ExpectedVerse should always be empty even if verse ref fails to parse if type is ExtraVerse; filter books; add project name

Author: Enkidu93

src/SIL.Machine/Corpora/ParatextProjectSettings.cs

@@ -104,13 +104,13 @@ public string GetBookFileName(string bookId)
             return FileNamePrefix + bookPart + FileNameSuffix;
         }
 
-        public IEnumerable<string> GetAllScriptureBookFileNames()
+        public IEnumerable<string> GetAllScriptureBookIds()
         {
             BookSet scriptureBooks = Canon.ScriptureBooks;
             scriptureBooks.SelectAll();
             foreach (string bookId in scriptureBooks.SelectedBookIds)
             {
-                yield return GetBookFileName(bookId);
+                yield return bookId;
             }
         }
 

src/SIL.Machine/Corpora/ParatextProjectVersificationErrorDetectorBase.cs

@@ -2,6 +2,7 @@
 using System.Collections.Generic;
 using System.IO;
 using System.Text;
+using SIL.Scripture;
 
 namespace SIL.Machine.Corpora
 {
@@ -20,15 +21,21 @@ ParatextProjectSettings settings
         }
 
         public IReadOnlyList<UsfmVersificationError> GetUsfmVersificationErrors(
-            UsfmVersificationErrorDetector handler = null
+            UsfmVersificationErrorDetector handler = null,
+            HashSet<int> books = null
         )
         {
-            handler = handler ?? new UsfmVersificationErrorDetector(_settings.Versification);
-            foreach (string fileName in _settings.GetAllScriptureBookFileNames())
+            handler = handler ?? new UsfmVersificationErrorDetector(_settings);
+            foreach (string bookId in _settings.GetAllScriptureBookIds())
             {
+                string fileName = _settings.GetBookFileName(bookId);
+
                 if (!_paratextProjectFileHandler.Exists(fileName))
                     continue;
 
+                if (books != null && !books.Contains(Canon.BookIdToNumber(bookId)))
+                    continue;
+
                 string usfm;
                 using (var reader = new StreamReader(_paratextProjectFileHandler.Open(fileName)))
                 {

src/SIL.Machine/Corpora/UsfmVersificationErrorDetector.cs

@@ -100,14 +100,15 @@ public string ExpectedVerseRef
         {
             get
             {
+                if (Type == UsfmVersificationErrorType.ExtraVerse)
+                    return "";
+
                 // We do not want to throw an exception here, and the VerseRef constructor can throw
                 // an exception with certain invalid verse data; use TryParse instead.
                 if (!VerseRef.TryParse($"{_bookNum} {_expectedChapter}:{_expectedVerse}", out VerseRef defaultVerseRef))
                 {
                     return DefaultVerse(_expectedChapter, _expectedVerse);
                 }
-                if (Type == UsfmVersificationErrorType.ExtraVerse)
-                    return "";
                 if (
                     Type == UsfmVersificationErrorType.MissingVerseSegment
                     && VerseRef.TryParse(
@@ -178,16 +179,18 @@ public class UsfmVersificationErrorDetector : UsfmParserHandlerBase
         private VerseRef _currentVerse;
         private readonly List<UsfmVersificationError> _errors;
 
-        public UsfmVersificationErrorDetector(ScrVers versification)
+        public UsfmVersificationErrorDetector(ParatextProjectSettings settings)
         {
-            _versification = versification;
+            ProjectName = settings.Name;
+            _versification = settings.Versification;
             _currentBook = 0;
             _currentChapter = 0;
             _currentVerse = new VerseRef();
             _errors = new List<UsfmVersificationError>();
         }
 
         public IReadOnlyList<UsfmVersificationError> Errors => _errors;
+        public string ProjectName { get; private set; }
 
         public override void EndUsfm(UsfmParserState state)
         {