Use language code from settings file

sillsdev · Aug 16, 2024 · df2e9ac · df2e9ac
1 parent 056c9fd
commit df2e9ac
Show file tree

Hide file tree

Showing 5 changed files with 69 additions and 46 deletions.
diff --git a/src/SIL.Machine/Corpora/ParatextBackupTermsCorpus.cs b/src/SIL.Machine/Corpora/ParatextBackupTermsCorpus.cs
@@ -32,48 +32,17 @@ public class ParatextBackupTermsCorpus : DictionaryTextCorpus
             { "pt", "SIL.Machine.Corpora.BiblicalTermsPt.xml" }
         };
 
+        private static readonly Regex ContentInBracketsRegex = new Regex(@"^\[(.+?)\]$", RegexOptions.Compiled);
+        private static readonly Regex NumericalInformationRegex = new Regex(@"\s+\d+(\.\d+)*$", RegexOptions.Compiled);
+
         public ParatextBackupTermsCorpus(
             string fileName,
             IEnumerable<string> termCategories,
-            string languageCode = null,
             bool preferTermsLocalization = false
         )
         {
             using (var archive = ZipFile.OpenRead(fileName))
             {
-                ZipArchiveEntry termsFileEntry = archive.GetEntry("TermRenderings.xml");
-                XDocument doc;
-                bool useTermsRenderingXml = !preferTermsLocalization && termsFileEntry != null;
-
-                if (!SupportedLanguageTermsLocalizationXmls.TryGetValue(languageCode, out string resourceName))
-                {
-                    if (termsFileEntry != null)
-                    {
-                        useTermsRenderingXml = true;
-                    }
-                    else
-                    {
-                        return;
-                    }
-                }
-
-                if (useTermsRenderingXml)
-                {
-                    using (Stream keyTermsFile = termsFileEntry.Open())
-                    {
-                        doc = XDocument.Load(keyTermsFile);
-                    }
-                }
-                else
-                {
-                    using (
-                        Stream keyTermsFile = Assembly.GetExecutingAssembly().GetManifestResourceStream(resourceName)
-                    )
-                    {
-                        doc = XDocument.Load(keyTermsFile);
-                    }
-                }
-
                 var settingsParser = new ZipParatextProjectSettingsParser(archive);
                 ParatextProjectSettings settings = settingsParser.Parse();
 
@@ -121,6 +90,40 @@ public ParatextBackupTermsCorpus(
                 {
                     termIdToCategoryDictionary = new Dictionary<string, string>();
                 }
+                ZipArchiveEntry termsFileEntry = archive.GetEntry("TermRenderings.xml");
+                XDocument doc;
+                bool useTermsRenderingXml =
+                    (!preferTermsLocalization || settings.BiblicalTermsListType != "Major") && termsFileEntry != null;
+
+                if (!SupportedLanguageTermsLocalizationXmls.TryGetValue(settings.LanguageCode, out string resourceName))
+                {
+                    if (termsFileEntry != null)
+                    {
+                        useTermsRenderingXml = true;
+                    }
+                    else
+                    {
+                        return;
+                    }
+                }
+
+                if (useTermsRenderingXml)
+                {
+                    using (Stream keyTermsFile = termsFileEntry.Open())
+                    {
+                        doc = XDocument.Load(keyTermsFile);
+                    }
+                }
+                else
+                {
+                    using (
+                        Stream keyTermsFile = Assembly.GetExecutingAssembly().GetManifestResourceStream(resourceName)
+                    )
+                    {
+                        doc = XDocument.Load(keyTermsFile);
+                    }
+                }
+
                 AddTexts(doc, settings, termCategories, termIdToCategoryDictionary);
             }
         }
@@ -168,19 +171,17 @@ IDictionary<string, string> termIdToCategoryDictionary
         public static IReadOnlyList<string> GetGlosses(string gloss)
         {
             //If entire term rendering is surrounded in square brackets, remove them
-            Regex rx = new Regex(@"^\[(.+?)\]$", RegexOptions.Compiled);
-            Match rx_match = rx.Match(gloss);
-            if (rx_match.Success)
-                gloss = rx_match.Groups[0].Value;
+            Match match = ContentInBracketsRegex.Match(gloss);
+            if (match.Success)
+                gloss = match.Groups[0].Value;
             gloss = gloss.Replace("?", "");
             gloss = gloss.Replace("*", "");
             gloss = gloss.Replace("/", " ");
             gloss = gloss.Trim();
             gloss = StripParens(gloss);
             gloss = StripParens(gloss, left: '[', right: ']');
-            // gloss = gloss.Trim();
-            Regex rx2 = new Regex(@"\s+\d+(\.\d+)*$", RegexOptions.Compiled);
-            foreach (Match m in rx2.Matches(gloss))
+            gloss = gloss.Trim();
+            foreach (Match m in NumericalInformationRegex.Matches(gloss))
             {
                 gloss.Replace(m.Value, "");
             }

diff --git a/src/SIL.Machine/Corpora/ParatextProjectSettings.cs b/src/SIL.Machine/Corpora/ParatextProjectSettings.cs
@@ -17,7 +17,8 @@ public ParatextProjectSettings(
             string fileNameSuffix,
             string biblicalTermsListType,
             string biblicalTermsProjectName,
-            string biblicalTermsFileName
+            string biblicalTermsFileName,
+            string languageCode
         )
         {
             Name = name;
@@ -31,6 +32,7 @@ string biblicalTermsFileName
             BiblicalTermsListType = biblicalTermsListType;
             BiblicalTermsProjectName = biblicalTermsProjectName;
             BiblicalTermsFileName = biblicalTermsFileName;
+            LanguageCode = languageCode;
         }
 
         public string Name { get; }
@@ -45,6 +47,8 @@ string biblicalTermsFileName
         public string BiblicalTermsProjectName { get; }
         public string BiblicalTermsFileName { get; }
 
+        public string LanguageCode { get; }
+
         public bool IsBookFileName(string fileName, out string bookId)
         {
             bookId = null;

diff --git a/src/SIL.Machine/Corpora/ParatextProjectSettingsParserBase.cs b/src/SIL.Machine/Corpora/ParatextProjectSettingsParserBase.cs
@@ -92,9 +92,25 @@ public ParatextProjectSettings Parse()
             {
                 throw new InvalidOperationException(
                     $"The BiblicalTermsListSetting element in Settings.xml in project {fullName}"
-                        + $" is not in the expected format (i.e., Major::BiblicalTerms.xml) but is {biblicalTermsListSetting}."
+                        + $" is not in the expected format (e.g., Major::BiblicalTerms.xml) but is {biblicalTermsListSetting}."
                 );
             }
+            string languageIsoCodeSetting = settingsDoc.Root.Element("LanguageIsoCode")?.Value;
+            if (languageIsoCodeSetting == null)
+            {
+                throw new InvalidOperationException(
+                    $"The LanguageIsoCode element in Settings.xml in project {fullName} does not exist."
+                );
+            }
+            string[] languageIsoCodeSettingParts = settingsDoc.Root.Element("LanguageIsoCode").Value.Split(':');
+            if (languageIsoCodeSettingParts.Length != 4)
+            {
+                throw new InvalidOperationException(
+                    $"The LanguageIsoCode element in Settings.xml in project {fullName}"
+                        + $" is not in the expected format (e.g., en:::) but is {languageIsoCodeSetting}."
+                );
+            }
+            string languageCode = languageIsoCodeSettingParts[0];
 
             return new ParatextProjectSettings(
                 name,
@@ -107,7 +123,8 @@ public ParatextProjectSettings Parse()
                 suffix,
                 parts[0],
                 parts[1],
-                parts[2]
+                parts[2],
+                languageCode
             );
         }
 

diff --git a/tests/SIL.Machine.Tests/Corpora/ParatextBackupTermsCorpusTests.cs b/tests/SIL.Machine.Tests/Corpora/ParatextBackupTermsCorpusTests.cs
@@ -58,7 +58,7 @@ private class TestEnvironment : DisposableBase
         public TestEnvironment(bool preferTermsLocalization = false)
         {
             _backupPath = CorporaTestHelpers.CreateTestParatextBackup();
-            Corpus = new ParatextBackupTermsCorpus(_backupPath, new string[] { "PN" }, "en", preferTermsLocalization);
+            Corpus = new ParatextBackupTermsCorpus(_backupPath, new string[] { "PN" }, preferTermsLocalization);
         }
 
         public ParatextBackupTermsCorpus Corpus { get; }

diff --git a/tests/SIL.Machine.Tests/Corpora/ParatextProjectSettingsTests.cs b/tests/SIL.Machine.Tests/Corpora/ParatextProjectSettingsTests.cs
@@ -184,7 +184,8 @@ private static ParatextProjectSettings CreateSettings(string fileNameForm)
             ".SFM",
             "Major",
             "",
-            "BiblicalTerms.xml"
+            "BiblicalTerms.xml",
+            "en"
         );
     }
 }