SabreTools/SabreTools.Library/DatItems/Media.cs

using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Xml.Serialization;

using SabreTools.Core;
using SabreTools.Filtering;
using SabreTools.Library.FileTypes;
using Newtonsoft.Json;

namespace SabreTools.Library.DatItems
{
    /// <summary>
    /// Represents Aaruformat images which use internal hashes
    /// </summary>
    [JsonObject("media"), XmlRoot("media")]
    public class Media : DatItem
    {
        #region Private instance variables

        private byte[] _md5; // 16 bytes
        private byte[] _sha1; // 20 bytes
        private byte[] _sha256; // 32 bytes
        private byte[] _spamsum; // variable bytes

        #endregion

        #region Fields

        /// <summary>
        /// Name of the item
        /// </summary>
        [JsonProperty("name")]
        [XmlElement("name")]
        public string Name { get; set; }

        /// <summary>
        /// Data MD5 hash
        /// </summary>
        [JsonProperty("md5", DefaultValueHandling = DefaultValueHandling.Ignore)]
        [XmlElement("md5")]
        public string MD5
        {
            get { return _md5.IsNullOrEmpty() ? null : Utilities.ByteArrayToString(_md5); }
            set { _md5 = Utilities.StringToByteArray(Sanitizer.CleanMD5(value)); }
        }

        /// <summary>
        /// Data SHA-1 hash
        /// </summary>
        [JsonProperty("sha1", DefaultValueHandling = DefaultValueHandling.Ignore)]
        [XmlElement("sha1")]
        public string SHA1
        {
            get { return _sha1.IsNullOrEmpty() ? null : Utilities.ByteArrayToString(_sha1); }
            set { _sha1 = Utilities.StringToByteArray(Sanitizer.CleanSHA1(value)); }
        }

        /// <summary>
        /// Data SHA-256 hash
        /// </summary>
        [JsonProperty("sha256", DefaultValueHandling = DefaultValueHandling.Ignore)]
        [XmlElement("sha256")]
        public string SHA256
        {
            get { return _sha256.IsNullOrEmpty() ? null : Utilities.ByteArrayToString(_sha256); }
            set { _sha256 = Utilities.StringToByteArray(Sanitizer.CleanSHA256(value)); }
        }

        /// <summary>
        /// File SpamSum fuzzy hash
        /// </summary>
        [JsonProperty("spamsum", DefaultValueHandling = DefaultValueHandling.Ignore)]
        [XmlElement("spamsum")]
        public string SpamSum
        {
            get { return _spamsum.IsNullOrEmpty() ? null : Encoding.UTF8.GetString(_spamsum); }
            set { _spamsum = Encoding.UTF8.GetBytes(value ?? string.Empty); }
        }

        #endregion

        #region Accessors

        /// <summary>
        /// Gets the name to use for a DatItem
        /// </summary>
        /// <returns>Name if available, null otherwise</returns>
        public override string GetName()
        {
            return Name;
        }

        /// <summary>
        /// Set fields with given values
        /// </summary>
        /// <param name="mappings">Mappings dictionary</param>
        public override void SetFields(Dictionary<Field, string> mappings)
        {
            // Set base fields
            base.SetFields(mappings);

            // Handle Media-specific fields
            if (mappings.Keys.Contains(Field.DatItem_Name))
                Name = mappings[Field.DatItem_Name];

            if (mappings.Keys.Contains(Field.DatItem_MD5))
                MD5 = mappings[Field.DatItem_MD5];

            if (mappings.Keys.Contains(Field.DatItem_SHA1))
                SHA1 = mappings[Field.DatItem_SHA1];

            if (mappings.Keys.Contains(Field.DatItem_SHA256))
                SHA256 = mappings[Field.DatItem_SHA256];

            if (mappings.Keys.Contains(Field.DatItem_SpamSum))
                SpamSum = mappings[Field.DatItem_SpamSum];
        }

        #endregion

        #region Constructors

        /// <summary>
        /// Create a default, empty Media object
        /// </summary>
        public Media()
        {
            Name = string.Empty;
            ItemType = ItemType.Media;
            DupeType = 0x00;
        }

        /// <summary>
        /// Create a Media object from a BaseFile
        /// </summary>
        /// <param name="baseFile"></param>
        public Media(BaseFile baseFile)
        {
            Name = baseFile.Filename;
            _md5 = baseFile.MD5;
            _sha1 = baseFile.SHA1;
            _sha256 = baseFile.SHA256;
            _spamsum = baseFile.SpamSum;

            ItemType = ItemType.Media;
            DupeType = 0x00;
        }

        #endregion

        #region Cloning Methods

        public override object Clone()
        {
            return new Media()
            {
                ItemType = this.ItemType,
                DupeType = this.DupeType,

                Machine = this.Machine.Clone() as Machine,
                Source = this.Source.Clone() as Source,
                Remove = this.Remove,

                Name = this.Name,
                _md5 = this._md5,
                _sha1 = this._sha1,
                _sha256 = this._sha256,
                _spamsum = this._spamsum,
            };
        }

        /// <summary>
        /// Convert Media object to a BaseFile
        /// </summary>
        public BaseFile ConvertToBaseFile()
        {
            return new BaseFile()
            {
                Filename = this.Name,
                Parent = this.Machine?.Name,
                MD5 = this._md5,
                SHA1 = this._sha1,
                SHA256 = this._sha256,
                SpamSum = this._spamsum,
            };
        }

        /// <summary>
        /// Convert a media to the closest Rom approximation
        /// </summary>
        /// <returns></returns>
        public Rom ConvertToRom()
        {
            var rom = new Rom()
            {
                ItemType = ItemType.Rom,
                DupeType = this.DupeType,

                Machine = this.Machine.Clone() as Machine,
                Source = this.Source.Clone() as Source,
                Remove = this.Remove,

                Name = this.Name + ".aif",
                MD5 = this.MD5,
                SHA1 = this.SHA1,
                SHA256 = this.SHA256,
                SpamSum = this.SpamSum,
            };

            return rom;
        }

        #endregion

        #region Comparision Methods

        public override bool Equals(DatItem other)
        {
            bool dupefound = false;

            // If we don't have a Media, return false
            if (ItemType != other.ItemType)
                return dupefound;

            // Otherwise, treat it as a Media
            Media newOther = other as Media;

            // If we get a partial match
            if (HashMatch(newOther))
                dupefound = true;

            return dupefound;
        }

        /// <summary>
        /// Fill any missing size and hash information from another Media
        /// </summary>
        /// <param name="other">Media to fill information from</param>
        public void FillMissingInformation(Media other)
        {
            if (_md5.IsNullOrEmpty() && !other._md5.IsNullOrEmpty())
                _md5 = other._md5;

            if (_sha1.IsNullOrEmpty() && !other._sha1.IsNullOrEmpty())
                _sha1 = other._sha1;

            if (_sha256.IsNullOrEmpty() && !other._sha256.IsNullOrEmpty())
                _sha256 = other._sha256;

            if (_spamsum.IsNullOrEmpty() && !other._spamsum.IsNullOrEmpty())
                _spamsum = other._spamsum;
        }

        /// <summary>
        /// Get unique duplicate suffix on name collision
        /// </summary>
        /// <returns>String representing the suffix</returns>
        public string GetDuplicateSuffix()
        {
             if (!_md5.IsNullOrEmpty())
                return $"_{MD5}";
            else if (!_sha1.IsNullOrEmpty())
                return $"_{SHA1}";
            else if (!_sha256.IsNullOrEmpty())
                return $"_{SHA256}";
            else if (!_spamsum.IsNullOrEmpty())
                return $"_{SpamSum}";
            else
                return "_1";
        }

        /// <summary>
        /// Returns if there are no, non-empty hashes in common with another Media
        /// </summary>
        /// <param name="other">Media to compare against</param>
        /// <returns>True if at least one hash is not mutually exclusive, false otherwise</returns>
        private bool HasCommonHash(Media other)
        {
            return !(_md5.IsNullOrEmpty() ^ other._md5.IsNullOrEmpty())
                || !(_sha1.IsNullOrEmpty() ^ other._sha1.IsNullOrEmpty())
                || !(_sha256.IsNullOrEmpty() ^ other._sha256.IsNullOrEmpty())
                || !(_spamsum.IsNullOrEmpty() ^ other._spamsum.IsNullOrEmpty());
        }

        /// <summary>
        /// Returns if the Media contains any hashes
        /// </summary>
        /// <returns>True if any hash exists, false otherwise</returns>
        private bool HasHashes()
        {
            return !_md5.IsNullOrEmpty()
                || !_sha1.IsNullOrEmpty()
                || !_sha256.IsNullOrEmpty()
                || !_spamsum.IsNullOrEmpty();
        }

        /// <summary>
        /// Returns if any hashes are common with another Media
        /// </summary>
        /// <param name="other">Media to compare against</param>
        /// <returns>True if any hashes are in common, false otherwise</returns>
        private bool HashMatch(Media other)
        {
            // If either have no hashes, we return false, otherwise this would be a false positive
            if (!HasHashes() || !other.HasHashes())
                return false;

            // If neither have hashes in common, we return false, otherwise this would be a false positive
            if (!HasCommonHash(other))
                return false;

            // Return if all hashes match according to merge rules
            return ConditionalHashEquals(_md5, other._md5)
                && ConditionalHashEquals(_sha1, other._sha1)
                && ConditionalHashEquals(_spamsum, other._spamsum);
        }

        #endregion

        #region Filtering

        /// <summary>
        /// Clean a DatItem according to the cleaner
        /// </summary>
        /// <param name="cleaner">Cleaner to implement</param>
        public override void Clean(Cleaner cleaner)
        {
            // Clean common items first
            base.Clean(cleaner);

            // If we're stripping unicode characters, strip item name
            if (cleaner?.RemoveUnicode == true)
                Name = Sanitizer.RemoveUnicodeCharacters(Name);

            // If we are in NTFS trim mode, trim the game name
            if (cleaner?.Trim == true)
            {
                // Windows max name length is 260
                int usableLength = 260 - Machine.Name.Length - (cleaner.Root?.Length ?? 0);
                if (Name.Length > usableLength)
                {
                    string ext = Path.GetExtension(Name);
                    Name = Name.Substring(0, usableLength - ext.Length);
                    Name += ext;
                }
            }
        }

        /// <summary>
        /// Check to see if a DatItem passes the filter
        /// </summary>
        /// <param name="filter">Filter to check against</param>
        /// <param name="sub">True if this is a subitem, false otherwise</param>
        /// <returns>True if the item passed the filter, false otherwise</returns>
        public override bool PassesFilter(Filter filter, bool sub = false)
        {
            // Check common fields first
            if (!base.PassesFilter(filter, sub))
                return false;

            // Filter on item name
            if (!filter.PassStringFilter(filter.DatItem_Name, Name))
                return false;

            // Filter on MD5
            if (!filter.PassStringFilter(filter.DatItem_MD5, MD5))
                return false;

            // Filter on SHA-1
            if (!filter.PassStringFilter(filter.DatItem_SHA1, SHA1))
                return false;

            // Filter on SHA-256
            if (!filter.PassStringFilter(filter.DatItem_SHA256, SHA256))
                return false;

            // Filter on SpamSum
            if (!filter.PassStringFilter(filter.DatItem_SpamSum, SpamSum))
                return false;

            return true;
        }

        /// <summary>
        /// Remove fields from the DatItem
        /// </summary>
        /// <param name="fields">List of Fields to remove</param>
        public override void RemoveFields(List<Field> fields)
        {
            // Remove common fields first
            base.RemoveFields(fields);

            // Remove the fields
            if (fields.Contains(Field.DatItem_Name))
                Name = null;

            if (fields.Contains(Field.DatItem_MD5))
                MD5 = null;

            if (fields.Contains(Field.DatItem_SHA1))
                SHA1 = null;

            if (fields.Contains(Field.DatItem_SHA256))
                SHA256 = null;

            if (fields.Contains(Field.DatItem_SpamSum))
                SpamSum = null;
        }

        /// <summary>
        /// Set internal names to match One Rom Per Game (ORPG) logic
        /// </summary>
        public override void SetOneRomPerGame()
        {
            string[] splitname = Name.Split('.');
            Machine.Name += $"/{string.Join(".", splitname.Take(splitname.Length > 1 ? splitname.Length - 1 : 1))}";
            Name = Path.GetFileName(Name);
        }

        #endregion

        #region Sorting and Merging

        /// <summary>
        /// Get the dictionary key that should be used for a given item and bucketing type
        /// </summary>
        /// <param name="bucketedBy">Field enum representing what key to get</param>
        /// <param name="lower">True if the key should be lowercased (default), false otherwise</param>
        /// <param name="norename">True if games should only be compared on game and file name, false if system and source are counted</param>
        /// <returns>String representing the key to be used for the DatItem</returns>
        public override string GetKey(Field bucketedBy, bool lower = true, bool norename = true)
        {
            // Set the output key as the default blank string
            string key = string.Empty;

            // Now determine what the key should be based on the bucketedBy value
            switch (bucketedBy)
            {
                case Field.DatItem_MD5:
                    key = MD5;
                    break;

                case Field.DatItem_SHA1:
                    key = SHA1;
                    break;

                case Field.DatItem_SHA256:
                    key = SHA256;
                    break;

                case Field.DatItem_SpamSum:
                    key = SpamSum;
                    break;

                // Let the base handle generic stuff
                default:
                    return base.GetKey(bucketedBy, lower, norename);
            }

            // Double and triple check the key for corner cases
            if (key == null)
                key = string.Empty;

            return key;
        }

        /// <summary>
        /// Replace fields from another item
        /// </summary>
        /// <param name="item">DatItem to pull new information from</param>
        /// <param name="fields">List of Fields representing what should be updated</param>
        public override void ReplaceFields(DatItem item, List<Field> fields)
        {
            // Replace common fields first
            base.ReplaceFields(item, fields);

            // If we don't have a Media to replace from, ignore specific fields
            if (item.ItemType != ItemType.Media)
                return;

            // Cast for easier access
            Media newItem = item as Media;

            // Replace the fields
            if (fields.Contains(Field.DatItem_Name))
                Name = newItem.Name;

            if (fields.Contains(Field.DatItem_MD5))
            {
                if (string.IsNullOrEmpty(MD5) && !string.IsNullOrEmpty(newItem.MD5))
                    MD5 = newItem.MD5;
            }

            if (fields.Contains(Field.DatItem_SHA1))
            {
                if (string.IsNullOrEmpty(SHA1) && !string.IsNullOrEmpty(newItem.SHA1))
                    SHA1 = newItem.SHA1;
            }

            if (fields.Contains(Field.DatItem_SHA256))
            {
                if (string.IsNullOrEmpty(SHA256) && !string.IsNullOrEmpty(newItem.SHA256))
                    SHA256 = newItem.SHA256;
            }

            if (fields.Contains(Field.DatItem_SpamSum))
            {
                if (string.IsNullOrEmpty(SpamSum) && !string.IsNullOrEmpty(newItem.SpamSum))
                    SpamSum = newItem.SpamSum;
            }
        }

        #endregion
    }
}