Port key terms updates

src/SIL.Machine/Corpora/CorporaUtils.cs

@@ -75,12 +75,14 @@ public static string MergeVerseRanges(string verse1, string verse2)
             return sb.ToString();
         }
 
-        internal static IEnumerable<(string Id, string FileName)> GetFiles(IEnumerable<string> filePatterns)
+        internal static IEnumerable<(string Id, string FileName, int PatternIndex)> GetFiles(
+            IEnumerable<string> filePatterns
+        )
         {
             string[] filePatternArray = filePatterns.ToArray();
             if (filePatternArray.Length == 1 && File.Exists(filePatternArray[0]))
             {
-                yield return ("*all*", filePatternArray[0]);
+                yield return ("*all*", filePatternArray[0], 0);
             }
             else
             {
@@ -89,7 +91,7 @@ public static string MergeVerseRanges(string verse1, string verse2)
                     string filePattern = filePatternArray[i];
                     if (File.Exists(filePattern))
                     {
-                        yield return (i.ToString(CultureInfo.InvariantCulture), filePattern);
+                        yield return (i.ToString(CultureInfo.InvariantCulture), filePattern, i);
                         continue;
                     }
 
@@ -145,7 +147,7 @@ public static string MergeVerseRanges(string verse1, string verse2)
                             if (sb.Length > 0)
                                 id = sb.ToString();
                         }
-                        yield return (id, fileName);
+                        yield return (id, fileName, i);
                     }
                 }
             }

src/SIL.Machine/Corpora/FileParatextProjectFileHandler.cs

@@ -14,37 +14,37 @@ public FileParatextProjectFileHandler(string projectDir)
 
         public bool Exists(string fileName)
         {
-            return Directory
-                .EnumerateFiles(_projectDir)
-                .Any(f => Path.GetFileName(f).Equals(fileName, System.StringComparison.InvariantCultureIgnoreCase));
+            return GetFileName(fileName) != null;
         }
 
         public Stream Open(string fileName)
         {
-            return File.OpenRead(
-                Path.Combine(
-                    _projectDir,
-                    Directory
-                        .EnumerateFiles(_projectDir)
-                        .FirstOrDefault(f =>
-                            Path.GetFileName(f).Equals(fileName, System.StringComparison.InvariantCultureIgnoreCase)
-                        )
-                )
-            );
+            fileName = GetFileName(fileName) ?? fileName;
+            return File.OpenRead(Path.Combine(_projectDir, fileName));
         }
 
         public UsfmStylesheet CreateStylesheet(string fileName)
         {
-            string customStylesheetFileName = Path.Combine(_projectDir, "custom.sty");
+            string customStylesheetFileName = GetFileName("custom.sty");
             return new UsfmStylesheet(
                 fileName,
-                File.Exists(customStylesheetFileName) ? customStylesheetFileName : null
+                customStylesheetFileName != null ? Path.Combine(_projectDir, customStylesheetFileName) : null
             );
         }
 
         public string Find(string extension)
         {
             return Directory.EnumerateFiles(_projectDir, "*" + extension).FirstOrDefault();
         }
+
+        private string GetFileName(string caseInsensitiveFileName)
+        {
+            return Directory
+                .EnumerateFiles(_projectDir)
+                .Select(p => Path.GetFileName(p))
+                .FirstOrDefault(f =>
+                    f.Equals(caseInsensitiveFileName, System.StringComparison.InvariantCultureIgnoreCase)
+                );
+        }
     }
 }

src/SIL.Machine/Corpora/KeyTerm.cs

@@ -0,0 +1,30 @@
+using System.Collections.Generic;
+using System.Linq;
+using SIL.Scripture;
+
+public class KeyTerm
+{
+    public string Id { get; }
+    public string Category { get; }
+    public string Domain { get; }
+    public IReadOnlyList<string> Renderings { get; }
+    public IReadOnlyList<VerseRef> References { get; }
+    public IReadOnlyList<string> RenderingsPatterns { get; }
+
+    public KeyTerm(
+        string id,
+        string category,
+        string domain,
+        IEnumerable<string> renderings,
+        IEnumerable<VerseRef> references,
+        IEnumerable<string> renderingsPatterns
+    )
+    {
+        Id = id;
+        Category = category;
+        Domain = domain;
+        Renderings = renderings.ToArray();
+        References = references.ToArray();
+        RenderingsPatterns = renderingsPatterns.ToArray();
+    }
+}

src/SIL.Machine/Corpora/NParallelTextCorpus.cs

@@ -294,6 +294,8 @@ private IEnumerable<NParallelTextRow> CreateRows(
                 throw new ArgumentNullException("A corpus row must be specified.");
 
             object[] defaultRefs = new object[] { rows.Where(r => r != null).Select(r => r.Ref).First() };
+            TextRowContentType contentType = TextRowContentType.Segment;
+
             string textId = null;
             object[][] refs = new object[N][];
             TextRowFlags[] flags = new TextRowFlags[N];
@@ -327,7 +329,7 @@ private IEnumerable<NParallelTextRow> CreateRows(
             }
             refs = refs.Select(r => r ?? defaultRefs).ToArray();
 
-            yield return new NParallelTextRow(textId, refs)
+            yield return new NParallelTextRow(textId, refs, contentType)
             {
                 NSegments = rows.Select(r => r?.Segment ?? Array.Empty<string>()).ToArray(),
                 NFlags = flags.ToReadOnlyList()
@@ -441,6 +443,7 @@ private class RangeRow
             public bool IsSentenceStart { get; set; } = false;
             public bool IsInRange => Refs.Count > 0;
             public bool IsEmpty => Segment.Count == 0;
+            public TextRowContentType ContentType { get; set; } = TextRowContentType.Segment;
         }
 
         private class NRangeInfo
@@ -451,6 +454,7 @@ private class NRangeInfo
             public IComparer<object> RowRefComparer { get; set; } = null;
             public List<RangeRow> Rows { get; }
             public bool IsInRange => Rows.Any(r => r.IsInRange);
+            public TextRowContentType ContentType { get; set; } = TextRowContentType.Segment;
 
             public NRangeInfo(int n)
             {
@@ -472,6 +476,7 @@ public void AddTextRow(TextRow row, int index)
                 }
                 TextId = row.TextId;
                 Rows[index].Refs.Add(row.Ref);
+                Rows[index].ContentType = row.ContentType;
                 if (Rows[index].IsEmpty)
                     Rows[index].IsSentenceStart = row.IsSentenceStart;
                 Rows[index].Segment.AddRange(row.Segment);
@@ -486,8 +491,8 @@ public NParallelTextRow CreateRow()
                     .ToList();
                 foreach (int i in Enumerable.Range(0, Rows.Count))
                 {
-                    var row = Rows[i];
-
+                    RangeRow row = Rows[i];
+                    ContentType = row.ContentType;
                     if (Versifications.All(v => v != null) && row.Refs.Count() == 0)
                     {
                         refs[i] = referenceRefs
@@ -502,7 +507,7 @@ public NParallelTextRow CreateRow()
                         refs[i] = row.Refs.ToArray();
                     }
                 }
-                var nParRow = new NParallelTextRow(TextId, refs)
+                var nParRow = new NParallelTextRow(TextId, refs, ContentType)
                 {
                     NSegments = Rows.Select(r => r.Segment.ToArray()).ToArray(),
                     NFlags = Rows.Select(r => r.IsSentenceStart ? TextRowFlags.SentenceStart : TextRowFlags.None)

src/SIL.Machine/Corpora/NParallelTextRow.cs

@@ -8,7 +8,11 @@ namespace SIL.Machine.Corpora
 {
     public class NParallelTextRow : IRow
     {
-        public NParallelTextRow(string textId, IEnumerable<IReadOnlyList<object>> nRefs)
+        public NParallelTextRow(
+            string textId,
+            IEnumerable<IReadOnlyList<object>> nRefs,
+            TextRowContentType contentType = TextRowContentType.Segment
+        )
         {
             if (string.IsNullOrEmpty(textId))
                 throw new ArgumentNullException(nameof(textId));
@@ -21,6 +25,7 @@ public NParallelTextRow(string textId, IEnumerable<IReadOnlyList<object>> nRefs)
             N = NRefs.Count;
             NSegments = Enumerable.Range(0, N).Select(_ => Array.Empty<string>()).ToImmutableArray();
             NFlags = Enumerable.Range(0, N).Select(_ => TextRowFlags.SentenceStart).ToImmutableArray();
+            ContentType = contentType;
         }
 
         public string TextId { get; }
@@ -32,6 +37,7 @@ public NParallelTextRow(string textId, IEnumerable<IReadOnlyList<object>> nRefs)
 
         public IReadOnlyList<IReadOnlyList<string>> NSegments { get; set; }
         public IReadOnlyList<TextRowFlags> NFlags { get; set; }
+        public TextRowContentType ContentType { get; }
 
         public bool IsSentenceStart(int i) =>
             NFlags.Count > i ? NFlags[i].HasFlag(TextRowFlags.SentenceStart) : throw new ArgumentOutOfRangeException();
@@ -48,7 +54,10 @@ public bool IsRangeStart(int i) =>
 
         public NParallelTextRow Invert()
         {
-            return new NParallelTextRow(TextId, NRefs.Reverse()) { NFlags = NFlags.Reverse().ToImmutableArray(), };
+            return new NParallelTextRow(TextId, NRefs.Reverse(), ContentType)
+            {
+                NFlags = NFlags.Reverse().ToImmutableArray(),
+            };
         }
     }
 }

src/SIL.Machine/Corpora/ParallelTextCorpus.cs

@@ -61,7 +61,8 @@ public override IEnumerable<ParallelTextRow> GetRows(IEnumerable<string> textIds
                     yield return new ParallelTextRow(
                         nRow.TextId,
                         nRow.NRefs[0].Count > 0 || !isScripture ? nRow.NRefs[0] : new object[] { nRow.Ref },
-                        nRow.NRefs[1].Count > 0 || !isScripture ? nRow.NRefs[1] : new object[] { nRow.Ref }
+                        nRow.NRefs[1].Count > 0 || !isScripture ? nRow.NRefs[1] : new object[] { nRow.Ref },
+                        nRow.ContentType
                     )
                     {
                         SourceFlags = nRow.NFlags[0],

src/SIL.Machine/Corpora/ParallelTextRow.cs

@@ -6,7 +6,12 @@ namespace SIL.Machine.Corpora
 {
     public class ParallelTextRow : IRow
     {
-        public ParallelTextRow(string textId, IReadOnlyList<object> sourceRefs, IReadOnlyList<object> targetRefs)
+        public ParallelTextRow(
+            string textId,
+            IReadOnlyList<object> sourceRefs,
+            IReadOnlyList<object> targetRefs,
+            TextRowContentType contentType = TextRowContentType.Segment
+        )
         {
             if (string.IsNullOrEmpty(textId))
                 throw new ArgumentNullException(nameof(textId));
@@ -17,6 +22,7 @@ public ParallelTextRow(string textId, IReadOnlyList<object> sourceRefs, IReadOnl
             TextId = textId;
             SourceRefs = sourceRefs;
             TargetRefs = targetRefs;
+            ContentType = contentType;
         }
 
         public string TextId { get; }
@@ -36,6 +42,7 @@ public ParallelTextRow(string textId, IReadOnlyList<object> sourceRefs, IReadOnl
 
         public TextRowFlags SourceFlags { get; set; } = TextRowFlags.SentenceStart;
         public TextRowFlags TargetFlags { get; set; } = TextRowFlags.SentenceStart;
+        public TextRowContentType ContentType { get; }
 
         public bool IsSourceSentenceStart => SourceFlags.HasFlag(TextRowFlags.SentenceStart);
         public bool IsSourceInRange => SourceFlags.HasFlag(TextRowFlags.InRange);
@@ -51,7 +58,7 @@ public ParallelTextRow(string textId, IReadOnlyList<object> sourceRefs, IReadOnl
 
         public ParallelTextRow Invert()
         {
-            return new ParallelTextRow(TextId, TargetRefs, SourceRefs)
+            return new ParallelTextRow(TextId, TargetRefs, SourceRefs, ContentType)
             {
                 SourceSegment = TargetSegment,
                 TargetSegment = SourceSegment,

src/SIL.Machine/Corpora/ParatextBackupTermsCorpus.cs

@@ -15,9 +15,9 @@ public ParatextBackupTermsCorpus(
         {
             using (var archive = ZipFile.OpenRead(fileName))
             {
-                IEnumerable<(string, IReadOnlyList<string>)> glosses = new ZipParatextProjectTermsParser(archive)
+                IEnumerable<KeyTerm> keyTerms = new ZipParatextProjectTermsParser(archive)
                     .Parse(termCategories, useTermGlosses, chapters)
-                    .OrderBy(g => g.TermId);
+                    .OrderBy(g => g.Id);
 
                 ParatextProjectSettings settings = ZipParatextProjectSettingsParser.Parse(archive);
 
@@ -26,8 +26,11 @@ public ParatextBackupTermsCorpus(
 
                 IText text = new MemoryText(
                     textId,
-                    glosses.SelectMany(kvp =>
-                        kvp.Item2.Select(gloss => new TextRow(textId, kvp.Item1) { Segment = new string[] { gloss } })
+                    keyTerms.SelectMany(keyTerm =>
+                        keyTerm.Renderings.Select(gloss => new TextRow(textId, keyTerm.Id, TextRowContentType.Word)
+                        {
+                            Segment = new string[] { gloss }
+                        })
                     )
                 );
                 AddText(text);