| Stephane Eranian | e9c4bcd | 2015-11-30 10:02:20 +0100 | [diff] [blame] | 1 | #include <sys/types.h> |
| 2 | #include <stdio.h> |
| 3 | #include <string.h> |
| 4 | #include "util.h" |
| 5 | #include "debug.h" |
| 6 | #include "symbol.h" |
| 7 | |
| 8 | #include "demangle-java.h" |
| 9 | |
| 10 | enum { |
| 11 | MODE_PREFIX = 0, |
| 12 | MODE_CLASS = 1, |
| 13 | MODE_FUNC = 2, |
| 14 | MODE_TYPE = 3, |
| 15 | MODE_CTYPE = 3, /* class arg */ |
| 16 | }; |
| 17 | |
| 18 | #define BASE_ENT(c, n) [c - 'A']=n |
| 19 | static const char *base_types['Z' - 'A' + 1] = { |
| 20 | BASE_ENT('B', "byte" ), |
| 21 | BASE_ENT('C', "char" ), |
| 22 | BASE_ENT('D', "double" ), |
| 23 | BASE_ENT('F', "float" ), |
| 24 | BASE_ENT('I', "int" ), |
| 25 | BASE_ENT('J', "long" ), |
| 26 | BASE_ENT('S', "short" ), |
| 27 | BASE_ENT('Z', "bool" ), |
| 28 | }; |
| 29 | |
| 30 | /* |
| 31 | * demangle Java symbol between str and end positions and stores |
| 32 | * up to maxlen characters into buf. The parser starts in mode. |
| 33 | * |
| 34 | * Use MODE_PREFIX to process entire prototype till end position |
| 35 | * Use MODE_TYPE to process return type if str starts on return type char |
| 36 | * |
| 37 | * Return: |
| 38 | * success: buf |
| 39 | * error : NULL |
| 40 | */ |
| 41 | static char * |
| 42 | __demangle_java_sym(const char *str, const char *end, char *buf, int maxlen, int mode) |
| 43 | { |
| 44 | int rlen = 0; |
| 45 | int array = 0; |
| 46 | int narg = 0; |
| 47 | const char *q; |
| 48 | |
| 49 | if (!end) |
| 50 | end = str + strlen(str); |
| 51 | |
| 52 | for (q = str; q != end; q++) { |
| 53 | |
| 54 | if (rlen == (maxlen - 1)) |
| 55 | break; |
| 56 | |
| 57 | switch (*q) { |
| 58 | case 'L': |
| 59 | if (mode == MODE_PREFIX || mode == MODE_CTYPE) { |
| 60 | if (mode == MODE_CTYPE) { |
| 61 | if (narg) |
| 62 | rlen += scnprintf(buf + rlen, maxlen - rlen, ", "); |
| 63 | narg++; |
| 64 | } |
| 65 | rlen += scnprintf(buf + rlen, maxlen - rlen, "class "); |
| 66 | if (mode == MODE_PREFIX) |
| 67 | mode = MODE_CLASS; |
| 68 | } else |
| 69 | buf[rlen++] = *q; |
| 70 | break; |
| 71 | case 'B': |
| 72 | case 'C': |
| 73 | case 'D': |
| 74 | case 'F': |
| 75 | case 'I': |
| 76 | case 'J': |
| 77 | case 'S': |
| 78 | case 'Z': |
| 79 | if (mode == MODE_TYPE) { |
| 80 | if (narg) |
| 81 | rlen += scnprintf(buf + rlen, maxlen - rlen, ", "); |
| 82 | rlen += scnprintf(buf + rlen, maxlen - rlen, "%s", base_types[*q - 'A']); |
| 83 | while (array--) |
| 84 | rlen += scnprintf(buf + rlen, maxlen - rlen, "[]"); |
| 85 | array = 0; |
| 86 | narg++; |
| 87 | } else |
| 88 | buf[rlen++] = *q; |
| 89 | break; |
| 90 | case 'V': |
| 91 | if (mode == MODE_TYPE) { |
| 92 | rlen += scnprintf(buf + rlen, maxlen - rlen, "void"); |
| 93 | while (array--) |
| 94 | rlen += scnprintf(buf + rlen, maxlen - rlen, "[]"); |
| 95 | array = 0; |
| 96 | } else |
| 97 | buf[rlen++] = *q; |
| 98 | break; |
| 99 | case '[': |
| 100 | if (mode != MODE_TYPE) |
| 101 | goto error; |
| 102 | array++; |
| 103 | break; |
| 104 | case '(': |
| 105 | if (mode != MODE_FUNC) |
| 106 | goto error; |
| 107 | buf[rlen++] = *q; |
| 108 | mode = MODE_TYPE; |
| 109 | break; |
| 110 | case ')': |
| 111 | if (mode != MODE_TYPE) |
| 112 | goto error; |
| 113 | buf[rlen++] = *q; |
| 114 | narg = 0; |
| 115 | break; |
| 116 | case ';': |
| 117 | if (mode != MODE_CLASS && mode != MODE_CTYPE) |
| 118 | goto error; |
| 119 | /* safe because at least one other char to process */ |
| 120 | if (isalpha(*(q + 1))) |
| 121 | rlen += scnprintf(buf + rlen, maxlen - rlen, "."); |
| 122 | if (mode == MODE_CLASS) |
| 123 | mode = MODE_FUNC; |
| 124 | else if (mode == MODE_CTYPE) |
| 125 | mode = MODE_TYPE; |
| 126 | break; |
| 127 | case '/': |
| 128 | if (mode != MODE_CLASS && mode != MODE_CTYPE) |
| 129 | goto error; |
| 130 | rlen += scnprintf(buf + rlen, maxlen - rlen, "."); |
| 131 | break; |
| 132 | default : |
| 133 | buf[rlen++] = *q; |
| 134 | } |
| 135 | } |
| 136 | buf[rlen] = '\0'; |
| 137 | return buf; |
| 138 | error: |
| 139 | return NULL; |
| 140 | } |
| 141 | |
| 142 | /* |
| 143 | * Demangle Java function signature (openJDK, not GCJ) |
| 144 | * input: |
| 145 | * str: string to parse. String is not modified |
| 146 | * flags: comobination of JAVA_DEMANGLE_* flags to modify demangling |
| 147 | * return: |
| 148 | * if input can be demangled, then a newly allocated string is returned. |
| 149 | * if input cannot be demangled, then NULL is returned |
| 150 | * |
| 151 | * Note: caller is responsible for freeing demangled string |
| 152 | */ |
| 153 | char * |
| 154 | java_demangle_sym(const char *str, int flags) |
| 155 | { |
| 156 | char *buf, *ptr; |
| 157 | char *p; |
| 158 | size_t len, l1 = 0; |
| 159 | |
| 160 | if (!str) |
| 161 | return NULL; |
| 162 | |
| 163 | /* find start of retunr type */ |
| 164 | p = strrchr(str, ')'); |
| 165 | if (!p) |
| 166 | return NULL; |
| 167 | |
| 168 | /* |
| 169 | * expansion factor estimated to 3x |
| 170 | */ |
| 171 | len = strlen(str) * 3 + 1; |
| 172 | buf = malloc(len); |
| 173 | if (!buf) |
| 174 | return NULL; |
| 175 | |
| 176 | buf[0] = '\0'; |
| 177 | if (!(flags & JAVA_DEMANGLE_NORET)) { |
| 178 | /* |
| 179 | * get return type first |
| 180 | */ |
| 181 | ptr = __demangle_java_sym(p + 1, NULL, buf, len, MODE_TYPE); |
| 182 | if (!ptr) |
| 183 | goto error; |
| 184 | |
| 185 | /* add space between return type and function prototype */ |
| 186 | l1 = strlen(buf); |
| 187 | buf[l1++] = ' '; |
| 188 | } |
| 189 | |
| 190 | /* process function up to return type */ |
| 191 | ptr = __demangle_java_sym(str, p + 1, buf + l1, len - l1, MODE_PREFIX); |
| 192 | if (!ptr) |
| 193 | goto error; |
| 194 | |
| 195 | return buf; |
| 196 | error: |
| 197 | free(buf); |
| 198 | return NULL; |
| 199 | } |