xTr1m/RolemasterDB
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Implement phase 5 critical branch extraction

Browse Source
This commit is contained in:
Frank Tovar
2026-03-14 10:21:26 +01:00
parent b2f61c3d73
commit 60c5d886a4
20 changed files with 589 additions and 399 deletions
Download Patch File Download Diff File Expand all files Collapse all files

141
src/RolemasterDb.ImportTool/Parsing/VariantColumnCriticalTableParser.cs
View File

@@ -28,13 +28,7 @@ public sealed class VariantColumnCriticalTableParser
.ToList();
var bodyStartTop = headerFragments.Max(item => item.Top) + CriticalTableParserSupport.HeaderToBodyMinimumGap;
var keyTop = fragments
.Where(item =>
string.Equals(item.Text, "Key:", StringComparison.OrdinalIgnoreCase) ||
item.Text.Contains("must parry", StringComparison.OrdinalIgnoreCase) ||
item.Text.Contains("attacker gets", StringComparison.OrdinalIgnoreCase))
.Select(item => (int?)item.Top)
.Min() ?? int.MaxValue;
var keyTop = CriticalTableParserSupport.FindKeyTop(fragments);
var affixLegendSymbols = CriticalTableParserSupport.DetectAffixLegendSymbols(fragments, keyTop);
var leftCutoff = headerFragments.Min(item => item.Left) - 10;
var rowLabelFragments = CriticalTableParserSupport.FindRowLabelFragments(
@@ -42,11 +36,7 @@ public sealed class VariantColumnCriticalTableParser
leftCutoff,
bodyStartTop,
keyTop);
var rowAnchors = rowLabelFragments
.OrderBy(item => item.Top)
.Select((item, index) => new RowAnchor(CriticalTableParserSupport.NormalizeRollBandLabel(item.Text), item.Top, index + 1))
.ToList();
var rowAnchors = CriticalTableParserSupport.CreateRowAnchors(rowLabelFragments);
if (rowAnchors.Count == 0)
{
@@ -57,33 +47,32 @@ public sealed class VariantColumnCriticalTableParser
.Select(item => (item.Key, item.CenterX))
.ToList();
var bodyFragments = fragments
.Where(item =>
item.Top >= bodyStartTop &&
item.Top < keyTop - CriticalTableParserSupport.TopGroupingTolerance &&
!CriticalTableParserSupport.IsFooterPageNumberFragment(item, keyTop) &&
!CriticalTableParserSupport.IsPotentialRowLabelFragment(item, leftCutoff) &&
!rowAnchors.Any(anchor => anchor.Top == item.Top && string.Equals(anchor.Label, CriticalTableParserSupport.NormalizeRollBandLabel(item.Text), StringComparison.OrdinalIgnoreCase)) &&
!headerFragments.Contains(item))
.ToList();
bodyFragments = CriticalTableParserSupport.SplitBoundaryCrossingAffixFragments(bodyFragments, columnCenters, affixLegendSymbols);
var bodyFragments = CriticalTableParserSupport.BuildBodyFragments(
fragments,
bodyStartTop,
keyTop,
leftCutoff,
rowAnchors,
headerFragments,
columnCenters,
affixLegendSymbols);
var bodyLines = CriticalTableParserSupport.BuildBodyLines(bodyFragments, columnCenters, affixLegendSymbols);
var parsedRollBands = rowAnchors
.Select(anchor => CriticalTableParserSupport.CreateRollBand(anchor.Label, anchor.SortOrder))
.ToList();
var cellEntries = new List<CellEntry>();
var cellEntries = new List<ColumnarCellEntry>();
for (var rowIndex = 0; rowIndex < rowAnchors.Count; rowIndex++)
{
var rowStart = rowIndex == 0
? bodyStartTop
: ResolveRowBoundaryTop(rowAnchors[rowIndex - 1], rowAnchors[rowIndex], bodyLines);
: CriticalTableParserSupport.ResolveRowBoundaryTop(rowAnchors[rowIndex - 1], rowAnchors[rowIndex], bodyLines);
var rowEnd = rowIndex == rowAnchors.Count - 1
? keyTop - 1
: ResolveRowBoundaryTop(rowAnchors[rowIndex], rowAnchors[rowIndex + 1], bodyLines);
: CriticalTableParserSupport.ResolveRowBoundaryTop(rowAnchors[rowIndex], rowAnchors[rowIndex + 1], bodyLines);
var rowFragments = bodyFragments
.Where(item => item.Top >= rowStart && item.Top < rowEnd)
@@ -103,7 +92,8 @@ public sealed class VariantColumnCriticalTableParser
continue;
}
cellEntries.Add(new CellEntry(
cellEntries.Add(new ColumnarCellEntry(
null,
rowAnchors[rowIndex].Label,
rowIndex,
columnAnchor.Key,
@@ -111,42 +101,11 @@ public sealed class VariantColumnCriticalTableParser
}
}
RepairLeadingAffixLeakage(cellEntries, affixLegendSymbols);
CriticalTableParserSupport.RepairLeadingAffixLeakage(cellEntries, affixLegendSymbols);
var parsedCells = new List<ParsedCriticalCellArtifact>();
var parsedResults = new List<ParsedCriticalResult>();
foreach (var cellEntry in cellEntries.OrderBy(item => item.RowIndex).ThenBy(item => item.ColumnKey, StringComparer.Ordinal))
{
var segmentCount = CriticalTableParserSupport.CountLineTypeSegments(cellEntry.Lines, affixLegendSymbols);
if (segmentCount > 2)
{
validationErrors.Add($"Cell '{cellEntry.RollBandLabel}/{cellEntry.ColumnKey}' interleaves prose and affix lines.");
}
var rawAffixLines = cellEntry.Lines.Where(line => CriticalTableParserSupport.IsAffixLikeLine(line, affixLegendSymbols)).ToList();
var descriptionLines = cellEntry.Lines.Where(line => !CriticalTableParserSupport.IsAffixLikeLine(line, affixLegendSymbols)).ToList();
var rawCellText = string.Join(Environment.NewLine, cellEntry.Lines);
var descriptionText = CriticalTableParserSupport.CollapseWhitespace(string.Join(' ', descriptionLines));
var rawAffixText = rawAffixLines.Count == 0 ? null : string.Join(Environment.NewLine, rawAffixLines);
parsedCells.Add(new ParsedCriticalCellArtifact(
null,
cellEntry.RollBandLabel,
cellEntry.ColumnKey,
cellEntry.Lines,
rawCellText,
descriptionText,
rawAffixText));
parsedResults.Add(new ParsedCriticalResult(
null,
cellEntry.ColumnKey,
cellEntry.RollBandLabel,
rawCellText,
descriptionText,
rawAffixText));
}
CriticalTableParserSupport.BuildParsedArtifacts(cellEntries, affixLegendSymbols, parsedCells, parsedResults, validationErrors);
if (columnAnchors.Count != ExpectedColumns.Length)
{
@@ -208,69 +167,5 @@ public sealed class VariantColumnCriticalTableParser
ExpectedColumns.SingleOrDefault(item => string.Equals(item.Label, value.Trim(), StringComparison.OrdinalIgnoreCase))
?? throw new InvalidOperationException($"Unsupported variant column label '{value}'.");
private static void RepairLeadingAffixLeakage(List<CellEntry> cellEntries, ISet<string> affixLegendSymbols)
{
var maxRowIndex = cellEntries.Count == 0 ? -1 : cellEntries.Max(item => item.RowIndex);
var columnKeys = cellEntries.Select(item => item.ColumnKey).Distinct(StringComparer.OrdinalIgnoreCase).ToList();
for (var rowIndex = 0; rowIndex < maxRowIndex; rowIndex++)
{
foreach (var columnKey in columnKeys)
{
var current = cellEntries.SingleOrDefault(item => item.RowIndex == rowIndex && item.ColumnKey == columnKey);
var next = cellEntries.SingleOrDefault(item => item.RowIndex == rowIndex + 1 && item.ColumnKey == columnKey);
if (current is null || next is null)
{
continue;
}
var leadingAffixCount = 0;
while (leadingAffixCount < next.Lines.Count && CriticalTableParserSupport.IsAffixLikeLine(next.Lines[leadingAffixCount], affixLegendSymbols))
{
leadingAffixCount++;
}
if (leadingAffixCount == 0 || leadingAffixCount == next.Lines.Count)
{
continue;
}
current.Lines.AddRange(next.Lines.Take(leadingAffixCount));
next.Lines.RemoveRange(0, leadingAffixCount);
}
}
}
private static int ResolveRowBoundaryTop(
RowAnchor current,
RowAnchor next,
IReadOnlyList<(int Top, bool IsAffixLike)> bodyLines)
{
var linesBetweenLabels = bodyLines
.Where(item => item.Top >= current.Top && item.Top < next.Top)
.OrderBy(item => item.Top)
.ToList();
for (var index = linesBetweenLabels.Count - 2; index >= 0; index--)
{
if (linesBetweenLabels[index].IsAffixLike && !linesBetweenLabels[index + 1].IsAffixLike)
{
return (int)Math.Floor((linesBetweenLabels[index].Top + linesBetweenLabels[index + 1].Top) / 2.0) + 1;
}
}
return (int)Math.Floor((current.Top + next.Top) / 2.0) + 1;
}
private sealed record ColumnDefinition(string Key, string Label);
private sealed record RowAnchor(string Label, int Top, int SortOrder);
private sealed class CellEntry(string rollBandLabel, int rowIndex, string columnKey, List<string> lines)
{
public string RollBandLabel { get; } = rollBandLabel;
public int RowIndex { get; } = rowIndex;
public string ColumnKey { get; } = columnKey;
public List<string> Lines { get; } = lines;
}
}