/* * See the file LICENSE for redistribution information. * * Copyright (c) 1996, 2010 Oracle and/or its affiliates. All rights reserved. * */ /* * This is the entry function of the db_sql command. Db_sql is a * utility program that translates a schema description written in a * SQL Data Definition Language dialect into C code that implements * the schema using Berkeley DB. */ #include #include #include #include #include "db_sql_codegen.h" extern int getopt(int, char *const [], const char *); static int usage(char *); static char * change_extension(char *path, char *extension); static int read_and_parse(FILE *fp); char *progname = "db_sql"; int line_number = 0; int debug = 0; int txnflag = 0; int main(argc,argv) int argc; char **argv; { extern char *optarg; extern int optind; int opt, free_ofilename, free_hfilename; FILE *ifile, *hfile, *ofile, *tfile, *vfile; char *ifilename, *hfilename, *ofilename, *tfilename, *vfilename; ifilename = hfilename = ofilename = tfilename = vfilename = NULL; free_ofilename = free_hfilename = 0; progname = argv[0]; /* parse the command line switches */ while ((opt = getopt(argc, argv, "i:t:o:h:dv:x")) != -1) { switch (opt) { case 'i': /* input file name */ ifilename = optarg; break; case 'h': /* header output file name */ hfilename = optarg; break; case 'o': /* output file name */ ofilename = optarg; break; case 't': /* test code output file name */ tfilename = optarg; break; case 'd': debug = 1; break; case 'v': /* verification code output file name */ vfilename = optarg; break; case 'x': txnflag = 1; break; default: return (usage(0)); } } argc -= optind; argv += optind; if (argc != 0) { fprintf(stderr, "extra argument %s after switch arguments\n", *argv); return (usage(0)); } if (ifilename == NULL) ifile = stdin; else if ((ifile = fopen(ifilename, "r")) == NULL) return (usage(ifilename)); /* if ofilename wasn't given, use ifilename with a .c extension */ if (ofilename == NULL && ifilename != NULL) { ofilename = change_extension(ifilename, "c"); free_ofilename = 1; } if (ofilename == NULL) ofile = stdout; else if ((ofile = fopen(ofilename, "w")) == NULL) return (usage(ofilename)); /* if hfilename wasn't given, use ofilename with a .h extension */ if (hfilename == NULL && ofilename != NULL) { hfilename = change_extension(ofilename, "h"); free_hfilename = 1; } if (hfilename == NULL) hfile = stdout; else if ((hfile = fopen(hfilename, "w")) == NULL) return (usage(hfilename)); /* * if tfile wasn't given, we won't generate the test code. * tfile == null turns off test code generation */ if (tfilename == NULL) tfile = 0; else { if (hfilename == NULL) { fprintf(stderr, "Can't produce test when streaming to stdout\n"); return (usage(0)); } if ((tfile = fopen(tfilename, "w")) == NULL) return (usage(tfilename)); } /* * Verification files are generated for internal testing purposes, * they are similar to the test output file. This functionality is * not targeted at end users, so is not documented. */ if (vfilename == NULL) vfile = 0; else { if (hfilename == NULL) { fprintf(stderr, "Can't produce verify when streaming to stdout\n"); return (usage(0)); } if ((vfile = fopen(vfilename, "w")) == NULL) return (usage(vfilename)); } if (read_and_parse(ifile)) exit(1); generate(hfile, ofile, tfile, vfile, hfilename); /* clean up the allocated memory */ if (free_ofilename) free(ofilename); if (free_hfilename) free(hfilename); return 0; } /* * Scan input buffer for a semicolon that is not in a comment. * Later, this may need to notice quotes as well. */ static char * scan_for_rightmost_semicolon(p) char *p; { static enum scanner_state { IDLE = 0, GOT_SLASH = 1, IN_SLASHSTAR_COMMENT = 2, GOT_STAR = 3, GOT_HYPHEN = 4, IN_HYPHHYPH_COMMENT = 5 } state = IDLE; char *result; result = NULL; if (p == NULL || *p == '\0') return result; do { switch (state) { case IDLE: switch (*p) { case '/': state = GOT_SLASH; break; case '*': state = GOT_STAR; break; case '-': state = GOT_HYPHEN; break; } break; case GOT_SLASH: switch (*p) { case '*': state = IN_SLASHSTAR_COMMENT; break; default: state = IDLE; } break; case IN_SLASHSTAR_COMMENT: switch (*p) { case '*': state = GOT_STAR; break; } break; case GOT_STAR: switch (*p) { case '/': state = IDLE; break; default: state = IN_SLASHSTAR_COMMENT; break; } break; case GOT_HYPHEN: switch (*p) { case '-': state = IN_HYPHHYPH_COMMENT; break; default: state = IDLE; break; } case IN_HYPHHYPH_COMMENT: switch (*p) { case '\n': state = IDLE; break; } break; } if (state == IDLE && *p == ';') result = p; } while (*p++); return result; } /* * read_and_parse reads lines from the input file (containing SQL DDL), * and sends the to the tokenizer and parser. Because of the way the * SQLite tokenizer works, the chunks sent to the tokenizer must * contain a multiple of whole SQL statements -- a partial statement * will produce a syntax error. Therefore, this function splits its * input at semicolons. */ static int read_and_parse(fp) FILE *fp; { size_t line_len, copy_len, collector_len; char *q, *collector, buf[256], *err_msg; collector = 0; collector_len = 0; err_msg = 0; /* line_number is global */ for (line_number = 1; fgets(buf, sizeof(buf), fp) != 0; line_number++) { line_len = strlen(buf); if (1 + strlen(buf) == sizeof(buf)) { fprintf(stderr, "%s: line %d is too long", progname, line_number); return 1; } /* * Does this line contain a semicolon? If so, copy * the line, up to and including its last semicolon, * into collector and parse it. Then reinitialize * collector with the remainer of the line */ if ((q = scan_for_rightmost_semicolon(buf)) != NULL) copy_len = 1 + q - buf; else copy_len = line_len; collector_len += 1 + copy_len; if (collector == NULL) collector = calloc(1, collector_len); else collector = realloc(collector, collector_len); strnconcat(collector, collector_len, buf, copy_len); if (q != 0) { if (do_parse(collector, &err_msg) != 0) { fprintf(stderr, "parsing error at line %d : %s\n", line_number, err_msg); return 1; } collector_len = 1 + line_len - copy_len; collector = realloc(collector, collector_len); memcpy(collector, buf + copy_len, collector_len); assert(collector[collector_len-1] == 0); } } /* * if there's anything after the final semicolon, send it on * to the tokenizer -- it might be a hint comment */ if (collector != 0) { if (strlen(collector) > 0 && do_parse(collector, &err_msg) != 0) { fprintf(stderr, "parsing error at end of file: %s\n", err_msg); return 1; } free (collector); } return 0; } /* * Basename isn't available everywhere, so we have our own version * which works on unix and windows. */ static char * final_component_of(path) char *path; { char *p; p = strrchr(path, '/'); if (p == NULL) p = strrchr(path, '\\'); if (p != NULL) return p + 1; return path; } /* * Return a new pathname in which any existing "extension" (the part * after ".") has been replaced by the given extension. If the * pathname has no extension, the new extension is simply appended. * Returns allocated memory */ static char * change_extension(path, extension) char *path, *extension; { size_t path_len, copy_len; char *p, *copy; const char dot = '.'; /* isolate the final component of the pathname, so that we can * examine it for the presence of a '.' without finding a '.' * in a directory name componenet of the pathname */ p = final_component_of(path); if (*p != 0) p++; /* skip initial char in basename, it could be a dot */ /* * Is there a dot in the basename? If so, then the path has * an extension that we'll elide before adding the new one. */ if (strrchr(p, dot) != 0) { p = strrchr(path, dot); path_len = p - path; } else path_len = strlen(path); copy_len = 2 + path_len + strlen(extension); copy = malloc(copy_len); memcpy(copy, path, path_len); copy[path_len] = 0; /* terminate the string */ strconcat(copy, copy_len, "."); strconcat(copy, copy_len, extension); return copy; } static int usage(char *error_tag) { if (error_tag != 0) perror(error_tag); fprintf(stderr, "\ Usage: %s [-i inputFile] [-h outputHeaderFile] [-o outputFile] \ [-t testOutputFile] [-d] [-v verificationOutputFile] [-x]\n", progname); return (1); }