From 62d87e063318cc4c842b1b2d8c184f48aeaf6659 Mon Sep 17 00:00:00 2001
From: Sebastiano Tronto <sebastiano@tronto.net>
Date: Mon, 19 May 2025 17:45:14 +0200
Subject: Refactored checkdata

Relevant changes include:
  - Changed the signature of nissy_checkdata().
  - Removed expected_distribution.h from tools; this data is now included
    in each solver's src/ code.
  - Removed distribution check for cocsep; may add back later.
---
 src/nissy.c                      | 120 ++------------------
 src/nissy.h                      |   2 +
 src/solvers/coord/checkdata.h    |  65 +++++++++++
 src/solvers/coord/coord.h        |   1 +
 src/solvers/coord/dr.h           |  16 +++
 src/solvers/coord/dreo.h         |  14 +++
 src/solvers/coord/eo.h           |  11 ++
 src/solvers/coord/gendata.h      |   2 +-
 src/solvers/coord/solve.h        |   3 +-
 src/solvers/coord/types_macros.h |   2 +
 src/solvers/coord/utils.h        |  28 ++---
 src/solvers/dispatch.h           |   6 +-
 src/solvers/distribution.h       |  27 +++++
 src/solvers/h48/checkdata.h      | 234 +++++++++++++++++++++++++++++++++++++++
 src/solvers/h48/gendata_eoesep.h |   4 +-
 src/solvers/h48/gendata_h48.h    |  90 ---------------
 src/solvers/h48/h48.h            |   1 +
 17 files changed, 401 insertions(+), 225 deletions(-)
 create mode 100644 src/solvers/coord/checkdata.h
 create mode 100644 src/solvers/h48/checkdata.h

(limited to 'src')

diff --git a/src/nissy.c b/src/nissy.c
index 1f929c0..aaa5c2a 100644
--- a/src/nissy.c
+++ b/src/nissy.c
@@ -12,9 +12,6 @@
 #include "core/core.h"
 #include "solvers/solvers.h"
 
-STATIC bool checkdata(const unsigned char *, const tableinfo_t [static 1]);
-STATIC bool distribution_equal(const uint64_t [static INFO_DISTRIBUTION_LEN],
-    const uint64_t [static INFO_DISTRIBUTION_LEN], uint8_t);
 STATIC long long write_result(oriented_cube_t, char [static NISSY_SIZE_CUBE]);
 STATIC long long nissy_dataid(const char *, char [static NISSY_SIZE_DATAID]);
 STATIC long long nissy_gendata_unsafe(
@@ -30,47 +27,6 @@ struct {
 	GETCUBE_OPTIONS(NULL, NULL)
 };
 
-STATIC bool
-checkdata(const unsigned char *buf, const tableinfo_t info[static 1])
-{
-	uint64_t distr[INFO_DISTRIBUTION_LEN];
-
-	if (info->type == TABLETYPE_PRUNING) {
-		getdistribution(buf + INFOSIZE, distr, info);
-		LOG("\n[checkdata] Checking distribution for %s\n", info->solver);
-		return distribution_equal(info->distribution, distr, info->maxvalue);
-	} else {
-		LOG("\n[checkdata] Skipping distribution check for "
-		    "special table %s\n", info->solver);
-		return true;
-	}
-}
-
-STATIC bool
-distribution_equal(
-	const uint64_t expected[static INFO_DISTRIBUTION_LEN],
-	const uint64_t actual[static INFO_DISTRIBUTION_LEN],
-	uint8_t maxvalue
-)
-{
-	int wrong;
-	uint8_t i;
-
-	for (i = 0, wrong = 0; i <= MIN(maxvalue, 20); i++) {
-		if (expected[i] != actual[i]) {
-			wrong++;
-			LOG("[checkdata] Value for depth %" PRIu8
-			    ": expected %" PRIu64 ", found %" PRIu64 "\n",
-			    i, expected[i], actual[i]);
-		} else {
-			LOG("[checkdata] Value for depth %" PRIu8
-			    " is correct (%" PRIu64 ")\n", i, actual[i]);
-		}
-	}
-
-	return wrong == 0;
-}
-
 STATIC long long
 write_result(oriented_cube_t cube, char result[static NISSY_SIZE_CUBE])
 {
@@ -227,55 +183,6 @@ nissy_getcube(
 	return write_result(oc, result);
 }
 
-long long
-nissy_datainfo(
-        uint64_t data_size,
-	const unsigned char data[data_size]
-)
-{
-	uint8_t i;
-	tableinfo_t info;
-	long long ret;
-
-	if ((size_t)data % 8 != 0) {
-		LOG("[datainfo] Error: buffer is not 8-byte aligned\n");
-		return NISSY_ERROR_DATA;
-	}
-
-	ret = readtableinfo(data_size, data, &info);
-	if (ret != 0)
-		return ret;
-
-	LOG("\n---------\n\n"
-	    "Table information for '%s'\n\n"
-	    "Size:      %" PRIu64 " bytes\n"
-	    "Entries:   %" PRIu64 " (%" PRIu8 " bits per entry)\n",
-	    info.solver, info.fullsize, info.entries, info.bits);
-
-	switch (info.type) {
-	case TABLETYPE_PRUNING:
-		LOG("\nTable distribution:\nValue\tPositions\n");
-		for (i = 0; i <= info.maxvalue; i++) {
-			LOG("%" PRIu8 "\t%" PRIu64 "\n",
-			    i + info.base, info.distribution[i]);
-		}
-		break;
-	case TABLETYPE_SPECIAL:
-		LOG("This is an ad-hoc table\n");
-		break;
-	default:
-		LOG("datainfo: unknown table type\n");
-		return NISSY_ERROR_DATA;
-	}
-
-	if (info.next != 0)
-		return nissy_datainfo(data_size - info.next, data + info.next);
-
-	LOG("\n---------\n");
-
-	return NISSY_OK;
-}
-
 STATIC long long
 nissy_dataid(const char *solver, char dataid[static NISSY_SIZE_DATAID])
 {
@@ -344,33 +251,20 @@ nissy_gendata_unsafe(
 
 long long
 nissy_checkdata(
+	const char *solver,
 	unsigned long long data_size,
 	const unsigned char data[data_size]
 )
 {
-	tableinfo_t info;
-	int64_t err;
-
-	if ((size_t)data % 8 != 0) {
-		LOG("[checkdata] Error: buffer is not 8-byte aligned\n");
-		return NISSY_ERROR_DATA;
-	}
+	solver_dispatch_t *dispatch;
 
-	for (const unsigned char *buf = data;
-	     (err = readtableinfo(data_size, buf, &info)) == NISSY_OK;
-	     buf += info.next, data_size -= info.next)
-	{
-		if (!checkdata(buf, &info)) {
-			LOG("[checkdata] Error: data for solver '%s' is "
-			    "corrupted!\n", info.solver);
-			return NISSY_ERROR_DATA;
-		}
-
-		if (info.next == 0)
-			break;
+	dispatch = match_solver(solver);
+	if (dispatch == NULL) {
+		LOG("[checkdata] Unknown solver %s\n", solver);
+		return NISSY_ERROR_INVALID_SOLVER;
 	}
 
-	return err;
+	return dispatch->checkdata(solver, data_size, data);
 }
 
 long long
diff --git a/src/nissy.h b/src/nissy.h
index 2fa4a75..0447d66 100644
--- a/src/nissy.h
+++ b/src/nissy.h
@@ -283,6 +283,7 @@ nissy_gendata(
 Check that the data is a valid data table for a solver.
 
 Parameters:
+   solver    - The name of the solver.
    data_size - The size of the data buffer.
    data      - The data for the solver. Can be computed with gendata.
                This buffer must have 8-byte alignment.
@@ -293,6 +294,7 @@ Return values:
 */
 long long
 nissy_checkdata(
+	const char *solver,
 	unsigned long long data_size,
 	const unsigned char data[data_size]
 );
diff --git a/src/solvers/coord/checkdata.h b/src/solvers/coord/checkdata.h
new file mode 100644
index 0000000..81394c7
--- /dev/null
+++ b/src/solvers/coord/checkdata.h
@@ -0,0 +1,65 @@
+STATIC long long checkdata_coord(
+    const char *, unsigned long long n, const unsigned char [n]);
+
+STATIC long long
+checkdata_coord(
+	const char *solver,
+	unsigned long long data_size,
+	const unsigned char data[data_size]
+)
+{
+	coord_t *coord;
+	const unsigned char *table;
+	tableinfo_t info;
+	int64_t err;
+	uint64_t actual_distribution[INFO_DISTRIBUTION_LEN];
+
+	if ((size_t)data % 8 != 0) {
+		LOG("[checkdata] Error: buffer is not 8-byte aligned\n");
+		return NISSY_ERROR_DATA;
+	}
+
+	parse_coord_and_axis(solver, &coord, NULL);
+	if (coord == NULL) {
+		LOG("[cehckdata] Unknown coordinate solver '%s'\n", solver);
+		return NISSY_ERROR_DATA;
+	}
+
+	table = data + INFOSIZE;
+	err = readtableinfo(data_size, data, &info);
+	if (err != NISSY_OK) {
+		LOG("[checkdata] Data is corrupt\n");
+		return err;
+	}
+
+	if (info.type != TABLETYPE_PRUNING) {
+		LOG("[checkdata] Skipping '%s'\n", info.solver);
+		table += info.next;
+		err = readtableinfo_n(data_size, data, 2, &info);
+		if (err != NISSY_OK) {
+			LOG("[checkdata] Data is corrupt\n");
+			return err;
+		}
+	}
+	
+	LOG("[checkdata] Checking distribution for '%s' from "
+	    "table preamble\n", info.solver);
+	if (!distribution_equal(coord->pruning_distribution,
+	    info.distribution, coord->pruning_max)) {
+		LOG("[checkdata] Distribution from the table preamble does "
+		    "not match the expected one\n");
+		return NISSY_ERROR_DATA;
+	}
+
+	LOG("\n[checkdata] Checking distribution for '%s' from "
+	    "actual table\n", info.solver);
+	getdistribution(table, actual_distribution, &info);
+	if (!distribution_equal(coord->pruning_distribution,
+	    actual_distribution, coord->pruning_max)) {
+		LOG("[checkdata] Distribution from the actual table does "
+		    "not match the expected one\n");
+		return NISSY_ERROR_DATA;
+	}
+
+	return NISSY_OK;
+}
diff --git a/src/solvers/coord/coord.h b/src/solvers/coord/coord.h
index 2f24035..da93439 100644
--- a/src/solvers/coord/coord.h
+++ b/src/solvers/coord/coord.h
@@ -6,4 +6,5 @@
 #include "list.h"
 #include "utils.h"
 #include "gendata.h"
+#include "checkdata.h"
 #include "solve.h"
diff --git a/src/solvers/coord/dr.h b/src/solvers/coord/dr.h
index 607cc39..d986742 100644
--- a/src/solvers/coord/dr.h
+++ b/src/solvers/coord/dr.h
@@ -32,6 +32,22 @@ STATIC coord_t coordinate_dr = {
 	},
 	.is_admissible = &solution_lastqt_cw,
 	.is_solvable = &is_eoco_solvable,
+	.pruning_distribution = {
+		[0] = 1,
+		[1] = 1,
+		[2] = 5,
+		[3] = 44,
+		[4] = 487,
+		[5] = 5841,
+		[6] = 68364,
+		[7] = 776568,
+		[8] = 7950748,
+		[9] = 52098876,
+		[10] = 76236234,
+		[11] = 3771112,
+		[12] = 129,
+	},
+	.pruning_max = 12,
 	.sym = {
 		.classes = DREOESEP_CLASSES,
 		.max = DREOESEP_MAX,
diff --git a/src/solvers/coord/dreo.h b/src/solvers/coord/dreo.h
index 1cce4e2..f6fa3e9 100644
--- a/src/solvers/coord/dreo.h
+++ b/src/solvers/coord/dreo.h
@@ -27,6 +27,20 @@ STATIC coord_t coordinate_dreo = {
 	},
 	.is_admissible = &solution_lastqt_cw,
 	.is_solvable = &is_dreo_solvable,
+	.pruning_distribution = {
+		[0] = 1,
+		[1] = 1,
+		[2] = 4,
+		[3] = 22,
+		[4] = 160,
+		[5] = 1286,
+		[6] = 8550,
+		[7] = 42152,
+		[8] = 90748,
+		[9] = 33466,
+		[10] = 757,
+	},
+	.pruning_max = 10,
 	.sym = {
 		.classes = DRESEP_CLASSES,
 		.max = COMB_12_4,
diff --git a/src/solvers/coord/eo.h b/src/solvers/coord/eo.h
index 51dfb29..7735319 100644
--- a/src/solvers/coord/eo.h
+++ b/src/solvers/coord/eo.h
@@ -20,6 +20,17 @@ STATIC coord_t coordinate_eo = {
 	},
 	.is_admissible = &solution_lastqt_cw,
 	.is_solvable = &is_eo_even,
+	.pruning_distribution = {
+		[0] = 1,
+		[1] = 2,
+		[2] = 25,
+		[3] = 202,
+		[4] = 620,
+		[5] = 900,
+		[6] = 285,
+		[7] = 13,
+	},
+	.pruning_max = 7,
 	.sym = {0},
 };
 
diff --git a/src/solvers/coord/gendata.h b/src/solvers/coord/gendata.h
index 1b65f49..0cc012e 100644
--- a/src/solvers/coord/gendata.h
+++ b/src/solvers/coord/gendata.h
@@ -22,7 +22,7 @@ gendata_coord_dispatch(
 {
 	coord_t *coord;
 
-	parse_coord_and_axis(strlen(coordstr), coordstr, &coord, NULL);
+	parse_coord_and_axis(coordstr, &coord, NULL);
 
 	if (coord == NULL) {
 		LOG("Error: could not parse coordinate '%s'\n", coordstr);
diff --git a/src/solvers/coord/solve.h b/src/solvers/coord/solve.h
index df0f6ff..659040c 100644
--- a/src/solvers/coord/solve.h
+++ b/src/solvers/coord/solve.h
@@ -222,8 +222,7 @@ solve_coord_dispatch(
 	coord_t *coord;
 	uint8_t axis;
 
-	parse_coord_and_axis(
-	    strlen(coord_and_axis), coord_and_axis, &coord, &axis);
+	parse_coord_and_axis(coord_and_axis, &coord, &axis);
 
 	if (coord == NULL) {
 		LOG("Error: could not parse coordinate from '%s'\n",
diff --git a/src/solvers/coord/types_macros.h b/src/solvers/coord/types_macros.h
index a5ac7f3..0c7a3a2 100644
--- a/src/solvers/coord/types_macros.h
+++ b/src/solvers/coord/types_macros.h
@@ -26,6 +26,8 @@ typedef struct {
 	uint8_t axistrans[3];
 	bool (*is_admissible)(const solution_moves_t[static 1]);
 	bool (*is_solvable)(cube_t);
+	uint64_t pruning_distribution[INFO_DISTRIBUTION_LEN];
+	uint8_t pruning_max;
 	struct {
 		size_t classes;
 		uint64_t max;
diff --git a/src/solvers/coord/utils.h b/src/solvers/coord/utils.h
index aa4d74c..6066049 100644
--- a/src/solvers/coord/utils.h
+++ b/src/solvers/coord/utils.h
@@ -1,7 +1,6 @@
 STATIC coord_t *parse_coord(size_t n, const char [n]);
-STATIC uint8_t parse_axis(size_t n, const char [n]);
-STATIC void parse_coord_and_axis(
-    size_t n, const char [n], coord_t **, uint8_t *);
+STATIC uint8_t parse_axis(const char [static 2]);
+STATIC void parse_coord_and_axis(const char *, coord_t **, uint8_t *);
 STATIC long long dataid_coord(const char *, char [static NISSY_SIZE_DATAID]);
 
 STATIC coord_t *
@@ -17,13 +16,13 @@ parse_coord(size_t n, const char coord[n])
 }
 
 STATIC uint8_t
-parse_axis(size_t n, const char axis[n])
+parse_axis(const char axis[static 2])
 {
-	if (!strncmp(axis, "UD", n) || !strncmp(axis, "DU", n)) {
+	if (!strcmp(axis, "UD") || !strcmp(axis, "DU")) {
 		return AXIS_UD;
-	} else if (!strncmp(axis, "RL", n) || !strncmp(axis, "LR", n)) {
+	} else if (!strcmp(axis, "RL") || !strcmp(axis, "LR")) {
 		return AXIS_RL;
-	} else if (!strncmp(axis, "FB", n) || !strncmp(axis, "BF", n)) {
+	} else if (!strcmp(axis, "FB") || !strcmp(axis, "BF")) {
 		return AXIS_FB;
 	}
 
@@ -32,25 +31,22 @@ parse_axis(size_t n, const char axis[n])
 
 STATIC void
 parse_coord_and_axis(
-	size_t n,
-	const char str[n],
+	const char *str,
 	coord_t **coord,
 	uint8_t *axis
 )
 {
-	const char *s;
 	size_t i;
 
-	s = str + 6;
-	for (i = 0; i < n; i++)
-		if (s[i] == '_')
+	for (i = 6; i < strlen(str); i++)
+		if (str[i] == '_')
 			break;
 
 	if (coord != NULL)
-		*coord = parse_coord(i, s);
+		*coord = parse_coord(i-6, str+6);
 
 	if (axis != NULL)
-		*axis = i == n ? UINT8_ERROR : parse_axis(n-i-1, s+i+1);
+		*axis = i == strlen(str) ? UINT8_ERROR : parse_axis(str+i+1);
 }
 
 STATIC long long
@@ -58,7 +54,7 @@ dataid_coord(const char *ca, char dataid[static NISSY_SIZE_DATAID])
 {
 	coord_t *c;
 
-	parse_coord_and_axis(strlen(ca), ca, &c, NULL);
+	parse_coord_and_axis(ca, &c, NULL);
 
 	if (c == NULL) {
 		LOG("Error: cannot parse coordinate from '%s'\n", ca);
diff --git a/src/solvers/dispatch.h b/src/solvers/dispatch.h
index 0cfbc3b..7cf4441 100644
--- a/src/solvers/dispatch.h
+++ b/src/solvers/dispatch.h
@@ -2,7 +2,9 @@ typedef struct {
 	const char *prefix;
 	long long (*dataid)(const char *, char [static NISSY_SIZE_DATAID]);
 	long long (*gendata)(
-	    const char *, unsigned long long, unsigned char *);
+	    const char *, unsigned long long n, unsigned char [n]);
+	long long (*checkdata)(
+	    const char *, unsigned long long, const unsigned char *);
 	long long (*solve)(oriented_cube_t, const char *, unsigned, unsigned,
 	    unsigned, unsigned, unsigned, unsigned, unsigned long long,
 	    const unsigned char *, unsigned, char *,
@@ -17,12 +19,14 @@ solver_dispatch_t solver_dispatchers[] = {
 	.prefix = "h48",
 	.dataid = dataid_h48,
 	.gendata = gendata_h48_dispatch,
+	.checkdata = checkdata_h48,
 	.solve = solve_h48_dispatch,
 },
 {
 	.prefix = "coord_",
 	.dataid = dataid_coord,
 	.gendata = gendata_coord_dispatch,
+	.checkdata = checkdata_coord,
 	.solve = solve_coord_dispatch,
 },
 {
diff --git a/src/solvers/distribution.h b/src/solvers/distribution.h
index 1a3719c..3c006a7 100644
--- a/src/solvers/distribution.h
+++ b/src/solvers/distribution.h
@@ -13,6 +13,8 @@ typedef struct {
 STATIC void *getdistribution_runthread(void *);
 STATIC void getdistribution(const unsigned char *,
     uint64_t [static INFO_DISTRIBUTION_LEN], const tableinfo_t [static 1]);
+STATIC bool distribution_equal(const uint64_t [static INFO_DISTRIBUTION_LEN],
+    const uint64_t [static INFO_DISTRIBUTION_LEN], uint8_t);
 
 STATIC void *
 getdistribution_runthread(void *arg)
@@ -75,3 +77,28 @@ getdistribution(
 		distr[pval]++;
 	}
 }
+
+STATIC bool
+distribution_equal(
+	const uint64_t expected[static INFO_DISTRIBUTION_LEN],
+	const uint64_t actual[static INFO_DISTRIBUTION_LEN],
+	uint8_t maxvalue
+)
+{
+	int wrong;
+	uint8_t i;
+
+	for (i = 0, wrong = 0; i <= MIN(maxvalue, 20); i++) {
+		if (expected[i] != actual[i]) {
+			wrong++;
+			LOG("[checkdata] Value for depth %" PRIu8
+			    ": expected %" PRIu64 ", found %" PRIu64 "\n",
+			    i, expected[i], actual[i]);
+		} else {
+			LOG("[checkdata] Value for depth %" PRIu8
+			    " is correct (%" PRIu64 ")\n", i, actual[i]);
+		}
+	}
+
+	return wrong == 0;
+}
diff --git a/src/solvers/h48/checkdata.h b/src/solvers/h48/checkdata.h
new file mode 100644
index 0000000..53e616b
--- /dev/null
+++ b/src/solvers/h48/checkdata.h
@@ -0,0 +1,234 @@
+STATIC long long checkdata_h48(
+    const char *, unsigned long long n, const unsigned char [n]);
+
+/*
+Currently unused.
+TODO: re-introduce check on cocsep table
+*/
+uint64_t expected_cocsep[21] = {
+	[0] = 1,
+	[1] = 6,
+	[2] = 63,
+	[3] = 468,
+	[4] = 3068,
+	[5] = 15438,
+	[6] = 53814,
+	[7] = 71352,
+	[8] = 8784,
+	[9] = 96
+};
+
+struct {
+	uint8_t max;
+	uint64_t table[21];
+} expected_h48[12][9] = {
+	[0] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 5473562,
+				[1] = 34776317,
+				[2] = 68566704,
+				[3] = 8750867,
+			},
+		},
+		[4] = {
+			.max = 12,
+			.table = {
+				[0] = 1,
+				[1] = 1,
+				[2] = 4,
+				[3] = 34,
+				[4] = 331,
+				[5] = 3612,
+				[6] = 41605,
+				[7] = 474128,
+				[8] = 4953846,
+				[9] = 34776317,
+				[10] = 68566704,
+				[11] = 8749194,
+				[12] = 1673,
+			},
+		},
+	},
+	[1] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 6012079,
+				[1] = 45822302,
+				[2] = 142018732,
+				[3] = 41281787,
+			},
+		},
+	},
+	[2] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 6391286,
+				[1] = 55494785,
+				[2] = 252389935,
+				[3] = 155993794,
+			},
+		},
+	},
+	[3] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 6686828,
+				[1] = 63867852,
+				[2] = 392789689,
+				[3] = 477195231,
+			},
+		},
+	},
+	[4] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 77147213,
+				[1] = 543379415,
+				[2] = 1139570251,
+				[3] = 120982321,
+			},
+		},
+	},
+	[5] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 82471284,
+				[1] = 687850732,
+				[2] = 2345840746,
+				[3] = 645995638,
+			},
+		},
+	},
+	[6] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 85941099,
+				[1] = 804752968,
+				[2] = 4077248182,
+				[3] = 2556374551,
+			},
+		},
+	},
+	[7] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 88529761,
+				[1] = 897323475,
+				[2] = 6126260791,
+				[3] = 7936519573,
+			},
+		},
+	},
+	[8] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 1051579940,
+				[1] = 8136021316,
+				[2] = 19024479822,
+				[3] = 18851861220,
+			},
+		},
+	},
+	[9] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 1102038189,
+				[1] = 9888265242,
+				[2] = 38299375805,
+				[3] = 10904855164,
+			},
+		},
+	},
+	[10] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 1133240039,
+				[1] = 11196285614,
+				[2] = 64164702961,
+				[3] = 43894840186,
+			},
+		},
+	},
+	[11] = {
+		[2] = {
+			.max = 3,
+			.table = {
+				[0] = 1150763161,
+				[1] = 12045845660,
+				[2] = 91163433330,
+				[3] = 136418095449,
+			},
+		},
+	},
+};
+
+STATIC long long
+checkdata_h48(
+	const char *solver,
+	unsigned long long data_size,
+	const unsigned char data[data_size]
+)
+{
+	const unsigned char *table;
+	tableinfo_t info;
+	int64_t err;
+	uint64_t actual_distribution[INFO_DISTRIBUTION_LEN];
+	uint8_t em;
+	uint64_t *ed;
+
+	if ((size_t)data % 8 != 0) {
+		LOG("[checkdata] Error: buffer is not 8-byte aligned\n");
+		return NISSY_ERROR_DATA;
+	}
+
+	do {
+		err = readtableinfo(data_size, data, &info);
+		if (err != NISSY_OK) {
+			LOG("[checkdata] Data is corrupt\n");
+			return NISSY_ERROR_DATA;
+		}
+
+		table = data + INFOSIZE;
+		data += info.next;
+		data_size -= info.next;
+
+		if (info.type != TABLETYPE_PRUNING) {
+			LOG("[checkdata] Skipping '%s'\n", info.solver);
+			continue;
+		}
+
+		ed = expected_h48[info.h48h][info.bits].table;
+		em = expected_h48[info.h48h][info.bits].max;
+
+		LOG("[checkdata] Checking distribution for '%s' from "
+		    "table preamble\n", info.solver);
+		if (!distribution_equal(ed, info.distribution, em)) {
+			LOG("[checkdata] Distribution from the table preamble "
+			    "does not matche the expected one\n");
+			return NISSY_ERROR_DATA;
+		}
+
+		LOG("[checkdata] Checking distribution for '%s' from "
+		    "actual table\n", info.solver);
+		getdistribution(table, actual_distribution, &info);
+		if (!distribution_equal(ed, actual_distribution, em)) {
+			LOG("[checkdata] Distribution from the actual table "
+			    "does not match the expected one\n");
+			return NISSY_ERROR_DATA;
+		}
+	} while (info.next != 0);
+
+	return NISSY_OK;
+}
diff --git a/src/solvers/h48/gendata_eoesep.h b/src/solvers/h48/gendata_eoesep.h
index f1e3d4c..52e2b8b 100644
--- a/src/solvers/h48/gendata_eoesep.h
+++ b/src/solvers/h48/gendata_eoesep.h
@@ -79,7 +79,7 @@ gendata_eoesep(unsigned char *buf, uint8_t maxdepth)
 	if (buf == NULL)
 		goto gendata_eoesep_return_size;
 
-	LOG("Computing eoesep data\n");
+	LOG("[H48 gendata] Computing eoesep data\n");
 	memset(buf, 0xFF, EOESEP_FULLSIZE);
 	esep_classes = (uint32_t *)(buf + INFOSIZE);
 	buf8 = buf + INFOSIZE + 4*ESEP_MAX;
@@ -111,7 +111,7 @@ gendata_eoesep(unsigned char *buf, uint8_t maxdepth)
 
 	writetableinfo(&info, EOESEP_FULLSIZE, buf);
 
-	LOG("eoesep data computed\n");
+	LOG("[H48 gendata] eoesep data computed\n");
 
 gendata_eoesep_return_size:
 	return EOESEP_FULLSIZE;
diff --git a/src/solvers/h48/gendata_h48.h b/src/solvers/h48/gendata_h48.h
index 644ddc2..d1ab3bd 100644
--- a/src/solvers/h48/gendata_h48.h
+++ b/src/solvers/h48/gendata_h48.h
@@ -25,8 +25,6 @@ STATIC_INLINE uint8_t get_h48_pval_atomic(
 STATIC_INLINE void set_h48_pval_atomic(
     _Atomic unsigned char *, int64_t, uint8_t, uint8_t);
 
-size_t gendata_h48_derive(uint8_t, const unsigned char *, unsigned char *);
-
 STATIC long long
 gendata_h48_dispatch(
 	const char *solver,
@@ -711,91 +709,3 @@ set_h48_pval_atomic(
 	table[H48_INDEX(i, k)] = (table[H48_INDEX(i, k)] & (~H48_MASK(i, k)))
 	    | (val << H48_SHIFT(i, k));
 }
-
-size_t
-gendata_h48_derive(uint8_t h, const unsigned char *fulltable, unsigned char *buf)
-{
-	size_t cocsepsize, h48size;
-	uint8_t val_full, val_derive;
-	const unsigned char *h48full;
-	unsigned char *h48derive;
-	int64_t i, j, h48max;
-	uint64_t bufsize;
-	gendata_h48_arg_t arg;
-	tableinfo_t cocsepinfo, fulltableinfo;
-
-	/* Initializing values in case of error */
-	/* TODO cleanup this */
-	fulltableinfo.h48h = 11;
-	fulltableinfo.bits = 2;
-	fulltableinfo.base = 8;
-
-	int64_t TODOlarge = 999999999999; /* TODO: cleanup here */
-
-	readtableinfo_n(TODOlarge, fulltable, 2, &fulltableinfo);
-	arg.h = h;
-	arg.k = fulltableinfo.bits;
-	arg.maxdepth = 20;
-	arg.buf = buf;
-	arg.cocsepdata = (uint32_t *)(buf + INFOSIZE);
-	arg.base = fulltableinfo.base;
-	arg.info = makeinfo_h48k2(&arg);
-
-	/* Technically this step is redundant, except that we
-	   need selfsim and crep */
-	cocsepsize = gendata_cocsep(buf, arg.selfsim, arg.crep);
-	arg.h48buf = (_Atomic unsigned char *)buf + cocsepsize;
-	h48size = H48_TABLESIZE(h, arg.k) + INFOSIZE;
-
-	if (buf == NULL)
-		goto gendata_h48_derive_return_size;
-
-	bufsize = COCSEP_FULLSIZE + INFOSIZE;
-	if (readtableinfo(bufsize, buf, &cocsepinfo) != NISSY_OK) {
-		LOG("[H48 derive gendata] Error: could not read info for "
-		    "cocsep table\n");
-		goto gendata_h48_derive_error;
-	}
-
-	cocsepinfo.next = cocsepsize;
-	bufsize = COCSEP_FULLSIZE + INFOSIZE;
-	if (writetableinfo(&cocsepinfo, bufsize, buf) != NISSY_OK) {
-		LOG("[H48 derive gendata] Error: could not write info for "
-		    "cocsep table with updated 'next' value\n");
-		goto gendata_h48_derive_error;
-	}
-
-	h48full = fulltable + cocsepsize + INFOSIZE;
-	h48derive = (unsigned char *)arg.h48buf + INFOSIZE;
-	memset(h48derive, 0xFF, H48_TABLESIZE(h, arg.k));
-	memset(arg.info.distribution, 0,
-	    INFO_DISTRIBUTION_LEN * sizeof(uint64_t));
-
-	h48max = H48_COORDMAX(fulltableinfo.h48h);
-	for (i = 0; i < h48max; i++) {
-		if (i % INT64_C(1000000000) == 0 && i > 0)
-			LOG("[H48 derive gendata] Processing %" PRId64
-			    "th coordinate\n", i);
-		j = i >> (int64_t)(fulltableinfo.h48h - h);
-		val_full = get_h48_pval(h48full, i, arg.k);
-		val_derive = get_h48_pval(h48derive, j, arg.k);
-		set_h48_pval(
-		    h48derive, j, arg.k, MIN(val_full, val_derive));
-	}
-
-	getdistribution(h48derive, arg.info.distribution, &arg.info);
-
-	bufsize = arg.buf_size - COCSEP_FULLSIZE - INFOSIZE;
-	if (writetableinfo(&arg.info, bufsize, (unsigned char *)arg.h48buf)
-	    != NISSY_OK) {
-		LOG("H48 derive gendata] Error: could not write info "
-		    "for table\n");
-		goto gendata_h48_derive_error;
-	}
-
-gendata_h48_derive_return_size:
-	return cocsepsize + h48size;
-
-gendata_h48_derive_error:
-	return 0;
-}
diff --git a/src/solvers/h48/h48.h b/src/solvers/h48/h48.h
index 79d2583..44fdb59 100644
--- a/src/solvers/h48/h48.h
+++ b/src/solvers/h48/h48.h
@@ -7,4 +7,5 @@
 #include "gendata_cocsep.h"
 #include "gendata_eoesep.h"
 #include "gendata_h48.h"
+#include "checkdata.h"
 #include "solve.h"
-- 
cgit v1.3