| 1 | #include "jsi.h" |
| 2 | #include "jsvalue.h" |
| 3 | #include "jsbuiltin.h" |
| 4 | #include "regexp.h" |
| 5 | |
| 6 | void js_newregexp(js_State *J, const char *pattern, int flags) |
| 7 | { |
| 8 | const char *error; |
| 9 | js_Object *obj; |
| 10 | Reprog *prog; |
| 11 | int opts; |
| 12 | |
| 13 | obj = jsV_newobject(J, JS_CREGEXP, J->RegExp_prototype); |
| 14 | |
| 15 | opts = 0; |
| 16 | if (flags & JS_REGEXP_I) opts |= REG_ICASE; |
| 17 | if (flags & JS_REGEXP_M) opts |= REG_NEWLINE; |
| 18 | |
| 19 | prog = js_regcompx(J->alloc, J->actx, pattern, opts, &error); |
| 20 | if (!prog) |
| 21 | js_syntaxerror(J, "regular expression: %s" , error); |
| 22 | |
| 23 | obj->u.r.prog = prog; |
| 24 | obj->u.r.source = js_strdup(J, pattern); |
| 25 | obj->u.r.flags = flags; |
| 26 | obj->u.r.last = 0; |
| 27 | js_pushobject(J, obj); |
| 28 | } |
| 29 | |
| 30 | void js_RegExp_prototype_exec(js_State *J, js_Regexp *re, const char *text) |
| 31 | { |
| 32 | int result; |
| 33 | int i; |
| 34 | int opts; |
| 35 | Resub m; |
| 36 | |
| 37 | opts = 0; |
| 38 | if (re->flags & JS_REGEXP_G) { |
| 39 | if (re->last > strlen(text)) { |
| 40 | re->last = 0; |
| 41 | js_pushnull(J); |
| 42 | return; |
| 43 | } |
| 44 | if (re->last > 0) { |
| 45 | text += re->last; |
| 46 | opts |= REG_NOTBOL; |
| 47 | } |
| 48 | } |
| 49 | |
| 50 | result = js_regexec(re->prog, text, &m, opts); |
| 51 | if (result < 0) |
| 52 | js_error(J, "regexec failed" ); |
| 53 | if (result == 0) { |
| 54 | js_newarray(J); |
| 55 | js_pushstring(J, text); |
| 56 | js_setproperty(J, -2, "input" ); |
| 57 | js_pushnumber(J, js_utfptrtoidx(text, m.sub[0].sp)); |
| 58 | js_setproperty(J, -2, "index" ); |
| 59 | for (i = 0; i < m.nsub; ++i) { |
| 60 | js_pushlstring(J, m.sub[i].sp, m.sub[i].ep - m.sub[i].sp); |
| 61 | js_setindex(J, -2, i); |
| 62 | } |
| 63 | if (re->flags & JS_REGEXP_G) |
| 64 | re->last = re->last + (m.sub[0].ep - text); |
| 65 | return; |
| 66 | } |
| 67 | |
| 68 | if (re->flags & JS_REGEXP_G) |
| 69 | re->last = 0; |
| 70 | |
| 71 | js_pushnull(J); |
| 72 | } |
| 73 | |
| 74 | static void Rp_test(js_State *J) |
| 75 | { |
| 76 | js_Regexp *re; |
| 77 | const char *text; |
| 78 | int result; |
| 79 | int opts; |
| 80 | Resub m; |
| 81 | |
| 82 | re = js_toregexp(J, 0); |
| 83 | text = js_tostring(J, 1); |
| 84 | |
| 85 | opts = 0; |
| 86 | if (re->flags & JS_REGEXP_G) { |
| 87 | if (re->last > strlen(text)) { |
| 88 | re->last = 0; |
| 89 | js_pushboolean(J, 0); |
| 90 | return; |
| 91 | } |
| 92 | if (re->last > 0) { |
| 93 | text += re->last; |
| 94 | opts |= REG_NOTBOL; |
| 95 | } |
| 96 | } |
| 97 | |
| 98 | result = js_regexec(re->prog, text, &m, opts); |
| 99 | if (result < 0) |
| 100 | js_error(J, "regexec failed" ); |
| 101 | if (result == 0) { |
| 102 | if (re->flags & JS_REGEXP_G) |
| 103 | re->last = re->last + (m.sub[0].ep - text); |
| 104 | js_pushboolean(J, 1); |
| 105 | return; |
| 106 | } |
| 107 | |
| 108 | if (re->flags & JS_REGEXP_G) |
| 109 | re->last = 0; |
| 110 | |
| 111 | js_pushboolean(J, 0); |
| 112 | } |
| 113 | |
| 114 | static void jsB_new_RegExp(js_State *J) |
| 115 | { |
| 116 | js_Regexp *old; |
| 117 | const char *pattern; |
| 118 | int flags; |
| 119 | |
| 120 | if (js_isregexp(J, 1)) { |
| 121 | if (js_isdefined(J, 2)) |
| 122 | js_typeerror(J, "cannot supply flags when creating one RegExp from another" ); |
| 123 | old = js_toregexp(J, 1); |
| 124 | pattern = old->source; |
| 125 | flags = old->flags; |
| 126 | } else if (js_isundefined(J, 1)) { |
| 127 | pattern = "(?:)" ; |
| 128 | flags = 0; |
| 129 | } else { |
| 130 | pattern = js_tostring(J, 1); |
| 131 | flags = 0; |
| 132 | } |
| 133 | |
| 134 | if (strlen(pattern) == 0) |
| 135 | pattern = "(?:)" ; |
| 136 | |
| 137 | if (js_isdefined(J, 2)) { |
| 138 | const char *s = js_tostring(J, 2); |
| 139 | int g = 0, i = 0, m = 0; |
| 140 | while (*s) { |
| 141 | if (*s == 'g') ++g; |
| 142 | else if (*s == 'i') ++i; |
| 143 | else if (*s == 'm') ++m; |
| 144 | else js_syntaxerror(J, "invalid regular expression flag: '%c'" , *s); |
| 145 | ++s; |
| 146 | } |
| 147 | if (g > 1) js_syntaxerror(J, "invalid regular expression flag: 'g'" ); |
| 148 | if (i > 1) js_syntaxerror(J, "invalid regular expression flag: 'i'" ); |
| 149 | if (m > 1) js_syntaxerror(J, "invalid regular expression flag: 'm'" ); |
| 150 | if (g) flags |= JS_REGEXP_G; |
| 151 | if (i) flags |= JS_REGEXP_I; |
| 152 | if (m) flags |= JS_REGEXP_M; |
| 153 | } |
| 154 | |
| 155 | js_newregexp(J, pattern, flags); |
| 156 | } |
| 157 | |
| 158 | static void jsB_RegExp(js_State *J) |
| 159 | { |
| 160 | if (js_isregexp(J, 1)) |
| 161 | return; |
| 162 | jsB_new_RegExp(J); |
| 163 | } |
| 164 | |
| 165 | static void Rp_toString(js_State *J) |
| 166 | { |
| 167 | js_Regexp *re; |
| 168 | char *out; |
| 169 | |
| 170 | re = js_toregexp(J, 0); |
| 171 | |
| 172 | out = js_malloc(J, strlen(re->source) + 6); /* extra space for //gim */ |
| 173 | strcpy(out, "/" ); |
| 174 | strcat(out, re->source); |
| 175 | strcat(out, "/" ); |
| 176 | if (re->flags & JS_REGEXP_G) strcat(out, "g" ); |
| 177 | if (re->flags & JS_REGEXP_I) strcat(out, "i" ); |
| 178 | if (re->flags & JS_REGEXP_M) strcat(out, "m" ); |
| 179 | |
| 180 | if (js_try(J)) { |
| 181 | js_free(J, out); |
| 182 | js_throw(J); |
| 183 | } |
| 184 | js_pop(J, 0); |
| 185 | js_pushstring(J, out); |
| 186 | js_endtry(J); |
| 187 | js_free(J, out); |
| 188 | } |
| 189 | |
| 190 | static void Rp_exec(js_State *J) |
| 191 | { |
| 192 | js_RegExp_prototype_exec(J, js_toregexp(J, 0), js_tostring(J, 1)); |
| 193 | } |
| 194 | |
| 195 | void jsB_initregexp(js_State *J) |
| 196 | { |
| 197 | js_pushobject(J, J->RegExp_prototype); |
| 198 | { |
| 199 | jsB_propf(J, "RegExp.prototype.toString" , Rp_toString, 0); |
| 200 | jsB_propf(J, "RegExp.prototype.test" , Rp_test, 0); |
| 201 | jsB_propf(J, "RegExp.prototype.exec" , Rp_exec, 0); |
| 202 | } |
| 203 | js_newcconstructor(J, jsB_RegExp, jsB_new_RegExp, "RegExp" , 1); |
| 204 | js_defglobal(J, "RegExp" , JS_DONTENUM); |
| 205 | } |
| 206 | |