/* * bpeg.c - Source code for the bpeg parser * * See `man ./bp.1` for more details */ #include #include #include #include #include #include #include #include "compiler.h" #include "file_loader.h" #include "grammar.h" #include "json.h" #include "utils.h" #include "viz.h" #include "vm.h" static const char *usage = ( "BP - a Parsing Expression Grammar command line tool\n\n" "Usage:\n" " bp [flags] [...]\n\n" "Flags:\n" " -h --help print the usage and quit\n" " -v --verbose print verbose debugging info\n" " -e --explain explain the matches\n" " -i --ignore-case preform matching case-insensitively\n" " -d --define : define a grammar rule\n" " -D --define-string : define a grammar rule (string-pattern)\n" " -p --pattern provide a pattern (equivalent to bp '\\()')\n" " -P --pattern-string provide a string pattern (may be useful if '' begins with a '-')\n" " -r --replace replace the input pattern with the given replacement\n" " -m --mode set the behavior mode (defult: find-all)\n" " -g --grammar use the specified file as a grammar\n"); static print_options_t print_options = 0; static char *getflag(const char *flag, char *argv[], int *i) { size_t n = strlen(flag); check(argv[*i], "Attempt to get flag from NULL argument"); if (strncmp(argv[*i], flag, n) == 0) { if (argv[*i][n] == '=') { return &argv[*i][n+1]; } else if (argv[*i][n] == '\0') { check(argv[*i+1], "Expected argument after '%s'\n\n%s", flag, usage); ++(*i); return argv[*i]; } } return NULL; } static int print_errors(file_t *f, match_t *m) { int ret = 0; if (m->op->op == VM_CAPTURE && m->value.name && streq(m->value.name, "!")) { printf("\033[31;1m"); print_match(f, m, print_options); printf("\033[0m\n"); fprint_line(stdout, f, m->start, m->end, " "); return 1; } if (m->child) ret += print_errors(f, m->child); if (m->nextsibling) ret += print_errors(f, m->nextsibling); return ret; } static int run_match(grammar_t *g, const char *filename, vm_op_t *pattern, unsigned int flags) { static int printed_matches = 0; file_t *f = load_file(filename); check(f, "Could not open file: %s", filename); match_t *m = match(g, f, f->contents, pattern, flags); if (m && print_errors(f, m) > 0) _exit(1); if (m != NULL && m->end > m->start + 1) { ++printed_matches; if (flags & BPEG_EXPLAIN) { if (filename) printf("\033[1;4m%s\033[0m\n", filename); visualize_match(m); } else if (flags & BPEG_JSON) { if (printed_matches > 1) fprintf(stdout, ",\n"); printf("{\"filename\":\"%s\",", filename ? filename : "-"); printf("\"tree\":{\"rule\":\"text\",\"start\":%d,\"end\":%ld,\"children\":[", 0, f->end - f->contents); json_match(f->contents, m, (flags & BPEG_VERBOSE) ? 1 : 0); printf("]}}\n"); } else { if (printed_matches > 1) fputc('\n', stdout); if (filename) { if (print_options & PRINT_COLOR) printf("\033[1;4;33m%s\033[0m\n", filename); else printf("%s:\n", filename); } print_match(f, m, print_options); } destroy_file(&f); return 0; } else { destroy_file(&f); return 1; } } #define FLAG(f) (flag=getflag((f), argv, &i)) int main(int argc, char *argv[]) { unsigned int flags = 0; char *flag = NULL; char path[PATH_MAX] = {0}; const char *rule = "find-all"; grammar_t *g = new_grammar(); // Load builtins: if (access("/etc/xdg/bp/builtins.bp", R_OK) != -1) load_grammar(g, load_file("/etc/xdg/bp/builtins.bp")); // Keep in memory for debugging output sprintf(path, "%s/.config/bp/builtins.bp", getenv("HOME")); if (access(path, R_OK) != -1) load_grammar(g, load_file(path)); // Keep in memory for debugging output int i, npatterns = 0; check(argc > 1, "%s", usage); for (i = 1; i < argc; i++) { if (streq(argv[i], "--")) { ++i; break; } else if (streq(argv[i], "--help") || streq(argv[i], "-h")) { printf("%s\n", usage); return 0; } else if (streq(argv[i], "--verbose") || streq(argv[i], "-v")) { flags |= BPEG_VERBOSE; } else if (streq(argv[i], "--explain") || streq(argv[i], "-e")) { flags |= BPEG_EXPLAIN; } else if (streq(argv[i], "--json") || streq(argv[i], "-j")) { flags |= BPEG_JSON; } else if (streq(argv[i], "--ignore-case") || streq(argv[i], "-i")) { flags |= BPEG_IGNORECASE; } else if (FLAG("--replace") || FLAG("-r")) { file_t *replace_file = spoof_file("", flag); vm_op_t *patref = bpeg_pattern(replace_file, "pattern"); vm_op_t *rep = bpeg_replacement(replace_file, patref, flag); check(rep, "Replacement failed to compile: %s", flag); add_def(g, replace_file, flag, "replacement", rep); rule = "replace-all"; } else if (FLAG("--grammar") || FLAG("-g")) { file_t *f = load_file(flag); if (f == NULL) { sprintf(path, "%s/.config/bp/%s.bp", getenv("HOME"), flag); f = load_file(path); } if (f == NULL) { sprintf(path, "/etc/xdg/bp/%s.bp", flag); f = load_file(path); } check(f != NULL, "Couldn't find grammar: %s", flag); load_grammar(g, f); // Keep in memory for debug output } else if (FLAG("--define") || FLAG("-d")) { char *def = flag; char *eq = strchr(def, ':'); check(eq, "Rule definitions must include an ':'\n\n%s", usage); *eq = '\0'; char *src = ++eq; file_t *def_file = spoof_file(def, flag); vm_op_t *pat = bpeg_pattern(def_file, src); check(pat, "Failed to compile pattern: %s", flag); add_def(g, def_file, src, def, pat); } else if (FLAG("--define-string") || FLAG("-D")) { char *def = flag; char *eq = strchr(def, ':'); check(eq, "Rule definitions must include an ':'\n\n%s", usage); *eq = '\0'; char *src = ++eq; file_t *def_file = spoof_file(def, flag); vm_op_t *pat = bpeg_stringpattern(def_file, src); check(pat, "Failed to compile pattern: %s", flag); add_def(g, def_file, src, def, pat); } else if (FLAG("--pattern") || FLAG("-p")) { check(npatterns == 0, "Cannot define multiple patterns"); file_t *arg_file = spoof_file("", flag); vm_op_t *p = bpeg_pattern(arg_file, flag); check(p, "Pattern failed to compile: %s", flag); add_def(g, arg_file, flag, "pattern", p); ++npatterns; } else if (FLAG("--pattern-string") || FLAG("-P")) { file_t *arg_file = spoof_file("", flag); vm_op_t *p = bpeg_stringpattern(arg_file, flag); check(p, "Pattern failed to compile: %s", flag); add_def(g, arg_file, flag, "pattern", p); ++npatterns; } else if (FLAG("--mode") || FLAG("-m")) { rule = flag; } else if (argv[i][0] != '-') { if (npatterns > 0) break; file_t *arg_file = spoof_file("", argv[i]); vm_op_t *p = bpeg_stringpattern(arg_file, argv[i]); check(p, "Pattern failed to compile: %s", argv[i]); add_def(g, arg_file, argv[i], "pattern", p); ++npatterns; } else { printf("Unrecognized flag: %s\n\n%s\n", argv[i], usage); return 1; } } if (isatty(STDOUT_FILENO)) { print_options |= PRINT_COLOR | PRINT_LINE_NUMBERS; } vm_op_t *pattern = lookup(g, rule); check(pattern != NULL, "No such rule: '%s'", rule); int ret = 1; if (flags & BPEG_JSON) printf("["); if (i < argc) { // Files pass in as command line args: for (int nfiles = 0; i < argc; nfiles++, i++) { ret &= run_match(g, argv[i], pattern, flags); } } 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++) { ret &= run_match(g, globbuf.gl_pathv[i], pattern, flags); } globfree(&globbuf); } else { // Piped in input: ret &= run_match(g, NULL, pattern, flags); } if (flags & BPEG_JSON) printf("]\n"); return ret; } // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1