add adbgen applet to generate databases from it's text dump

14 files changed, 430 insertions, 8 deletions

diff --git a/src/adb.c b/src/adb.c
index e86e243..b33799c 100644
--- a/src/adb.c
+++ b/src/adb.c
@@ -252,6 +252,14 @@ struct adb_obj *adb_r_rootobj(struct adb *db, struct adb_obj *obj, const struct
 	return adb_r_obj(db, adb_r_root(db), obj, schema);
 }
 
+const uint8_t *adb_ro_kind(const struct adb_obj *o, unsigned i)
+{
+	if (o->schema->kind == ADB_KIND_ADB ||
+	    o->schema->kind == ADB_KIND_ARRAY)
+		i = 1;
+	return o->schema->fields[i-1].kind;
+}
+
 adb_val_t adb_ro_val(const struct adb_obj *o, unsigned i)
 {
 	if (i >= o->num) return ADB_NULL;
@@ -593,6 +601,7 @@ adb_val_t adb_w_fromstring(struct adb *db, const uint8_t *kind, apk_blob_t val)
 		struct adb_obj obj;
 		struct adb_object_schema *schema = container_of(kind, struct adb_object_schema, kind);
 		adb_wo_alloca(&obj, schema, db);
+		if (!schema->fromstring) return ADB_ERROR(EAPKDBFORMAT);
 		r = schema->fromstring(&obj, val);
 		if (r) return ADB_ERROR(r);
 		return adb_w_obj(&obj);
@@ -618,6 +627,25 @@ struct adb_obj *adb_wo_init(struct adb_obj *o, adb_val_t *p, const struct adb_ob
 	return o;
 }
 
+struct adb_obj *adb_wo_init_val(struct adb_obj *o, adb_val_t *p, const struct adb_obj *parent, unsigned i)
+{
+	const uint8_t *kind = adb_ro_kind(parent, i);
+	const struct adb_object_schema *schema = 0;
+	switch (*kind) {
+	case ADB_KIND_OBJECT:
+	case ADB_KIND_ARRAY:
+		schema = container_of(kind, struct adb_object_schema, kind);
+		break;
+	case ADB_KIND_ADB:
+		schema = container_of(kind, struct adb_adb_schema, kind)->schema;
+		break;
+	default:
+		assert(1);
+	}
+
+	return adb_wo_init(o, p, schema, parent->db);
+}
+
 void adb_wo_reset(struct adb_obj *o)
 {
 	uint32_t max = o->obj[ADBI_NUM_ENTRIES];
@@ -765,9 +793,17 @@ void adb_wa_sort_unique(struct adb_obj *arr)
 }
 
 /* Schema helpers */
+int adb_s_field_by_name_blob(const struct adb_object_schema *schema, apk_blob_t blob)
+{
+	for (int i = 0; i < schema->num_fields-1 && schema->fields[i].name; i++)
+		if (apk_blob_compare(APK_BLOB_STR(schema->fields[i].name), blob) == 0)
+			return i + 1;
+	return 0;
+}
+
 int adb_s_field_by_name(const struct adb_object_schema *schema, const char *name)
 {
-	for (int i = 0; i < schema->num_fields; i++)
+	for (int i = 0; i < schema->num_fields-1 && schema->fields[i].name; i++)
 		if (strcmp(schema->fields[i].name, name) == 0)
 			return i + 1;
 	return 0;
diff --git a/src/adb.h b/src/adb.h
index cf3da69..10b46ac 100644
--- a/src/adb.h
+++ b/src/adb.h
@@ -172,6 +172,7 @@ struct adb_obj *adb_r_obj(struct adb *, adb_val_t, struct adb_obj *o, const stru
 static inline uint32_t adb_ro_num(const struct adb_obj *o) { return o->num; }
 static inline uint32_t adb_ra_num(const struct adb_obj *o) { return (o->num ?: 1) - 1; }
 
+const uint8_t *adb_ro_kind(const struct adb_obj *o, unsigned i);
 adb_val_t adb_ro_val(const struct adb_obj *o, unsigned i);
 uint32_t adb_ro_int(const struct adb_obj *o, unsigned i);
 apk_blob_t adb_ro_blob(const struct adb_obj *o, unsigned i);
@@ -192,6 +193,7 @@ adb_val_t adb_w_fromstring(struct adb *, const uint8_t *kind, apk_blob_t);
 #define adb_wo_alloca(o, schema, db) adb_wo_init(o, alloca(sizeof(adb_val_t[(schema)->num_fields])), schema, db)
 
 struct adb_obj *adb_wo_init(struct adb_obj *, adb_val_t *, const struct adb_object_schema *, struct adb *);
+struct adb_obj *adb_wo_init_val(struct adb_obj *, adb_val_t *, const struct adb_obj *, unsigned i);
 void adb_wo_reset(struct adb_obj *);
 void adb_wo_resetdb(struct adb_obj *);
 adb_val_t adb_w_obj(struct adb_obj *);
@@ -210,6 +212,7 @@ void adb_wa_sort(struct adb_obj *);
 void adb_wa_sort_unique(struct adb_obj *);
 
 /* Schema helpers */
+int adb_s_field_by_name_blob(const struct adb_object_schema *schema, apk_blob_t blob);
 int adb_s_field_by_name(const struct adb_object_schema *, const char *);
 
 /* Creation */
@@ -254,7 +257,7 @@ struct adb_walk_ops {
 	int (*scalar)(struct adb_walk *, apk_blob_t scalar, int multiline);
 };
 
-extern const struct adb_walk_ops adb_walk_gentext_ops;
+extern const struct adb_walk_ops adb_walk_gentext_ops, adb_walk_genadb_ops;
 
 struct adb_walk {
 	const struct adb_walk_ops *ops;
@@ -269,6 +272,20 @@ struct adb_walk_gentext {
 	int key_printed : 1;
 };
 
+#define ADB_WALK_GENADB_MAX_NESTING	32
+#define ADB_WALK_GENADB_MAX_VALUES	100000
+
+struct adb_walk_genadb {
+	struct adb_walk d;
+	struct adb db;
+	struct adb idb[2];
+	int nest, nestdb, num_vals;
+	struct adb_obj objs[ADB_WALK_GENADB_MAX_NESTING];
+	unsigned int curkey[ADB_WALK_GENADB_MAX_NESTING];
+	adb_val_t vals[ADB_WALK_GENADB_MAX_VALUES];
+};
+
 int adb_walk_adb(struct adb_walk *d, struct adb *db, struct apk_trust *trust);
+int adb_walk_istream(struct adb_walk *d, struct apk_istream *is);
 
 #endif
diff --git a/src/adb_walk_adb.c b/src/adb_walk_adb.c
index 713d5a7..4f99d85 100644
--- a/src/adb_walk_adb.c
+++ b/src/adb_walk_adb.c
@@ -51,10 +51,15 @@ static int dump_item(struct adb_walk_ctx *ctx, const char *name, const uint8_t *
 		d->ops->end(d);
 		ctx->db = origdb;
 		break;
-	case ADB_KIND_OBJECT:
-		d->ops->start_object(d);
-		dump_object(ctx, container_of(kind, struct adb_object_schema, kind), v);
-		d->ops->end(d);
+	case ADB_KIND_OBJECT:;
+		struct adb_object_schema *object = container_of(kind, struct adb_object_schema, kind);
+		if (!object->tostring) {
+			d->ops->start_object(d);
+			dump_object(ctx, object, v);
+			d->ops->end(d);
+		} else {
+			dump_object(ctx, object, v);
+		}
 		break;
 	case ADB_KIND_BLOB:
 	case ADB_KIND_INT:;
@@ -162,5 +167,6 @@ int adb_walk_adb(struct adb_walk *d, struct adb *db, struct apk_trust *trust)
 		.db = db,
 		.trust = trust,
 	};
+	d->ops->schema(d, db->hdr.schema);
 	return dump_adb(&ctx);
 }
diff --git a/src/adb_walk_genadb.c b/src/adb_walk_genadb.c
new file mode 100644
index 0000000..a15aa82
--- /dev/null
+++ b/src/adb_walk_genadb.c
@@ -0,0 +1,138 @@
+#include <errno.h>
+#include "adb.h"
+#include "apk_print.h"
+
+static int adb_walk_genadb_schema(struct adb_walk *d, uint32_t schema_id)
+{
+	struct adb_walk_genadb *dt = container_of(d, struct adb_walk_genadb, d);
+	const struct adb_db_schema *s;
+
+	dt->db.hdr.schema = htole32(schema_id);
+	for (s = d->schemas; s->magic; s++)
+		if (s->magic == schema_id) break;
+	if (!s) return -EAPKDBFORMAT;
+
+	adb_wo_init(&dt->objs[0], &dt->vals[0], s->root, &dt->db);
+	dt->num_vals += s->root->num_fields;
+	if (dt->num_vals >= ARRAY_SIZE(dt->vals)) return -E2BIG;
+	dt->nest = 0;
+
+	return 0;
+}
+
+static int adb_walk_genadb_comment(struct adb_walk *d, apk_blob_t comment)
+{
+	return 0;
+}
+
+static int adb_walk_genadb_start_object(struct adb_walk *d)
+{
+	struct adb_walk_genadb *dt = container_of(d, struct adb_walk_genadb, d);
+
+	if (!dt->db.hdr.schema) return -EAPKDBFORMAT;
+	if (dt->nest >= ARRAY_SIZE(dt->objs)) return -EAPKDBFORMAT;
+
+	if (dt->curkey[dt->nest] == 0 &&
+	    dt->objs[dt->nest].schema->kind == ADB_KIND_OBJECT)
+		return -EAPKDBFORMAT;
+
+	dt->nest++;
+	adb_wo_init_val(
+		&dt->objs[dt->nest], &dt->vals[dt->num_vals],
+		&dt->objs[dt->nest-1], dt->curkey[dt->nest-1]);
+
+	if (*adb_ro_kind(&dt->objs[dt->nest-1], dt->curkey[dt->nest-1]) == ADB_KIND_ADB) {
+		struct adb_adb_schema *schema = container_of(&dt->objs[dt->nest-1].schema->kind, struct adb_adb_schema, kind);
+		adb_reset(&dt->idb[dt->nestdb]);
+		dt->idb[dt->nestdb].hdr.schema = htole32(schema->schema_id);
+		dt->objs[dt->nest].db = &dt->idb[dt->nestdb];
+		dt->nestdb++;
+	}
+
+	dt->num_vals += dt->objs[dt->nest].schema->num_fields;
+	if (dt->num_vals >= ARRAY_SIZE(dt->vals)) return -E2BIG;
+
+	return 0;
+}
+
+static int adb_walk_genadb_start_array(struct adb_walk *d, unsigned int num)
+{
+	return adb_walk_genadb_start_object(d);
+}
+
+static int adb_walk_genadb_end(struct adb_walk *d)
+{
+	struct adb_walk_genadb *dt = container_of(d, struct adb_walk_genadb, d);
+	adb_val_t val;
+
+	val = adb_w_obj(&dt->objs[dt->nest]);
+	if (ADB_IS_ERROR(val))
+		return -ADB_VAL_VALUE(val);
+
+	dt->curkey[dt->nest] = 0;
+	dt->num_vals -= dt->objs[dt->nest].schema->num_fields;
+
+	if (dt->nest == 0) {
+		adb_w_root(&dt->db, val);
+		return 0;
+	}
+
+	dt->nest--;
+
+	if (*adb_ro_kind(&dt->objs[dt->nest], dt->curkey[dt->nest]) == ADB_KIND_ADB) {
+		dt->nestdb--;
+		adb_w_root(&dt->idb[dt->nestdb], val);
+		val = adb_w_adb(&dt->db, &dt->idb[dt->nestdb]);
+	}
+
+	if (dt->curkey[dt->nest] == 0) {
+		adb_wa_append(&dt->objs[dt->nest], val);
+	} else {
+		adb_wo_val(&dt->objs[dt->nest], dt->curkey[dt->nest], val);
+		dt->curkey[dt->nest] = 0;
+	}
+
+	return 0;
+}
+
+static int adb_walk_genadb_key(struct adb_walk *d, apk_blob_t key)
+{
+	struct adb_walk_genadb *dt = container_of(d, struct adb_walk_genadb, d);
+	uint8_t kind = dt->objs[dt->nest].schema->kind;
+
+	if (kind != ADB_KIND_OBJECT && kind != ADB_KIND_ADB)
+		return -EAPKDBFORMAT;
+
+	dt->curkey[dt->nest] = adb_s_field_by_name_blob(dt->objs[dt->nest].schema, key);
+	if (dt->curkey[dt->nest] == 0)
+		return -EAPKDBFORMAT;
+
+	return 0;
+}
+
+static int adb_walk_genadb_scalar(struct adb_walk *d, apk_blob_t scalar, int multiline)
+{
+	struct adb_walk_genadb *dt = container_of(d, struct adb_walk_genadb, d);
+
+	if (dt->objs[dt->nest].schema->kind == ADB_KIND_ARRAY) {
+		adb_wa_append_fromstring(&dt->objs[dt->nest], scalar);
+	} else {
+		if (dt->curkey[dt->nest] == 0)
+			adb_wo_fromstring(&dt->objs[dt->nest], scalar);
+		else
+			adb_wo_val_fromstring(&dt->objs[dt->nest], dt->curkey[dt->nest], scalar);
+	}
+	dt->curkey[dt->nest] = 0;
+
+	return 0;
+}
+
+const struct adb_walk_ops adb_walk_genadb_ops = {
+	.schema = adb_walk_genadb_schema,
+	.comment = adb_walk_genadb_comment,
+	.start_array = adb_walk_genadb_start_array,
+	.start_object = adb_walk_genadb_start_object,
+	.end = adb_walk_genadb_end,
+	.key = adb_walk_genadb_key,
+	.scalar = adb_walk_genadb_scalar,
+};
diff --git a/src/adb_walk_gentext.c b/src/adb_walk_gentext.c
index 995dda3..477cacd 100644
--- a/src/adb_walk_gentext.c
+++ b/src/adb_walk_gentext.c
@@ -111,6 +111,7 @@ static int adb_walk_gentext_scalar(struct adb_walk *d, apk_blob_t scalar, int mu
 
 		fprintf(out, "|\n");
 		adb_walk_gentext_newline(dt);
+
 		dt->nest++;
 		while (apk_blob_split(scalar, nl, &l, &scalar)) {
 			adb_walk_gentext_indent(dt);
diff --git a/src/adb_walk_istream.c b/src/adb_walk_istream.c
new file mode 100644
index 0000000..b85565f
--- /dev/null
+++ b/src/adb_walk_istream.c
@@ -0,0 +1,131 @@
+#include <errno.h>
+#include "adb.h"
+
+//#define DEBUG_PRINT
+#ifdef DEBUG_PRINT
+#include <stdio.h>
+#define dbg_printf(args...) fprintf(stderr, args)
+#else
+#define dbg_printf(args...)
+#endif
+
+int adb_walk_istream(struct adb_walk *d, struct apk_istream *is)
+{
+	const apk_blob_t token = APK_BLOB_STR("\n");
+	const apk_blob_t comment = APK_BLOB_STR(" #");
+	const apk_blob_t key_sep = APK_BLOB_STR(": ");
+	char mblockdata[1024*4];
+	apk_blob_t l, comm, mblock = APK_BLOB_BUF(mblockdata);
+	int r = 0, i, multi_line = 0, nesting = 0, new_item = 0;
+	uint8_t started[64] = {0};
+
+	if (IS_ERR_OR_NULL(is)) return PTR_ERR(is);
+	l = apk_istream_get_delim(is, token);
+	apk_blob_pull_blob_match(&l, APK_BLOB_STR("#%SCHEMA: "));
+	if ((r = d->ops->schema(d, apk_blob_pull_uint(&l, 16))) != 0) goto err;
+
+	started[0] = 1;
+	while (!APK_BLOB_IS_NULL(l = apk_istream_get_delim(is, token))) {
+		for (i = 0; l.len >= 2 && l.ptr[0] == ' ' && l.ptr[1] == ' '; i++, l.ptr += 2, l.len -= 2)
+			if (multi_line && i >= multi_line) break;
+
+		for (; nesting > i; nesting--) {
+			if (multi_line) {
+				apk_blob_t data = apk_blob_pushed(APK_BLOB_BUF(mblockdata), mblock);
+				if (APK_BLOB_IS_NULL(data)) {
+					r = -E2BIG;
+					goto err;
+				}
+				if (data.len && data.ptr[data.len-1] == '\n') data.len--;
+				dbg_printf("Multiline-Scalar >%d> "BLOB_FMT"\n", nesting, BLOB_PRINTF(data));
+				if ((r = d->ops->scalar(d, data, 1)) != 0) goto err;
+				mblock = APK_BLOB_BUF(mblockdata);
+				multi_line = 0;
+			}
+			if (started[nesting]) {
+				dbg_printf("End %d\n", nesting);
+				if ((r = d->ops->end(d)) != 0) goto err;
+			}
+		}
+		if (l.ptr[0] == '-' && l.ptr[1] == ' ') {
+			l.ptr += 2, l.len -= 2;
+			if (!started[nesting]) {
+				dbg_printf("Array %d\n", nesting);
+				if ((r = d->ops->start_array(d, 0)) != 0) goto err;
+				started[nesting] = 1;
+			}
+			new_item = 1;
+		}
+		dbg_printf(" >%d/%d< >"BLOB_FMT"<\n", nesting, i, BLOB_PRINTF(l));
+
+		if (multi_line) {
+			dbg_printf("Scalar-Block:>%d> "BLOB_FMT"\n", nesting, BLOB_PRINTF(l));
+			apk_blob_push_blob(&mblock, l);
+			apk_blob_push_blob(&mblock, APK_BLOB_STR("\n"));
+			new_item = 0;
+			continue;
+		}
+
+		if (l.ptr[0] == '#') {
+			if ((r = d->ops->comment(d, l)) != 0) goto err;
+			continue;
+		}
+
+		// contains ' #' -> comment
+		if (!apk_blob_split(l, comment, &l, &comm))
+			comm.len = 0;
+
+		if (l.len) {
+			apk_blob_t key = APK_BLOB_NULL, scalar = APK_BLOB_NULL;
+			int start = 0;
+
+			if (apk_blob_split(l, key_sep, &key, &scalar)) {
+				// contains ': ' -> key + scalar
+			} else if (l.ptr[l.len-1] == ':') {
+				// ends ':' -> key + indented object/array
+				key = APK_BLOB_PTR_LEN(l.ptr, l.len-1);
+				start = 1;
+			} else {
+				scalar = l;
+			}
+			if (key.len) {
+				if (new_item) {
+					started[++nesting] = 0;
+					dbg_printf("Array-Object %d\n", nesting);
+				}
+				if (!started[nesting]) {
+					dbg_printf("Object %d\n", nesting);
+					if ((r = d->ops->start_object(d)) != 0) goto err;
+					started[nesting] = 1;
+				}
+				dbg_printf("Key >%d> "BLOB_FMT"\n", nesting, BLOB_PRINTF(key));
+				if ((r = d->ops->key(d, key)) != 0) goto err;
+				if (start) started[++nesting] = 0;
+			}
+
+			if (scalar.len) {
+				if (scalar.ptr[0] == '|') {
+					dbg_printf("Scalar-block >%d> "BLOB_FMT"\n", nesting, BLOB_PRINTF(scalar));
+					// scalar '|' -> starts string literal block
+					started[++nesting] = 0;
+					multi_line = nesting;
+				} else {
+					dbg_printf("Scalar >%d> "BLOB_FMT"\n", nesting, BLOB_PRINTF(scalar));
+					if ((r = d->ops->scalar(d, scalar, 0)) != 0) goto err;
+				}
+			}
+			new_item = 0;
+		}
+
+		if (comm.len) {
+			if ((r = d->ops->comment(d, comm)) != 0) goto err;
+		}
+
+		//fprintf(stderr, ">%d> "BLOB_FMT"\n", indent, BLOB_PRINTF(l));
+	}
+	d->ops->end(d);
+
+err:
+	if (r) apk_istream_error(is, r);
+	return apk_istream_close(is);
+}
diff --git a/src/apk_adb.c b/src/apk_adb.c
index 0e08848..0785455 100644
--- a/src/apk_adb.c
+++ b/src/apk_adb.c
@@ -146,9 +146,27 @@ static apk_blob_t hexblob_tostring(struct adb *db, adb_val_t val, char *buf, siz
 	return APK_BLOB_PTR_LEN(buf, snprintf(buf, bufsz, "(%ld bytes)", b.len));
 }
 
+static adb_val_t hexblob_fromstring(struct adb *db, apk_blob_t val)
+{
+	char buf[256];
+
+	if (val.len & 1)
+		return ADB_ERROR(EINVAL);
+	if (val.len > sizeof buf)
+		return ADB_ERROR(E2BIG);
+
+	apk_blob_t b = APK_BLOB_PTR_LEN(buf, val.len / 2);
+	apk_blob_pull_hexdump(&val, b);
+	if (APK_BLOB_IS_NULL(val))
+		return ADB_ERROR(EINVAL);
+
+	return adb_w_blob(db, b);
+}
+
 static struct adb_scalar_schema scalar_hexblob = {
 	.kind = ADB_KIND_BLOB,
 	.tostring = hexblob_tostring,
+	.fromstring = hexblob_fromstring,
 };
 
 static apk_blob_t int_tostring(struct adb *db, adb_val_t val, char *buf, size_t bufsz)
@@ -184,9 +202,17 @@ static apk_blob_t oct_tostring(struct adb *db, adb_val_t val, char *buf, size_t
 	return APK_BLOB_PTR_LEN(buf, snprintf(buf, bufsz, "%o", adb_r_int(db, val)));
 }
 
+static adb_val_t oct_fromstring(struct adb *db, apk_blob_t val)
+{
+	uint32_t n = apk_blob_pull_uint(&val, 8);
+	if (val.len) return ADB_ERROR(EINVAL);
+	return adb_w_int(db, n) ?: ADB_VAL_NULL;
+}
+
 static struct adb_scalar_schema scalar_oct = {
 	.kind = ADB_KIND_INT,
 	.tostring = oct_tostring,
+	.fromstring = oct_fromstring,
 };
 
 static apk_blob_t hsize_tostring(struct adb *db, adb_val_t val, char *buf, size_t bufsz)
@@ -197,10 +223,23 @@ static apk_blob_t hsize_tostring(struct adb *db, adb_val_t val, char *buf, size_
 	return APK_BLOB_PTR_LEN(buf, snprintf(buf, bufsz, "%jd %s", (intmax_t)v, unit));
 }
 
+static adb_val_t hsize_fromstring(struct adb *db, apk_blob_t val)
+{
+	apk_blob_t l, r;
+
+	if (!apk_blob_split(val, APK_BLOB_STR(" "), &l, &r))
+		return int_fromstring(db, val);
+
+	uint64_t n = apk_blob_pull_uint(&l, 10);
+	int sz = apk_get_human_size_unit(r);
+	n *= sz;
+	return adb_w_int(db, n);
+}
+
 static struct adb_scalar_schema scalar_hsize = {
 	.kind = ADB_KIND_INT,
 	.tostring = hsize_tostring,
-	.fromstring = int_fromstring,
+	.fromstring = hsize_fromstring,
 	.compare = int_compare,
 };
 
diff --git a/src/apk_blob.h b/src/apk_blob.h
index 9426721..e222739 100644
--- a/src/apk_blob.h
+++ b/src/apk_blob.h
@@ -97,6 +97,7 @@ unsigned long apk_blob_hash_seed(apk_blob_t, unsigned long seed);
 unsigned long apk_blob_hash(apk_blob_t str);
 int apk_blob_compare(apk_blob_t a, apk_blob_t b);
 int apk_blob_sort(apk_blob_t a, apk_blob_t b);
+int apk_blob_starts_with(apk_blob_t a, apk_blob_t b);
 int apk_blob_ends_with(apk_blob_t str, apk_blob_t suffix);
 int apk_blob_for_each_segment(apk_blob_t blob, const char *split,
 			      apk_blob_cb cb, void *ctx);
diff --git a/src/apk_io.h b/src/apk_io.h
index eafb12f..8a243ac 100644
--- a/src/apk_io.h
+++ b/src/apk_io.h
@@ -81,6 +81,7 @@ struct apk_istream *apk_istream_from_file(int atfd, const char *file);
 struct apk_istream *apk_istream_from_file_gz(int atfd, const char *file);
 struct apk_istream *apk_istream_from_fd(int fd);
 struct apk_istream *apk_istream_from_fd_url_if_modified(int atfd, const char *url, time_t since);
+static inline int apk_istream_error(struct apk_istream *is, int err) { if (!is->err) is->err = err; return err; }
 ssize_t apk_istream_read(struct apk_istream *is, void *ptr, size_t size);
 apk_blob_t apk_istream_get(struct apk_istream *is, size_t len);
 apk_blob_t apk_istream_get_max(struct apk_istream *is, size_t size);
diff --git a/src/apk_print.h b/src/apk_print.h
index 132fd8c..a8e734a 100644
--- a/src/apk_print.h
+++ b/src/apk_print.h
@@ -13,6 +13,7 @@
 #include "apk_blob.h"
 
 const char *apk_error_str(int error);
+int apk_get_human_size_unit(apk_blob_t b);
 const char *apk_get_human_size(off_t size, off_t *dest);
 
 struct apk_url_print {
diff --git a/src/app_adbdump.c b/src/app_adbdump.c
index 5020314..958c90f 100644
--- a/src/app_adbdump.c
+++ b/src/app_adbdump.c
@@ -53,3 +53,37 @@ static struct apk_applet apk_adbdump = {
 };
 APK_DEFINE_APPLET(apk_adbdump);
 
+
+static int adbgen_main(void *pctx, struct apk_ctx *ac, struct apk_string_array *args)
+{
+	struct apk_out *out = &ac->out;
+	char **arg;
+	int r;
+	struct adb_walk_genadb genadb = {
+		.d.ops = &adb_walk_genadb_ops,
+		.d.schemas = dbschemas,
+	};
+
+	adb_w_init_alloca(&genadb.db, 0, 1000);
+	adb_w_init_alloca(&genadb.idb[0], 0, 100);
+	foreach_array_item(arg, args) {
+		adb_reset(&genadb.db);
+		r = adb_walk_istream(&genadb.d, apk_istream_from_file(AT_FDCWD, *arg));
+		if (!r) {
+			r = adb_c_create(apk_ostream_to_fd(STDOUT_FILENO), &genadb.db,
+				apk_ctx_get_trust(ac));
+		}
+		adb_free(&genadb.db);
+		adb_free(&genadb.idb[0]);
+		if (r) apk_err(out, "%s: %s", *arg, apk_error_str(r));
+	}
+
+	return 0;
+}
+
+static struct apk_applet apk_adbgen = {
+	.name = "adbgen",
+	.main = adbgen_main,
+};
+APK_DEFINE_APPLET(apk_adbgen);
+
diff --git a/src/blob.c b/src/blob.c
index aa7ca83..32cd92e 100644
--- a/src/blob.c
+++ b/src/blob.c
@@ -263,6 +263,12 @@ int apk_blob_sort(apk_blob_t a, apk_blob_t b)
 	return a.len - b.len;
 }
 
+int apk_blob_starts_with(apk_blob_t a, apk_blob_t b)
+{
+	if (a.len < b.len) return 0;
+	return memcmp(a.ptr, b.ptr, b.len) == 0;
+}
+
 int apk_blob_ends_with(apk_blob_t a, apk_blob_t b)
 {
 	if (a.len < b.len) return 0;
diff --git a/src/meson.build b/src/meson.build
index 43f6c38..6c5f6ca 100644
--- a/src/meson.build
+++ b/src/meson.build
@@ -2,7 +2,9 @@ libapk_so_version = '2.99.0'
 libapk_src = [
 	'adb.c',
 	'adb_walk_adb.c',
+	'adb_walk_genadb.c',
 	'adb_walk_gentext.c',
+	'adb_walk_istream.c',
 	'apk_adb.c',
 	'atom.c',
 	'blob.c',
diff --git a/src/print.c b/src/print.c
index 0203cbb..26e31e2 100644
--- a/src/print.c
+++ b/src/print.c
@@ -65,9 +65,18 @@ const char *apk_error_str(int error)
 	}
 }
 
+static const char *size_units[] = {"B", "KiB", "MiB", "GiB", "TiB"};
+
+int apk_get_human_size_unit(apk_blob_t b)
+{
+	for (int i = 0, s = 1; i < ARRAY_SIZE(size_units); i++, s *= 1024)
+		if (apk_blob_compare(b, APK_BLOB_STR(size_units[i])) == 0)
+			return s;
+	return 1;
+}
+
 const char *apk_get_human_size(off_t size, off_t *dest)
 {
-	static const char *size_units[] = {"B", "KiB", "MiB", "GiB", "TiB"};
 	size_t i;
 	off_t s;