Files
SabreTools/SabreTools.DatFiles/ItemDictionary.cs

1092 lines
40 KiB
C#
Raw Normal View History

using System;
using System.Collections;
2024-02-28 22:54:56 -05:00
#if NET40_OR_GREATER || NETCOREAPP
2020-08-31 15:54:53 -07:00
using System.Collections.Concurrent;
2024-02-28 22:54:56 -05:00
#endif
2020-07-26 22:34:45 -07:00
using System.Collections.Generic;
2024-03-19 16:18:35 -04:00
using System.IO;
2020-07-26 22:34:45 -07:00
using System.Linq;
2024-03-05 03:04:47 -05:00
#if NET40_OR_GREATER || NETCOREAPP
2020-07-26 22:34:45 -07:00
using System.Threading.Tasks;
2024-03-05 03:04:47 -05:00
#endif
2020-09-08 10:12:41 -07:00
using System.Xml.Serialization;
2024-02-29 21:20:44 -05:00
using Newtonsoft.Json;
2020-12-08 13:23:59 -08:00
using SabreTools.Core;
using SabreTools.Core.Filter;
using SabreTools.Core.Tools;
2020-12-08 15:15:41 -08:00
using SabreTools.DatItems;
2021-02-02 10:23:43 -08:00
using SabreTools.DatItems.Formats;
2024-03-04 23:56:05 -05:00
using SabreTools.Hashing;
2020-12-07 14:29:45 -08:00
using SabreTools.Logging;
2024-02-29 21:20:44 -05:00
using SabreTools.Matching;
2020-07-26 22:34:45 -07:00
namespace SabreTools.DatFiles
2020-07-26 22:34:45 -07:00
{
/// <summary>
/// Item dictionary with statistics, bucketing, and sorting
/// </summary>
2020-09-08 10:12:41 -07:00
[JsonObject("items"), XmlRoot("items")]
public class ItemDictionary : IDictionary<string, ConcurrentList<DatItem>?>
2020-07-26 22:34:45 -07:00
{
#region Private instance variables
/// <summary>
/// Determine the bucketing key for all items
/// </summary>
2020-12-14 15:31:28 -08:00
private ItemKey bucketedBy;
2020-07-26 22:34:45 -07:00
/// <summary>
/// Determine merging type for all items
/// </summary>
private DedupeType mergedBy;
/// <summary>
/// Internal dictionary for the class
/// </summary>
2024-02-28 22:54:56 -05:00
#if NET40_OR_GREATER || NETCOREAPP
private readonly ConcurrentDictionary<string, ConcurrentList<DatItem>?> items;
2024-02-28 22:54:56 -05:00
#else
private readonly Dictionary<string, ConcurrentList<DatItem>?> items;
#endif
2020-07-26 22:34:45 -07:00
/// <summary>
/// Logging object
/// </summary>
private readonly Logger logger;
2020-07-26 22:34:45 -07:00
#endregion
#region Publically available fields
2020-07-27 01:39:32 -07:00
#region Keys
2020-07-26 22:34:45 -07:00
/// <summary>
/// Get the keys from the file dictionary
/// </summary>
/// <returns>List of the keys</returns>
2020-09-08 10:12:41 -07:00
[JsonIgnore, XmlIgnore]
2020-07-26 22:34:45 -07:00
public ICollection<string> Keys
{
get { return items.Keys; }
}
/// <summary>
/// Get the keys in sorted order from the file dictionary
/// </summary>
/// <returns>List of the keys in sorted order</returns>
2020-09-08 10:12:41 -07:00
[JsonIgnore, XmlIgnore]
2020-07-26 22:34:45 -07:00
public List<string> SortedKeys
{
get
{
var keys = items.Keys.ToList();
keys.Sort(new NaturalComparer());
return keys;
}
}
2020-07-27 01:39:32 -07:00
#endregion
#region Statistics
/// <summary>
/// DAT statistics
2020-07-27 01:39:32 -07:00
/// </summary>
2020-09-08 10:12:41 -07:00
[JsonIgnore, XmlIgnore]
public DatStatistics DatStatistics { get; } = new DatStatistics();
2020-08-30 17:02:07 -07:00
2020-07-26 22:34:45 -07:00
#endregion
2020-07-27 01:39:32 -07:00
#endregion
2024-03-13 11:21:38 -04:00
#region Constructors
/// <summary>
/// Generic constructor
/// </summary>
public ItemDictionary()
{
bucketedBy = ItemKey.NULL;
mergedBy = DedupeType.None;
#if NET40_OR_GREATER || NETCOREAPP
items = new ConcurrentDictionary<string, ConcurrentList<DatItem>?>();
#else
items = new Dictionary<string, ConcurrentList<DatItem>?>();
#endif
logger = new Logger(this);
}
#endregion
2020-07-26 22:34:45 -07:00
#region Accessors
/// <summary>
/// Passthrough to access the file dictionary
/// </summary>
/// <param name="key">Key in the dictionary to reference</param>
public ConcurrentList<DatItem>? this[string key]
2020-07-26 22:34:45 -07:00
{
get
{
// Explicit lock for some weird corner cases
lock (key)
{
// Ensure the key exists
EnsureKey(key);
// Now return the value
return items[key];
}
}
set
{
2020-08-30 17:02:07 -07:00
Remove(key);
if (value == null)
items[key] = null;
else
AddRange(key, value);
2020-07-26 22:34:45 -07:00
}
}
/// <summary>
/// Add a value to the file dictionary
/// </summary>
/// <param name="key">Key in the dictionary to add to</param>
/// <param name="value">Value to add to the dictionary</param>
public void Add(string key, DatItem value)
{
// Explicit lock for some weird corner cases
lock (key)
{
// Ensure the key exists
EnsureKey(key);
// If item is null, don't add it
if (value == null)
return;
// Now add the value
items[key]!.Add(value);
2020-07-26 22:34:45 -07:00
// Now update the statistics
DatStatistics.AddItemStatistics(value);
2020-07-26 22:34:45 -07:00
}
}
/// <summary>
/// Add a range of values to the file dictionary
/// </summary>
/// <param name="key">Key in the dictionary to add to</param>
/// <param name="value">Value to add to the dictionary</param>
public void Add(string key, ConcurrentList<DatItem>? value)
2020-07-26 22:34:45 -07:00
{
AddRange(key, value);
}
/// <summary>
/// Add a DatItem to the dictionary after checking
/// </summary>
/// <param name="item">Item data to check against</param>
/// <param name="statsOnly">True to only add item statistics while parsing, false otherwise</param>
/// <returns>The key for the item</returns>
public string AddItem(DatItem item, bool statsOnly)
{
string key;
// If we have a Disk, Media, or Rom, clean the hash data
if (item is Disk disk)
{
// If the file has aboslutely no hashes, skip and log
if (disk.GetStringFieldValue(Models.Metadata.Disk.StatusKey).AsEnumValue<ItemStatus>() != ItemStatus.Nodump
&& string.IsNullOrEmpty(disk.GetStringFieldValue(Models.Metadata.Disk.MD5Key))
&& string.IsNullOrEmpty(disk.GetStringFieldValue(Models.Metadata.Disk.SHA1Key)))
{
logger.Verbose($"Incomplete entry for '{disk.GetName()}' will be output as nodump");
disk.SetFieldValue<string?>(Models.Metadata.Disk.StatusKey, ItemStatus.Nodump.AsStringValue());
}
item = disk;
}
if (item is Media media)
{
// If the file has aboslutely no hashes, skip and log
if (string.IsNullOrEmpty(media.GetStringFieldValue(Models.Metadata.Media.MD5Key))
&& string.IsNullOrEmpty(media.GetStringFieldValue(Models.Metadata.Media.SHA1Key))
&& string.IsNullOrEmpty(media.GetStringFieldValue(Models.Metadata.Media.SHA256Key))
&& string.IsNullOrEmpty(media.GetStringFieldValue(Models.Metadata.Media.SpamSumKey)))
{
logger.Verbose($"Incomplete entry for '{media.GetName()}' will be output as nodump");
}
item = media;
}
else if (item is Rom rom)
{
long? size = rom.GetInt64FieldValue(Models.Metadata.Rom.SizeKey);
// If we have the case where there is SHA-1 and nothing else, we don't fill in any other part of the data
if (size == null && !rom.HasHashes())
{
// No-op, just catch it so it doesn't go further
//logger.Verbose($"{Header.GetStringFieldValue(DatHeader.FileNameKey)}: Entry with only SHA-1 found - '{rom.GetName()}'");
}
// If we have a rom and it's missing size AND the hashes match a 0-byte file, fill in the rest of the info
else if ((size == 0 || size == null)
&& (string.IsNullOrEmpty(rom.GetStringFieldValue(Models.Metadata.Rom.CRCKey)) || rom.HasZeroHash()))
{
rom.SetFieldValue<string?>(Models.Metadata.Rom.SizeKey, Constants.SizeZero.ToString());
rom.SetFieldValue<string?>(Models.Metadata.Rom.CRCKey, Constants.CRCZero);
rom.SetFieldValue<string?>(Models.Metadata.Rom.MD5Key, Constants.MD5Zero);
rom.SetFieldValue<string?>(Models.Metadata.Rom.SHA1Key, Constants.SHA1Zero);
rom.SetFieldValue<string?>(Models.Metadata.Rom.SHA256Key, null); // Constants.SHA256Zero;
rom.SetFieldValue<string?>(Models.Metadata.Rom.SHA384Key, null); // Constants.SHA384Zero;
rom.SetFieldValue<string?>(Models.Metadata.Rom.SHA512Key, null); // Constants.SHA512Zero;
rom.SetFieldValue<string?>(Models.Metadata.Rom.SpamSumKey, null); // Constants.SpamSumZero;
}
// If the file has no size and it's not the above case, skip and log
else if (rom.GetStringFieldValue(Models.Metadata.Rom.StatusKey).AsEnumValue<ItemStatus>() != ItemStatus.Nodump && (size == 0 || size == null))
{
//logger.Verbose($"{Header.GetStringFieldValue(DatHeader.FileNameKey)}: Incomplete entry for '{rom.GetName()}' will be output as nodump");
rom.SetFieldValue<string?>(Models.Metadata.Rom.StatusKey, ItemStatus.Nodump.AsStringValue());
}
// If the file has a size but aboslutely no hashes, skip and log
else if (rom.GetStringFieldValue(Models.Metadata.Rom.StatusKey).AsEnumValue<ItemStatus>() != ItemStatus.Nodump
&& size != null && size > 0
&& !rom.HasHashes())
{
//logger.Verbose($"{Header.GetStringFieldValue(DatHeader.FileNameKey)}: Incomplete entry for '{rom.GetName()}' will be output as nodump");
rom.SetFieldValue<string?>(Models.Metadata.Rom.StatusKey, ItemStatus.Nodump.AsStringValue());
}
item = rom;
}
// Get the key and add the file
key = item.GetKey(ItemKey.Machine);
// If only adding statistics, we add an empty key for games and then just item stats
if (statsOnly)
{
EnsureKey(key);
DatStatistics.AddItemStatistics(item);
}
else
{
Add(key, item);
}
return key;
}
2020-07-26 22:34:45 -07:00
/// <summary>
/// Add a range of values to the file dictionary
/// </summary>
/// <param name="key">Key in the dictionary to add to</param>
/// <param name="value">Value to add to the dictionary</param>
public void AddRange(string key, ConcurrentList<DatItem>? value)
2020-07-26 22:34:45 -07:00
{
// Explicit lock for some weird corner cases
lock (key)
{
// If the value is null or empty, just return
if (value == null || value.Count == 0)
return;
2020-07-26 22:34:45 -07:00
// Ensure the key exists
EnsureKey(key);
// Now add the value
items[key]!.AddRange(value);
2020-07-26 22:34:45 -07:00
// Now update the statistics
foreach (DatItem item in value)
{
DatStatistics.AddItemStatistics(item);
2020-07-26 22:34:45 -07:00
}
}
}
2024-03-13 11:21:38 -04:00
/// <summary>
/// Remove any keys that have null or empty values
/// </summary>
public void ClearEmpty()
{
var keys = items.Keys.Where(k => k != null).ToList();
foreach (string key in keys)
{
// If the key doesn't exist, skip
if (!items.ContainsKey(key))
continue;
// If the value is null, remove
else if (items[key] == null)
#if NET40_OR_GREATER || NETCOREAPP
items.TryRemove(key, out _);
#else
items.Remove(key);
#endif
// If there are no non-blank items, remove
else if (!items[key]!.Any(i => i != null && i is not Blank))
#if NET40_OR_GREATER || NETCOREAPP
items.TryRemove(key, out _);
#else
items.Remove(key);
#endif
}
}
/// <summary>
/// Remove all items marked for removal
/// </summary>
public void ClearMarked()
{
var keys = items.Keys.ToList();
foreach (string key in keys)
{
ConcurrentList<DatItem>? oldItemList = items[key];
ConcurrentList<DatItem>? newItemList = oldItemList?.Where(i => i.GetBoolFieldValue(DatItem.RemoveKey) != true)?.ToConcurrentList();
Remove(key);
AddRange(key, newItemList);
}
}
2020-07-26 22:34:45 -07:00
/// <summary>
/// Get if the file dictionary contains the key
/// </summary>
/// <param name="key">Key in the dictionary to check</param>
/// <returns>True if the key exists, false otherwise</returns>
public bool ContainsKey(string key)
{
// If the key is null, we return false since keys can't be null
if (key == null)
return false;
// Explicit lock for some weird corner cases
lock (key)
{
return items.ContainsKey(key);
}
}
/// <summary>
/// Get if the file dictionary contains the key and value
/// </summary>
/// <param name="key">Key in the dictionary to check</param>
/// <param name="value">Value in the dictionary to check</param>
/// <returns>True if the key exists, false otherwise</returns>
public bool Contains(string key, DatItem value)
{
// If the key is null, we return false since keys can't be null
if (key == null)
return false;
// Explicit lock for some weird corner cases
lock (key)
{
if (items.ContainsKey(key) && items[key] != null)
return items[key]!.Contains(value);
2020-07-26 22:34:45 -07:00
}
return false;
}
/// <summary>
/// Ensure the key exists in the items dictionary
/// </summary>
/// <param name="key">Key to ensure</param>
public void EnsureKey(string key)
{
// If the key is missing from the dictionary, add it
if (!items.ContainsKey(key))
2024-02-28 22:54:56 -05:00
#if NET40_OR_GREATER || NETCOREAPP
2024-02-28 19:19:50 -05:00
items.TryAdd(key, []);
2024-02-28 22:54:56 -05:00
#else
items[key] = [];
#endif
}
2020-08-28 15:06:07 -07:00
/// <summary>
/// Get a list of filtered items for a given key
/// </summary>
/// <param name="key">Key in the dictionary to retrieve</param>
public ConcurrentList<DatItem> FilteredItems(string key)
2020-08-28 15:06:07 -07:00
{
lock (key)
{
// Get the list, if possible
ConcurrentList<DatItem>? fi = items[key];
2020-08-28 15:06:07 -07:00
if (fi == null)
2024-02-28 19:19:50 -05:00
return [];
2020-08-28 15:06:07 -07:00
// Filter the list
return fi.Where(i => i != null)
.Where(i => i.GetBoolFieldValue(DatItem.RemoveKey) != true)
2024-03-11 14:31:02 -04:00
.Where(i => i.GetFieldValue<Machine>(DatItem.MachineKey) != null)
.ToConcurrentList();
2020-08-28 15:06:07 -07:00
}
}
2020-07-26 22:34:45 -07:00
/// <summary>
/// Remove a key from the file dictionary if it exists
/// </summary>
/// <param name="key">Key in the dictionary to remove</param>
public bool Remove(string key)
{
// Explicit lock for some weird corner cases
lock (key)
2020-07-26 22:34:45 -07:00
{
// If the key doesn't exist, return
if (!ContainsKey(key) || items[key] == null)
return false;
2020-07-26 22:34:45 -07:00
// Remove the statistics first
foreach (DatItem item in items[key]!)
{
DatStatistics.RemoveItemStatistics(item);
}
// Remove the key from the dictionary
2024-02-28 22:54:56 -05:00
#if NET40_OR_GREATER || NETCOREAPP
return items.TryRemove(key, out _);
2024-02-28 22:54:56 -05:00
#else
return items.Remove(key);
#endif
}
2020-07-26 22:34:45 -07:00
}
/// <summary>
/// Remove the first instance of a value from the file dictionary if it exists
/// </summary>
/// <param name="key">Key in the dictionary to remove from</param>
/// <param name="value">Value to remove from the dictionary</param>
public bool Remove(string key, DatItem value)
{
// Explicit lock for some weird corner cases
lock (key)
{
// If the key and value doesn't exist, return
if (!Contains(key, value) || items[key] == null)
return false;
2020-07-26 22:34:45 -07:00
// Remove the statistics first
DatStatistics.RemoveItemStatistics(value);
2020-07-26 22:34:45 -07:00
return items[key]!.Remove(value);
}
2020-07-26 22:34:45 -07:00
}
2020-08-28 01:13:55 -07:00
/// <summary>
/// Reset a key from the file dictionary if it exists
/// </summary>
/// <param name="key">Key in the dictionary to reset</param>
public bool Reset(string key)
{
// If the key doesn't exist, return
if (!ContainsKey(key) || items[key] == null)
2020-08-28 01:13:55 -07:00
return false;
// Remove the statistics first
foreach (DatItem item in items[key]!)
2020-08-28 01:13:55 -07:00
{
DatStatistics.RemoveItemStatistics(item);
2020-08-28 01:13:55 -07:00
}
// Remove the key from the dictionary
2024-02-28 19:19:50 -05:00
items[key] = [];
2020-08-28 01:13:55 -07:00
return true;
}
2020-07-26 22:34:45 -07:00
/// <summary>
2020-12-14 15:31:28 -08:00
/// Override the internal ItemKey value
2020-07-26 22:34:45 -07:00
/// </summary>
/// <param name="newBucket"></param>
2020-12-14 15:31:28 -08:00
public void SetBucketedBy(ItemKey newBucket)
2020-07-26 22:34:45 -07:00
{
bucketedBy = newBucket;
}
#endregion
2024-03-13 11:07:53 -04:00
#region Bucketing
2020-07-26 22:34:45 -07:00
/// <summary>
/// Take the arbitrarily bucketed Files Dictionary and convert to one bucketed by a user-defined method
/// </summary>
2020-12-14 15:31:28 -08:00
/// <param name="bucketBy">ItemKey enum representing how to bucket the individual items</param>
2020-07-26 22:34:45 -07:00
/// <param name="dedupeType">Dedupe type that should be used</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>
2020-12-14 15:31:28 -08:00
public void BucketBy(ItemKey bucketBy, DedupeType dedupeType, bool lower = true, bool norename = true)
2020-07-26 22:34:45 -07:00
{
// If we have a situation where there's no dictionary or no keys at all, we skip
2024-02-28 22:54:56 -05:00
#if NET40_OR_GREATER || NETCOREAPP
if (items == null || items.IsEmpty)
2024-02-28 22:54:56 -05:00
#else
if (items == null || items.Count == 0)
#endif
2020-07-26 22:34:45 -07:00
return;
// If the sorted type isn't the same, we want to sort the dictionary accordingly
if (bucketedBy != bucketBy && bucketBy != ItemKey.NULL)
2020-07-26 22:34:45 -07:00
{
logger.User($"Organizing roms by {bucketBy}");
2024-03-13 11:00:02 -04:00
PerformBucketing(bucketBy, lower, norename);
2020-07-26 22:34:45 -07:00
}
// If the merge type isn't the same, we want to merge the dictionary accordingly
2020-07-27 01:39:32 -07:00
if (mergedBy != dedupeType)
2020-07-26 22:34:45 -07:00
{
logger.User($"Deduping roms by {dedupeType}");
2024-03-13 11:00:02 -04:00
PerformDeduplication(bucketBy, dedupeType);
2020-07-26 22:34:45 -07:00
}
// If the merge type is the same, we want to sort the dictionary to be consistent
else
{
2024-03-13 11:00:02 -04:00
logger.User($"Sorting roms by {bucketBy}");
PerformSorting();
2020-07-26 22:34:45 -07:00
}
}
2024-03-13 11:07:53 -04:00
2020-07-26 22:34:45 -07:00
/// <summary>
/// List all duplicates found in a DAT based on a DatItem
/// </summary>
/// <param name="datItem">Item to try to match</param>
/// <param name="sorted">True if the DAT is already sorted accordingly, false otherwise (default)</param>
/// <returns>List of matched DatItem objects</returns>
public ConcurrentList<DatItem> GetDuplicates(DatItem datItem, bool sorted = false)
2020-07-26 22:34:45 -07:00
{
2024-02-28 19:19:50 -05:00
ConcurrentList<DatItem> output = [];
2020-07-26 22:34:45 -07:00
// Check for an empty rom list first
if (DatStatistics.TotalCount == 0)
2020-07-26 22:34:45 -07:00
return output;
// We want to get the proper key for the DatItem
string key = SortAndGetKey(datItem, sorted);
// If the key doesn't exist, return the empty list
if (!ContainsKey(key))
return output;
// Try to find duplicates
ConcurrentList<DatItem>? roms = this[key];
if (roms == null)
return output;
2024-02-28 19:19:50 -05:00
ConcurrentList<DatItem> left = [];
2020-07-26 22:34:45 -07:00
for (int i = 0; i < roms.Count; i++)
{
DatItem other = roms[i];
if (other.GetBoolFieldValue(DatItem.RemoveKey) == true)
2020-08-28 22:38:10 -07:00
continue;
2020-07-26 22:34:45 -07:00
if (datItem.Equals(other))
{
2024-03-11 14:31:02 -04:00
other.SetFieldValue<bool?>(DatItem.RemoveKey, true);
2020-07-26 22:34:45 -07:00
output.Add(other);
}
else
{
left.Add(other);
}
}
2020-08-28 22:38:10 -07:00
// Add back all roms with the proper flags
Remove(key);
AddRange(key, output);
AddRange(key, left);
2020-07-26 22:34:45 -07:00
return output;
}
/// <summary>
/// Check if a DAT contains the given DatItem
/// </summary>
/// <param name="datItem">Item to try to match</param>
/// <param name="sorted">True if the DAT is already sorted accordingly, false otherwise (default)</param>
/// <returns>True if it contains the rom, false otherwise</returns>
public bool HasDuplicates(DatItem datItem, bool sorted = false)
{
// Check for an empty rom list first
if (DatStatistics.TotalCount == 0)
2020-07-26 22:34:45 -07:00
return false;
// We want to get the proper key for the DatItem
string key = SortAndGetKey(datItem, sorted);
// If the key doesn't exist, return the empty list
if (!ContainsKey(key))
return false;
// Try to find duplicates
ConcurrentList<DatItem>? roms = this[key];
return roms?.Any(r => datItem.Equals(r)) == true;
2020-07-26 22:34:45 -07:00
}
2020-12-10 23:24:09 -08:00
/// <summary>
/// Get the highest-order Field value that represents the statistics
/// </summary>
2020-12-14 15:31:28 -08:00
private ItemKey GetBestAvailable()
2020-12-10 23:24:09 -08:00
{
2024-03-04 23:17:13 -05:00
// Get the required counts
long diskCount = DatStatistics.GetItemCount(ItemType.Disk);
long mediaCount = DatStatistics.GetItemCount(ItemType.Media);
long romCount = DatStatistics.GetItemCount(ItemType.Rom);
long nodumpCount = DatStatistics.GetStatusCount(ItemStatus.Nodump);
2024-03-04 22:52:03 -05:00
2020-12-10 23:24:09 -08:00
// If all items are supposed to have a SHA-512, we bucket by that
if (diskCount + mediaCount + romCount - nodumpCount == DatStatistics.GetHashCount(HashType.SHA512))
2020-12-14 15:31:28 -08:00
return ItemKey.SHA512;
2020-12-10 23:24:09 -08:00
// If all items are supposed to have a SHA-384, we bucket by that
else if (diskCount + mediaCount + romCount - nodumpCount == DatStatistics.GetHashCount(HashType.SHA384))
2020-12-14 15:31:28 -08:00
return ItemKey.SHA384;
2020-12-10 23:24:09 -08:00
// If all items are supposed to have a SHA-256, we bucket by that
else if (diskCount + mediaCount + romCount - nodumpCount == DatStatistics.GetHashCount(HashType.SHA256))
2020-12-14 15:31:28 -08:00
return ItemKey.SHA256;
2020-12-10 23:24:09 -08:00
// If all items are supposed to have a SHA-1, we bucket by that
else if (diskCount + mediaCount + romCount - nodumpCount == DatStatistics.GetHashCount(HashType.SHA1))
2020-12-14 15:31:28 -08:00
return ItemKey.SHA1;
2020-12-10 23:24:09 -08:00
// If all items are supposed to have a MD5, we bucket by that
else if (diskCount + mediaCount + romCount - nodumpCount == DatStatistics.GetHashCount(HashType.MD5))
2020-12-14 15:31:28 -08:00
return ItemKey.MD5;
2020-12-10 23:24:09 -08:00
// Otherwise, we bucket by CRC
else
2020-12-14 15:31:28 -08:00
return ItemKey.CRC;
2020-12-10 23:24:09 -08:00
}
2024-03-13 11:00:02 -04:00
/// <summary>
/// Perform bucketing based on the item key provided
/// </summary>
/// <param name="bucketBy">ItemKey enum representing how to bucket the individual items</param>
/// <param name="lower">True if the key should be lowercased, 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>
private void PerformBucketing(ItemKey bucketBy, bool lower, bool norename)
{
// Set the sorted type
bucketedBy = bucketBy;
// Reset the merged type since this might change the merge
mergedBy = DedupeType.None;
// First do the initial sort of all of the roms inplace
List<string> oldkeys = [.. Keys];
#if NET452_OR_GREATER || NETCOREAPP
Parallel.For(0, oldkeys.Count, Globals.ParallelOptions, k =>
#elif NET40_OR_GREATER
Parallel.For(0, oldkeys.Count, k =>
#else
for (int k = 0; k < oldkeys.Count; k++)
#endif
{
string key = oldkeys[k];
if (this[key] == null)
Remove(key);
// Now add each of the roms to their respective keys
for (int i = 0; i < this[key]!.Count; i++)
{
DatItem item = this[key]![i];
if (item == null)
continue;
// We want to get the key most appropriate for the given sorting type
string newkey = item.GetKey(bucketBy, lower, norename);
// If the key is different, move the item to the new key
if (newkey != key)
{
Add(newkey, item);
Remove(key, item);
i--; // This make sure that the pointer stays on the correct since one was removed
}
}
// If the key is now empty, remove it
if (this[key]!.Count == 0)
Remove(key);
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
/// <summary>
/// Perform deduplication based on the deduplication type provided
/// </summary>
/// <param name="bucketBy">ItemKey enum representing how to bucket the individual items</param>
/// <param name="dedupeType">Dedupe type that should be used</param>
private void PerformDeduplication(ItemKey bucketBy, DedupeType dedupeType)
{
// Set the sorted type
mergedBy = dedupeType;
List<string> keys = [.. Keys];
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(keys, key =>
#else
foreach (var key in keys)
#endif
{
// Get the possibly unsorted list
ConcurrentList<DatItem>? sortedlist = this[key]?.ToConcurrentList();
if (sortedlist == null)
#if NET40_OR_GREATER || NETCOREAPP
return;
#else
continue;
#endif
// Sort the list of items to be consistent
DatItem.Sort(ref sortedlist, false);
// If we're merging the roms, do so
if (dedupeType == DedupeType.Full || (dedupeType == DedupeType.Game && bucketBy == ItemKey.Machine))
sortedlist = DatItem.Merge(sortedlist);
// Add the list back to the dictionary
Reset(key);
AddRange(key, sortedlist);
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
/// <summary>
/// Perform inplace sorting of the dictionary
/// </summary>
private void PerformSorting()
{
List<string> keys = [.. Keys];
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(keys, key =>
#else
foreach (var key in keys)
#endif
{
// Get the possibly unsorted list
ConcurrentList<DatItem>? sortedlist = this[key];
// Sort the list of items to be consistent
if (sortedlist != null)
DatItem.Sort(ref sortedlist, false);
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
2020-07-26 22:34:45 -07:00
/// <summary>
/// Sort the input DAT and get the key to be used by the item
/// </summary>
/// <param name="datItem">Item to try to match</param>
/// <param name="sorted">True if the DAT is already sorted accordingly, false otherwise (default)</param>
/// <returns>Key to try to use</returns>
private string SortAndGetKey(DatItem datItem, bool sorted = false)
{
// If we're not already sorted, take care of it
if (!sorted)
2020-07-27 01:39:32 -07:00
BucketBy(GetBestAvailable(), DedupeType.None);
2020-07-26 22:34:45 -07:00
// Now that we have the sorted type, we get the proper key
return datItem.GetKey(bucketedBy);
}
#endregion
#region Filtering
/// <summary>
/// Execute all filters in a filter runner on the items in the dictionary
/// </summary>
/// <param name="filterRunner">Preconfigured filter runner to use</param>
public void ExecuteFilters(FilterRunner filterRunner)
{
List<string> keys = [.. Keys];
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(keys, key =>
#else
foreach (var key in keys)
#endif
{
ConcurrentList<DatItem>? items = this[key];
if (items == null)
#if NET40_OR_GREATER || NETCOREAPP
return;
#else
continue;
#endif
// Filter all items in the current key
var newItems = new ConcurrentList<DatItem>();
foreach (var item in items)
{
if (item.PassesFilter(filterRunner))
newItems.Add(item);
}
// Set the value in the key to the new set
this[key] = newItems;
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
/// <summary>
/// Use game descriptions as names, updating cloneof/romof/sampleof
/// </summary>
/// <param name="throwOnError">True if the error that is thrown should be thrown back to the caller, false otherwise</param>
public void MachineDescriptionToName(bool throwOnError = false)
{
try
{
// First we want to get a mapping for all games to description
#if NET40_OR_GREATER || NETCOREAPP
ConcurrentDictionary<string, string> mapping = new();
#else
Dictionary<string, string> mapping = [];
#endif
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(Keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(Keys, key =>
#else
foreach (var key in Keys)
#endif
{
var items = this[key];
if (items == null)
#if NET40_OR_GREATER || NETCOREAPP
return;
#else
continue;
#endif
foreach (DatItem item in items)
{
// If the key mapping doesn't exist, add it
#if NET40_OR_GREATER || NETCOREAPP
mapping.TryAdd(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey)!, item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.DescriptionKey)!.Replace('/', '_').Replace("\"", "''").Replace(":", " -"));
#else
mapping[item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey)!] = item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.DescriptionKey)!.Replace('/', '_').Replace("\"", "''").Replace(":", " -");
#endif
}
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
// Now we loop through every item and update accordingly
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(Keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(Keys, key =>
#else
foreach (var key in Keys)
#endif
{
var items = this[key];
if (items == null)
#if NET40_OR_GREATER || NETCOREAPP
return;
#else
continue;
#endif
ConcurrentList<DatItem> newItems = [];
foreach (DatItem item in items)
{
// Update machine name
if (!string.IsNullOrEmpty(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey)) && mapping.ContainsKey(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey)!))
item.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.NameKey, mapping[item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey)!]);
// Update cloneof
if (!string.IsNullOrEmpty(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.CloneOfKey)) && mapping.ContainsKey(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.CloneOfKey)!))
item.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.CloneOfKey, mapping[item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.CloneOfKey)!]);
// Update romof
if (!string.IsNullOrEmpty(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.RomOfKey)) && mapping.ContainsKey(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.RomOfKey)!))
item.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.RomOfKey, mapping[item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.RomOfKey)!]);
// Update sampleof
if (!string.IsNullOrEmpty(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.SampleOfKey)) && mapping.ContainsKey(item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.SampleOfKey)!))
item.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.SampleOfKey, mapping[item.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.SampleOfKey)!]);
// Add the new item to the output list
newItems.Add(item);
}
// Replace the old list of roms with the new one
Remove(key);
AddRange(key, newItems);
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
catch (Exception ex) when (!throwOnError)
{
logger.Warning(ex.ToString());
}
}
2024-03-19 16:18:35 -04:00
/// <summary>
/// Ensure that all roms are in their own game (or at least try to ensure)
/// </summary>
public void SetOneRomPerGame()
{
// For each rom, we want to update the game to be "<game name>/<rom name>"
#if NET452_OR_GREATER || NETCOREAPP
Parallel.ForEach(Keys, Globals.ParallelOptions, key =>
#elif NET40_OR_GREATER
Parallel.ForEach(Keys, key =>
#else
foreach (var key in Keys)
#endif
{
var items = this[key];
if (items == null)
#if NET40_OR_GREATER || NETCOREAPP
return;
#else
continue;
#endif
for (int i = 0; i < items.Count; i++)
{
SetOneRomPerGame(items[i]);
}
#if NET40_OR_GREATER || NETCOREAPP
});
#else
}
#endif
}
/// <summary>
/// Set internal names to match One Rom Per Game (ORPG) logic
/// </summary>
/// <param name="datItem">DatItem to run logic on</param>
internal static void SetOneRomPerGame(DatItem datItem)
{
if (datItem.GetName() == null)
return;
string[] splitname = datItem.GetName()!.Split('.');
#if NET20 || NET35
datItem.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.NameKey, datItem.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey) + $"/{string.Join(".", splitname.Take(splitname.Length > 1 ? splitname.Length - 1 : 1).ToArray())}");
#else
datItem.GetFieldValue<Machine>(DatItem.MachineKey)!.SetFieldValue<string?>(Models.Metadata.Machine.NameKey, datItem.GetFieldValue<Machine>(DatItem.MachineKey)!.GetStringFieldValue(Models.Metadata.Machine.NameKey) + $"/{string.Join(".", splitname.Take(splitname.Length > 1 ? splitname.Length - 1 : 1))}");
#endif
datItem.SetName(Path.GetFileName(datItem.GetName()));
}
#endregion
2024-03-13 11:07:53 -04:00
#region Statistics
/// <summary>
/// Recalculate the statistics for the Dat
/// </summary>
public void RecalculateStats()
{
// Wipe out any stats already there
DatStatistics.ResetStatistics();
// If we have a blank Dat in any way, return
if (items == null)
return;
// Loop through and add
foreach (string key in items.Keys)
{
ConcurrentList<DatItem>? datItems = items[key];
if (datItems == null)
continue;
foreach (DatItem item in datItems)
{
DatStatistics.AddItemStatistics(item);
}
}
}
#endregion
2020-07-26 22:34:45 -07:00
#region IDictionary Implementations
public ICollection<ConcurrentList<DatItem>?> Values => ((IDictionary<string, ConcurrentList<DatItem>?>)items).Values;
2020-07-26 22:34:45 -07:00
public int Count => ((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).Count;
2020-07-26 22:34:45 -07:00
public bool IsReadOnly => ((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).IsReadOnly;
2020-07-26 22:34:45 -07:00
public bool TryGetValue(string key, out ConcurrentList<DatItem>? value)
2020-07-26 22:34:45 -07:00
{
return ((IDictionary<string, ConcurrentList<DatItem>?>)items).TryGetValue(key, out value);
2020-07-26 22:34:45 -07:00
}
public void Add(KeyValuePair<string, ConcurrentList<DatItem>?> item)
2020-07-26 22:34:45 -07:00
{
((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).Add(item);
2020-07-26 22:34:45 -07:00
}
public void Clear()
{
((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).Clear();
2020-07-26 22:34:45 -07:00
}
public bool Contains(KeyValuePair<string, ConcurrentList<DatItem>?> item)
2020-07-26 22:34:45 -07:00
{
return ((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).Contains(item);
2020-07-26 22:34:45 -07:00
}
public void CopyTo(KeyValuePair<string, ConcurrentList<DatItem>?>[] array, int arrayIndex)
2020-07-26 22:34:45 -07:00
{
((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).CopyTo(array, arrayIndex);
2020-07-26 22:34:45 -07:00
}
public bool Remove(KeyValuePair<string, ConcurrentList<DatItem>?> item)
2020-07-26 22:34:45 -07:00
{
return ((ICollection<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).Remove(item);
2020-07-26 22:34:45 -07:00
}
public IEnumerator<KeyValuePair<string, ConcurrentList<DatItem>?>> GetEnumerator()
2020-07-26 22:34:45 -07:00
{
return ((IEnumerable<KeyValuePair<string, ConcurrentList<DatItem>?>>)items).GetEnumerator();
2020-07-26 22:34:45 -07:00
}
IEnumerator IEnumerable.GetEnumerator()
{
return ((IEnumerable)items).GetEnumerator();
}
#endregion
}
}