Navmesher_old/espReader/ESPReader.h

#pragma once

/* For reading structured data out of Creation Engine esp/esm files.
 * Based on information from:
 *     https://en.uesp.net/wiki/Skyrim_Mod:Mod_File_Format
 *
 * esp/esm files generally have a tree-like structure where:
 * - top level is a a TES4 record concatenated with each of the top-level groups
 * - groups contain a concatenation of 0 or more groups or records
 * - records contain a concatenation of 0 or more fields
 * - fields contain a structure of statically or dynamically sized data
 * - groups, records, and fields all have headers
 *
 * === NOTE ===
 * Assumptions:
 * - Little endian system
 * - Compiled with MSVC
 */

#include <stdint.h>
#include "msh.h"

 // Guards for C++ usage
#ifdef __cplusplus
extern "C" {
#endif

#define RFS_INNER_SIZE 32

#define RT_SIZE 128

#define RT_HASH_SIZE 512
#define RT_HASH_BITS 9
#define RT_HASH_SEED 131261257

#define COMPRESSED_FLAG (((uint32_t)1) << 18)

#define GO_SIZE 118
#define GTS_SIZE 10


//
// === FORWARD DEFS ===
//

	typedef union type4 Type4;
	typedef struct timestamp Timestamp;
	typedef struct group Group;
	typedef struct record Record;
	typedef struct field Field;

//
// === SIMPLE TYPES ===
//

	// Basic types
	typedef uint32_t formid;

	// char[4] with uint32_t access
	union type4 {
		char bytes[4];
		uint32_t uint;
	};

	// indexed by flag bit
	typedef const char *const rfs_inner[RFS_INNER_SIZE];

//
// === ENUMS ===
//

	// Tag for generic node tagged union
	enum node_type { // NT_ prefix
		NT_GROUP,
		NT_RECORD,
	};

	// Record type enum
	enum record_type {
		_NONE,
		AACT, ACHR, ACTI, ADDN, ALCH, AMMO,
		ANIO, APPA, ARMA, ARMO, ARTO, ASPC,
		ASTP, AVIF, BOOK, BPTD, CAMS, CELL,
		CLAS, CLDC, CLFM, CLMT, COBJ, COLL,
		CONT, CPTH, CSTY, DEBR, DIAL, DLBR,
		DLVW, DOBJ, DOOR, DUAL, ECZN, EFSH,
		ENCH, EQUP, EXPL, EYES, FACT, FLOR,
		FLST, FSTP, FSTS, FURN, GLOB, GMST,
		GRAS, GRUP, HAIR, HAZD, HDPT, IDLE,
		IDLM, IMAD, IMGS, INFO, INGR, IPCT,
		IPDS, KEYM, KYWD, LAND, LCRT, LCTN,
		LGTM, LIGH, LSCR, LTEX, LVLI, LVLN,
		LVSP, MATO, MATT, MESG, MGEF, MISC,
		MOVT, MSTT, MUSC, MUST, NAVI, NAVM,
		NOTE, NPC_, OTFT, PACK, PERK, PGRE,
		PHZD, PROJ, PWAT, QUST, RACE, REFR,
		REGN, RELA, REVB, RFCT, RGDL, SCEN,
		SCOL, SCPT, SCRL, SHOU, SLGM, SMBN,
		SMEN, SMQN, SNCT, SNDR, SOPM, SOUN,
		SPEL, SPGD, STAT, TACT, TES4, TREE,
		TXST, VTYP, WATR, WEAP, WOOP, WRLD,
		WTHR,
	};

	// Enums of perfect hash values for 
	enum record_type_hash {
		RT_AACT = 496, RT_ACHR = 249, RT_ACTI = 293, RT_ADDN = 316,
		RT_ALCH = 312, RT_AMMO = 157, RT_ANIO = 297, RT_APPA = 230,
		RT_ARMA = 222, RT_ARMO = 218, RT_ARTO = 328, RT_ASPC = 252,
		RT_ASTP = 384, RT_AVIF = 309, RT_BOOK = 318, RT_BPTD = 454,
		RT_CAMS = 319, RT_CELL = 18, RT_CLAS = 16, RT_CLDC = 68,
		RT_CLFM = 389, RT_CLMT = 497, RT_COBJ = 422, RT_COLL = 140,
		RT_CONT = 169, RT_CPTH = 30, RT_CSTY = 193, RT_DEBR = 460,
		RT_DIAL = 304, RT_DLBR = 340, RT_DLVW = 434, RT_DOBJ = 437,
		RT_DOOR = 347, RT_DUAL = 246, RT_ECZN = 229, RT_EFSH = 509,
		RT_ENCH = 194, RT_EQUP = 57, RT_EXPL = 153, RT_EYES = 181,
		RT_FACT = 62, RT_FLOR = 137, RT_FLST = 199, RT_FSTP = 462,
		RT_FSTS = 388, RT_FURN = 105, RT_GLOB = 376, RT_GMST = 125,
		RT_GRAS = 49, RT_GRUP = 511, RT_HAIR = 481, RT_HAZD = 21,
		RT_HDPT = 100, RT_IDLE = 204, RT_IDLM = 348, RT_IMAD = 390,
		RT_IMGS = 187, RT_INFO = 82, RT_INGR = 463, RT_IPCT = 292,
		RT_IPDS = 89, RT_KEYM = 54, RT_KYWD = 123, RT_LAND = 261,
		RT_LCRT = 138, RT_LCTN = 172, RT_LGTM = 177, RT_LIGH = 13,
		RT_LSCR = 288, RT_LTEX = 447, RT_LVLI = 235, RT_LVLN = 453,
		RT_LVSP = 343, RT_MATO = 2, RT_MATT = 220, RT_MESG = 66,
		RT_MGEF = 32, RT_MISC = 145, RT_MOVT = 378, RT_MSTT = 132,
		RT_MUSC = 87, RT_MUST = 9, RT_NAVI = 50, RT_NAVM = 122,
		RT_NOTE = 366, RT_NPC_ = 440, RT_OTFT = 365, RT_PACK = 441,
		RT_PERK = 243, RT_PGRE = 210, RT_PHZD = 26, RT_PROJ = 120,
		RT_PWAT = 397, RT_QUST = 71, RT_RACE = 108, RT_REFR = 449,
		RT_REGN = 320, RT_RELA = 182, RT_REVB = 266, RT_RFCT = 311,
		RT_RGDL = 19, RT_SCEN = 119, RT_SCOL = 276, RT_SCPT = 363,
		RT_SCRL = 104, RT_SHOU = 115, RT_SLGM = 70, RT_SMBN = 413,
		RT_SMEN = 240, RT_SMQN = 63, RT_SNCT = 117, RT_SNDR = 280,
		RT_SOPM = 306, RT_SOUN = 165, RT_SPEL = 190, RT_SPGD = 443,
		RT_STAT = 202, RT_TACT = 282, RT_TES4 = 474, RT_TREE = 27,
		RT_TXST = 359, RT_VTYP = 335, RT_WATR = 84, RT_WEAP = 10,
		RT_WOOP = 352, RT_WRLD = 38, RT_WTHR = 83,
	};

	// GRUP type values
	enum group_type { // GT_ prefix
		GT_TOP = 0,
		GT_WORLD_CHILDREN = 1,
		GT_INTERIOR_CELL_BLOCK = 2,
		GT_INTERIOR_CELL_SUBBLOCK = 3,
		GT_EXTERIOR_CELL_BLOCK = 4,
		GT_EXTERIOR_CELL_SUBBLOCK = 5,
		GT_CELL_CHILDREN = 6,
		GT_TOPIC_CHILDREN = 7,
		GT_CELL_PERSISTENT_CHILDREN = 8,
		GT_CELL_TEMPORARY_CHILDREN = 9,
	};

//
// === COMPOSITE TYPES ===
//

	// Generic node
	typedef struct node Node;
	struct node {
		union {
			Group *group;
			Record *record;
		} header;
		char *const data;
		enum node_type type;
		uint32_t _pad;
	};

	// calculated timestamp
	struct timestamp {
		uint16_t year;
		uint8_t month;
		uint8_t day;
	};

//
// === BINARY DATA OVERLAYS ===
//

#pragma pack(push, 1)

	// Group header overlay
	struct group {
		Type4 grup; // always RT_GRUP
		uint32_t size; // uncludes the 24 byte group header
		union {
			Type4 type; // this may be mangled and should not be relied on
			formid formid;
			int32_t number;
			int16_t coord[2];
		} label; // access determined by the `type` below
		int32_t type; // group_type enum
		uint16_t timestamp;
		uint16_t vcinfo;
		uint32_t unknown;
	};

	// Record header overlay
	struct record {
		Type4 type;
		uint32_t size;
		uint32_t flags;
		uint32_t formid;
		uint16_t timestamp;
		uint16_t vcinfo;
		uint16_t version;
		uint16_t unknown;
	};

	// Field header overlay
	struct field {
		Type4 type;
		uint16_t size;
	};

#pragma pack(pop)

//
// === LUTs ===
//

	// record type enum to fourcc value
	extern const uint32_t rt[RT_SIZE];

	// for converting between record_type and record_type_hash enums
	extern const uint16_t rt2rth[RT_SIZE];
	extern const uint8_t rth2rt[RT_HASH_SIZE];

	// type -> flag mappings
	// NULL table pointers indicate no flags
	// NULL string pointers indicate invalid flag
	extern rfs_inner *const rfs[RT_HASH_SIZE];

	extern rfs_inner *const rfs_refr[RT_HASH_SIZE];

	// Expected (probably) order of top level groups in an esp/esm
	extern const enum record_type group_order[GO_SIZE];

	// Printable strings for group types
	extern const char *const group_type_strings[GTS_SIZE];


//
// === FUNCTIONS ===
//

	// hashes type value into RT_ hash value
	inline uint32_t rt_hash(uint32_t type) {
		return uint32_t_msh(type, RT_HASH_BITS, RT_HASH_SEED);
	}

	/* `espr_walk` walks through the tree structure of the esp/esm binary data 
	 * starting at `data` of `size` bytes. 
	 *
	 * `cb` is a callback that takes a `Node` to process. `pt` is a pointer to
	 *  arbitrary data that is passed on to `cb` whenever it is called.
	 *
	 * Data is walked sequentially. Nodes passed to `cb` will be strictly increasing
	 * in terms of memory location within the buffer.
	 */
	void espr_walk(char *data, size_t size, void (*cb)(Node n, void *pt), void *pt);

	/* `espr_print` prints the header of every group and record in the given 
	 * esp/esm binary data.
	 */
	void espr_print(char *data, size_t size);

	/* Calculates the size of the esp data if all of the compressed records are
	 * decompressed.
	 */
	size_t espr_decompressed_size(char *data, size_t size);

	/* Counts the number of formids present in the esp/esm data. This should be
	 * equal to the number of records.
	 */
	size_t espr_formid_count(char *data, size_t size);

// End C++ guard
#ifdef __cplusplus
}
#endif