| 1 | // SPDX-License-Identifier: GPL-3.0-or-later |
| 2 | |
| 3 | #include "log2journal.h" |
| 4 | |
| 5 | #define LOGFMT_ERROR_LINE_MAX 1024 |
| 6 | #define LOGFMT_KEY_MAX 1024 |
| 7 | |
| 8 | struct logfmt_state { |
| 9 | LOG_JOB *jb; |
| 10 | |
| 11 | const char *line; |
| 12 | uint32_t pos; |
| 13 | uint32_t key_start; |
| 14 | |
| 15 | char key[LOGFMT_KEY_MAX]; |
| 16 | char msg[LOGFMT_ERROR_LINE_MAX]; |
| 17 | }; |
| 18 | |
| 19 | #define logfmt_current_pos(lfs) &(lfs)->line[(lfs)->pos] |
| 20 | #define logfmt_consume_char(lfs) ++(lfs)->pos |
| 21 | |
| 22 | static inline void logfmt_process_key_value(LOGFMT_STATE *lfs, const char *value, size_t len) { |
| 23 | log_job_send_extracted_key_value(lfs->jb, lfs->key, value, len); |
| 24 | } |
| 25 | |
| 26 | static inline void logfmt_skip_spaces(LOGFMT_STATE *lfs) { |
| 27 | const char *s = logfmt_current_pos(lfs); |
| 28 | const char *start = s; |
| 29 | |
| 30 | while(isspace(*s)) s++; |
| 31 | |
| 32 | lfs->pos += s - start; |
| 33 | } |
| 34 | |
| 35 | static inline void copy_newline(LOGFMT_STATE *lfs __maybe_unused, char **d, size_t *remaining) { |
| 36 | if(*remaining > 3) { |
| 37 | *(*d)++ = '\\'; |
| 38 | *(*d)++ = 'n'; |
| 39 | (*remaining) -= 2; |
| 40 | } |
| 41 | } |
| 42 | |
| 43 | static inline void copy_tab(LOGFMT_STATE *lfs __maybe_unused, char **d, size_t *remaining) { |
| 44 | if(*remaining > 3) { |
| 45 | *(*d)++ = '\\'; |
| 46 | *(*d)++ = 't'; |
| 47 | (*remaining) -= 2; |
| 48 | } |
| 49 | } |
| 50 | |
| 51 | static inline bool logftm_parse_value(LOGFMT_STATE *lfs) { |
| 52 | static __thread char value[JOURNAL_MAX_VALUE_LEN]; |
| 53 | |
| 54 | char quote = '\0'; |
| 55 | const char *s = logfmt_current_pos(lfs); |
| 56 | if(*s == '"' || *s == '\'') { |
| 57 | quote = *s; |
| 58 | logfmt_consume_char(lfs); |
| 59 | } |
| 60 | |
| 61 | value[0] = '\0'; |
| 62 | char *d = value; |
| 63 | s = logfmt_current_pos(lfs); |
| 64 | size_t remaining = sizeof(value); |
| 65 | |
| 66 | char end_char = (char)(quote == '\0' ? ' ' : quote); |
| 67 | while (*s && *s != end_char) { |
| 68 | char c; |
| 69 | |
| 70 | if (*s == '\\') { |
| 71 | s++; |
| 72 | |
| 73 | if(!*s) |
| 74 | c = '\\'; |
| 75 | else { |
| 76 | switch (*s) { |
| 77 | case 'n': |
| 78 | copy_newline(lfs, &d, &remaining); |
| 79 | s++; |
| 80 | continue; |
| 81 | |
| 82 | case 't': |
| 83 | copy_tab(lfs, &d, &remaining); |
| 84 | s++; |
| 85 | continue; |
| 86 | |
| 87 | case 'f': |
| 88 | case 'b': |
| 89 | case 'r': |
| 90 | c = ' '; |
| 91 | s++; |
| 92 | break; |
| 93 | |
| 94 | default: |
| 95 | c = *s++; |
| 96 | break; |
| 97 | } |
| 98 | } |
| 99 | } |
| 100 | else |
| 101 | c = *s++; |
| 102 | |
| 103 | if(remaining < 2) { |
| 104 | snprintf(lfs->msg, sizeof(lfs->msg), |
| 105 | "LOGFMT PARSER: truncated string value at position %u", lfs->pos); |
| 106 | return false; |
| 107 | } |
| 108 | else { |
| 109 | *d++ = c; |
| 110 | remaining--; |
| 111 | } |
| 112 | } |
| 113 | *d = '\0'; |
| 114 | lfs->pos += s - logfmt_current_pos(lfs); |
| 115 | |
| 116 | s = logfmt_current_pos(lfs); |
| 117 | |
| 118 | if(quote != '\0') { |
| 119 | if (*s != quote) { |
| 120 | snprintf(lfs->msg, sizeof(lfs->msg), |
| 121 | "LOGFMT PARSER: missing quote at position %u: '%s'", |
| 122 | lfs->pos, s); |
| 123 | return false; |
| 124 | } |
| 125 | else |
| 126 | logfmt_consume_char(lfs); |
| 127 | } |
| 128 | |
| 129 | if(d > value) |
| 130 | logfmt_process_key_value(lfs, value, d - value); |
| 131 | |
| 132 | return true; |
| 133 | } |
| 134 | |
| 135 | static inline bool logfmt_parse_key(LOGFMT_STATE *lfs) { |
| 136 | logfmt_skip_spaces(lfs); |
| 137 | |
| 138 | char *d = &lfs->key[lfs->key_start]; |
| 139 | |
| 140 | size_t remaining = sizeof(lfs->key) - (d - lfs->key); |
| 141 | |
| 142 | const char *s = logfmt_current_pos(lfs); |
| 143 | char last_c = '\0'; |
| 144 | while(*s && *s != '=') { |
| 145 | char c; |
| 146 | |
| 147 | if (*s == '\\') { |
| 148 | s++; |
| 149 | |
| 150 | if(!*s) |
| 151 | c = journal_key_characters_map[(unsigned char)'\\']; |
| 152 | else |
| 153 | c = journal_key_characters_map[(unsigned char)*s++]; |
| 154 | } |
| 155 | else |
| 156 | c = journal_key_characters_map[(unsigned char)*s++]; |
| 157 | |
| 158 | if(c == '_' && last_c == '_') |
| 159 | continue; |
| 160 | else { |
| 161 | if(remaining < 2) { |
| 162 | snprintf(lfs->msg, sizeof(lfs->msg), |
| 163 | "LOGFMT PARSER: key buffer full - keys are too long, at position %u", lfs->pos); |
| 164 | return false; |
| 165 | } |
| 166 | *d++ = c; |
| 167 | remaining--; |
| 168 | } |
| 169 | |
| 170 | last_c = c; |
| 171 | } |
| 172 | *d = '\0'; |
| 173 | lfs->pos += s - logfmt_current_pos(lfs); |
| 174 | |
| 175 | s = logfmt_current_pos(lfs); |
| 176 | if(*s != '=') { |
| 177 | snprintf(lfs->msg, sizeof(lfs->msg), |
| 178 | "LOGFMT PARSER: key is missing the equal sign, at position %u", lfs->pos); |
| 179 | return false; |
| 180 | } |
| 181 | |
| 182 | logfmt_consume_char(lfs); |
| 183 | |
| 184 | return true; |
| 185 | } |
| 186 | |
| 187 | LOGFMT_STATE *logfmt_parser_create(LOG_JOB *jb) { |
| 188 | LOGFMT_STATE *lfs = mallocz(sizeof(LOGFMT_STATE)); |
| 189 | memset(lfs, 0, sizeof(LOGFMT_STATE)); |
| 190 | lfs->jb = jb; |
| 191 | |
| 192 | if(jb->prefix) |
| 193 | lfs->key_start = copy_to_buffer(lfs->key, sizeof(lfs->key), lfs->jb->prefix, strlen(lfs->jb->prefix)); |
| 194 | |
| 195 | return lfs; |
| 196 | } |
| 197 | |
| 198 | void logfmt_parser_destroy(LOGFMT_STATE *lfs) { |
| 199 | if(lfs) |
| 200 | freez(lfs); |
| 201 | } |
| 202 | |
| 203 | const char *logfmt_parser_error(LOGFMT_STATE *lfs) { |
| 204 | return lfs->msg; |
| 205 | } |
| 206 | |
| 207 | bool logfmt_parse_document(LOGFMT_STATE *lfs, const char *txt) { |
| 208 | lfs->line = txt; |
| 209 | lfs->pos = 0; |
| 210 | lfs->msg[0] = '\0'; |
| 211 | |
| 212 | const char *s; |
| 213 | do { |
| 214 | if(!logfmt_parse_key(lfs)) |
| 215 | return false; |
| 216 | |
| 217 | if(!logftm_parse_value(lfs)) |
| 218 | return false; |
| 219 | |
| 220 | logfmt_skip_spaces(lfs); |
| 221 | |
| 222 | s = logfmt_current_pos(lfs); |
| 223 | } while(*s); |
| 224 | |
| 225 | return true; |
| 226 | } |
| 227 | |
| 228 | |
| 229 | void logfmt_test(void) { |
| 230 | LOG_JOB jb = { 0 }; |
| 231 | log_job_init(&jb); |
| 232 | log_job_key_prefix_set(&jb, "NGINX_", 6); |
| 233 | LOGFMT_STATE *logfmt = logfmt_parser_create(&jb); |
| 234 | |
| 235 | logfmt_parse_document(logfmt, "x=1 y=2 z=\"3 \\ 4\" 5 "); |
| 236 | |
| 237 | logfmt_parser_destroy(logfmt); |
| 238 | log_job_cleanup(&jb); |
| 239 | } |