* selector is a simple command line utility for selection of strings
* with a dynamic pattern-matching.
*
- * Copyright (c) 2009, 2010 Francois Fleuret
+ * Copyright (c) 2009, 2010, 2011, 2012 Francois Fleuret
* Written by Francois Fleuret <francois@fleuret.org>
*
* This file is part of selector.
#include <getopt.h>
#include <limits.h>
-#define VERSION "1.1.2"
+#define VERSION "1.1.6"
#define BUFFER_SIZE 4096
int case_sensitive = 0;
char *title = 0;
int error_flash = 0;
+int upper_caps_makes_case_sensitive = 0;
+int show_long_lines = 0;
+int show_hits = 0;
-int attr_modeline, attr_focus_line, attr_error;
+int attr_modeline, attr_focus_line, attr_error, attr_hits;
/********************************************************************/
void *safe_malloc(size_t n) {
void *p = malloc(n);
- if (!p && n != 0) {
- printf("Can not allocate memory: %s\n", strerror(errno));
+ if(!p && n != 0) {
+ fprintf(stderr,
+ "selector: can not allocate memory: %s\n", strerror(errno));
exit(EXIT_FAILURE);
}
return p;
fprintf(out, " start in regexp mode\n");
fprintf(out, " -a, --case-sensitive\n");
fprintf(out, " start in case sensitive mode\n");
+ fprintf(out, " -j, --show-long-lines\n");
+ fprintf(out, " print a long-line indicator at the end of truncated lines\n");
+ fprintf(out, " -y, --show-hits\n");
+ fprintf(out, " highlight the matching substrings\n");
+ fprintf(out, " -u, --upper-case-makes-case-sensitive\n");
+ fprintf(out, " using an upper case character in the matching string makes\n");
+ fprintf(out, " the matching case-sensitive\n");
fprintf(out, " -m, --monochrome\n");
fprintf(out, " monochrome mode\n");
fprintf(out, " -q, --no-beep\n");
fprintf(out, " make a flash instead of a beep on an edition error\n");
fprintf(out, " --bash\n");
fprintf(out, " setting for bash history search, same as -b -i -d -v -w -l ${HISTSIZE}\n");
- fprintf(out, " --, --rest-are-files\n");
- fprintf(out, " all following arguments are filenames\n");
+ fprintf(out, " -- all following arguments are filenames\n");
fprintf(out, " -t <title>, --title <title>\n");
fprintf(out, " add a title in the modeline\n");
fprintf(out, " -c <colors>, --colors <colors>\n");
/* A quick and dirty hash table */
+#define MAGIC_HASH_MULTIPLIER 387433
+
/* The table itself stores indexes of the strings taken in a char**
table. When a string is added, if it was already in the table, the
new index replaces the previous one. */
const char *new_string, int new_index,
char **strings) {
- unsigned int code = 0;
+ unsigned int code = 0, start;
int k;
/* This is my recipe. I checked, it seems to work (as long as
- hash_table->size is not a multiple of 387433 that should be
- okay) */
+ hash_table->size is not a multiple of MAGIC_HASH_MULTIPLIER that
+ should be okay) */
for(k = 0; new_string[k]; k++) {
- code = code * 387433 + (unsigned int) (new_string[k]);
+ code = code * MAGIC_HASH_MULTIPLIER + (unsigned int) (new_string[k]);
}
code = code % hash_table->size;
+ start = code;
while(hash_table->entries[code] >= 0) {
/* There is a string with that code */
/* This collision was not the same string, let's move to the next
in the table */
code = (code + 1) % hash_table->size;
+ /* We came back to our original code, which means that the table
+ is full */
+ if(code == start) {
+ fprintf(stderr,
+ "Full hash table (that should not happen)\n");
+ exit(EXIT_FAILURE);
+ }
}
/* This string was not already in there, store the index in the
A matcher matches either with a collection of substrings, or with a
regexp */
-typedef struct {
+struct matcher {
regex_t preg;
int regexp_error;
int nb_patterns;
int case_sensitive;
char *splitted_patterns, **patterns;
-} matcher_t;
+};
+
+/* Routine to add an interval to a sorted list of intervals
+ extremities. Returns the resulting number of extremities.
+
+ This routine is an effing nightmare */
+
+int add_interval(int n, int *switches, int start, int end) {
+ int f, g, k;
+
+ if(start == end) { return n; }
+
+ f = 0;
+ while(f < n && switches[f] <= start) { f++; }
+ g = f;
+ while(g < n && switches[g] <= end) { g++; }
+
+ if(f == n) {
+ /* switches[n-1] start end */
+ /* XXXXXXXXXXXX| */
+ switches[f] = start;
+ switches[f+1] = end;
+ return n + 2;
+ }
+
+ if(f % 2) {
+
+ if(g % 2) {
+ /* switches[f-1] start switches[f] switches[g-1] end switches[g] */
+ /* |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX| ... |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX| */
+ for(k = f; k < n; k++) { switches[k] = switches[k + (g - f)]; }
+ return n - (g - f);
+ } else {
+ /* switches[f-1] start switches[f] switches[g-1] end switches[g] */
+ /* |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX| ... XXXXXXXXXXXX| |XXXXXXXXXX */
+ switches[g - 1] = end;
+ for(k = f; k < n; k++) { switches[k] = switches[k + ((g - 1) - f)]; }
+ return n - ((g - 1) - f);
+ }
+
+ } else {
+
+ if(f == g) {
+ /* switches[f-1] start end switches[f] */
+ /* XXXXXXXXXXXX| |XXXXXXXXXX */
+ for(k = n - 1; k >= f; k--) {
+ switches[k + 2] = switches[k];
+ }
+ switches[f] = start;
+ switches[f + 1] = end;
+ return n + 2;
+ }
+
+ if(g % 2) {
+ /* switches[f-1] start switches[f] switches[g-1] end switches[g] */
+ /* XXXXXXXXXXXX| |XXXXXXXXXX ... |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX| */
+ switches[f] = start;
+ for(k = f + 1; k < n; k++) { switches[k] = switches[k + (g - (f + 1))]; }
+ return n - (g - (f + 1));
+ } else {
+ /* switches[f-1] start switches[f] switches[g-1] end switches[g] */
+ /* XXXXXXXXXXXX| |XXXXXXXXXX ... XXXXXXXXXXXX| |XXXXXXXXXX */
+ switches[f] = start;
+ switches[g - 1] = end;
+ for(k = f + 1; k < n; k++) { switches[k] = switches[k + ((g - 1) - (f + 1))]; }
+ return n - ((g - 1) - (f + 1));
+ }
+ }
+}
-int match(char *string, matcher_t *matcher) {
+int match(struct matcher *matcher, char *string, int *nb_switches, int *switches) {
int n;
+ char *where;
+ regmatch_t matches;
+
+ if(nb_switches) { *nb_switches = 0; }
+
if(matcher->nb_patterns >= 0) {
if(matcher->case_sensitive) {
for(n = 0; n < matcher->nb_patterns; n++) {
- if(strstr(string, matcher->patterns[n]) == 0) return 0;
+ if((where = strstr(string, matcher->patterns[n])) == 0) return 0;
+ if(switches) {
+ *nb_switches = add_interval(*nb_switches, switches,
+ (int) (where - string),
+ (int) (where - string) + strlen(matcher->patterns[n]));
+ }
}
} else {
for(n = 0; n < matcher->nb_patterns; n++) {
- if(strcasestr(string, matcher->patterns[n]) == 0) return 0;
+ if((where = strcasestr(string, matcher->patterns[n])) == 0) return 0;
+ if(switches) {
+ *nb_switches = add_interval(*nb_switches, switches,
+ (int) (where - string),
+ (int) (where - string) + strlen(matcher->patterns[n]));
+ }
}
}
return 1;
} else {
- return regexec(&matcher->preg, string, 0, 0, 0) == 0;
+ if(switches) {
+ if(regexec(&matcher->preg, string, 1, &matches, 0) == 0) {
+ *nb_switches = 2;
+ switches[0] = matches.rm_so;
+ switches[1] = matches.rm_eo;
+ return 1;
+ } else {
+ return 0;
+ }
+ } else {
+ return regexec(&matcher->preg, string, 0, 0, 0) == 0;
+ }
}
}
-void free_matcher(matcher_t *matcher) {
+void free_matcher(struct matcher *matcher) {
if(matcher->nb_patterns < 0) {
if(!matcher->regexp_error) regfree(&matcher->preg);
} else {
}
}
-void initialize_matcher(int use_regexp, int case_sensitive,
- matcher_t *matcher, const char *pattern) {
+void initialize_matcher(struct matcher *matcher,
+ int use_regexp, int case_sensitive,
+ const char *pattern) {
const char *s;
char *t, *last_pattern_start;
int n;
if(use_regexp) {
+ matcher->case_sensitive = case_sensitive;
matcher->nb_patterns = -1;
matcher->regexp_error = regcomp(&matcher->preg, pattern,
case_sensitive ? 0 : REG_ICASE);
} else {
matcher->regexp_error = 0;
matcher->nb_patterns = 1;
+
+ if(upper_caps_makes_case_sensitive) {
+ for(s = pattern; *s && !case_sensitive; s++) {
+ case_sensitive = (*s >= 'A' && *s <= 'Z');
+ }
+ }
+
matcher->case_sensitive = case_sensitive;
for(s = pattern; *s; s++) {
/*********************************************************************/
-int previous_visible(int current_line, char **lines, matcher_t *matcher) {
+int previous_visible(int current_line, char **lines, struct matcher *matcher) {
int line = current_line - 1;
- while(line >= 0 && !match(lines[line], matcher)) line--;
+ while(line >= 0 && !match(matcher, lines[line], 0, 0)) line--;
return line;
}
int next_visible(int current_line, int nb_lines, char **lines,
- matcher_t *matcher) {
+ struct matcher *matcher) {
int line = current_line + 1;
- while(line < nb_lines && !match(lines[line], matcher)) line++;
+ while(line < nb_lines && !match(matcher, lines[line], 0, 0)) line++;
if(line < nb_lines)
return line;
/*********************************************************************/
+void print_string_with_switches(char *buffer, int line_width,
+ int nb_patterns, int *switches) {
+ int w, current = 0, next;
+ if(switches) {
+ for(w = 0; w < nb_patterns && switches[2 * w] < line_width; w++) {
+ if(switches[2 * w] < switches[2 * w + 1]) {
+ next = switches[2 * w];
+ if(next > line_width) { next = line_width; }
+ if(next > current) { addnstr(buffer + current, next - current); }
+ attron(attr_hits);
+ current = next;
+ next = switches[2 * w + 1];
+ if(next > line_width) { next = line_width; }
+ if(next > current) { addnstr(buffer + current, next - current); }
+ attroff(attr_hits);
+ current = next;
+ }
+ }
+ if(current < line_width) {
+ addnstr(buffer + current, line_width - current);
+ }
+ } else {
+ addnstr(buffer, line_width);
+ }
+}
+
/* The line highlighted is the first one matching the matcher in that
order: (1) current_focus_line after motion, if it does not match,
then (2) the first with a greater index, if none matches, then (3)
int nb_lines, char **lines,
int cursor_position,
char *pattern) {
-
+ int *switches;
char buffer[BUFFER_SIZE];
- matcher_t matcher;
+ struct matcher matcher;
int k, l, m;
int console_width, console_height;
int nb_printed_lines = 0;
int cursor_x;
+ int nb_switches;
+
+ initialize_matcher(&matcher, use_regexp, case_sensitive, pattern);
- initialize_matcher(use_regexp, case_sensitive, &matcher, pattern);
+ if(show_hits) {
+ if(matcher.nb_patterns >= 0) {
+ switches = safe_malloc(sizeof(int) * matcher.nb_patterns * 2);
+ } else {
+ switches = safe_malloc(sizeof(int) * 2);
+ }
+ } else {
+ switches = 0;
+ }
console_width = getmaxx(stdscr);
console_height = getmaxy(stdscr);
else if(nb_lines > 0) {
int new_focus_line;
- if(match(lines[*current_focus_line], &matcher)) {
+ if(match(&matcher, lines[*current_focus_line], 0, 0)) {
new_focus_line = *current_focus_line;
} else {
new_focus_line = next_visible(*current_focus_line, nb_lines, lines,
if(first_line > 0) {
first_line--;
- while(first_line > 0 && !match(lines[first_line], &matcher)) {
+ while(first_line > 0 && !match(&matcher, lines[first_line], 0, 0)) {
first_line--;
}
- if(match(lines[first_line], &matcher)) {
+ if(match(&matcher, lines[first_line], 0, 0)) {
nb_match++;
}
}
if(nb_match < console_height - 1 && last_line < nb_lines - 1) {
last_line++;
- while(last_line < nb_lines - 1 && !match(lines[last_line], &matcher)) {
+ while(last_line < nb_lines - 1 && !match(&matcher, lines[last_line], 0, 0)) {
last_line++;
}
- if(match(lines[last_line], &matcher)) {
+ if(match(&matcher, lines[last_line], 0, 0)) {
nb_match++;
}
}
/* Now we display them */
for(l = first_line; l <= last_line; l++) {
- if(match(lines[l], &matcher)) {
+ if(match(&matcher, lines[l], &nb_switches, switches)) {
int k = 0;
- while(lines[l][k] && k < BUFFER_SIZE - 2 && k < console_width - 2) {
+ while(lines[l][k] && k < BUFFER_SIZE - 2 && k < console_width) {
buffer[k] = lines[l][k];
k++;
}
- /* We fill the rest of the line with blanks if this is the
- highlighted line */
-
- if(l == new_focus_line) {
- while(k < console_width) {
- buffer[k++] = ' ';
- }
- }
-
- buffer[k++] = '\n';
- buffer[k++] = '\0';
-
- clrtoeol();
-
/* Highlight the highlighted line ... */
if(l == new_focus_line) {
- attron(attr_focus_line);
- addnstr(buffer, console_width);
- attroff(attr_focus_line);
+ if(show_long_lines && k >= console_width) {
+ attron(attr_focus_line);
+ print_string_with_switches(buffer, console_width-1,
+ nb_switches / 2, switches);
+ /* attron(attr_error); */
+ addnstr("\\", 1);
+ /* attroff(attr_error); */
+ attroff(attr_focus_line);
+ } else {
+ while(k < console_width) {
+ buffer[k++] = ' ';
+ }
+ attron(attr_focus_line);
+ print_string_with_switches(buffer, k,
+ nb_switches / 2, switches);
+ attroff(attr_focus_line);
+ }
} else {
- addnstr(buffer, console_width);
+ if(show_long_lines && k >= console_width) {
+ print_string_with_switches(buffer, console_width-1,
+ nb_switches / 2, switches);
+ attron(attr_focus_line);
+ addnstr("\\", 1);
+ attroff(attr_focus_line);
+ } else {
+ buffer[k++] = '\n';
+ buffer[k++] = '\0';
+
+ print_string_with_switches(buffer, k,
+ nb_switches / 2, switches);
+ }
}
nb_printed_lines++;
/* Add a few info about the mode we are in (regexp and/or case
sensitive) */
- if(use_regexp || case_sensitive) {
+ if(use_regexp || matcher.case_sensitive) {
addstr(" [");
if(use_regexp) {
addstr("regexp");
}
- if(case_sensitive) {
+ if(matcher.case_sensitive) {
if(use_regexp) {
addstr(",");
}
/* We are done */
refresh();
+ if(switches) { free(switches); }
free_matcher(&matcher);
}
int nb_lines_max, int *nb_lines, char **lines) {
char raw_line[BUFFER_SIZE];
- int start, end, eol, k;
+ char *s;
FILE *file;
file = fopen(input_filename, "r");
exit(EXIT_FAILURE);
}
- start = 0;
- end = 0;
-
- while(*nb_lines < nb_lines_max && (end > start || !feof(file))) {
- eol = start;
-
- /* Look for the end of a line in what is already in the buffer */
- while(eol < end && raw_line[eol] != '\n') eol++;
-
- /* if we did not find the of a line, move what has not been
- processed and is in the buffer to the beginning of the buffer,
- fill the buffer with new data from the file, and look for the
- end of a line */
- if(eol == end) {
- for(k = 0; k < end - start; k++) {
- raw_line[k] = raw_line[k + start];
- }
- end -= start;
- eol -= start;
- start = 0;
- end += fread(raw_line + end, sizeof(char), BUFFER_SIZE - end, file);
- while(eol < end && raw_line[eol] != '\n') eol++;
- }
-
- /* The end of the line is the buffer size, which means the line is
- too long */
-
- if(eol == BUFFER_SIZE) {
- raw_line[BUFFER_SIZE - 1] = '\0';
- fprintf(stderr, "selector: Line too long (max is %d characters):\n",
- BUFFER_SIZE);
- fprintf(stderr, raw_line);
- fprintf(stderr, "\n");
- exit(EXIT_FAILURE);
+ while(*nb_lines < nb_lines_max && fgets(raw_line, BUFFER_SIZE, file)) {
+ for(s = raw_line + strlen(raw_line) - 1; s > raw_line && *s == '\n'; s--) {
+ *s = '\0';
}
-
- /* If we got a line, we replace the carriage return by a \0 to
- finish the string */
-
- raw_line[eol] = '\0';
-
- store_line(hash_table, raw_line + start,
- nb_lines, lines);
-
- start = eol + 1;
+ store_line(hash_table, raw_line, nb_lines, lines);
}
fclose(file);
{ "remove-duplicates", no_argument, 0, 'd' },
{ "regexp", no_argument, 0, 'e' },
{ "case-sensitive", no_argument, 0, 'a' },
+ { "show-long-lines", no_argument, 0, 'j'},
+ { "show-hits", no_argument, 0, 'j'},
+ { "upper-case-makes-case-sensitive", no_argument, 0, 'u' },
{ "title", 1, 0, 't' },
{ "number-of-lines", 1, 0, 'l' },
{ "colors", 1, 0, 'c' },
- { "rest-are-files", no_argument, 0, '-' },
{ "bash", no_argument, 0, OPT_BASH_MODE },
{ "help", no_argument, 0, 'h' },
{ 0, 0, 0, 0 }
char pattern[BUFFER_SIZE];
int c, k, l, n;
int cursor_position;
- int error = 0, show_help = 0;
- int rest_are_files = 0;
+ int error = 0, show_help = 0, done = 0;
int key;
int current_focus_line, displayed_focus_line;
strcpy(output_filename, "");
- while (!rest_are_files &&
- (c = getopt_long(argc, argv, "o:s:x:vwmqf:ibzdeat:l:c:-h",
+ while ((c = getopt_long(argc, argv, "o:s:x:vwmqf:ibzdeajyunt:l:c:-h",
long_options, NULL)) != -1) {
switch(c) {
case_sensitive = 1;
break;
+ case 'j':
+ show_long_lines = 1;
+ break;
+
+ case 'y':
+ show_hits = 1;
+ break;
+
+ case 'u':
+ upper_caps_makes_case_sensitive = 1;
+ break;
+
case 't':
free(title);
title = safe_malloc((strlen(optarg) + 1) * sizeof(char));
color_bg_highlight = colors[3];
break;
- case '-':
- rest_are_files = 1;
- break;
-
case 'h':
show_help = 1;
break;
}
}
- if(show_help || error) {
- if(error) {
- usage(stderr);
- exit(EXIT_FAILURE);
- } else {
- usage(stdout);
- exit(EXIT_SUCCESS);
- }
+ if(error) {
+ usage(stderr);
+ exit(EXIT_FAILURE);
+ }
+
+ if(show_help) {
+ usage(stdout);
+ exit(EXIT_SUCCESS);
}
lines = safe_malloc(nb_lines_max * sizeof(char *));
attr_error = A_STANDOUT;
attr_modeline = A_REVERSE;
attr_focus_line = A_STANDOUT;
+ attr_hits = A_BOLD;
if(with_colors && has_colors()) {
clear();
}
+ else if(key == '\007' || /* ^G */
+ key == '\033' || /* ^[ (escape) */
+ key == '\n' ||
+ key == KEY_ENTER) {
+ done = 1;
+ }
+
+ else if(key == KEY_RESIZE || key == -1) {
+ /* Do nothing when the tty is resized */
+ }
+
+ else {
+ /* Unknown key */
+ error_feedback();
+ }
+
update_screen(¤t_focus_line, &displayed_focus_line,
motion,
nb_lines, labels, cursor_position, pattern);
- } while(key != '\007' && /* ^G */
- key != '\033' && /* ^[ (escape) */
- key != '\n' &&
- key != KEY_ENTER);
+ } while(!done);
echo();
endwin();
/* Here we come back to standard display */
- if((key == KEY_ENTER || key == '\n')) {
+ if(key == KEY_ENTER || key == '\n') {
char *t;
FILE *out = fopen(output_filename, "w");
if(out) {
if(t) {
- fprintf(out, t);
+ fprintf(out, "%s", t);
}
fprintf(out, "\n");
} else {
projects / selector.git / blobdiff