| | | 1 | | using OzonDomains.Nomenclature; |
| | | 2 | | |
| | | 3 | | namespace Services.ApiServices._1CApi.Nomenclature; |
| | | 4 | | |
| | | 5 | | /// <summary>Строка справочника производителей (<c>et_producers</c>): <see cref="RealId"/> объединяет разные написания о |
| | | 6 | | public sealed record ProducerRow(int Id, int RealId, string Name, string? Prefix, string? ExistName, string? MarketPrefi |
| | | 7 | | |
| | | 8 | | /// <summary>Ручная связь написаний (<c>OneCProducerAliases</c>): обе стороны считаются одним брендом.</summary> |
| | | 9 | | public sealed record ProducerLink(string AliasName, string CanonicalName); |
| | | 10 | | |
| | | 11 | | public enum ProducerResolutionKind |
| | | 12 | | { |
| | | 13 | | /// <summary>Бренд определён (одна или несколько групп-кандидатов с равным правом).</summary> |
| | | 14 | | Known, |
| | | 15 | | |
| | | 16 | | /// <summary>Бренд определить нечем (нет префикса и имени, либо имя — заглушка).</summary> |
| | | 17 | | Unknown, |
| | | 18 | | |
| | | 19 | | /// <summary>Префикс общий у нескольких брендов (HND, MTL…) и выбрать между ними нечем.</summary> |
| | | 20 | | Ambiguous |
| | | 21 | | } |
| | | 22 | | |
| | | 23 | | /// <param name="Groups">Ключи групп бренда; для <see cref="ProducerResolutionKind.Unknown"/> пусто.</param> |
| | | 24 | | /// <param name="Source">Чем определено: <c>prefix</c>, <c>prefix+name</c>, <c>name</c>, <c>placeholder</c>, <c>none</c> |
| | | 25 | | public sealed record ProducerResolution(ProducerResolutionKind Kind, IReadOnlySet<string> Groups, string Source) |
| | | 26 | | { |
| | | 27 | | public static readonly IReadOnlySet<string> NoGroups = new HashSet<string>(); |
| | | 28 | | } |
| | | 29 | | |
| | | 30 | | /// <summary> |
| | | 31 | | /// Группы эквивалентных написаний бренда. Основа — <c>et_producers.realid</c> (19 тыс. имён → 16 тыс. групп), |
| | | 32 | | /// сверху ручные связи (<c>OneCProducerAliases</c>); связь двусторонняя: <c>KAYABA ↔ KYB</c>, а не «KAYABA → KYB». |
| | | 33 | | /// Чистый класс без ввода-вывода: сборка в <see cref="Build"/>, чтение БД — в <c>ProducerRegistry</c>. |
| | | 34 | | /// </summary> |
| | | 35 | | public sealed class ProducerDirectory |
| | | 36 | | { |
| | | 37 | | /// <summary> |
| | | 38 | | /// Имя производителя, которым заказ помечается, когда бренд не известен (ключ без префикса бренда). |
| | | 39 | | /// Это заглушка, а не бренд: искать по ней нельзя, иначе карточки Nissan, Mazda, Samson… «другого бренда». |
| | | 40 | | /// </summary> |
| | 1 | 41 | | private static readonly HashSet<string> PlaceholderNames = new(StringComparer.Ordinal) { "DAIHATSU" }; |
| | | 42 | | |
| | 21 | 43 | | private readonly Dictionary<string, string> _parent = new(StringComparer.Ordinal); |
| | 21 | 44 | | private readonly Dictionary<string, HashSet<string>> _nameToRawGroups = new(StringComparer.Ordinal); |
| | 21 | 45 | | private readonly Dictionary<string, HashSet<string>> _prefixToRawGroups = new(StringComparer.Ordinal); |
| | 21 | 46 | | private readonly Dictionary<string, HashSet<string>> _namesByGroup = new(StringComparer.Ordinal); |
| | | 47 | | |
| | 21 | 48 | | private ProducerDirectory() |
| | | 49 | | { |
| | 21 | 50 | | } |
| | | 51 | | |
| | 0 | 52 | | public int GroupCount => _namesByGroup.Count; |
| | | 53 | | |
| | | 54 | | public static ProducerDirectory Build(IEnumerable<ProducerRow> rows, IEnumerable<ProducerLink>? links = null) |
| | | 55 | | { |
| | 21 | 56 | | var dir = new ProducerDirectory(); |
| | | 57 | | |
| | 648 | 58 | | foreach (var row in rows) |
| | | 59 | | { |
| | 303 | 60 | | var group = "g" + (row.RealId != 0 ? row.RealId : row.Id); |
| | 303 | 61 | | dir.EnsureGroup(group); |
| | | 62 | | |
| | 2424 | 63 | | foreach (var name in new[] { row.Name, row.Prefix, row.ExistName }) |
| | 909 | 64 | | dir.AddName(name, group); |
| | | 65 | | |
| | 303 | 66 | | var prefix = row.MarketPrefix?.Trim().ToUpperInvariant(); |
| | 303 | 67 | | if (!string.IsNullOrEmpty(prefix) && prefix != "0") |
| | 213 | 68 | | GetOrAdd(dir._prefixToRawGroups, prefix).Add(group); |
| | | 69 | | } |
| | | 70 | | |
| | 46 | 71 | | foreach (var link in links ?? []) |
| | | 72 | | { |
| | 2 | 73 | | var alias = Norm(link.AliasName); |
| | 2 | 74 | | var canonical = Norm(link.CanonicalName); |
| | 2 | 75 | | if (alias.Length == 0 || canonical.Length == 0 || alias == canonical) |
| | | 76 | | continue; |
| | | 77 | | |
| | 2 | 78 | | var aliasGroups = dir.RawGroupsOfNorm(alias) ?? dir.CreateNameGroup(alias, link.AliasName); |
| | 2 | 79 | | var canonicalGroups = dir.RawGroupsOfNorm(canonical) ?? dir.CreateNameGroup(canonical, link.CanonicalName); |
| | 8 | 80 | | foreach (var a in aliasGroups.ToList()) |
| | 8 | 81 | | foreach (var c in canonicalGroups.ToList()) |
| | 2 | 82 | | dir.Union(a, c); |
| | | 83 | | } |
| | | 84 | | |
| | 21 | 85 | | return dir; |
| | | 86 | | } |
| | | 87 | | |
| | | 88 | | /// <summary>Группы бренда по имени (с учётом диакритики, регистра и ручных связей); пусто, если имя не известно.</s |
| | | 89 | | public IReadOnlySet<string> GroupsOfName(string? name) |
| | | 90 | | { |
| | 141 | 91 | | var raw = RawGroupsOfNorm(Norm(name)); |
| | 141 | 92 | | return raw == null ? ProducerResolution.NoGroups : Representatives(raw); |
| | | 93 | | } |
| | | 94 | | |
| | | 95 | | /// <summary>Группы бренда по трёхбуквенному коду ключа (<c>FEB</c>, <c>KYB</c>, <c>MET</c>).</summary> |
| | | 96 | | public IReadOnlySet<string> GroupsOfPrefix(string? prefix) |
| | | 97 | | { |
| | 57 | 98 | | var key = prefix?.Trim().ToUpperInvariant(); |
| | 57 | 99 | | return !string.IsNullOrEmpty(key) && _prefixToRawGroups.TryGetValue(key, out var raw) |
| | 57 | 100 | | ? Representatives(raw) |
| | 57 | 101 | | : ProducerResolution.NoGroups; |
| | | 102 | | } |
| | | 103 | | |
| | | 104 | | /// <summary> |
| | | 105 | | /// Бренд товара из заказа. Однозначный префикс ключа сильнее имени <c>EtProducer</c> (<c>=MET</c> — это Metaco, |
| | | 106 | | /// хотя в заказе производитель записан как «MET»). Имя-заглушка без префикса бренда не считается. |
| | | 107 | | /// </summary> |
| | | 108 | | public ProducerResolution Resolve(ParsedProductRef parsed, string? etProducerName) |
| | | 109 | | { |
| | 57 | 110 | | var byPrefix = GroupsOfPrefix(parsed.BrandPrefix); |
| | 57 | 111 | | if (byPrefix.Count == 1) |
| | 39 | 112 | | return new ProducerResolution(ProducerResolutionKind.Known, byPrefix, "prefix"); |
| | | 113 | | |
| | 18 | 114 | | if (byPrefix.Count > 1) |
| | | 115 | | { |
| | 3 | 116 | | var narrowed = byPrefix.Intersect(GroupsOfName(etProducerName)).ToHashSet(StringComparer.Ordinal); |
| | 3 | 117 | | return narrowed.Count > 0 |
| | 3 | 118 | | ? new ProducerResolution(ProducerResolutionKind.Known, narrowed, "prefix+name") |
| | 3 | 119 | | : new ProducerResolution(ProducerResolutionKind.Ambiguous, byPrefix, "prefix"); |
| | | 120 | | } |
| | | 121 | | |
| | 15 | 122 | | var name = Norm(etProducerName); |
| | 15 | 123 | | if (name.Length == 0) |
| | 6 | 124 | | return new ProducerResolution(ProducerResolutionKind.Unknown, ProducerResolution.NoGroups, "none"); |
| | | 125 | | |
| | 9 | 126 | | if (parsed.BrandPrefix.Length == 0 && PlaceholderNames.Contains(name)) |
| | 2 | 127 | | return new ProducerResolution(ProducerResolutionKind.Unknown, ProducerResolution.NoGroups, "placeholder"); |
| | | 128 | | |
| | 7 | 129 | | var byName = GroupsOfName(etProducerName); |
| | 7 | 130 | | return byName.Count > 0 |
| | 7 | 131 | | ? new ProducerResolution(ProducerResolutionKind.Known, byName, "name") |
| | 7 | 132 | | : new ProducerResolution(ProducerResolutionKind.Unknown, ProducerResolution.NoGroups, "none"); |
| | | 133 | | } |
| | | 134 | | |
| | | 135 | | /// <summary> |
| | | 136 | | /// Упомянут ли бренд группы в названии товара («NARVA 17048 Лампа», «FEBI 06161 …»). Сравниваются целые слова |
| | | 137 | | /// и пары соседних слов («FEBI BILSTEIN»); имена короче трёх знаков не участвуют — слишком много ложных совпадений. |
| | | 138 | | /// </summary> |
| | | 139 | | public bool NameMentionedIn(string? title, IReadOnlySet<string> groups) |
| | | 140 | | { |
| | 12 | 141 | | if (string.IsNullOrWhiteSpace(title) || groups.Count == 0) |
| | 2 | 142 | | return false; |
| | | 143 | | |
| | 10 | 144 | | var tokens = title.Split([' ', ',', '.', ';', ':', '(', ')', '/', '"', '«', '»', '-', '_'], StringSplitOptions.R |
| | 10 | 145 | | .Select(Norm) |
| | 29 | 146 | | .Where(t => t.Length > 0) |
| | 10 | 147 | | .ToList(); |
| | 10 | 148 | | if (tokens.Count == 0) |
| | 0 | 149 | | return false; |
| | | 150 | | |
| | 10 | 151 | | var phrases = new HashSet<string>(tokens, StringComparer.Ordinal); |
| | 58 | 152 | | for (var i = 0; i + 1 < tokens.Count; i++) |
| | 19 | 153 | | phrases.Add(tokens[i] + tokens[i + 1]); |
| | | 154 | | |
| | 36 | 155 | | foreach (var group in groups) |
| | | 156 | | { |
| | 10 | 157 | | if (!_namesByGroup.TryGetValue(group, out var names)) |
| | | 158 | | continue; |
| | | 159 | | |
| | 24 | 160 | | if (names.Any(n => n.Length >= 3 && phrases.Contains(n))) |
| | 4 | 161 | | return true; |
| | | 162 | | } |
| | | 163 | | |
| | 6 | 164 | | return false; |
| | 4 | 165 | | } |
| | | 166 | | |
| | | 167 | | /// <summary>Написания (нормализованные), объединённые в группу; для показа в диагностике.</summary> |
| | | 168 | | public IReadOnlyCollection<string> NamesOfGroup(string group) |
| | 9 | 169 | | => _namesByGroup.TryGetValue(group, out var names) ? names.OrderBy(n => n, StringComparer.Ordinal).ToList() : [] |
| | | 170 | | |
| | 1099 | 171 | | private static string Norm(string? value) => ArticleKey.Key(value); |
| | | 172 | | |
| | | 173 | | private void AddName(string? name, string group) |
| | | 174 | | { |
| | 910 | 175 | | var norm = Norm(name); |
| | 910 | 176 | | if (norm.Length == 0) |
| | 303 | 177 | | return; |
| | | 178 | | |
| | 607 | 179 | | GetOrAdd(_nameToRawGroups, norm).Add(group); |
| | 607 | 180 | | GetOrAdd(_namesByGroup, group).Add(norm); |
| | 607 | 181 | | } |
| | | 182 | | |
| | | 183 | | private HashSet<string>? RawGroupsOfNorm(string norm) |
| | 145 | 184 | | => norm.Length > 0 && _nameToRawGroups.TryGetValue(norm, out var raw) ? raw : null; |
| | | 185 | | |
| | | 186 | | private HashSet<string> CreateNameGroup(string norm, string? name) |
| | | 187 | | { |
| | 1 | 188 | | var group = "n:" + norm; |
| | 1 | 189 | | EnsureGroup(group); |
| | 1 | 190 | | AddName(name, group); |
| | 1 | 191 | | return _nameToRawGroups[norm]; |
| | | 192 | | } |
| | | 193 | | |
| | | 194 | | private IReadOnlySet<string> Representatives(IEnumerable<string> rawGroups) |
| | 163 | 195 | | => rawGroups.Select(Find).ToHashSet(StringComparer.Ordinal); |
| | | 196 | | |
| | | 197 | | private void EnsureGroup(string group) |
| | | 198 | | { |
| | 304 | 199 | | _parent.TryAdd(group, group); |
| | 304 | 200 | | GetOrAdd(_namesByGroup, group); |
| | 304 | 201 | | } |
| | | 202 | | |
| | | 203 | | private string Find(string group) |
| | | 204 | | { |
| | 170 | 205 | | var root = group; |
| | 173 | 206 | | while (_parent.TryGetValue(root, out var parent) && parent != root) |
| | 3 | 207 | | root = parent; |
| | | 208 | | |
| | 170 | 209 | | return root; |
| | | 210 | | } |
| | | 211 | | |
| | | 212 | | private void Union(string a, string b) |
| | | 213 | | { |
| | 2 | 214 | | var rootA = Find(a); |
| | 2 | 215 | | var rootB = Find(b); |
| | 2 | 216 | | if (rootA == rootB) |
| | 0 | 217 | | return; |
| | | 218 | | |
| | | 219 | | // Имена объединённой группы нужны для NameMentionedIn под представителем. |
| | 2 | 220 | | var winner = string.CompareOrdinal(rootA, rootB) <= 0 ? rootA : rootB; |
| | 2 | 221 | | var loser = winner == rootA ? rootB : rootA; |
| | 2 | 222 | | _parent[loser] = winner; |
| | 2 | 223 | | GetOrAdd(_namesByGroup, winner).UnionWith(GetOrAdd(_namesByGroup, loser)); |
| | 2 | 224 | | } |
| | | 225 | | |
| | | 226 | | private static HashSet<string> GetOrAdd(Dictionary<string, HashSet<string>> map, string key) |
| | | 227 | | { |
| | 1735 | 228 | | if (!map.TryGetValue(key, out var set)) |
| | 715 | 229 | | map[key] = set = new HashSet<string>(StringComparer.Ordinal); |
| | | 230 | | |
| | 1735 | 231 | | return set; |
| | | 232 | | } |
| | | 233 | | } |