feat(commons): quote-aware OverrideCsvParser (T16 CSV)

2026-06-18 02:01:38 -04:00
parent a87bf8a459
commit 77a31ba994
2 changed files with 369 additions and 0 deletions
@@ -0,0 +1,197 @@
+namespace ZB.MOM.WW.ScadaBridge.Commons.Types;
+
+/// <summary>
+/// One parsed instance-attribute-override CSV row: an attribute name plus its
+/// (canonical-string) value and optional list element type. A null
+/// <see cref="Value"/> means "clear the override". <see cref="LineNumber"/> is the
+/// 1-based source line (the header is line 1) so downstream errors can point back
+/// at the operator's file.
+/// </summary>
+public sealed record OverrideCsvRow(string AttributeName, string? Value, string? ElementType, int LineNumber);
+
+/// <summary>
+/// Outcome of parsing an override CSV: the successfully-parsed <see cref="Rows"/>
+/// plus per-line <see cref="Errors"/>. Parsing never throws — malformed rows are
+/// reported and excluded, valid rows still flow through. Downstream callers
+/// validate names/types against the instance schema; this parser is purely
+/// syntactic.
+/// </summary>
+public sealed record OverrideCsvParseResult(IReadOnlyList<OverrideCsvRow> Rows, IReadOnlyList<string> Errors);
+
+/// <summary>
+/// Pure, dependency-free, quote-aware parser turning instance-attribute-override
+/// CSV text into structured rows plus per-line errors. Callers supply the text
+/// (no file I/O). The header row is required and case-insensitive
+/// (<c>AttributeName,Value,ElementType</c>); the <c>ElementType</c> column is
+/// optional. Fields follow RFC-4180 quoting: a double-quoted field may embed
+/// commas and doubled quotes (<c>""</c> → <c>"</c>); only unquoted fields are
+/// whitespace-trimmed.
+/// </summary>
+public static class OverrideCsvParser
+{
+    private const string HeaderError =
+        "Missing or invalid header row. Expected 'AttributeName,Value,ElementType' " +
+        "(ElementType column optional).";
+
+    /// <summary>
+    /// Parses override CSV <paramref name="csvText"/>. Returns parsed rows and any
+    /// per-line errors; never throws. On a missing/unrecognized header returns zero
+    /// rows and a single header error.
+    /// </summary>
+    public static OverrideCsvParseResult Parse(string csvText)
+    {
+        var rows = new List<OverrideCsvRow>();
+        var errors = new List<string>();
+
+        // Split into physical lines; \r\n and \r are normalized to \n boundaries.
+        var lines = (csvText ?? string.Empty).Replace("\r\n", "\n").Replace('\r', '\n').Split('\n');
+
+        var headerSeen = false;
+        var hasElementTypeColumn = false;
+
+        for (var i = 0; i < lines.Length; i++)
+        {
+            var lineNumber = i + 1;
+            var rawLine = lines[i];
+
+            // Skip fully-blank lines (whitespace-only included) without error.
+            if (string.IsNullOrWhiteSpace(rawLine))
+                continue;
+
+            var fields = SplitFields(rawLine);
+
+            if (!headerSeen)
+            {
+                if (!TryMatchHeader(fields, out hasElementTypeColumn))
+                {
+                    errors.Add(HeaderError);
+                    return new OverrideCsvParseResult(rows, errors);
+                }
+
+                headerSeen = true;
+                continue;
+            }
+
+            var expectedColumns = hasElementTypeColumn ? 3 : 2;
+            if (fields.Count != expectedColumns)
+            {
+                errors.Add(
+                    $"Line {lineNumber}: expected {expectedColumns} columns but found {fields.Count}.");
+                continue;
+            }
+
+            var attributeName = fields[0];
+            if (string.IsNullOrWhiteSpace(attributeName))
+            {
+                errors.Add($"Line {lineNumber}: AttributeName must not be blank.");
+                continue;
+            }
+
+            var value = NullIfEmpty(fields[1]);
+            var elementType = hasElementTypeColumn ? NullIfEmpty(fields[2]) : null;
+
+            rows.Add(new OverrideCsvRow(attributeName, value, elementType, lineNumber));
+        }
+
+        if (!headerSeen)
+            errors.Add(HeaderError);
+
+        return new OverrideCsvParseResult(rows, errors);
+    }
+
+    /// <summary>Matches the required header (case-insensitive); reports whether the optional ElementType column is present.</summary>
+    private static bool TryMatchHeader(IReadOnlyList<string> fields, out bool hasElementTypeColumn)
+    {
+        hasElementTypeColumn = false;
+
+        var matchesTwoColumn =
+            fields.Count == 2 &&
+            HeaderEquals(fields[0], "AttributeName") &&
+            HeaderEquals(fields[1], "Value");
+
+        var matchesThreeColumn =
+            fields.Count == 3 &&
+            HeaderEquals(fields[0], "AttributeName") &&
+            HeaderEquals(fields[1], "Value") &&
+            HeaderEquals(fields[2], "ElementType");
+
+        if (matchesThreeColumn)
+        {
+            hasElementTypeColumn = true;
+            return true;
+        }
+
+        return matchesTwoColumn;
+    }
+
+    private static bool HeaderEquals(string field, string expected) =>
+        string.Equals(field, expected, StringComparison.OrdinalIgnoreCase);
+
+    private static string? NullIfEmpty(string field) => field.Length == 0 ? null : field;
+
+    /// <summary>
+    /// RFC-4180-ish field splitter for a single physical line: quoted fields may
+    /// embed commas and doubled quotes (<c>""</c> → <c>"</c>); unquoted fields are
+    /// whitespace-trimmed.
+    /// </summary>
+    private static List<string> SplitFields(string line)
+    {
+        var fields = new List<string>();
+        var field = new System.Text.StringBuilder();
+        var inQuotes = false;
+        var quoted = false; // this field was (at least partly) quoted → preserve whitespace
+
+        for (var i = 0; i < line.Length; i++)
+        {
+            var c = line[i];
+
+            if (inQuotes)
+            {
+                if (c == '"')
+                {
+                    // Doubled quote inside a quoted field → a single literal quote.
+                    if (i + 1 < line.Length && line[i + 1] == '"')
+                    {
+                        field.Append('"');
+                        i++;
+                    }
+                    else
+                    {
+                        inQuotes = false;
+                    }
+                }
+                else
+                {
+                    field.Append(c);
+                }
+
+                continue;
+            }
+
+            switch (c)
+            {
+                case '"':
+                    inQuotes = true;
+                    quoted = true;
+                    break;
+                case ',':
+                    fields.Add(Finalize(field, quoted));
+                    field.Clear();
+                    quoted = false;
+                    break;
+                default:
+                    field.Append(c);
+                    break;
+            }
+        }
+
+        fields.Add(Finalize(field, quoted));
+        return fields;
+    }
+
+    private static string Finalize(System.Text.StringBuilder field, bool quoted)
+    {
+        var text = field.ToString();
+        return quoted ? text : text.Trim(); // only unquoted whitespace is trimmed
+    }
+}
@@ -0,0 +1,172 @@
+using ZB.MOM.WW.ScadaBridge.Commons.Types;
+
+namespace ZB.MOM.WW.ScadaBridge.Commons.Tests;
+
+public class OverrideCsvParserTests
+{
+    [Fact]
+    public void Parse_SimpleThreeColumnFile_ReturnsTwoRowsNoErrors()
+    {
+        const string csv = "AttributeName,Value,ElementType\nSetpoint,42,Int32\nName,Pump A,\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        Assert.Equal(2, result.Rows.Count);
+
+        Assert.Equal("Setpoint", result.Rows[0].AttributeName);
+        Assert.Equal("42", result.Rows[0].Value);
+        Assert.Equal("Int32", result.Rows[0].ElementType);
+        Assert.Equal(2, result.Rows[0].LineNumber);
+
+        Assert.Equal("Name", result.Rows[1].AttributeName);
+        Assert.Equal("Pump A", result.Rows[1].Value);
+        Assert.Null(result.Rows[1].ElementType);
+        Assert.Equal(3, result.Rows[1].LineNumber);
+    }
+
+    [Fact]
+    public void Parse_TwoColumnFileWithoutElementType_RowsHaveNullElementType()
+    {
+        const string csv = "AttributeName,Value\nSetpoint,42\nName,Pump A\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        Assert.Equal(2, result.Rows.Count);
+        Assert.All(result.Rows, r => Assert.Null(r.ElementType));
+        Assert.Equal("42", result.Rows[0].Value);
+        Assert.Equal("Pump A", result.Rows[1].Value);
+    }
+
+    [Fact]
+    public void Parse_QuotedValueWithComma_PreservesEmbeddedComma()
+    {
+        const string csv = "AttributeName,Value,ElementType\nName,\"a,b,c\",\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        var row = Assert.Single(result.Rows);
+        Assert.Equal("Name", row.AttributeName);
+        Assert.Equal("a,b,c", row.Value);
+        Assert.Null(row.ElementType);
+    }
+
+    [Fact]
+    public void Parse_DoubledQuoteEscape_UnescapesToSingleQuote()
+    {
+        const string csv = "AttributeName,Value,ElementType\nName,\"he said \"\"hi\"\"\",\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        var row = Assert.Single(result.Rows);
+        Assert.Equal("he said \"hi\"", row.Value);
+    }
+
+    [Fact]
+    public void Parse_EmptyValueField_YieldsNullValue()
+    {
+        const string csv = "AttributeName,Value,ElementType\nSetpoint,,\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        var row = Assert.Single(result.Rows);
+        Assert.Equal("Setpoint", row.AttributeName);
+        Assert.Null(row.Value);
+        Assert.Null(row.ElementType);
+    }
+
+    [Fact]
+    public void Parse_RowWithTooFewColumns_ProducesLineNumberedErrorAndExcludesRow()
+    {
+        const string csv = "AttributeName,Value,ElementType\nSetpoint\nName,Pump A,\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        // Bad row on line 2 excluded; good row on line 3 retained.
+        var row = Assert.Single(result.Rows);
+        Assert.Equal("Name", row.AttributeName);
+        Assert.Equal(3, row.LineNumber);
+
+        var error = Assert.Single(result.Errors);
+        Assert.Contains("2", error);
+    }
+
+    [Fact]
+    public void Parse_BlankAttributeName_ProducesLineNumberedError()
+    {
+        const string csv = "AttributeName,Value,ElementType\n,42,Int32\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Rows);
+        var error = Assert.Single(result.Errors);
+        Assert.Contains("2", error);
+    }
+
+    [Fact]
+    public void Parse_BlankLines_AreSkippedWithoutError()
+    {
+        const string csv = "AttributeName,Value,ElementType\n\nSetpoint,42,Int32\n   \nName,Pump A,\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        Assert.Equal(2, result.Rows.Count);
+        // LineNumber reflects the true source line (blank line 2 skipped).
+        Assert.Equal(3, result.Rows[0].LineNumber);
+        Assert.Equal(5, result.Rows[1].LineNumber);
+    }
+
+    [Fact]
+    public void Parse_MissingHeader_ReturnsZeroRowsAndHeaderError()
+    {
+        const string csv = "Setpoint,42,Int32\nName,Pump A,\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Rows);
+        var error = Assert.Single(result.Errors);
+        Assert.Contains("header", error, StringComparison.OrdinalIgnoreCase);
+    }
+
+    [Fact]
+    public void Parse_HeaderIsCaseInsensitiveAndTrimsWhitespace()
+    {
+        const string csv = "attributename, value , elementtype\nSetpoint,42,Int32\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        var row = Assert.Single(result.Rows);
+        Assert.Equal("Setpoint", row.AttributeName);
+        Assert.Equal("42", row.Value);
+        Assert.Equal("Int32", row.ElementType);
+    }
+
+    [Fact]
+    public void Parse_EmptyInput_ReturnsHeaderError()
+    {
+        var result = OverrideCsvParser.Parse(string.Empty);
+
+        Assert.Empty(result.Rows);
+        Assert.Single(result.Errors);
+    }
+
+    [Fact]
+    public void Parse_UnquotedWhitespace_IsTrimmedButQuotedWhitespacePreserved()
+    {
+        const string csv = "AttributeName,Value,ElementType\n  Setpoint ,  42  ,Int32\nName,\"  spaced  \",\n";
+
+        var result = OverrideCsvParser.Parse(csv);
+
+        Assert.Empty(result.Errors);
+        Assert.Equal(2, result.Rows.Count);
+        Assert.Equal("Setpoint", result.Rows[0].AttributeName);
+        Assert.Equal("42", result.Rows[0].Value);
+        Assert.Equal("  spaced  ", result.Rows[1].Value);
+    }
+}