2020-09-07 23:22:43 -07:00
|
|
|
/*
|
2020-09-10 02:02:40 -07:00
|
|
|
* bpeg.c - Source code for the bpeg parser
|
2020-09-07 23:22:43 -07:00
|
|
|
*
|
2020-09-13 23:31:38 -07:00
|
|
|
* See `man ./bpeg.1` for more details
|
2020-09-07 23:22:43 -07:00
|
|
|
*/
|
2020-09-11 01:28:06 -07:00
|
|
|
#include <fcntl.h>
|
2020-09-13 15:55:09 -07:00
|
|
|
#include <glob.h>
|
2020-09-12 15:49:51 -07:00
|
|
|
#include <limits.h>
|
2020-09-11 01:28:06 -07:00
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <unistd.h>
|
|
|
|
|
|
|
|
#include "compiler.h"
|
|
|
|
#include "grammar.h"
|
|
|
|
#include "utils.h"
|
|
|
|
#include "vm.h"
|
|
|
|
|
|
|
|
static const char *usage = (
|
2020-09-14 12:39:31 -07:00
|
|
|
"BPEG - a Parsing Expression Grammar command line tool\n\n"
|
2020-09-11 01:28:06 -07:00
|
|
|
"Usage:\n"
|
|
|
|
" bpeg [flags] <pattern> [<input files>...]\n\n"
|
|
|
|
"Flags:\n"
|
2020-09-14 12:39:31 -07:00
|
|
|
" -h --help print the usage and quit\n"
|
|
|
|
" -v --verbose print verbose debugging info\n"
|
|
|
|
" -i --ignore-case preform matching case-insensitively\n"
|
|
|
|
" -d --define <name>=<def> define a grammar rule\n"
|
|
|
|
" -D --define-string <name>=<def> define a grammar rule (string-pattern)\n"
|
|
|
|
" -p --pattern <pat> provide a pattern (equivalent to bpeg '\\(<pat>)')\n"
|
|
|
|
" -P --pattern-string <pat> provide a string pattern (may be useful if '<pat>' begins with a '-')\n"
|
|
|
|
" -r --replace <replacement> replace the input pattern with the given replacement\n"
|
|
|
|
" -m --mode <mode> set the behavior mode (defult: find-all)\n"
|
|
|
|
" -g --grammar <grammar file> use the specified file as a grammar\n");
|
2020-09-07 23:05:38 -07:00
|
|
|
|
2020-09-11 02:55:15 -07:00
|
|
|
static char *getflag(const char *flag, char *argv[], int *i)
|
|
|
|
{
|
|
|
|
size_t n = strlen(flag);
|
2020-09-14 12:39:31 -07:00
|
|
|
check(argv[*i], "Attempt to get flag from NULL argument");
|
2020-09-11 02:55:15 -07:00
|
|
|
if (strncmp(argv[*i], flag, n) == 0) {
|
|
|
|
if (argv[*i][n] == '=') {
|
|
|
|
return &argv[*i][n+1];
|
|
|
|
} else if (argv[*i][n] == '\0') {
|
2020-09-14 12:39:31 -07:00
|
|
|
check(argv[*i+1], "Expected argument after '%s'\n\n%s", flag, usage);
|
2020-09-11 02:55:15 -07:00
|
|
|
++(*i);
|
|
|
|
return argv[*i];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
2020-09-13 15:55:09 -07:00
|
|
|
|
2020-09-14 12:16:15 -07:00
|
|
|
static int run_match(grammar_t *g, const char *filename, vm_op_t *pattern, unsigned int flags)
|
2020-09-13 15:55:09 -07:00
|
|
|
{
|
|
|
|
char *input;
|
|
|
|
if (filename == NULL || streq(filename, "-")) {
|
|
|
|
input = readfile(STDIN_FILENO);
|
|
|
|
} else {
|
|
|
|
int fd = open(filename, O_RDONLY);
|
|
|
|
check(fd >= 0, "Couldn't open file: %s", filename);
|
|
|
|
input = readfile(fd);
|
|
|
|
}
|
2020-09-14 12:16:15 -07:00
|
|
|
match_t *m = match(g, input, pattern, flags);
|
2020-09-13 15:55:09 -07:00
|
|
|
if (m != NULL && m->end > m->start + 1) {
|
2020-09-13 20:33:11 -07:00
|
|
|
if (filename != NULL) {
|
|
|
|
if (isatty(STDOUT_FILENO)) printf("\033[1;4;33m%s\033[0m\n", filename);
|
|
|
|
else printf("%s\n", filename);
|
|
|
|
}
|
2020-09-14 12:16:15 -07:00
|
|
|
print_match(m, isatty(STDOUT_FILENO) ? "\033[0m" : NULL, (flags & BPEG_VERBOSE) != 0);
|
2020-09-13 15:55:09 -07:00
|
|
|
freefile(input);
|
|
|
|
return 0;
|
|
|
|
} else {
|
|
|
|
freefile(input);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-09-11 02:55:15 -07:00
|
|
|
#define FLAG(f) (flag=getflag((f), argv, &i))
|
2020-09-09 22:29:09 -07:00
|
|
|
|
|
|
|
int main(int argc, char *argv[])
|
|
|
|
{
|
2020-09-14 12:16:15 -07:00
|
|
|
unsigned int flags = 0;
|
2020-09-11 02:55:15 -07:00
|
|
|
char *flag = NULL;
|
2020-09-13 23:31:38 -07:00
|
|
|
char path[PATH_MAX] = {0};
|
2020-09-11 02:55:15 -07:00
|
|
|
const char *rule = "find-all";
|
2020-09-09 22:29:09 -07:00
|
|
|
|
2020-09-11 01:28:06 -07:00
|
|
|
grammar_t *g = new_grammar();
|
|
|
|
|
2020-09-13 23:31:38 -07:00
|
|
|
// Load builtins:
|
2020-09-12 15:11:44 -07:00
|
|
|
int fd;
|
|
|
|
if ((fd=open("/etc/xdg/bpeg/builtins.bpeg", O_RDONLY)) >= 0)
|
2020-09-12 20:05:55 -07:00
|
|
|
load_grammar(g, readfile(fd)); // Keep in memory for debugging output
|
2020-09-12 15:49:51 -07:00
|
|
|
sprintf(path, "%s/.config/bpeg/builtins.bpeg", getenv("HOME"));
|
|
|
|
if ((fd=open(path, O_RDONLY)) >= 0)
|
2020-09-12 20:05:55 -07:00
|
|
|
load_grammar(g, readfile(fd)); // Keep in memory for debugging output
|
2020-09-12 15:11:44 -07:00
|
|
|
|
2020-09-11 02:55:15 -07:00
|
|
|
int i, npatterns = 0;
|
2020-09-14 12:39:31 -07:00
|
|
|
check(argc > 1, "%s", usage);
|
2020-09-11 02:55:15 -07:00
|
|
|
for (i = 1; i < argc; i++) {
|
|
|
|
if (streq(argv[i], "--")) {
|
|
|
|
++i;
|
|
|
|
break;
|
2020-09-13 23:31:38 -07:00
|
|
|
} else if (streq(argv[i], "--help") || streq(argv[i], "-h")) {
|
2020-09-09 22:29:09 -07:00
|
|
|
printf("%s\n", usage);
|
|
|
|
return 0;
|
2020-09-13 23:31:38 -07:00
|
|
|
} else if (streq(argv[i], "--verbose") || streq(argv[i], "-v")) {
|
2020-09-14 12:16:15 -07:00
|
|
|
flags |= BPEG_VERBOSE;
|
|
|
|
} else if (streq(argv[i], "--ignore-case") || streq(argv[i], "-i")) {
|
|
|
|
flags |= BPEG_IGNORECASE;
|
2020-09-11 02:55:15 -07:00
|
|
|
} else if (FLAG("--replace") || FLAG("-r")) {
|
|
|
|
vm_op_t *p = bpeg_replacement(bpeg_pattern("pattern"), flag);
|
|
|
|
check(p, "Replacement failed to compile");
|
|
|
|
add_def(g, flag, "replacement", p);
|
|
|
|
rule = "replace-all";
|
|
|
|
} else if (FLAG("--grammar") || FLAG("-g")) {
|
2020-09-12 15:11:44 -07:00
|
|
|
int fd;
|
2020-09-12 15:49:51 -07:00
|
|
|
if (streq(flag, "-")) {
|
2020-09-12 15:11:44 -07:00
|
|
|
fd = STDIN_FILENO;
|
2020-09-11 02:55:15 -07:00
|
|
|
} else {
|
2020-09-12 15:49:51 -07:00
|
|
|
fd = open(flag, O_RDONLY);
|
|
|
|
if (fd < 0) {
|
|
|
|
sprintf(path, "%s/.config/bpeg/%s.bpeg", getenv("HOME"), flag);
|
|
|
|
fd = open(path, O_RDONLY);
|
|
|
|
}
|
|
|
|
if (fd < 0) {
|
|
|
|
sprintf(path, "/etc/xdg/bpeg/%s.bpeg", flag);
|
|
|
|
fd = open(path, O_RDONLY);
|
|
|
|
}
|
|
|
|
check(fd >= 0, "Couldn't find grammar: %s", flag);
|
2020-09-11 02:55:15 -07:00
|
|
|
}
|
2020-09-12 20:05:55 -07:00
|
|
|
load_grammar(g, readfile(fd)); // Keep in memory for debug output
|
2020-09-11 02:55:15 -07:00
|
|
|
} else if (FLAG("--define") || FLAG("-d")) {
|
|
|
|
char *def = flag;
|
2020-09-10 22:50:49 -07:00
|
|
|
char *eq = strchr(def, '=');
|
2020-09-14 12:39:31 -07:00
|
|
|
check(eq, "Rule definitions must include an '='\n\n%s", usage);
|
2020-09-10 22:50:49 -07:00
|
|
|
*eq = '\0';
|
2020-09-11 01:28:06 -07:00
|
|
|
char *src = ++eq;
|
|
|
|
vm_op_t *pat = bpeg_pattern(src);
|
|
|
|
check(pat, "Failed to compile pattern");
|
|
|
|
add_def(g, src, def, pat);
|
2020-09-13 20:33:11 -07:00
|
|
|
} else if (FLAG("--define-string") || FLAG("-D")) {
|
|
|
|
char *def = flag;
|
|
|
|
char *eq = strchr(def, '=');
|
2020-09-14 12:39:31 -07:00
|
|
|
check(eq, "Rule definitions must include an '='\n\n%s", usage);
|
2020-09-13 20:33:11 -07:00
|
|
|
*eq = '\0';
|
|
|
|
char *src = ++eq;
|
|
|
|
vm_op_t *pat = bpeg_stringpattern(src);
|
|
|
|
check(pat, "Failed to compile pattern");
|
|
|
|
add_def(g, src, def, pat);
|
2020-09-13 23:31:38 -07:00
|
|
|
} else if (FLAG("--pattern") || FLAG("-p")) {
|
2020-09-12 18:20:13 -07:00
|
|
|
check(npatterns == 0, "Cannot define multiple patterns");
|
2020-09-12 20:05:55 -07:00
|
|
|
vm_op_t *p = bpeg_pattern(flag);
|
|
|
|
check(p, "Pattern failed to compile: '%s'", flag);
|
|
|
|
add_def(g, flag, "pattern", p);
|
|
|
|
++npatterns;
|
2020-09-13 23:31:38 -07:00
|
|
|
} else if (FLAG("--pattern-string") || FLAG("-P")) {
|
2020-09-12 20:05:55 -07:00
|
|
|
vm_op_t *p = bpeg_stringpattern(flag);
|
2020-09-12 18:20:13 -07:00
|
|
|
check(p, "Pattern failed to compile");
|
2020-09-12 20:05:55 -07:00
|
|
|
add_def(g, flag, "pattern", p);
|
2020-09-12 18:20:13 -07:00
|
|
|
++npatterns;
|
2020-09-13 00:37:17 -07:00
|
|
|
} else if (FLAG("--mode") || FLAG("-m")) {
|
|
|
|
rule = flag;
|
2020-09-11 02:55:15 -07:00
|
|
|
} else if (argv[i][0] != '-') {
|
|
|
|
if (npatterns > 0) break;
|
|
|
|
vm_op_t *p = bpeg_stringpattern(argv[i]);
|
|
|
|
check(p, "Pattern failed to compile");
|
|
|
|
add_def(g, argv[i], "pattern", p);
|
|
|
|
++npatterns;
|
2020-09-12 20:05:55 -07:00
|
|
|
} else {
|
2020-09-14 12:39:31 -07:00
|
|
|
printf("Unrecognized flag: %s\n\n%s\n", argv[i], usage);
|
2020-09-12 20:05:55 -07:00
|
|
|
return 1;
|
2020-09-09 22:29:09 -07:00
|
|
|
}
|
|
|
|
}
|
2020-09-07 23:05:38 -07:00
|
|
|
|
2020-09-11 02:55:15 -07:00
|
|
|
vm_op_t *pattern = lookup(g, rule);
|
2020-09-12 01:32:59 -07:00
|
|
|
check(pattern != NULL, "No such rule: '%s'", rule);
|
2020-09-09 22:29:09 -07:00
|
|
|
|
2020-09-13 15:55:09 -07:00
|
|
|
int ret = 0;
|
|
|
|
if (i < argc) {
|
|
|
|
// Files pass in as command line args:
|
|
|
|
for (int nfiles = 0; i < argc; nfiles++, i++) {
|
2020-09-14 12:16:15 -07:00
|
|
|
ret |= run_match(g, argv[i], pattern, flags);
|
2020-09-11 02:55:15 -07:00
|
|
|
}
|
2020-09-13 15:55:09 -07:00
|
|
|
} else if (isatty(STDIN_FILENO)) {
|
|
|
|
// No files, no piped in input, so use * **/*:
|
|
|
|
glob_t globbuf;
|
|
|
|
glob("*", 0, NULL, &globbuf);
|
|
|
|
glob("**/*", GLOB_APPEND, NULL, &globbuf);
|
|
|
|
for (size_t i = 0; i < globbuf.gl_pathc; i++) {
|
2020-09-14 12:16:15 -07:00
|
|
|
ret |= run_match(g, globbuf.gl_pathv[i], pattern, flags);
|
2020-09-11 02:55:15 -07:00
|
|
|
}
|
2020-09-13 15:55:09 -07:00
|
|
|
globfree(&globbuf);
|
|
|
|
} else {
|
|
|
|
// Piped in input:
|
2020-09-14 12:16:15 -07:00
|
|
|
ret |= run_match(g, NULL, pattern, flags);
|
2020-09-07 23:05:38 -07:00
|
|
|
}
|
|
|
|
|
2020-09-13 15:55:09 -07:00
|
|
|
|
|
|
|
return ret;
|
2020-09-07 23:05:38 -07:00
|
|
|
}
|
2020-09-07 23:34:41 -07:00
|
|
|
|
2020-09-11 01:38:44 -07:00
|
|
|
// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1
|