123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839 |
- /*
- * Copyright 2017 Yann Weber
- *
- * This file is part of Ttail.
- *
- * Ttail is free software: you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation, either version 3 of the License, or
- * any later version.
- *
- * Ttail is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with Ttail. If not, see <http://www.gnu.org/licenses/>.
- */
- #include "ttail_search_files.h"
-
- int _ttail_search_closest_files(ttail_t* t)
- {
- int ret;
- size_t i, prev;
- struct tm **ftm; /* t->logfile_sz len of struct tm[2] */
- char prev_found;
- ret = _ttail_search_closest_files_init(t);
- if(ret)
- {
- return ret;
- }
- /* Storing min & max of each files in ftm and checking that files are
- * sorted well */
- ftm = malloc(sizeof(struct tm*) * t->logfile_sz);
- if(!ftm)
- {
- perror("Unable to allocate memory");
- goto _ttail_search_closest_files_alloc_err;
- }
- prev_found = 0;
- prev = 0;
- for(i=0; i<t->logfile_sz; i++)
- {
- if(!t->logfile[i])
- {
- ftm[i] = NULL;
- continue;
- }
- ftm[i] = malloc(sizeof(struct tm)*2);
- if(!ftm[i])
- {
- perror("Unable to allocate memory for time");
- goto _ttail_search_closest_files_alloc_loop_err;
- }
- ret = _ttail_file_minmax(t, i, ftm[i]);
- if(ret < 0)
- {
- fprintf(stderr, "Minmax error\n");
- goto _ttail_search_closest_files_loop_err;
- }
- else if (ret == 1)
- {
- fprintf(stderr, "Warning : unable to find a valid date \
- in '%s'\n", t->logfile_name[i]);
- free(ftm[i]);
- ftm[i] = NULL;
- fclose(t->logfile[i]);
- t->logfile[i] = NULL;
- continue;
- }
- if(i && prev_found &&
- ttail_tm_cmp(&(ftm[prev][1]), &(ftm[i][0])) > 0)
- {
- /* files are not sorted */
- fprintf(stderr, "Files do not seems to be sorted. \
- File sorting not implemented yet\n");
- goto _ttail_search_closest_files_loop_err;
- }
- prev_found = 1;
- prev = i;
- }
- if(t->flag & TTAIL_FLAG_DATE_MIN)
- {
- ret = _ttail_search_files_binary_search(t, &(t->date_min),
- (const struct tm**)ftm, 1);
- if(ret)
- {
- if(t->verbose > 2)
- {
- fprintf(stderr, "Error while looking for \
- date-max\n");
- }
- goto _ttail_search_closest_files_err;
- }
- }
- for(i=0; i<t->logfile_sz; i++)
- {
- if(!t->logfile[i])
- {
- continue;
- }
- if(fseek(t->logfile[i], 0, SEEK_SET) < 0)
- {
- perror("Error setting position in file");
- goto _ttail_search_closest_files_err;
- }
- }
- if(t->flag & TTAIL_FLAG_DATE_MAX)
- {
- ret = _ttail_search_files_binary_search(t, &(t->date_max),
- (const struct tm**)ftm, 0);
- if(ret)
- {
- if(t->verbose > 2)
- {
- fprintf(stderr, "Error while looking for \
- date-max\n");
- }
- goto _ttail_search_closest_files_err;
- }
- t->session->file.off_max.off++;
- }
- for(i=0; i<t->logfile_sz; i++)
- {
- if(ftm) { free(ftm[i]); }
- }
- free(ftm);
- return ret;
-
- goto _ttail_search_closest_files_err;
- _ttail_search_closest_files_err:
- i = t->logfile_sz;
- do
- {
- _ttail_search_closest_files_alloc_loop_err:
- i--;
- _ttail_search_closest_files_loop_err:
- if(ftm[i])
- {
- free(ftm[i]);
- }
- }while(i);
- free(ftm);
- _ttail_search_closest_files_alloc_err:
- return -1;
- }
-
- void _ttail_search_print_files(ttail_t* t, int out)
- {
- size_t i;
- int fd;
- char buf[8192];
- int r;
- off_t cur_off;
-
- /* if no date_min t->session->file.id == 0 */
- for(i=t->session->file.off_min.id; i<t->logfile_sz; i++)
- {
- if(!t->logfile[i])
- {
- continue;
- }
- fd = fileno(t->logfile[i]);
- if(t->flag & TTAIL_FLAG_DATE_MIN &&
- i==t->session->file.off_min.id)
- {
- if(t->flag & TTAIL_FLAG_DATE_MAX)
- {
- fseek(t->logfile[i],
- t->session->file.off_min.off,
- SEEK_SET);
- }
- else
- {
- lseek(fd, t->session->file.off_min.off,
- SEEK_SET);
- }
- }
- else
- {
- if(t->flag & TTAIL_FLAG_DATE_MAX)
- {
- fseek(t->logfile[i], 0, SEEK_SET);
- }
- else
- {
- lseek(fd, 0, SEEK_SET);
- }
- }
- cur_off = 0;
- while(1)
- {
- if(t->flag & TTAIL_FLAG_DATE_MAX)
- {
- if(i >= t->session->file.off_max.id &&
- cur_off >= t->session->file.off_max.off)
- {
- return;
- }
- if((r = ttail_file_getline(t, i)) < 0)
- {
- break;
- }
- cur_off += r;
- write(out, ttail_file_getline_buf(t),
- strlen(ttail_file_getline_buf(t)));
- }
- else
- {
- r = read(fd, buf, sizeof(buf));
- if(r == -1)
- {
- perror("unable to read file");
- return;
- }
- else if(r == 0)
- {
- break;
- }
- cur_off += r;
- r = write(out, buf, r);
- if(r == -1 || r == 0)
- {
- perror("Unable to write result");
- return;
- }
- }
- }
- }
- }
-
- int _ttail_search_files_binary_search(ttail_t* t, const struct tm* in,
- const struct tm** ftm, short min)
- {
- int cmin, cmax, ret;
- size_t valid;
- int cmpres;
- off_t tmpoff;
- ttail_files_off_t *files_off;
- size_t *id;
- off_t *off;
-
- files_off = min?&(t->session->file.off_min):&(t->session->file.off_max);
- id = &(files_off->id),
- off = &(files_off->off);
-
- *id = *off = 0;
- valid = 0;
- while(*id < t->logfile_sz)
- {
- if(!ftm[*id])
- {
- (*id)++;
- continue;
- }
- valid++;
- cmin = ttail_tm_cmp(&(ftm[*id][0]), in);
- cmax = ttail_tm_cmp(&(ftm[*id][1]), in);
- if(!cmin)
- {
- /* found at the begining of the file */
- return 0;
- }
- else if (cmax == 0)
- {
- /* found at EOF */
- if(min)
- {
- *off = _ttail_from_search_from_end(t, *id, in);
- if(*off < 0)
- {
- *off = 0;
- return -1;
- }
- }
- else
- {
- *off = t->session->file.file_sz[*id];
- }
- return 0;
- }
- else if(cmin > 0)
- {
- /* found at start of file */
- off = 0;
- return 0;
- }
- else if(cmax > 0)
- {
- /* somewhere in current file */
- ret = _ttail_search_file_binary_search(t, in, ftm, min);
- if(ret)
- {
- if(t->verbose > 2)
- {
- fprintf(stderr, "Error while \
- running binary search algorithm in '%s'\n", t->logfile_name[*id]);
- }
- *id = 0;
- return ret;
- }
- /* searching for equivalent */
- tmpoff = *off;
- while(1)
- {
- if(fseek(t->logfile[*id], tmpoff, SEEK_SET) < 0)
- {
- return -1;
- }
- tmpoff = min?
- _ttail_file_start_line(t->logfile[*id]):
- _ttail_file_next_line(t->logfile[*id]);
- if(tmpoff < 0)
- {
- break;
- }
- cmpres=0;
- ret = _ttail_file_cur_cmp(t, *id, in, &cmpres);
- if((min && cmpres < 0) || (!min && cmpres > 0))
- {
- break;
- }
- *off = tmpoff;
- if(min)
- {
- tmpoff--;
- }
- }
- return 0;
- }
- else if(*id == t->logfile_sz - 1 && cmax < 0)
- {
- *off = t->session->file.file_sz[*id];
- return 0;
- }
- (*id)++;
- }
- if(!valid)
- {
- fprintf(stderr, "No files to scan\n");
- return 0;
- }
- /* the answer is somewhere in *id file */
- *off = _ttail_from_search_from_end(t, *id, in);
- return 0;
- }
-
- inline int _ttail_search_file_binary_search(ttail_t* t, const struct tm* in,
- const struct tm** ftm, short min)
- {
- off_t cur, sz, d, prev, tmp;
- int ret, cmpres;
- ttail_files_off_t *files_off;
- size_t id;
- off_t *off;
-
- files_off = min?&(t->session->file.off_min):&(t->session->file.off_max);
- id = files_off->id,
- off = &(files_off->off);
-
- sz = t->session->file.file_sz[id];
- d = cur = sz / 2;
- prev = 0;
- cmpres = 0;
- while(1)
- {
- if(fseek(t->logfile[id], cur, SEEK_SET) < 0)
- {
- perror("Unable to move in the file");
- return -1;
- }
- if(cur > prev)
- {
- cur = _ttail_file_next_line(t->logfile[id]);
- if(cur < 0)
- {
- /* not sure errno is really set */
- perror("Error searching previous line");
- return -1;
- }
- }
- else
- {
- cur = _ttail_file_start_line(t->logfile[id]);
- if(cur < 0)
- {
- /* not sure errno is really set */
- perror("Error searching previous line");
- return -1;
- }
- }
- if(cur == prev)
- {
- *off = cur;
- return 0;
- }
- prev = cur;
- ret = _ttail_file_cur_cmp(t, id, in, &cmpres);
- if(ret < 0)
- {
- if(t->verbose > 2)
- {
- fprintf(stderr, "Error comparing a logline \
- to a date directly from a file\n");
- }
- return -1;
- }
- else if(cmpres == 0)
- {
- *off = cur;
- break;
- }
- else if(cmpres < 0)
- {
- tmp = _ttail_file_next_line(t->logfile[id]);
- ret = _ttail_file_cur_cmp(t, id, in, &cmpres);
- if(cmpres >=0)
- {
- *off = tmp;
- break;
- }
- d/=2;
- cur += d;
- cur %= t->session->file.file_sz[id];
- }
- else
- {
- d/=2;
- cur -= d;
- if(cur < 0)
- {
- cur = 0;
- }
- }
- }
- return 0;
- }
-
- int _ttail_search_closest_files_init(ttail_t* t)
- {
- struct stat stat;
- FILE *fp;
- int fd;
- size_t i;
- off_t *file_sz;
-
- t->session = (ttail_search_t*)malloc(sizeof(ttail_search_file_t));
- if(!t->session)
- {
- perror("Unable to allocate memory for search session");
- goto _ttail_search_closest_files_alloc_session_err;
- }
- memset(t->session, 0, sizeof(ttail_search_file_t));
-
- file_sz = malloc(sizeof(off_t)*t->logfile_sz);
- if(!file_sz)
- {
- perror("Unable to allocate memory for file sizes");
- goto _ttail_search_closest_files_alloc_err;
- }
- t->session->file.file_sz = file_sz;
- #ifdef HUGEFILE
- t->session->file.sz_div = 0;
- #endif
-
- for(i=0;i<t->logfile_sz;i++)
- {
- fp = t->logfile[i];
- if(!fp)
- {
- file_sz[i] = 0;
- continue;
- }
- if((fd = fileno(fp)) < 0)
- {
- perror("Unable to get fp");
- goto _ttail_search_closest_files_err;
- }
- if(fstat(fileno(fp), &stat))
- {
- perror("Unable to get file size");
- goto _ttail_search_closest_files_err;
- }
- file_sz[i] = stat.st_size;
- }
- /* we got all real size, determining if we need a divisor */
- /*
- * not implemented
- */
- if(_ttail_search_closest_files_set_fsizes(t))
- {
- goto _ttail_search_closest_files_err;
- }
-
- t->session->file.buf_sz = 128;
- t->session->file.buf = malloc(t->session->file.buf_sz);
- if(!t->session->file.buf)
- {
- goto _ttail_search_closest_files_err;
- }
-
- return 0;
-
- _ttail_search_closest_files_err:
- free(file_sz);
- t->session->file.file_sz = NULL;
- _ttail_search_closest_files_alloc_err:
- free(t->session);
- _ttail_search_closest_files_alloc_session_err:
- return -1;
- }
-
- int _ttail_search_closest_files_set_fsizes(ttail_t* t)
- {
- size_t i;
- off_t *vfile, *vsz;
- vfile = malloc(sizeof(off_t)*t->logfile_sz);
- if(!vfile)
- {
- perror("Unable to allocate memory for file size sum");
- return -1;
- }
- t->session->file.vfile = vfile;
- vsz = &(t->session->file.vsz);
- *vsz = 0;
- for(i=0; i<t->logfile_sz;i++)
- {
- vfile[i] = *vsz;
- #ifdef HUGEFILE
- *vsz += t->session->file.file_sz[i] >> t->session->file.sz_div;
- #else
- *vsz += t->session->file.file_sz[i];
- #endif
- }
- t->session->file.vpos = 0;
- return 0;
- }
-
- inline int _ttail_file_minmax(ttail_t* t, size_t id, struct tm tm[2])
- {
- FILE *fp;
- long cur;
- memset(tm, 0, sizeof(struct tm)*2);
- fp = t->logfile[id];
- if(!fp)
- {
- return 1;
- }
- if(fseek(fp, 0, SEEK_SET) < 0)
- {
- perror("Unable to manipulate fp");
- return -1;
- }
- while(1)
- {
- if(ttail_file_getline(t, id) < 0)
- {
- return 1;
- }
- if(!ttail_logline2date(t, ttail_file_getline_buf(t), tm))
- {
- break;
- }
- }
- if(fseek(fp, -1, SEEK_END) < 0)
- {
- perror("Unable to manipulate fp");
- return -1;
- }
- while(1)
- {
- if((cur = _ttail_file_start_line(fp)) < 0)
- {
- return -1;
- }
- if(ttail_file_getline(t, id) < 0)
- {
- return 1;
- }
- if(!ttail_logline2date(t, ttail_file_getline_buf(t), tm+1))
- {
- break;
- }
- if(!cur)
- {
- return 1;
- }
- if(fseek(fp, cur-1, SEEK_SET) < 0)
- {
- perror("Unable to manipulate fp");
- return -1;
- }
- }
- return 0;
- }
-
- int _ttail_file_reopen(ttail_t* t, size_t id)
- {
- if(t->logfile[id])
- {
- fclose(t->logfile[id]);
- }
- t->logfile[id] = fopen(t->logfile_name[id], "r");
- if(!t->logfile[id] && t->verbose > 2)
- {
- fprintf(stderr, "Unable to reopen '%s'\n",
- t->logfile_name[id]);
- }
- return t->logfile[id]?0:-1;
- }
-
- inline long _ttail_file_next_line(FILE* f)
- {
- ssize_t s;
- size_t r;
- char *buff;
- long res;
- int c;
-
- r=0;
- buff = NULL;
- s = getline(&buff, &r, f);
- if(s < 0)
- {
- goto _ttail_file_next_line_err;
- }
- while(1)
- {
- c = getc(f);
- if(c == EOF)
- {
- return 0;
- }
- else if(c!='\n')
- {
- if(fseek(f, -1, SEEK_CUR)<0)
- {
- perror("Unable to set position in file");
- goto _ttail_file_next_line_err;
- }
- break;
- }
- }
-
- res = ftell(f);
- free(buff);
- return res;
-
- _ttail_file_next_line_err:
- free(buff);
- return -1;
- }
-
- inline long _ttail_file_start_line(FILE* f)
- {
- #define _STARTLN_BUFFLEN 32
- long res; /* function result */
- long read_beg, cur, last, start;
- int read_sz;
- int c;
-
- if((start = ftell(f)) < 0)
- {
- perror("Unable to get position in file");
- return -1;
- }
- res = 0;
- read_beg = start;
- while(!res && start)
- {
- if(fseek(f, read_beg, SEEK_SET) < 0)
- {
- perror("Unable to set position in file");
- return -1;
- }
- start = read_beg;
- read_sz = start <= _STARTLN_BUFFLEN?start:_STARTLN_BUFFLEN;
- read_beg = start - read_sz;
- if(fseek(f, read_beg, SEEK_SET) < 0)
- {
- perror("Unable to set position in file");
- return -1;
- }
- last = -1; /* last pos we saw a '\n' */
- cur = read_beg;
- while(cur <= start)
- {
- c = getc(f);
- if(c == EOF)
- {
- if(!res)
- {
- return 0;
- }
- break;
- }
- else if (c =='\n')
- {
- last = cur;
- }
- else if(last >= 0)
- {
- res = cur;
- last = -1;
- }
- cur++;
- }
- if(!read_beg)
- {
- break;
- }
- }
- if(fseek(f, res, SEEK_SET) < 0)
- {
- perror("Unable to set position in file");
- return -1;
- }
- return res;
- }
-
- inline off_t _ttail_from_search_from_end(ttail_t* t , size_t id ,
- const struct tm* tm)
- {
- FILE *f;
- struct tm curtm;
- off_t last;
- int ret, cmpret;
- off_t result;
-
- result = -1;
- f = t->logfile[id];
- if(fseek(f, -1, SEEK_END) < 0)
- {
- return -1;
- }
- while(1)
- {
- last = _ttail_file_start_line(f);
- if(last < 0)
- {
- break;
- }
- if(ttail_file_getline(t, id) < 0)
- {
- break;
- }
- ret = ttail_logline2date(t, ttail_file_getline_buf(t), &curtm);
- if(ret < 0)
- {
- break;
- }
- cmpret = ttail_tm_cmp(&curtm, tm);
- if(!ret)
- {
- if(cmpret >= 0)
- {
- /* found but continue to search the first one*/
- result = last;
- }
- else
- {
- if(cmpret < 0 && result < 0)
- {
- result = last;
- }
- break;
- }
- }
- if(last == 0)
- {
- /* considere the begining of the file as the answer */
- return 0;
- }
- if(fseek(f, last-1, SEEK_SET))
- {
- return -1;
- }
- }
- return result;
- }
-
- inline int _ttail_file_off_cmp(ttail_t* t, size_t id, off_t off,
- const struct tm* tm, int *res)
- {
- if(fseek(t->logfile[id], off, SEEK_SET))
- {
- perror("Unable to set position in file");
- return -1;
- }
- if(ttail_file_getline(t, id) < 0)
- {
- perror("Unable to read a line from file");
- return -1;
- }
- return _ttail_file_cur_cmp(t, id, tm, res);
- }
- inline int _ttail_file_cur_cmp(ttail_t* t, size_t id, const struct tm* tm ,
- int* res)
- {
- int ret;
- struct tm ctm;
- if(ttail_file_getline(t, id) < 0)
- {
- return -1;
- }
- ret = ttail_logline2date(t, ttail_file_getline_buf(t), &ctm);
- if(ret)
- {
- return -1;
- }
- else if(ret == 1)
- {
- return 1;
- }
- *res = ttail_tm_cmp(&ctm, tm);
- return 0;
-
- }
-
- void _ttail_search_file_free(ttail_t* t)
- {
- if(!t->session)
- {
- return;
- }
- if(t->session->file.buf)
- {
- free(t->session->file.buf);
- }
- if(t->session->file.file_sz)
- {
- free(t->session->file.file_sz);
- }
- if(t->session->file.vfile)
- {
- free(t->session->file.vfile);
- }
- free(t->session);
- }
|