changeset 226:28441920fb35

beginning of number database version 2
author Mychaela Falconia <falcon@freecalypso.org>
date Sun, 13 Aug 2023 22:01:25 -0800
parents 243ed87880a1
children a349ae9d90fa
files .hgignore include/number_db_v2.h libutil/Makefile libutil/digit_groups.c utils/Makefile utils/themwi-update-numdb2.c
diffstat 6 files changed, 559 insertions(+), 3 deletions(-) [+]
line wrap: on
line diff
--- a/.hgignore	Sat Aug 05 12:52:33 2023 -0800
+++ b/.hgignore	Sun Aug 13 22:01:25 2023 -0800
@@ -31,4 +31,5 @@
 ^utils/themwi-dump-numdb$
 ^utils/themwi-short-dial$
 ^utils/themwi-update-numdb$
+^utils/themwi-update-numdb2$
 ^utils/themwi-update-outrt$
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/include/number_db_v2.h	Sun Aug 13 22:01:25 2023 -0800
@@ -0,0 +1,36 @@
+/*
+ * This header file defines version 2 of ThemWi number database
+ * binary file format.
+ */
+
+struct numdb_file_hdr {
+	uint32_t	owned_number_count;
+	uint32_t	short_number_count;
+};
+
+struct owned_number_rec {
+	uint16_t	number[3];	/* NPA-exchange-suffix breakdown */
+	uint8_t		number_flags;	/* properties of outside number */
+	uint8_t		usage;		/* usage inside ThemWi */
+	uint16_t	remap[3];	/* secondary remap number */
+};
+
+#define	NUMBER_FLAG_SMSPROV	0x01	/* provisioned for outside SMS */
+#define	NUMBER_FLAG_E911PROV	0x02	/* provisioned for E911 */
+
+#define	NUMBER_USAGE_MASK		0x0F
+#define	NUMBER_USAGE_TYPE_RSVD		0x00
+#define	NUMBER_USAGE_TYPE_GSM_SUB	0x01
+#define	NUMBER_USAGE_TYPE_ALIAS		0x02
+#define	NUMBER_USAGE_FLAG_E911_VIA	0x10
+
+struct short_number_rec {
+	uint16_t	short_num;
+	uint8_t		short_num_type;
+	uint8_t		fullnum_flags;
+	uint16_t	fullnum_prefix[2];
+};
+
+#define	SHORT_NUM_TYPE_ABBREV		0x01
+#define	SHORT_NUM_TYPE_ITN		0x02
+#define	SHORT_NUM_TYPE_TEST_SINK	0x03
--- a/libutil/Makefile	Sat Aug 05 12:52:33 2023 -0800
+++ b/libutil/Makefile	Sun Aug 13 22:01:25 2023 -0800
@@ -1,7 +1,8 @@
 CC=	gcc
 CFLAGS=	-O2
-OBJS=	bitfunc.o crc8gen.o dtmf_valid.o extdigits.o imsi_entry.o mncc_debug.o \
-	mncc_utils.o nanp_valid.o numstring.o sockinit.o tfo_msg_enc.o
+OBJS=	bitfunc.o crc8gen.o digit_groups.o dtmf_valid.o extdigits.o \
+	imsi_entry.o mncc_debug.o mncc_utils.o nanp_valid.o numstring.o \
+	sockinit.o tfo_msg_enc.o
 LIB=	libutil.a
 
 all:	${LIB}
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/libutil/digit_groups.c	Sun Aug 13 22:01:25 2023 -0800
@@ -0,0 +1,29 @@
+/*
+ * In version 2 of ThemWi owned number database, NANP numbers are stored as 3
+ * uint16_t words: NPA, exchange and prefix, each uint16_t encoding a group of
+ * 3 or 4 digits of the full telephone number.  This library module provides
+ * functions for turning groups of 3 or 4 digits into uint16_t words.
+ */
+
+digits3_to_uint16(str)
+	char *str;
+{
+	int acc;
+
+	acc = (str[0] - '0') * 100;
+	acc += (str[1] - '0') * 10;
+	acc += str[2] - '0';
+	return acc;
+}
+
+digits4_to_uint16(str)
+	char *str;
+{
+	int acc;
+
+	acc = (str[0] - '0') * 1000;
+	acc += (str[1] - '0') * 100;
+	acc += (str[2] - '0') * 10;
+	acc += str[3] - '0';
+	return acc;
+}
--- a/utils/Makefile	Sat Aug 05 12:52:33 2023 -0800
+++ b/utils/Makefile	Sun Aug 13 22:01:25 2023 -0800
@@ -2,7 +2,7 @@
 CFLAGS=	-O2
 PROGS=	sip-out-test sip-rx-test sip-udp-dump themwi-check-own \
 	themwi-dump-numdb themwi-short-dial themwi-update-numdb \
-	themwi-update-outrt
+	themwi-update-numdb2 themwi-update-outrt
 NOINST=	rtp-alloc-test smpp-test1 smpp-test2 tcpserv-dump
 LIBNUMDB=../libnumdb/libnumdb.a
 LIBRTPA=../librtpalloc/librtpalloc.a
@@ -45,6 +45,9 @@
 themwi-update-numdb:	themwi-update-numdb.o ${LIBUTIL}
 	${CC} ${CFLAGS} -o $@ $@.o ${LIBUTIL}
 
+themwi-update-numdb2:	themwi-update-numdb2.o ${LIBUTIL}
+	${CC} ${CFLAGS} -o $@ $@.o ${LIBUTIL}
+
 themwi-update-outrt:	themwi-update-outrt.o ${LIBUTIL}
 	${CC} ${CFLAGS} -o $@ $@.o ${LIBUTIL}
 
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/utils/themwi-update-numdb2.c	Sun Aug 13 22:01:25 2023 -0800
@@ -0,0 +1,486 @@
+/*
+ * This program reads (parses) ThemWi config file /var/gsm/number-db2,
+ * generates the compiled binary form of this database, and then makes
+ * it live via atomic rename.
+ */
+
+#include <ctype.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <strings.h>
+#include <unistd.h>
+#include "../include/number_db_v2.h"
+
+#define	MAX_OWNED_NUMBERS	1000
+#define	MAX_SHORT_NUMBERS	1000
+
+static struct owned_number_rec owned_number_buf[MAX_OWNED_NUMBERS];
+static struct short_number_rec short_number_buf[MAX_SHORT_NUMBERS];
+static unsigned owned_number_count, short_number_count;
+
+static char *system_dir;
+static FILE *inf;
+static int lineno;
+static char linebuf[256];
+static int prefix_set, prefix_allows_abbrev;
+static uint16_t prefix_buf[2];
+
+static void
+enter_owned_number(rec)
+	struct owned_number_rec *rec;
+{
+	if (owned_number_count >= MAX_OWNED_NUMBERS) {
+		fprintf(stderr, "error: MAX_OWNED_NUMBERS exceeded\n");
+		exit(1);
+	}
+	bcopy(rec, owned_number_buf + owned_number_count,
+		sizeof(struct owned_number_rec));
+	owned_number_count++;
+}
+
+static void
+enter_short_number(rec)
+	struct short_number_rec *rec;
+{
+	if (short_number_count >= MAX_SHORT_NUMBERS) {
+		fprintf(stderr, "error: MAX_SHORT_NUMBERS exceeded\n");
+		exit(1);
+	}
+	bcopy(rec, short_number_buf + short_number_count,
+		sizeof(struct short_number_rec));
+	short_number_count++;
+}
+
+static void
+handle_prefix_line(cp)
+	char *cp;
+{
+	char *np, prefix[7];
+
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (grok_number_string(np, 1) != 6) {
+		fprintf(stderr,
+		      "number-db2 line %d: prefix requires 6-digit argument\n",
+			lineno);
+		exit(1);
+	}
+	dehyphen_number_string(np, prefix);
+	if (!is_nanp_valid_prefix(prefix)) {
+		fprintf(stderr,
+			"number-db2 line %d: prefix violates NANP rules\n",
+			lineno);
+		exit(1);
+	}
+	prefix_buf[0] = digits3_to_uint16(prefix);
+	prefix_buf[1] = digits3_to_uint16(prefix + 3);
+	prefix_set = 1;
+	while (isspace(*cp))
+		cp++;
+	if (*cp == '\0' || *cp == '#') {
+		prefix_allows_abbrev = 0;
+		return;
+	}
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (!strcmp(np, "allow-abbrev")) {
+		prefix_allows_abbrev = 1;
+		return;
+	}
+	fprintf(stderr,
+	    "number-db2 line %d: non-understood notation \"%s\" after prefix\n",
+		lineno, np);
+	exit(1);
+}
+
+static int
+parse_extra_number(rec, numstr)
+	struct owned_number_rec *rec;
+	char *numstr;
+{
+	char buf[11];
+	int rc;
+
+	rc = grok_number_string(numstr, 1);
+	switch (rc) {
+	case 4:
+		if (!prefix_set)
+			return(-1);
+		dehyphen_number_string(numstr, buf);
+		rec->remap[0] = prefix_buf[0];
+		rec->remap[1] = prefix_buf[1];
+		rec->remap[2] = digits4_to_uint16(buf);
+		return(0);
+	case 10:
+		dehyphen_number_string(numstr, buf);
+		rec->remap[0] = digits3_to_uint16(buf);
+		rec->remap[1] = digits3_to_uint16(buf + 3);
+		rec->remap[2] = digits4_to_uint16(buf + 6);
+		return(0);
+	default:
+		return(-1);
+	}
+}
+
+static void
+handle_number_attr(rec, tail)
+	struct owned_number_rec *rec;
+	char *tail;
+{
+	char *cp, *np;
+	int rc;
+
+	for (cp = tail; ; ) {
+		while (isspace(*cp))
+			cp++;
+		if (*cp == '\0' || *cp == '#')
+			return;
+		for (np = cp; *cp && !isspace(*cp); cp++)
+			;
+		if (*cp)
+			*cp++ = '\0';
+		if (!strcmp(np, "sms"))
+			rec->number_flags |= NUMBER_FLAG_SMSPROV;
+		else if (!strcmp(np, "e911"))
+			rec->number_flags |= NUMBER_FLAG_E911PROV;
+		else if (!strcmp(np, "gsm-sub"))
+			rec->usage = NUMBER_USAGE_TYPE_GSM_SUB;
+		else if (!strcmp(np, "map-to")) {
+			rec->usage = NUMBER_USAGE_TYPE_ALIAS;
+			while (isspace(*cp))
+				cp++;
+			if (*cp == '\0' || *cp == '#') {
+				fprintf(stderr,
+			"number-db2 line %d: map-to requires an argument\n",
+					lineno);
+				exit(1);
+			}
+			for (np = cp; *cp && !isspace(*cp); cp++)
+				;
+			if (*cp)
+				*cp++ = '\0';
+			rc = parse_extra_number(rec, np);
+			if (rc < 0) {
+				fprintf(stderr,
+			"number-db2 line %d: map-to argument is invalid\n",
+					lineno);
+				exit(1);
+			}
+		} else if (!strcmp(np, "e911-via")) {
+			if (rec->usage != NUMBER_USAGE_TYPE_GSM_SUB) {
+				fprintf(stderr,
+			"number-db2 line %d: invalid usage of e911-via\n",
+					lineno);
+				exit(1);
+			}
+			rec->usage |= NUMBER_USAGE_FLAG_E911_VIA;
+			while (isspace(*cp))
+				cp++;
+			if (*cp == '\0' || *cp == '#') {
+				fprintf(stderr,
+			"number-db2 line %d: e911-via requires an argument\n",
+					lineno);
+				exit(1);
+			}
+			for (np = cp; *cp && !isspace(*cp); cp++)
+				;
+			if (*cp)
+				*cp++ = '\0';
+			rc = parse_extra_number(rec, np);
+			if (rc < 0) {
+				fprintf(stderr,
+			"number-db2 line %d: e911-via argument is invalid\n",
+					lineno);
+				exit(1);
+			}
+		} else {
+			fprintf(stderr,
+			"number-db2 line %d: invalid attribute \"%s\"\n",
+				lineno, np);
+			exit(1);
+		}
+	}
+}
+
+static void
+handle_suffix_line(cp)
+	char *cp;
+{
+	char *np;
+	uint16_t suffix;
+	struct owned_number_rec own_rec;
+	struct short_number_rec short_rec;
+
+	if (!prefix_set) {
+		fprintf(stderr,
+	      "number-db2 line %d: suffix not valid without preceding prefix\n",
+			lineno);
+		exit(1);
+	}
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (grok_number_string(np, 0) != 4) {
+		fprintf(stderr,
+		      "number-db2 line %d: suffix requires 4-digit argument\n",
+			lineno);
+		exit(1);
+	}
+	suffix = digits4_to_uint16(np);
+	bzero(&own_rec, sizeof own_rec);
+	own_rec.number[0] = prefix_buf[0];
+	own_rec.number[1] = prefix_buf[1];
+	own_rec.number[2] = suffix;
+	handle_number_attr(&own_rec, cp);
+	enter_owned_number(&own_rec);
+	if (!prefix_allows_abbrev || own_rec.usage != NUMBER_USAGE_TYPE_GSM_SUB)
+		return;
+	bzero(&short_rec, sizeof short_rec);
+	short_rec.short_num = suffix;
+	short_rec.short_num_type = SHORT_NUM_TYPE_ABBREV;
+	short_rec.fullnum_flags = own_rec.number_flags;
+	short_rec.fullnum_prefix[0] = prefix_buf[0];
+	short_rec.fullnum_prefix[1] = prefix_buf[1];
+	enter_short_number(&short_rec);
+}
+
+static void
+handle_full10_line(cp)
+	char *cp;
+{
+	char *np, full10[11];
+	struct owned_number_rec own_rec;
+
+	prefix_set = 0;		/* cancel any previous prefix line */
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (grok_number_string(np, 1) != 10) {
+		fprintf(stderr,
+		"number-db2 line %d: full10 requires 10-digit argument\n",
+			lineno);
+		exit(1);
+	}
+	dehyphen_number_string(np, full10);
+	if (!is_nanp_valid_prefix(full10)) {
+		fprintf(stderr,
+			"number-db2 line %d: number violates NANP rules\n",
+			lineno);
+		exit(1);
+	}
+	bzero(&own_rec, sizeof own_rec);
+	own_rec.number[0] = digits3_to_uint16(full10);
+	own_rec.number[1] = digits3_to_uint16(full10 + 3);
+	own_rec.number[2] = digits4_to_uint16(full10 + 6);
+	handle_number_attr(&own_rec, cp);
+	enter_owned_number(&own_rec);
+}
+
+static void
+handle_itn_line(cp)
+	char *cp;
+{
+	char *np;
+	struct short_number_rec short_rec;
+
+	prefix_set = 0;		/* cancel any previous prefix line */
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (grok_number_string(np, 0) != 4) {
+		fprintf(stderr,
+			"number-db2 line %d: itn requires 4-digit argument\n",
+			lineno);
+		exit(1);
+	}
+	bzero(&short_rec, sizeof short_rec);
+	short_rec.short_num = digits4_to_uint16(np);
+	short_rec.short_num_type = SHORT_NUM_TYPE_ITN;
+	enter_short_number(&short_rec);
+}
+
+static void
+process_line()
+{
+	char *cp, *np;
+	void (*handler)();
+
+	if (!index(linebuf, '\n')) {
+		fprintf(stderr,
+			"number-db2 line %d: too long or missing newline\n",
+			lineno);
+		exit(1);
+	}
+	for (cp = linebuf; isspace(*cp); cp++)
+		;
+	if (*cp == '\0' || *cp == '#')
+		return;
+	for (np = cp; *cp && !isspace(*cp); cp++)
+		;
+	if (*cp)
+		*cp++ = '\0';
+	if (!strcmp(np, "prefix"))
+		handler = handle_prefix_line;
+	else if (!strcmp(np, "suffix"))
+		handler = handle_suffix_line;
+	else if (!strcmp(np, "full10"))
+		handler = handle_full10_line;
+	else if (!strcmp(np, "itn"))
+		handler = handle_itn_line;
+	else {
+		fprintf(stderr,
+			"number-db2 line %d: non-understood keyword \"%s\"\n",
+			lineno, np);
+		exit(1);
+	}
+	while (isspace(*cp))
+		cp++;
+	if (*cp == '\0' || *cp == '#') {
+		fprintf(stderr,
+		"number-db2 line %d: missing argument after \"%s\" keyword\n",
+			lineno, np);
+		exit(1);
+	}
+	handler(cp);
+}
+
+static int
+compare_owned_num(p1, p2)
+	struct owned_number_rec *p1, *p2;
+{
+	if (p1->number[0] < p2->number[0])
+		return(-1);
+	if (p1->number[0] > p2->number[0])
+		return(1);
+	if (p1->number[1] < p2->number[1])
+		return(-1);
+	if (p1->number[1] > p2->number[1])
+		return(1);
+	if (p1->number[2] < p2->number[2])
+		return(-1);
+	if (p1->number[2] > p2->number[2])
+		return(1);
+	return(0);
+}
+
+static int
+compare_short_num(p1, p2)
+	struct short_number_rec *p1, *p2;
+{
+	if (p1->short_num < p2->short_num)
+		return(-1);
+	if (p1->short_num > p2->short_num)
+		return(1);
+	return(0);
+}
+
+static void
+owned_num_check_dup()
+{
+	struct owned_number_rec *p, *endp;
+
+	endp = owned_number_buf + owned_number_count - 1;
+	for (p = owned_number_buf; p < endp; p++) {
+		if (p[0].number[0] == p[1].number[0] &&
+		    p[0].number[1] == p[1].number[1] &&
+		    p[0].number[2] == p[1].number[2]) {
+			fprintf(stderr,
+		"error: NANP number %03u-%03u-%04u appears more than once\n",
+				p[0].number[0], p[0].number[1], p[0].number[2]);
+			exit(1);
+		}
+	}
+}
+
+static void
+short_num_check_dup()
+{
+	struct short_number_rec *p, *endp;
+
+	endp = short_number_buf + short_number_count - 1;
+	for (p = short_number_buf; p < endp; p++) {
+		if (p[0].short_num == p[1].short_num) {
+			fprintf(stderr,
+			"error: short number %04u appears more than once\n",
+				(unsigned) p->short_num);
+			exit(1);
+		}
+	}
+}
+
+static void
+emit_output()
+{
+	FILE *outf;
+	struct numdb_file_hdr hdr;
+
+	outf = fopen("number-db2.newbin", "w");
+	if (!outf) {
+		perror("creating number-db2.newbin");
+		exit(1);
+	}
+	hdr.owned_number_count = owned_number_count;
+	hdr.short_number_count = short_number_count;
+	if (fwrite(&hdr, sizeof hdr, 1, outf) != 1) {
+write_err:	fprintf(stderr, "error writing to new binary file\n");
+		exit(1);
+	}
+	if (fwrite(owned_number_buf, sizeof(owned_number_buf[0]),
+		   owned_number_count, outf) != owned_number_count)
+		goto write_err;
+	if (fwrite(short_number_buf, sizeof(short_number_buf[0]),
+		   short_number_count, outf) != short_number_count)
+		goto write_err;
+	fclose(outf);
+}
+
+main(argc, argv)
+	char **argv;
+{
+	if (argc > 2) {
+		fprintf(stderr, "usage: %s [directory]\n", argv[0]);
+		exit(1);
+	}
+	if (argv[1])
+		system_dir = argv[1];
+	else
+		system_dir = "/var/gsm";
+	if (chdir(system_dir) < 0) {
+		perror(system_dir);
+		exit(1);
+	}
+	inf = fopen("number-db2", "r");
+	if (!inf) {
+		perror("opening number-db2");
+		exit(1);
+	}
+	for (lineno = 1; fgets(linebuf, sizeof linebuf, inf); lineno++)
+		process_line();
+	fclose(inf);
+	if (owned_number_count >= 2) {
+		qsort(owned_number_buf, owned_number_count,
+			sizeof(owned_number_buf[0]), compare_owned_num);
+		owned_num_check_dup();
+	}
+	if (short_number_count >= 2) {
+		qsort(short_number_buf, short_number_count,
+			sizeof(short_number_buf[0]), compare_short_num);
+		short_num_check_dup();
+	}
+	emit_output();
+	/* make it live */
+	if (rename("number-db2.newbin", "number-db2.bin") < 0) {
+		perror("rename");
+		exit(1);
+	}
+	exit(0);
+}