
            return RZ_BIN_LANGUAGE_DLANG;
        } else if (check_kotlin(sym)) {
            info->lang = "kotlin";
            return RZ_BIN_LANGUAGE_KOTLIN;
        } else if (check_groovy(sym)) {
            info->lang = "groovy";
            return RZ_BIN_LANGUAGE_GROOVY;
        } else if (check_msvc(sym)) {
            info->lang = "c";
            return RZ_BIN_LANGUAGE_MSVC;
        } else if (check_dart(sym)) {
            info->lang = "dart";
            return RZ_BIN_LANGUAGE_DART;
        }
    }
 
    if (is_macho || is_elf) {
        rz_list_foreach (o->sections, iter, section) {
            if (!section->name) {
                continue;
            }
            if (strstr(section->name, "note.go.buildid") ||
                strstr(section->name, "gopclntab") ||
                strstr(section->name, "go_export")) {
                info->lang = "go";
                return RZ_BIN_LANGUAGE_GO;
            }
        }
    }
    if (is_cpp) {
        info->lang = "c++";
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_CXX, is_blocks);
    } else if (!info->lang) {
        info->lang = "c";
    } else if (strstr(info->lang, "java")) {
        return RZ_BIN_LANGUAGE_JAVA;
    }
    return language_apply_blocks_mask(RZ_BIN_LANGUAGE_C, is_blocks);
}

References check_cxx(), check_dart(), check_dlang(), check_golang(), check_groovy(), check_kotlin(), check_msvc(), check_objc(), check_rust(), check_swift(), rz_bin_object_t::imports, info(), rz_bin_object_t::info, rz_bin_info_t::lang, language_apply_blocks_mask, rz_bin_object_t::libs, rz_bin_symbol_t::name, NULL, rz_bin_file_t::o, rz_bin_info_t::rclass, RZ_BIN_LANGUAGE_C, RZ_BIN_LANGUAGE_CXX, RZ_BIN_LANGUAGE_DART, RZ_BIN_LANGUAGE_DLANG, RZ_BIN_LANGUAGE_GO, RZ_BIN_LANGUAGE_GROOVY, RZ_BIN_LANGUAGE_JAVA, RZ_BIN_LANGUAGE_KOTLIN, RZ_BIN_LANGUAGE_MSVC, RZ_BIN_LANGUAGE_OBJC, RZ_BIN_LANGUAGE_RUST, RZ_BIN_LANGUAGE_SWIFT, rz_bin_language_to_id(), RZ_BIN_LANGUAGE_UNKNOWN, rz_return_val_if_fail, rz_bin_object_t::sections, and rz_bin_object_t::symbols.

Referenced by rz_bin_object_set_items().

rz_bin_language_to_id()

RZ_API RzBinLanguage rz_bin_language_to_id

(

const char *

language

)

returns the language identifier based on the given lang name

Definition at line 193 of file bin_language.c.

                                                                 {
    if (RZ_STR_ISEMPTY(language)) {
        return RZ_BIN_LANGUAGE_UNKNOWN;
    }
    bool has_blocks = strstr(language, LANGUAGE_WITH_BLOCKS);
    if (strstr(language, "swift")) {
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_SWIFT, has_blocks);
    } else if (strstr(language, "java")) {
        return RZ_BIN_LANGUAGE_JAVA;
    } else if (strstr(language, "groovy")) {
        return RZ_BIN_LANGUAGE_GROOVY;
    } else if (strstr(language, "kotlin")) {
        return RZ_BIN_LANGUAGE_KOTLIN;
    } else if (strstr(language, "objc")) {
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_OBJC, has_blocks);
    } else if (strstr(language, "cxx") || strstr(language, "c++")) {
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_CXX, has_blocks);
    } else if (strstr(language, "dlang")) {
        return RZ_BIN_LANGUAGE_DLANG;
    } else if (strstr(language, "msvc")) {
        return RZ_BIN_LANGUAGE_MSVC;
    } else if (strstr(language, "rust")) {
        return RZ_BIN_LANGUAGE_RUST;
    } else if (strstr(language, "dart")) {
        return RZ_BIN_LANGUAGE_DART;
    } else if (!strcmp(language, "c") || !strcmp(language, "c" LANGUAGE_WITH_BLOCKS)) {
        return language_apply_blocks_mask(RZ_BIN_LANGUAGE_C, has_blocks);
    } else if (!strcmp(language, "go")) {
        return RZ_BIN_LANGUAGE_GO;
    }
    return RZ_BIN_LANGUAGE_UNKNOWN;
}

References language_apply_blocks_mask, LANGUAGE_WITH_BLOCKS, RZ_BIN_LANGUAGE_C, RZ_BIN_LANGUAGE_CXX, RZ_BIN_LANGUAGE_DART, RZ_BIN_LANGUAGE_DLANG, RZ_BIN_LANGUAGE_GO, RZ_BIN_LANGUAGE_GROOVY, RZ_BIN_LANGUAGE_JAVA, RZ_BIN_LANGUAGE_KOTLIN, RZ_BIN_LANGUAGE_MSVC, RZ_BIN_LANGUAGE_OBJC, RZ_BIN_LANGUAGE_RUST, RZ_BIN_LANGUAGE_SWIFT, RZ_BIN_LANGUAGE_UNKNOWN, and RZ_STR_ISEMPTY.

Referenced by rz_bin_demangle(), and rz_bin_language_detect().

rz_bin_language_to_string()

RZ_API const char* rz_bin_language_to_string

(

RzBinLanguage

language

)

returns the language name based on the given language identifier

Definition at line 229 of file bin_language.c.

                                                                     {
    switch (RZ_BIN_LANGUAGE_MASK(language)) {
    case RZ_BIN_LANGUAGE_SWIFT:
        return language_apply_blocks_string(language, "swift");
    case RZ_BIN_LANGUAGE_GO:
        return "go";
    case RZ_BIN_LANGUAGE_JAVA:
        return "java";
    case RZ_BIN_LANGUAGE_KOTLIN:
        return "kotlin";
    case RZ_BIN_LANGUAGE_C:
        return language_apply_blocks_string(language, "c");
    case RZ_BIN_LANGUAGE_CXX:
        return language_apply_blocks_string(language, "c++");
    case RZ_BIN_LANGUAGE_DLANG:
        return "dlang";
    case RZ_BIN_LANGUAGE_OBJC:
        return language_apply_blocks_string(language, "objc");
    case RZ_BIN_LANGUAGE_MSVC:
        return "msvc";
    case RZ_BIN_LANGUAGE_RUST:
        return "rust";
    case RZ_BIN_LANGUAGE_GROOVY:
        return "groovy";
    case RZ_BIN_LANGUAGE_DART:
        return "dart";
    default:
        return NULL;
    }
}

References language_apply_blocks_string, NULL, RZ_BIN_LANGUAGE_C, RZ_BIN_LANGUAGE_CXX, RZ_BIN_LANGUAGE_DART, RZ_BIN_LANGUAGE_DLANG, RZ_BIN_LANGUAGE_GO, RZ_BIN_LANGUAGE_GROOVY, RZ_BIN_LANGUAGE_JAVA, RZ_BIN_LANGUAGE_KOTLIN, RZ_BIN_LANGUAGE_MASK, RZ_BIN_LANGUAGE_MSVC, RZ_BIN_LANGUAGE_OBJC, RZ_BIN_LANGUAGE_RUST, and RZ_BIN_LANGUAGE_SWIFT.

Referenced by rz_bin_demangle(), and rz_bin_file_set_obj().