3 * selector is a simple command line utility for selection of strings
4 * with a dynamic pattern-matching.
6 * Copyright (c) 2009 Francois Fleuret
7 * Written by Francois Fleuret <francois@fleuret.org>
9 * This file is part of selector.
11 * selector is free software: you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License version 3 as
13 * published by the Free Software Foundation.
15 * selector is distributed in the hope that it will be useful, but
16 * WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with selector. If not, see <http://www.gnu.org/licenses/>.
25 // To use it as a super-history-search for bash:
26 // alias h='selector -d -i -b -v -f <(history)'
36 #include <sys/ioctl.h>
44 const int buffer_size = 4096;
46 // Yeah, global variables!
48 int nb_lines_max = 1000;
49 char pattern_separator = ';';
50 int output_to_vt_buffer = 0;
52 int zsh_history = 0, bash_history = 0;
53 int inverse_order = 0;
54 int remove_duplicates = 0;
56 int case_sensitive = 0;
60 #define COLOR_MODELINE 1
61 #define COLOR_HIGHLIGHTED_LINE 2
63 //////////////////////////////////////////////////////////////////////
65 void inject_into_tty_buffer(char *string) {
66 struct termios oldtio, newtio;
67 tcgetattr(STDIN_FILENO, &oldtio);
68 memset(&newtio, 0, sizeof(newtio));
69 // Set input mode (non-canonical, *no echo*,...)
70 tcsetattr(STDIN_FILENO, TCSANOW, &newtio);
71 // Put the selected string in the tty input buffer
72 for(char *k = string; *k; k++) {
73 ioctl(STDIN_FILENO, TIOCSTI, k);
75 // Restore the old settings
76 tcsetattr(STDIN_FILENO, TCSANOW, &oldtio);
79 //////////////////////////////////////////////////////////////////////
81 void check_opt(int argc, char **argv, int n_opt, int n, const char *help) {
82 if(n_opt + n >= argc) {
83 cerr << "Missing argument for " << argv[n_opt] << "."
85 << "Expecting " << help << "."
91 int string_to_positive_integer(char *string) {
96 for(char *s = string; *s; s++) {
97 if(*s >= '0' && *s <= '9') {
98 result = result * 10 + int(*s - '0');
104 cerr << "Value `" << string << "' is not a positive integer." << endl;
111 void error_feedback() {
119 //////////////////////////////////////////////////////////////////////
120 // A quick and dirty hash table
122 // The table itself stores index of the strings in a char
123 // **table. When a string is added, if it was already in the table,
124 // the new index replaces the previous one.
126 int *new_hash_table(int hash_table_size) {
128 result = new int[hash_table_size];
129 for(int k = 0; k < hash_table_size; k++) {
135 // Adds new_string in the table, associated to new_index. If this
136 // string was not already in the table, returns -1. Otherwise, returns
137 // the previous index it had.
139 int test_and_add(char *new_string, int new_index,
140 char **strings, int *hash_table, int hash_table_size) {
142 unsigned int code = 0;
144 // This is my recipe. I checked, it seems to work (as long as
145 // hash_table_size is not a multiple of 387433 that should be okay)
147 for(int k = 0; new_string[k]; k++) {
148 code = code * 387433 + (unsigned int) (new_string[k]);
151 code = code % hash_table_size;
153 while(hash_table[code] >= 0) {
154 // There is a string with that code
155 if(strcmp(new_string, strings[hash_table[code]]) == 0) {
156 // It is the same string, we keep a copy of the stored index
157 int result = hash_table[code];
159 hash_table[code] = new_index;
160 // And return the previous one
163 // This collision was not the same string, let's move to the next
165 code = (code + 1) % hash_table_size;
168 // This string was not already in there, store the index in the
169 // table and return -1
170 hash_table[code] = new_index;
174 //////////////////////////////////////////////////////////////////////
175 // A matcher matches either with a collection of substrings, or with a
183 char *splitted_patterns, **patterns;
186 int match(char *string, matcher_t *matcher) {
187 if(matcher->nb_patterns >= 0) {
188 if(matcher->case_sensitive) {
189 for(int n = 0; n < matcher->nb_patterns; n++) {
190 if(strstr(string, matcher->patterns[n]) == 0) return 0;
193 for(int n = 0; n < matcher->nb_patterns; n++) {
194 if(strcasestr(string, matcher->patterns[n]) == 0) return 0;
199 return regexec(&matcher->preg, string, 0, 0, 0) == 0;
203 void free_matcher(matcher_t *matcher) {
204 if(matcher->nb_patterns < 0) {
205 if(!matcher->regexp_error) regfree(&matcher->preg);
207 delete[] matcher->splitted_patterns;
208 delete[] matcher->patterns;
212 void initialize_matcher(int use_regexp, int case_sensitive,
213 matcher_t *matcher, const char *pattern) {
216 matcher->nb_patterns = -1;
217 matcher->regexp_error = regcomp(&matcher->preg, pattern, case_sensitive ? 0 : REG_ICASE);
219 matcher->regexp_error = 0;
220 matcher->nb_patterns = 1;
221 matcher->case_sensitive = case_sensitive;
223 for(const char *s = pattern; *s; s++) {
224 if(*s == pattern_separator) {
225 matcher->nb_patterns++;
229 matcher->splitted_patterns = new char[strlen(pattern) + 1];
230 matcher->patterns = new char*[matcher->nb_patterns];
232 strcpy(matcher->splitted_patterns, pattern);
235 char *last_pattern_start = matcher->splitted_patterns;
236 for(char *s = matcher->splitted_patterns; n < matcher->nb_patterns; s++) {
237 if(*s == pattern_separator || *s == '\0') {
239 matcher->patterns[n++] = last_pattern_start;
240 last_pattern_start = s + 1;
246 //////////////////////////////////////////////////////////////////////
249 void delete_char(char *buffer, int *position) {
250 if(buffer[*position]) {
252 while(c < buffer_size && buffer[c]) {
253 buffer[c] = buffer[c+1];
256 } else error_feedback();
259 void backspace_char(char *buffer, int *position) {
261 if(buffer[*position]) {
262 int c = *position - 1;
264 buffer[c] = buffer[c+1];
268 buffer[*position - 1] = '\0';
272 } else error_feedback();
275 void insert_char(char *buffer, int *position, char character) {
276 if(strlen(buffer) < buffer_size - 1) {
278 char t = buffer[c], u;
287 buffer[(*position)++] = character;
288 } else error_feedback();
291 void kill_before_cursor(char *buffer, int *position) {
293 while(buffer[*position + s]) {
294 buffer[s] = buffer[*position + s];
301 void kill_after_cursor(char *buffer, int *position) {
302 buffer[*position] = '\0';
305 //////////////////////////////////////////////////////////////////////
307 int previous_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
308 int line = current_line - 1;
309 while(line >= 0 && !match(lines[line], matcher)) line--;
313 int next_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
314 int line = current_line + 1;
315 while(line < nb_lines && !match(lines[line], matcher)) line++;
323 //////////////////////////////////////////////////////////////////////
325 void update_screen(int *current_line, int *temporary_line, int motion,
326 int nb_lines, char **lines,
330 char buffer[buffer_size];
333 initialize_matcher(use_regexp, case_sensitive, &matcher, pattern);
335 // We now take care of printing the lines per se
337 int console_width = getmaxx(stdscr);
338 int console_height = getmaxy(stdscr);
340 // First, we find a visible line. In priority: The current, or the
341 // first visible after it, or the first visible before it.
343 int nb_printed_lines = 0;
346 use_default_colors();
349 if(matcher.regexp_error) {
350 addstr("[regexp error]");
351 } else if(nb_lines > 0) {
353 if(match(lines[*current_line], &matcher)) {
354 new_line = *current_line;
356 new_line = next_visible(*current_line, nb_lines, lines, &matcher);
358 new_line = previous_visible(*current_line, nb_lines, lines, &matcher);
362 // If we found a visible line and we should move, let's move
364 if(new_line >= 0 && motion != 0) {
367 // We want to go down, let's find the first visible line below
368 for(int m = 0; l >= 0 && m < motion; m++) {
369 l = next_visible(l, nb_lines, lines, &matcher);
375 // We want to go up, let's find the first visible line above
376 for(int m = 0; l >= 0 && m < -motion; m++) {
377 l = previous_visible(l, nb_lines, lines, &matcher);
385 // Here new_line is either a line number matching the patterns, or -1
389 int first_line = new_line, last_line = new_line, nb_match = 1;
391 // We find the first and last line to show, so that the total of
392 // visible lines between them (them included) is console_height-1
394 while(nb_match < console_height-1 && (first_line > 0 || last_line < nb_lines - 1)) {
398 while(first_line > 0 && !match(lines[first_line], &matcher)) {
401 if(match(lines[first_line], &matcher)) {
406 if(nb_match < console_height - 1 && last_line < nb_lines - 1) {
408 while(last_line < nb_lines - 1 && !match(lines[last_line], &matcher)) {
412 if(match(lines[last_line], &matcher)) {
418 // Now we display them
420 for(int l = first_line; l <= last_line; l++) {
421 if(match(lines[l], &matcher)) {
424 while(lines[l][k] && k < buffer_size - 2 && k < console_width - 2) {
425 buffer[k] = lines[l][k];
429 // We fill the rest of the line with blanks if this is the
433 while(k < console_width) {
441 // Highlight the highlighted line ...
445 attron(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
446 addnstr(buffer, console_width);
447 attroff(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
450 addnstr(buffer, console_width);
454 addnstr(buffer, console_width);
462 *current_line = new_line;
466 *temporary_line = new_line;
468 if(nb_printed_lines == 0) {
469 addnstr("[no selection]\n", console_width);
472 addnstr("[empty choice]\n", console_width);
480 attron(COLOR_PAIR(COLOR_MODELINE));
485 for(int k = 0; k < console_width; k++) buffer[k] = ' ';
486 buffer[console_width] = '\0';
487 addnstr(buffer, console_width);
491 // There must be a more elegant way of moving the cursor at a
492 // location met during display
499 cursor_x += strlen(title) + 1;
502 sprintf(buffer, "%d/%d ", nb_printed_lines, nb_lines);
504 cursor_x += strlen(buffer);
506 addnstr(pattern, cursor_position);
507 cursor_x += cursor_position;
509 if(pattern[cursor_position]) {
510 addstr(pattern + cursor_position);
515 if(use_regexp || case_sensitive) {
533 attroff(COLOR_PAIR(COLOR_MODELINE));
541 free_matcher(&matcher);
544 //////////////////////////////////////////////////////////////////////
546 void read_file(const char *input_filename,
547 int nb_lines_max, int *nb_lines, char **lines,
548 int hash_table_size, int *hash_table) {
550 char buffer[buffer_size], raw_line[buffer_size];;
552 ifstream file(input_filename);
555 cerr << "Can not open " << input_filename << endl;
559 while(*nb_lines < nb_lines_max && !file.eof()) {
561 file.getline(raw_line, buffer_size);
566 cerr << "Line too long:" << endl;
567 cerr << raw_line << endl;
576 // Remove the zsh history prefix
578 if(zsh_history && *t == ':') {
579 while(*t && *t != ';') s++;
583 // Remove the bash history prefix
586 while(*t == ' ') t++;
587 while(*t >= '0' && *t <= '9') t++;
588 while(*t == ' ') t++;
591 // Copy the string while transforming the ctrl characters into
592 // printable characters
598 while(*u) { *s++ = *u++; }
602 // Check for duplicates with the hash table and insert the line
603 // in the list if necessary
608 dup = test_and_add(buffer, *nb_lines, lines, hash_table, hash_table_size);
614 lines[*nb_lines] = new char[strlen(buffer) + 1];
615 strcpy(lines[*nb_lines], buffer);
617 // The string was already in there, so we do not allocate a
618 // new string but use the pointer to the first occurence of it
619 lines[*nb_lines] = lines[dup];
628 //////////////////////////////////////////////////////////////////////
630 int main(int argc, char **argv) {
632 if(!ttyname(STDIN_FILENO)) {
633 cerr << "The standard input is not a tty." << endl;
637 int color_fg_modeline, color_bg_modeline;
638 int color_fg_highlight, color_bg_highlight;
640 color_fg_modeline = COLOR_WHITE;
641 color_bg_modeline = COLOR_BLACK;
642 color_fg_highlight = COLOR_BLACK;
643 color_bg_highlight = COLOR_YELLOW;
645 setlocale(LC_ALL, "");
647 char input_filename[buffer_size], output_filename[buffer_size];
649 strcpy(input_filename, "");
650 strcpy(output_filename, "");
653 int error = 0, show_help = 0;
654 int rest_are_files = 0;
656 while(!error && !show_help && i < argc && argv[i][0] == '-' && !rest_are_files) {
658 if(strcmp(argv[i], "-o") == 0) {
659 check_opt(argc, argv, i, 1, "<output filename>");
660 strncpy(output_filename, argv[i+1], buffer_size);
664 else if(strcmp(argv[i], "-s") == 0) {
665 check_opt(argc, argv, i, 1, "<pattern separator>");
666 pattern_separator = argv[i+1][0];
670 else if(strcmp(argv[i], "-v") == 0) {
671 output_to_vt_buffer = 1;
675 else if(strcmp(argv[i], "-m") == 0) {
680 else if(strcmp(argv[i], "-q") == 0) {
685 else if(strcmp(argv[i], "-f") == 0) {
686 check_opt(argc, argv, i, 1, "<input filename>");
687 strncpy(input_filename, argv[i+1], buffer_size);
691 else if(strcmp(argv[i], "-i") == 0) {
696 else if(strcmp(argv[i], "-b") == 0) {
701 else if(strcmp(argv[i], "-z") == 0) {
706 else if(strcmp(argv[i], "-d") == 0) {
707 remove_duplicates = 1;
711 else if(strcmp(argv[i], "-e") == 0) {
716 else if(strcmp(argv[i], "-a") == 0) {
721 else if(strcmp(argv[i], "-t") == 0) {
722 check_opt(argc, argv, i, 1, "<title>");
724 title = new char[strlen(argv[i+1]) + 1];
725 strcpy(title, argv[i+1]);
729 else if(strcmp(argv[i], "-l") == 0) {
730 check_opt(argc, argv, i, 1, "<maximum number of lines>");
731 nb_lines_max = string_to_positive_integer(argv[i+1]);
735 else if(strcmp(argv[i], "-c") == 0) {
736 check_opt(argc, argv, i, 4, "<fg modeline> <bg modeline> <fg highlight> <bg highlight>");
737 color_fg_modeline = string_to_positive_integer(argv[i + 1]);
738 color_bg_modeline = string_to_positive_integer(argv[i + 2]);
739 color_fg_highlight = string_to_positive_integer(argv[i + 3]);
740 color_bg_highlight = string_to_positive_integer(argv[i + 4]);
744 else if(strcmp(argv[i], "--") == 0) {
749 else if(strcmp(argv[i], "-h") == 0) {
755 cerr << "Unknown option " << argv[i] << "." << endl;
760 if(show_help || error) {
761 cerr << "Selector version " << VERSION << "-R" << REVISION_NUMBER
763 << "Written by Francois Fleuret <francois@fleuret.org>."
766 << "Usage: " << argv[0] << " [options] [<filename1> [<filename2> ...]]" << endl
768 << " -h show this help" << endl
769 << " -v inject the selected line in the tty" << endl
770 << " -d remove duplicated lines" << endl
771 << " -b remove the bash history line prefix" << endl
772 << " -z remove the zsh history line prefix" << endl
773 << " -i invert the order of lines" << endl
774 << " -e start in regexp mode" << endl
775 << " -a case sensitive" << endl
776 << " -m monochrome mode" << endl
777 << " -q make a flash instead of a beep on an edition error" << endl
778 << " -- rest of the arguments are filenames" << endl
779 << " -t <title>" << endl
780 << " add a title in the modeline" << endl
781 << " -c <fg modeline> <bg modeline> <fg highlight> <bg highlight>" << endl
782 << " set the display colors" << endl
783 << " -o <output filename>" << endl
784 << " set a file to write the selected line to" << endl
785 << " -s <pattern separator>" << endl
786 << " set the symbol to separate substrings in the pattern" << endl
787 << " -l <max number of lines>" << endl
788 << " set the maximum number of lines to take into account" << endl
794 char **lines = new char *[nb_lines_max];
797 int hash_table_size = nb_lines_max * 10;
800 if(remove_duplicates) {
801 hash_table = new_hash_table(hash_table_size);
804 if(input_filename[0]) {
805 read_file(input_filename,
806 nb_lines_max, &nb_lines, lines,
807 hash_table_size, hash_table);
812 nb_lines_max, &nb_lines, lines,
813 hash_table_size, hash_table);
819 // Now remove the null strings
822 for(int k = 0; k < nb_lines; k++) {
824 lines[n++] = lines[k];
831 for(int i = 0; i < nb_lines / 2; i++) {
832 char *s = lines[nb_lines - 1 - i];
833 lines[nb_lines - 1 - i] = lines[i];
838 char pattern[buffer_size];
843 //////////////////////////////////////////////////////////////////////
844 // Here we start to display with curse
850 // So that the arrow keys work
851 keypad(stdscr, TRUE);
859 if(color_fg_modeline < 0 || color_fg_modeline >= COLORS ||
860 color_bg_modeline < 0 || color_bg_modeline >= COLORS ||
861 color_fg_highlight < 0 || color_bg_highlight >= COLORS ||
862 color_bg_highlight < 0 || color_bg_highlight >= COLORS) {
865 cerr << "Color numbers have to be between 0 and " << COLORS - 1 << "." << endl;
869 init_pair(COLOR_MODELINE, color_fg_modeline, color_bg_modeline);
870 init_pair(COLOR_HIGHLIGHTED_LINE, color_fg_highlight, color_bg_highlight);
878 int current_line = 0, temporary_line = 0;
880 update_screen(¤t_line, &temporary_line, 0, nb_lines, lines, cursor_position, pattern);
888 if(key >= ' ' && key <= '~') { // Insert character
889 insert_char(pattern, &cursor_position, key);
892 else if(key == KEY_BACKSPACE ||
893 key == '\010' || // ^H
894 key == '\177') { // ^?
895 backspace_char(pattern, &cursor_position);
898 else if(key == KEY_DC ||
899 key == '\004') { // ^D
900 delete_char(pattern, &cursor_position);
903 else if(key == KEY_HOME) {
907 else if(key == KEY_END) {
908 current_line = nb_lines - 1;
911 else if(key == KEY_NPAGE) {
915 else if(key == KEY_PPAGE) {
919 else if(key == KEY_DOWN ||
920 key == '\016') { // ^N
924 else if(key == KEY_UP ||
925 key == '\020') { // ^P
929 else if(key == KEY_LEFT ||
930 key == '\002') { // ^B
931 if(cursor_position > 0) cursor_position--;
932 else error_feedback();
935 else if(key == KEY_RIGHT ||
936 key == '\006') { // ^F
937 if(pattern[cursor_position]) cursor_position++;
938 else error_feedback();
941 else if(key == '\001') { // ^A
945 else if(key == '\005') { // ^E
946 cursor_position = strlen(pattern);
949 else if(key == '\022') { // ^R
950 use_regexp = !use_regexp;
953 else if(key == '\011') { // ^I
954 case_sensitive = !case_sensitive;
957 else if(key == '\025') { // ^U
958 kill_before_cursor(pattern, &cursor_position);
961 else if(key == '\013') { // ^K
962 kill_after_cursor(pattern, &cursor_position);
965 update_screen(¤t_line, &temporary_line, motion,
966 nb_lines, lines, cursor_position, pattern);
968 } while(key != '\007' && // ^G
969 key != '\033' && // ^[ (escape)
976 //////////////////////////////////////////////////////////////////////
977 // Here we come back to standard display
979 if((key == KEY_ENTER || key == '\n')) {
981 if(output_to_vt_buffer) {
982 if(temporary_line >= 0 && temporary_line < nb_lines) {
983 inject_into_tty_buffer(lines[temporary_line]);
987 if(output_filename[0]) {
988 ofstream out(output_filename);
990 cerr << "Can not open " << output_filename << " for writing." << endl;
993 if(temporary_line >= 0 && temporary_line < nb_lines) {
994 out << lines[temporary_line] << endl;
1002 cout << "Aborted." << endl;
1005 for(int l = 0; l < nb_lines; l++) {