root / src / sylfilter.c @ cfadedec
History | View | Annotate | Download (5.1 kB)
| 1 | 4012ec30 | Hiroyuki Yamamoto | #include <stdio.h> |
|---|---|---|---|
| 2 | faa55f07 | Hiroyuki Yamamoto | #include <string.h> |
| 3 | 5a583287 | Hiroyuki Yamamoto | #include <locale.h> |
| 4 | 4012ec30 | Hiroyuki Yamamoto | |
| 5 | 4012ec30 | Hiroyuki Yamamoto | #include "filter.h" |
| 6 | 4012ec30 | Hiroyuki Yamamoto | #include "filter-manager.h" |
| 7 | ff656913 | Hiroyuki Yamamoto | #include "filter-utils.h" |
| 8 | 761f2955 | Hiroyuki Yamamoto | #include "filter-kvs.h" |
| 9 | 761f2955 | Hiroyuki Yamamoto | |
| 10 | 761f2955 | Hiroyuki Yamamoto | #define USE_QDBM 1 |
| 11 | 761f2955 | Hiroyuki Yamamoto | #ifdef USE_QDBM
|
| 12 | 761f2955 | Hiroyuki Yamamoto | # include "filter-kvs-qdbm.h" |
| 13 | 761f2955 | Hiroyuki Yamamoto | #endif
|
| 14 | 761f2955 | Hiroyuki Yamamoto | |
| 15 | 4012ec30 | Hiroyuki Yamamoto | #include "textcontent-filter.h" |
| 16 | 4012ec30 | Hiroyuki Yamamoto | #include "blacklist-filter.h" |
| 17 | 4012ec30 | Hiroyuki Yamamoto | #include "whitelist-filter.h" |
| 18 | 4012ec30 | Hiroyuki Yamamoto | #include "wordsep-filter.h" |
| 19 | 4012ec30 | Hiroyuki Yamamoto | #include "ngram-filter.h" |
| 20 | 4012ec30 | Hiroyuki Yamamoto | #include "bayes-filter.h" |
| 21 | 4012ec30 | Hiroyuki Yamamoto | |
| 22 | 4cab1247 | Hiroyuki Yamamoto | enum {
|
| 23 | 4cab1247 | Hiroyuki Yamamoto | MODE_TEST_JUNK, |
| 24 | 4cab1247 | Hiroyuki Yamamoto | MODE_LEARN_JUNK, |
| 25 | 4cab1247 | Hiroyuki Yamamoto | MODE_LEARN_CLEAN, |
| 26 | af7ecefa | Hiroyuki Yamamoto | MODE_SHOW_STATUS |
| 27 | 4cab1247 | Hiroyuki Yamamoto | }; |
| 28 | 4cab1247 | Hiroyuki Yamamoto | |
| 29 | 0113a62c | Hiroyuki Yamamoto | |
| 30 | 0113a62c | Hiroyuki Yamamoto | static int verbose = 0; |
| 31 | 0113a62c | Hiroyuki Yamamoto | |
| 32 | 4cab1247 | Hiroyuki Yamamoto | static int learn_filter(int mode, const char *file); |
| 33 | 4cab1247 | Hiroyuki Yamamoto | static int test_filter(int mode, const char *file); |
| 34 | 295ccf38 | Hiroyuki Yamamoto | static void print_message_data(XMessageData *msgdata); |
| 35 | 4cab1247 | Hiroyuki Yamamoto | |
| 36 | 4cab1247 | Hiroyuki Yamamoto | |
| 37 | 4012ec30 | Hiroyuki Yamamoto | int main(int argc, char *argv[]) |
| 38 | 4012ec30 | Hiroyuki Yamamoto | {
|
| 39 | 4cab1247 | Hiroyuki Yamamoto | int retval = 0; |
| 40 | 4cab1247 | Hiroyuki Yamamoto | int i;
|
| 41 | 4cab1247 | Hiroyuki Yamamoto | int mode = MODE_TEST_JUNK;
|
| 42 | 4cab1247 | Hiroyuki Yamamoto | |
| 43 | 4cab1247 | Hiroyuki Yamamoto | setlocale(LC_ALL, "");
|
| 44 | 4cab1247 | Hiroyuki Yamamoto | |
| 45 | 4cab1247 | Hiroyuki Yamamoto | xfilter_init(); |
| 46 | 4cab1247 | Hiroyuki Yamamoto | |
| 47 | 4cab1247 | Hiroyuki Yamamoto | #ifdef USE_QDBM
|
| 48 | 4cab1247 | Hiroyuki Yamamoto | xfilter_kvs_qdbm_set_engine(); |
| 49 | 4cab1247 | Hiroyuki Yamamoto | #endif
|
| 50 | 4cab1247 | Hiroyuki Yamamoto | xfilter_bayes_db_init(); |
| 51 | 4cab1247 | Hiroyuki Yamamoto | |
| 52 | 4cab1247 | Hiroyuki Yamamoto | for (i = 1; i < argc; i++) { |
| 53 | 4cab1247 | Hiroyuki Yamamoto | if (!strcmp(argv[i], "-j")) |
| 54 | 4cab1247 | Hiroyuki Yamamoto | mode = MODE_LEARN_JUNK; |
| 55 | 4cab1247 | Hiroyuki Yamamoto | else if (!strcmp(argv[i], "-c")) |
| 56 | 4cab1247 | Hiroyuki Yamamoto | mode = MODE_LEARN_CLEAN; |
| 57 | 4cab1247 | Hiroyuki Yamamoto | else if (!strcmp(argv[i], "-t")) |
| 58 | 4cab1247 | Hiroyuki Yamamoto | mode = MODE_TEST_JUNK; |
| 59 | af7ecefa | Hiroyuki Yamamoto | else if (!strcmp(argv[i], "-s")) |
| 60 | af7ecefa | Hiroyuki Yamamoto | mode = MODE_SHOW_STATUS; |
| 61 | 0113a62c | Hiroyuki Yamamoto | else if (!strcmp(argv[i], "-v")) |
| 62 | 0113a62c | Hiroyuki Yamamoto | verbose = 1;
|
| 63 | 0113a62c | Hiroyuki Yamamoto | else if (!strcmp(argv[i], "-d")) |
| 64 | 0113a62c | Hiroyuki Yamamoto | xfilter_set_debug_mode(1);
|
| 65 | 4cab1247 | Hiroyuki Yamamoto | } |
| 66 | 4cab1247 | Hiroyuki Yamamoto | |
| 67 | af7ecefa | Hiroyuki Yamamoto | if (mode == MODE_SHOW_STATUS) {
|
| 68 | af7ecefa | Hiroyuki Yamamoto | retval = xfilter_bayes_db_show_contents(); |
| 69 | e260dbfd | Hiroyuki Yamamoto | } else if (mode == MODE_LEARN_JUNK || mode == MODE_LEARN_CLEAN) { |
| 70 | e260dbfd | Hiroyuki Yamamoto | for (i = 1; i < argc; i++) { |
| 71 | e260dbfd | Hiroyuki Yamamoto | if (argv[i][0] != '-') { |
| 72 | e260dbfd | Hiroyuki Yamamoto | retval = learn_filter(mode, argv[i]); |
| 73 | e260dbfd | Hiroyuki Yamamoto | if (retval != 0) |
| 74 | e260dbfd | Hiroyuki Yamamoto | break;
|
| 75 | e260dbfd | Hiroyuki Yamamoto | } |
| 76 | e260dbfd | Hiroyuki Yamamoto | } |
| 77 | e260dbfd | Hiroyuki Yamamoto | } else {
|
| 78 | e260dbfd | Hiroyuki Yamamoto | for (i = 1; i < argc; i++) { |
| 79 | e260dbfd | Hiroyuki Yamamoto | if (argv[i][0] != '-') { |
| 80 | e260dbfd | Hiroyuki Yamamoto | retval = test_filter(mode, argv[i]); |
| 81 | 704a6ae7 | Hiroyuki Yamamoto | if (retval == 127) |
| 82 | 704a6ae7 | Hiroyuki Yamamoto | break;
|
| 83 | e260dbfd | Hiroyuki Yamamoto | } |
| 84 | e260dbfd | Hiroyuki Yamamoto | } |
| 85 | e260dbfd | Hiroyuki Yamamoto | } |
| 86 | 4cab1247 | Hiroyuki Yamamoto | |
| 87 | 4cab1247 | Hiroyuki Yamamoto | xfilter_bayes_db_done(); |
| 88 | 4cab1247 | Hiroyuki Yamamoto | xfilter_done(); |
| 89 | 4cab1247 | Hiroyuki Yamamoto | |
| 90 | 4cab1247 | Hiroyuki Yamamoto | return retval;
|
| 91 | 4cab1247 | Hiroyuki Yamamoto | } |
| 92 | 4cab1247 | Hiroyuki Yamamoto | |
| 93 | 4cab1247 | Hiroyuki Yamamoto | static int learn_filter(int mode, const char *file) |
| 94 | 4cab1247 | Hiroyuki Yamamoto | {
|
| 95 | 4012ec30 | Hiroyuki Yamamoto | XFilterManager *mgr; |
| 96 | 4012ec30 | Hiroyuki Yamamoto | XMessageData *msgdata; |
| 97 | 4012ec30 | Hiroyuki Yamamoto | XMessageData *resdata; |
| 98 | 4012ec30 | Hiroyuki Yamamoto | XFilterResult *res; |
| 99 | 4012ec30 | Hiroyuki Yamamoto | XFilterStatus status; |
| 100 | cfadedec | Hiroyuki Yamamoto | int retval = 0; |
| 101 | 4012ec30 | Hiroyuki Yamamoto | |
| 102 | faa55f07 | Hiroyuki Yamamoto | XFilterConstructorFunc learn_junk_ctors[] = {
|
| 103 | 4012ec30 | Hiroyuki Yamamoto | xfilter_textcontent_new, |
| 104 | 4012ec30 | Hiroyuki Yamamoto | xfilter_wordsep_new, |
| 105 | 4012ec30 | Hiroyuki Yamamoto | xfilter_ngram_new, |
| 106 | faa55f07 | Hiroyuki Yamamoto | xfilter_bayes_learn_junk_new, |
| 107 | faa55f07 | Hiroyuki Yamamoto | NULL
|
| 108 | faa55f07 | Hiroyuki Yamamoto | }; |
| 109 | faa55f07 | Hiroyuki Yamamoto | |
| 110 | faa55f07 | Hiroyuki Yamamoto | XFilterConstructorFunc learn_nojunk_ctors[] = {
|
| 111 | faa55f07 | Hiroyuki Yamamoto | xfilter_textcontent_new, |
| 112 | faa55f07 | Hiroyuki Yamamoto | xfilter_wordsep_new, |
| 113 | faa55f07 | Hiroyuki Yamamoto | xfilter_ngram_new, |
| 114 | faa55f07 | Hiroyuki Yamamoto | xfilter_bayes_learn_nojunk_new, |
| 115 | 4012ec30 | Hiroyuki Yamamoto | NULL
|
| 116 | 4012ec30 | Hiroyuki Yamamoto | }; |
| 117 | 4012ec30 | Hiroyuki Yamamoto | |
| 118 | 704a6ae7 | Hiroyuki Yamamoto | if (verbose)
|
| 119 | 704a6ae7 | Hiroyuki Yamamoto | printf("learning message file: %s\n", file);
|
| 120 | 704a6ae7 | Hiroyuki Yamamoto | |
| 121 | 4012ec30 | Hiroyuki Yamamoto | mgr = xfilter_manager_new(); |
| 122 | 4012ec30 | Hiroyuki Yamamoto | |
| 123 | 4cab1247 | Hiroyuki Yamamoto | if (mode == MODE_LEARN_JUNK)
|
| 124 | faa55f07 | Hiroyuki Yamamoto | xfilter_manager_add_filters(mgr, learn_junk_ctors); |
| 125 | faa55f07 | Hiroyuki Yamamoto | else
|
| 126 | faa55f07 | Hiroyuki Yamamoto | xfilter_manager_add_filters(mgr, learn_nojunk_ctors); |
| 127 | 4012ec30 | Hiroyuki Yamamoto | |
| 128 | 4cab1247 | Hiroyuki Yamamoto | msgdata = xfilter_message_data_read_file(file, "message/rfc822");
|
| 129 | 4012ec30 | Hiroyuki Yamamoto | |
| 130 | 4012ec30 | Hiroyuki Yamamoto | res = xfilter_manager_run(mgr, msgdata); |
| 131 | 0113a62c | Hiroyuki Yamamoto | if (verbose)
|
| 132 | 0113a62c | Hiroyuki Yamamoto | xfilter_result_print(res); |
| 133 | 4012ec30 | Hiroyuki Yamamoto | status = xfilter_result_get_status(res); |
| 134 | 704a6ae7 | Hiroyuki Yamamoto | if (status == XF_UNSUPPORTED_TYPE || status == XF_ERROR) {
|
| 135 | 704a6ae7 | Hiroyuki Yamamoto | printf("%s: Error on learning mail\n", file);
|
| 136 | 704a6ae7 | Hiroyuki Yamamoto | retval = 127;
|
| 137 | 704a6ae7 | Hiroyuki Yamamoto | } |
| 138 | 4012ec30 | Hiroyuki Yamamoto | |
| 139 | 704a6ae7 | Hiroyuki Yamamoto | if (xfilter_get_debug_mode()) {
|
| 140 | 0113a62c | Hiroyuki Yamamoto | resdata = xfilter_result_get_message_data(res); |
| 141 | 0113a62c | Hiroyuki Yamamoto | print_message_data(resdata); |
| 142 | 0113a62c | Hiroyuki Yamamoto | } |
| 143 | 4012ec30 | Hiroyuki Yamamoto | |
| 144 | 4012ec30 | Hiroyuki Yamamoto | xfilter_result_free(res); |
| 145 | 4012ec30 | Hiroyuki Yamamoto | xfilter_message_data_free(msgdata); |
| 146 | 4012ec30 | Hiroyuki Yamamoto | xfilter_manager_free(mgr); |
| 147 | 4012ec30 | Hiroyuki Yamamoto | |
| 148 | 704a6ae7 | Hiroyuki Yamamoto | return retval;
|
| 149 | 4cab1247 | Hiroyuki Yamamoto | } |
| 150 | 4012ec30 | Hiroyuki Yamamoto | |
| 151 | 4cab1247 | Hiroyuki Yamamoto | static int test_filter(int mode, const char *file) |
| 152 | 4cab1247 | Hiroyuki Yamamoto | {
|
| 153 | 4cab1247 | Hiroyuki Yamamoto | XFilterManager *mgr; |
| 154 | 4cab1247 | Hiroyuki Yamamoto | XMessageData *msgdata; |
| 155 | 4cab1247 | Hiroyuki Yamamoto | XMessageData *resdata; |
| 156 | 4cab1247 | Hiroyuki Yamamoto | XFilterResult *res; |
| 157 | 4cab1247 | Hiroyuki Yamamoto | XFilterStatus status; |
| 158 | 4cab1247 | Hiroyuki Yamamoto | int retval = 0; |
| 159 | 4012ec30 | Hiroyuki Yamamoto | |
| 160 | 4cab1247 | Hiroyuki Yamamoto | XFilterConstructorFunc ctors[] = {
|
| 161 | 4cab1247 | Hiroyuki Yamamoto | xfilter_textcontent_new, |
| 162 | 4cab1247 | Hiroyuki Yamamoto | xfilter_blacklist_new, |
| 163 | 4cab1247 | Hiroyuki Yamamoto | xfilter_whitelist_new, |
| 164 | 4cab1247 | Hiroyuki Yamamoto | xfilter_wordsep_new, |
| 165 | 4cab1247 | Hiroyuki Yamamoto | xfilter_ngram_new, |
| 166 | 4cab1247 | Hiroyuki Yamamoto | xfilter_bayes_new, |
| 167 | 4cab1247 | Hiroyuki Yamamoto | NULL
|
| 168 | 4cab1247 | Hiroyuki Yamamoto | }; |
| 169 | 4012ec30 | Hiroyuki Yamamoto | |
| 170 | 704a6ae7 | Hiroyuki Yamamoto | if (verbose)
|
| 171 | 704a6ae7 | Hiroyuki Yamamoto | printf("testing message file: %s\n", file);
|
| 172 | 704a6ae7 | Hiroyuki Yamamoto | |
| 173 | 4cab1247 | Hiroyuki Yamamoto | mgr = xfilter_manager_new(); |
| 174 | 4cab1247 | Hiroyuki Yamamoto | xfilter_manager_add_filters(mgr, ctors); |
| 175 | 4012ec30 | Hiroyuki Yamamoto | |
| 176 | 4cab1247 | Hiroyuki Yamamoto | msgdata = xfilter_message_data_read_file(file, "message/rfc822");
|
| 177 | 4012ec30 | Hiroyuki Yamamoto | |
| 178 | 4012ec30 | Hiroyuki Yamamoto | res = xfilter_manager_run(mgr, msgdata); |
| 179 | 0113a62c | Hiroyuki Yamamoto | if (verbose)
|
| 180 | 0113a62c | Hiroyuki Yamamoto | xfilter_result_print(res); |
| 181 | 4012ec30 | Hiroyuki Yamamoto | status = xfilter_result_get_status(res); |
| 182 | 4cab1247 | Hiroyuki Yamamoto | if (status == XF_JUNK) {
|
| 183 | 704a6ae7 | Hiroyuki Yamamoto | printf("%s: This is a junk mail (prob: %f)\n", file, xfilter_result_get_probability(res));
|
| 184 | f45a6905 | Hiroyuki Yamamoto | retval = 0;
|
| 185 | 4cab1247 | Hiroyuki Yamamoto | } else if (status == XF_UNCERTAIN) { |
| 186 | 704a6ae7 | Hiroyuki Yamamoto | printf("%s: This mail could not be classified (prob: %f)\n", file, xfilter_result_get_probability(res));
|
| 187 | f45a6905 | Hiroyuki Yamamoto | retval = 2;
|
| 188 | 4cab1247 | Hiroyuki Yamamoto | } else if (status == XF_UNSUPPORTED_TYPE || status == XF_ERROR) { |
| 189 | 704a6ae7 | Hiroyuki Yamamoto | printf("%s: Error on testing mail\n", file);
|
| 190 | f45a6905 | Hiroyuki Yamamoto | retval = 127;
|
| 191 | 4cab1247 | Hiroyuki Yamamoto | } else {
|
| 192 | 704a6ae7 | Hiroyuki Yamamoto | printf("%s: This is a clean mail (prob: %f)\n", file, xfilter_result_get_probability(res));
|
| 193 | f45a6905 | Hiroyuki Yamamoto | retval = 1;
|
| 194 | 4cab1247 | Hiroyuki Yamamoto | } |
| 195 | 4012ec30 | Hiroyuki Yamamoto | |
| 196 | 704a6ae7 | Hiroyuki Yamamoto | if (xfilter_get_debug_mode()) {
|
| 197 | 0113a62c | Hiroyuki Yamamoto | resdata = xfilter_result_get_message_data(res); |
| 198 | 0113a62c | Hiroyuki Yamamoto | print_message_data(resdata); |
| 199 | 0113a62c | Hiroyuki Yamamoto | } |
| 200 | 4012ec30 | Hiroyuki Yamamoto | |
| 201 | 4012ec30 | Hiroyuki Yamamoto | xfilter_result_free(res); |
| 202 | 4012ec30 | Hiroyuki Yamamoto | xfilter_message_data_free(msgdata); |
| 203 | 4012ec30 | Hiroyuki Yamamoto | |
| 204 | 4012ec30 | Hiroyuki Yamamoto | xfilter_manager_free(mgr); |
| 205 | 4012ec30 | Hiroyuki Yamamoto | |
| 206 | f45a6905 | Hiroyuki Yamamoto | return retval;
|
| 207 | 4012ec30 | Hiroyuki Yamamoto | } |
| 208 | 295ccf38 | Hiroyuki Yamamoto | |
| 209 | 295ccf38 | Hiroyuki Yamamoto | static void print_message_data(XMessageData *msgdata) |
| 210 | 295ccf38 | Hiroyuki Yamamoto | {
|
| 211 | 295ccf38 | Hiroyuki Yamamoto | const char *content; |
| 212 | 295ccf38 | Hiroyuki Yamamoto | |
| 213 | 295ccf38 | Hiroyuki Yamamoto | if (!msgdata)
|
| 214 | 295ccf38 | Hiroyuki Yamamoto | return;
|
| 215 | 295ccf38 | Hiroyuki Yamamoto | |
| 216 | 295ccf38 | Hiroyuki Yamamoto | printf("\n");
|
| 217 | 295ccf38 | Hiroyuki Yamamoto | |
| 218 | 295ccf38 | Hiroyuki Yamamoto | content = xfilter_message_data_get_attribute(msgdata, XM_FROM); |
| 219 | 295ccf38 | Hiroyuki Yamamoto | if (content)
|
| 220 | 295ccf38 | Hiroyuki Yamamoto | printf("from: %s\n", content);
|
| 221 | 295ccf38 | Hiroyuki Yamamoto | content = xfilter_message_data_get_attribute(msgdata, XM_TO); |
| 222 | 295ccf38 | Hiroyuki Yamamoto | if (content)
|
| 223 | 295ccf38 | Hiroyuki Yamamoto | printf("to: %s\n", content);
|
| 224 | 295ccf38 | Hiroyuki Yamamoto | content = xfilter_message_data_get_attribute(msgdata, XM_CC); |
| 225 | 295ccf38 | Hiroyuki Yamamoto | if (content)
|
| 226 | 295ccf38 | Hiroyuki Yamamoto | printf("cc: %s\n", content);
|
| 227 | 295ccf38 | Hiroyuki Yamamoto | content = xfilter_message_data_get_attribute(msgdata, XM_SUBJECT); |
| 228 | 295ccf38 | Hiroyuki Yamamoto | if (content)
|
| 229 | 295ccf38 | Hiroyuki Yamamoto | printf("subject: %s\n", content);
|
| 230 | 295ccf38 | Hiroyuki Yamamoto | content = xfilter_message_data_get_content(msgdata); |
| 231 | 295ccf38 | Hiroyuki Yamamoto | printf("content: %s\n", content);
|
| 232 | 295ccf38 | Hiroyuki Yamamoto | } |