public BlockPage Process(BlockPage page)
{
var result = new BlockPage();
var columnSequence = page.AllBlocks.Select(block =>
{
int columnId = FindColumnId(block);
if (columnId < 0)
{
PdfReaderException.Warning("Invalid blockset column assigned -- review stage 2 and 3");
return(null);
}
return(new ColumnSequence
{
ColumnId = columnId,
H = block.GetH() + block.GetHeight(),
Block = block
});
})
.Where(bl => bl != null)
.OrderBy(block => block);
var dbg = columnSequence.ToArray();
result.AddRange(columnSequence.Select(b => b.Block));
return(result);
}
public BlockPage Process(BlockPage page)
{
var orange = page.AllBlocks.Cast <MarkLine>().Where(l => l.Color == MarkLine.ORANGE);
var result = new BlockPage();
result.AddRange(orange);
bool overlap = HasTableOverlap(result);
if (overlap)
{
PdfReaderException.Warning("MarkOrangeNoOverlap: Overlap");
return(result);
}
// column
var bset = new BlockSet <IBlock>();
bset.Add(new BlockLine()
{
X = 1, H = 1, Width = 1, Height = 1, Text = "MarkOrange"
});
var almostEmpty = new BlockPage();
almostEmpty.Add(bset);
return(almostEmpty);
}
public BlockPage Process(BlockPage page)
{
if (page.AllBlocks.Count() == 0)
{
return(page);
}
float err = 1f;
float minH = page.AllBlocks.Min(b => b.GetH()) + err;
var blocksAtFooter = page.AllBlocks.Where(b => b.GetH() <= minH);
var bottomPage = new BlockPage();
bottomPage.AddRange(blocksAtFooter);
if (!HasFooter(bottomPage))
{
return(page);
}
// remove blockset that corresponds to footer
var result = new BlockPage();
var blocksAboveFooter = page.AllBlocks.Where(b => b.GetH() > minH);
result.AddRange(blocksAboveFooter);
return(result);
}
public BlockPage Validate(BlockPage page)
{
if (page.AllBlocks.Count() == 0)
{
return(page);
}
float err = 1f;
float maxH = page.AllBlocks.Max(b => b.GetH()) - err;
var blocksAtHeader = page.AllBlocks.Where(b => b.GetH() >= maxH);
var result = new BlockPage();
result.AddRange(blocksAtHeader);
float height = result.AllBlocks.GetHeight();
if (height > statRegionTooLarge)
{
PdfReaderException.AlwaysThrow("height > statRegionTooLarge");
}
return(result);
}
public void RemoveImage(IBlock block)
{
if (!(block is ImageBlock))
{
PdfReaderException.AlwaysThrow("Block is not ImageBlock");
}
if (Images == null)
{
PdfReaderException.AlwaysThrow("Images == null");
}
int before = Images.AllBlocks.Count();
var allBlocksMinusOne = Images.AllBlocks.Except(new IBlock[] { block });
Images = new BlockPage();
Images.AddRange(allBlocksMinusOne);
int after = Images.AllBlocks.Count();
if (after == before)
{
PdfReaderException.AlwaysThrow("after == before");
}
}
BlockPage BreakElements(BlockPage page)
{
var blocks = page.AllBlocks.ToList();
var replacements = new IBlock[blocks.Count][];
var result = new BlockPage();
for (int i = 0; i < blocks.Count; i++)
{
if (blocks[i] == null)
{
continue;
}
for (int j = 0; j < blocks.Count; j++)
{
if (blocks[j] == null)
{
continue;
}
// same block
if (i == j)
{
continue;
}
if (OverlapContains(blocks[i], blocks[j]))
{
bool doesntApply = !(blocks[i] is BlockSet <IBlock>);
if (doesntApply)
{
PdfReaderException.Throw("BreakinlineElements: try to break image/table");
continue;
}
var elems = BreakElements(blocks[i], blocks[j]);
if (elems == null)
{
PdfReaderException.Warning("(elems == null)");
continue;
}
// has to do replacement in place
blocks[i] = null;
blocks.AddRange(elems);
//replacements[i] = elems;
break;
}
}
}
result.AddRange(blocks.Where(b => b != null));
return(result);
}
BlockPage MergeElements(BlockPage page)
{
var blocks = page.AllBlocks.ToList();
var replacements = new IBlock[blocks.Count][];
var result = new BlockPage();
for (int i = 0; i < blocks.Count; i++)
{
if (blocks[i] == null)
{
continue;
}
for (int j = 0; j < blocks.Count; j++)
{
if (blocks[j] == null)
{
continue;
}
// same block
if (i == j)
{
continue;
}
bool doesntApplyI = !(blocks[i] is BlockSet <IBlock>);
bool doesntApplyJ = !(blocks[j] is BlockSet <IBlock>);
if (doesntApplyI || doesntApplyJ)
{
continue;
}
if (HasOverlap(blocks[i], blocks[j]))
{
var elems = BreakElements(blocks[i], blocks[j]);
if (elems == null || elems.Length != 2)
{
PdfReaderException.AlwaysThrow("merge: (elems == null || elems.Length != 2 )");
}
// has to do replacement in place
blocks[i] = elems[0];
blocks[j] = elems[1];
//blocks.AddRange(elems);
break;
}
}
}
result.AddRange(blocks.Where(b => b != null));
return(result);
}
public BlockPage Process(BlockPage page)
{
var newpage = new BlockPage();
newpage.AddRange(page.AllBlocks.AsEnumerable());
this.Images = newpage;
LastResult = newpage;
return(newpage);
}
public void SetPageTables(IEnumerable <IBlock> tables)
{
var page = new BlockPage();
page.AddRange(tables);
if (HasTableOverlap(page))
{
PdfReaderException.AlwaysThrow("blocks already have overlapped elements");
}
_pageResult = page;
}
public BlockPage FindBlocksAtHeader(BlockPage page)
{
float err = 1f;
float maxH = page.AllBlocks.Max(b => b.GetH()) - err;
var blocksAtHeader = page.AllBlocks.Where(b => b.GetH() >= maxH);
var result = new BlockPage();
result.AddRange(blocksAtHeader);
return(result);
}
public BlockPage Validate(BlockPage page)
{
var result = new BlockPage();
if (this._tables == null)
{
PdfReaderException.AlwaysThrow("MergeTableText requires IdentifyTables");
}
var tables = MergeTables(page, _tables);
result.AddRange(tables);
return(result);
}
public BlockPage Process(BlockPage page)
{
var overlappedImages = FindInlineElements(page);
foreach (var image in overlappedImages.AllBlocks)
{
_parse.RemoveImage(image);
}
var blocks = page.AllBlocks.Except(overlappedImages.AllBlocks);
var result = new BlockPage();
result.AddRange(blocks);
return(result);
}
public BlockPage Process(BlockPage page)
{
if (page.AllBlocks.Count() == 0)
{
return(page);
}
float err = 1f;
float maxH = page.AllBlocks.Max(b => b.GetH()) - err;
var blocksAtHeader = page.AllBlocks.Where(b => b.GetH() < maxH);
var result = new BlockPage();
result.AddRange(blocksAtHeader);
return(result);
}
public BlockPage Process(BlockPage page2)
{
var page = page2 as BlockPage2;
if (page == null)
{
PdfReaderException.AlwaysThrow("ShowBlocksets must execute AFTER OrganizePageLayout");
}
var blocksets = new BlockPage();
foreach (var seg in page.Segments)
{
blocksets.AddRange(seg.Columns);
}
return(blocksets);
}
public BlockPage Process(BlockPage page)
{
var overlappedImages = FindInlineElements(page);
foreach (var image in overlappedImages.AllBlocks)
{
if (!(image is ImageBlock))
{
PdfReaderException.AlwaysThrow("RemoveOverlapedImages2 should be used only with images");
}
_parse.RemoveImage(image);
}
var blocks = page.AllBlocks.Except(overlappedImages.AllBlocks);
var result = new BlockPage();
result.AddRange(blocks);
return(result);
}
public BlockPage Validate(BlockPage page)
{
if (page.AllBlocks.Count() == 0)
{
return(page);
}
float err = 1f;
float minH = page.AllBlocks.Min(b => b.GetH()) + err;
var blocksAtFooter = page.AllBlocks.Where(b => b.GetH() <= minH);
var result = new BlockPage();
result.AddRange(blocksAtFooter);
if (!HasFooter(result))
{
var emptyResult = new BlockPage();
return(emptyResult);
}
return(result);
}
public BlockPage Process(BlockPage page)
{
var blocksets = page.AllBlocks.ToList();
if (blocksets.Count == 0)
{
return(page);
}
float x1 = page.AllBlocks.GetX();
float x2 = page.AllBlocks.GetX() + page.AllBlocks.GetWidth();
float dx = page.AllBlocks.GetWidth() + 2;
float h1 = page.AllBlocks.GetH();
float h2 = page.AllBlocks.GetH() + page.AllBlocks.GetHeight();
float dh = page.AllBlocks.GetHeight() + 2;
// Prepare the values order by X
int id = 0;
this.Values = page.AllBlocks.Select(b => new Data
{
ID = id++,
X = (int)(6.0 * ((b.GetX() - x1) / dx) + 0.5),
X2 = (int)(6.0 * ((b.GetX() + b.GetWidth() - x1) / dx) + 0.5),
Y = (int)(1000 * (b.GetH() - h1) / (dh)),
Y1 = (int)(1000 * (b.GetH() + b.GetHeight() - h1) / (dh)),
W = (int)(6.0 * (b.GetWidth() / dx) + 0.5),
B = b
})
.OrderBy(p => 10000 * p.X - p.Y)
.ToList();
VERTICAL_DIFFERENCE_INT = (int)(1000 * VERTICAL_DIFFERENCE / dh);
var checkInvalidW = Values.Where(v => v.X2 - v.X != v.W).ToList();
// sometimes W is miscalculated - need to investigate
// it is related to smaller size than the expected
// check ResizeBlocksets as well
if (checkInvalidW.Count > 0)
{
// warn the issue
PdfReaderException.Warning("checkInvalidW failed");
// workaround: recalculate W in terms of X and X2
checkInvalidW.Select(t => { var inv = Values.Where(t1 => t1.ID == t.ID).First(); inv.W = inv.X2 - inv.X; return(0); }).ToList();
checkInvalidW = Values.Where(v => v.X2 - v.X != v.W).ToList();
if (checkInvalidW.Count > 0)
{
PdfReaderException.Throw("checkInvalidW failed");
}
}
var checkOverW = Values.Where(v => v.W < 0 || v.W > 6).ToList();
if (checkOverW.Count > 0)
{
PdfReaderException.Warning("checkOverW failed");
Values = Values.Where(t => t.W >= 0 && t.W <= 6)
.OrderBy(p => 10000 * p.X - p.Y)
.ToList();
}
// re-implement in ResizeBlocksets (column)
//// if column is narrow (W=1 and X=2, then W <- 2)
//int fixCount = Values.Where(v => v.W == 1 && v.X == 2).Select(v => v.W = 2).Count();
var checkOddW = Values.Where(v => v.W == 1 || v.W == 5).ToList();
if (checkOddW.Count > 0)
{
PdfReaderException.Warning("checkOddW failed");
Values = Values.Where(t => t.W != 1 && t.W != 5)
.OrderBy(p => 10000 * p.X - p.Y)
.ToList();
}
// very weird bug: causes infinite loop!
var checkZeroW = Values.Where(v => v.W == 0).ToList();
if (checkZeroW.Count > 0)
{
// try to set to 2
checkZeroW.Where(t => t.X == 4).Select(t => { var inv = Values.Where(t1 => t1.ID == t.ID).First(); inv.W = 2; inv.X2 = 6; return(0); }).ToList();
Values = Values.OrderBy(p => 10000 * p.X - p.Y).ToList();
checkZeroW = Values.Where(v => v.W == 0).ToList();
if (checkZeroW.Count > 0)
{
PdfReaderException.Warning("checkZeroW failed");
Values = Values.Where(t => t.W != 0)
.OrderBy(p => 10000 * p.X - p.Y)
.ToList();
}
}
var checkOddX = Values.Where(v => v.X != 2 && v.X != 3 && v.X != 4 && v.X != 0).ToList();
if (checkOddX.Count > 0)
{
PdfReaderException.Warning("check X failed");
}
// Prepare the values order by Y
this.ValuesY = Values.OrderBy(p => - 100 * p.Y + p.X).ToList();
this.ValuesB = new bool[Values.Count];
OrderedBlocks = new List <IBlock>();
scan();
var result = new BlockPage();
//result.AddRange(Values.Select(p => (IBlock)p.B));
result.AddRange(OrderedBlocks);
return(result);
}
public BlockPage Process(BlockPage page)
{
var blocksets = page.AllBlocks.ToList();
if (blocksets.Count == 0)
{
return(page);
}
// implemented ONLY for 3 columns
if (blocksets.Count != 3)
{
return(page);
}
var columns = page.AllBlocks.OrderBy(b => b.GetX()).ToArray();
float maxColumn = page.AllBlocks.Max(b => b.GetWidth());
float x1 = page.AllBlocks.GetX();
float x2 = page.AllBlocks.GetX() + page.AllBlocks.GetWidth();
float dx = page.AllBlocks.GetWidth() + 2;
int id = 0;
var resizedColumns = columns.Select(b => new
{
ID = id++,
X = (int)(6.0 * ((b.GetX() - x1) / dx) + 0.5),
W = (int)(6.0 * (b.GetWidth() / dx) + 0.5),
B = b
})
.Select(d =>
{
// may receive multiples - confusing...
var original = (IEnumerable <IBlock>)d.B;
if ((original is TableSet) || (original is ImageBlock))
{
return(d.B);
}
int nextId = d.ID + 1;
if (d.ID >= 3) // only first and second
{
return(d.B);
}
if (d.W == 1) // small column
{
var block = d.B;
float new_x2 = columns[nextId].GetX() - COLUMN_DISTANCE;
float old_x2 = block.GetX() + block.GetWidth();
float diff = new_x2 - old_x2;
if (diff < 0)
{
PdfReaderException.Warning("decreasing the column size");
}
var replace = new BlockSet2 <IBlock>(original, block.GetX(), block.GetH(), new_x2, block.GetH() + block.GetHeight());
return(replace);
}
return(d.B);
}).ToArray();
var newpage = new BlockPage();
newpage.AddRange(resizedColumns);
return(newpage);
}
public BlockPage ProcessTable(BlockPage page)
{
// try to improve processing time
var cellList = page.AllBlocks.Where(b => TableCell.HasDarkColor((TableCell)b)).ToList();
var blockArray = new TableSet[cellList.Count];
bool hasModification = true;
while (hasModification)
{
hasModification = false;
// iterate every line found
for (int i = 0; i < cellList.Count; i++)
{
var c = cellList[i];
if (blockArray[i] == null)
{
// create a fresh blockset
blockArray[i] = new TableSet();
// add the current element to the blockset
blockArray[i].Add(c);
}
var currentBlockset = blockArray[i];
// assume that currentBlockset ALWAYS contains c
// -- it was added during blockArray assignment
// look for connected lines
for (int j = i + 1; j < cellList.Count; j++)
{
// skip if it already has block array assigned
if (blockArray[j] == currentBlockset)
{
continue;
}
var last = cellList[j];
// check if blockSet contains c (two rectangles)
float b_x1 = last.GetX();
float b_x2 = last.GetX() + last.GetWidth();
float b_y1 = last.GetH();
float b_y2 = last.GetH() + last.GetHeight();
var blockSet = currentBlockset;
bool b1 = HasOverlap(blockSet, b_x1, b_y1);
bool b2 = HasOverlap(blockSet, b_x1, b_y2);
bool b3 = HasOverlap(blockSet, b_x2, b_y2);
bool b4 = HasOverlap(blockSet, b_x2, b_y1);
bool hasOverlap = b1 || b2 || b3 || b4;
// for some reason, hasOverlap is not 100% guarantee to work
if (blockArray[j] != null)
{
if (currentBlockset == null)
{
PdfReaderException.AlwaysThrow("currentBlockset == null");
}
bool bb = Block.HasOverlap(blockArray[j], currentBlockset);
if ((!hasOverlap) && bb)
{
hasOverlap = true;
}
}
// FOUND A CONNECTED LINE!
if (hasOverlap)
{
hasModification = true;
var nextBlockset = blockArray[j];
if (nextBlockset == null)
{
if (nextBlockset == currentBlockset)
{
PdfReaderException.AlwaysThrow("infinite loop?");
}
// assign the blockarray
blockArray[j] = currentBlockset;
// and add the element
blockArray[j].Add(last);
}
else
{
// has to merge changes
currentBlockset.MergeWith(nextBlockset);
// assign the blockarray
blockArray[j] = currentBlockset;
// assume nextBlockset already contains j
// remove all other references to nextBlockset
for (int k = 0; k < blockArray.Length; k++)
{
if (blockArray[k] == nextBlockset)
{
blockArray[k] = currentBlockset;
}
}
}
}
else
{
// do nothing
}
}
}
// infinite loop?
}
// transform blockArray into blockList
var blockList = blockArray.Distinct().ToList();
int count1 = blockArray.Length;
int count2 = blockList.Count;
var tables = new BlockPage();
var lines = new BlockPage();
var background = new BlockPage();
foreach (var b in blockList)
{
// does not add line segments
if ((b.Count() == 1) || (b.GetWidth() < MAXIMUM_LIZE_WIDTH) || (b.GetHeight() < MAXIMUM_LIZE_WIDTH))
{
lines.Add(b);
}
else
{
tables.Add(b);
}
}
// add background
var dark = page.AllBlocks
.Where(b => !TableCell.HasDarkColor((TableCell)b))
.Where(b => b.GetWidth() > MINIMUM_BACKGROUND_SIZE && b.GetHeight() > MINIMUM_BACKGROUND_SIZE)
.Select(b => new TableSet()
{
b
});
background.AddRange(dark);
this._pageResult = tables;
this._pageLines = lines;
this._pageBackground = background;
var result = new BlockPage();
result.AddRange(tables.AllBlocks);
result.AddRange(lines.AllBlocks);
return(result);
}
public BlockPage BreakPage(BlockPage page)
{
var blocks = page.AllBlocks.ToList();
var result = new BlockPage();
var splitted = blocks.Select(b => SplitBlock((BlockSet <IBlock>)b)).ToList();
for (int i = 0; i < blocks.Count; i++)
{
for (int j = i + 1; j < blocks.Count; j++)
{
if (blocks[i] == null)
{
continue;
}
if (blocks[j] == null)
{
continue;
}
if (Block.HasOverlap(blocks[i], blocks[j]))
{
// precheck: contained block?
bool blockContainsA = BlockContains(blocks[i], blocks[j]);
bool blockContainsB = BlockContains(blocks[j], blocks[i]);
if (blockContainsA || blockContainsB)
{
}
int k = SelectBlock(splitted, blocks, i, j);
bool breakInTheMiddle = false;
if ((k == -1) && (blockContainsA || blockContainsB))
{
k = (blockContainsA) ? i : k;
k = (blockContainsB) ? j : k;
breakInTheMiddle = true;
}
if (k == -1)
{
// the blocks can merge?
float wdiff = Math.Abs(blocks[i].GetWidth() - blocks[j].GetWidth());
float xdiff = Math.Abs(blocks[i].GetX() - blocks[j].GetX());
// ignore?
if (wdiff < 10f && xdiff < 10f)
{
continue;
}
// breakcolumns have a poor performance when
// tables and images get removed.
// we could retry after adding them back to the doc
// so far it is not supported yet
// very likely to have A contains B in Y axis, but not in X
// in this case, we need to break both blocks at the same operation
PdfReaderException.AlwaysThrow("true overlap?");
// throw new NotImplementedException("merge blockLines");
// cannot break the blocks ?!?!?!?!
//throw new InvalidOperationException("should be handled previously in precheck");
//continue;
}
var selected_block = blocks[k];
var selected_block_split = splitted[k];
IBlock otherBlock = (selected_block == blocks[i]) ? blocks[j] : blocks[i];
float middle = otherBlock.GetH() + otherBlock.GetHeight() / 2;
int size = -1;
if (breakInTheMiddle)
{
size = SelectSize(selected_block, middle);
}
else
{
size = SelectSize(blocks[i], blocks[j], selected_block_split);
if (size == -1)
{
size = SelectSize(selected_block, middle);
}
}
if (size == 0)
{
PdfReaderException.AlwaysThrow("size == 0");
}
if (size == -1)
{
PdfReaderException.AlwaysThrow("size == -1");
}
if (size == ((BlockSet <IBlock>)selected_block).Count())
{
PdfReaderException.AlwaysThrow("size > total_blocks");
}
var newblocks = CreateNewBlocks((BlockSet <IBlock>)selected_block, size);
if (breakInTheMiddle)
{
// Check if newblocks has collision
bool checkOverlap = CheckOverlapCrossIntersection(newblocks, otherBlock);
if (checkOverlap)
{
PdfReaderException.AlwaysThrow("checkOverlap");
}
}
// replace
blocks[k] = null;
blocks.Add(newblocks[0]);
blocks.Add(newblocks[1]);
splitted[k] = null;
splitted.Add(SplitBlock(newblocks[0]));
splitted.Add(SplitBlock(newblocks[1]));
}
}
}
result.AddRange(blocks.Where(b => b != null));
return(result);
}
public BlockPage BreakPage(BlockPage page)
{
var blocks = page.AllBlocks.ToList();
var result = new BlockPage();
var splitted = blocks.Select(b => SplitBlock((BlockSet <IBlock>)b)).ToList();
for (int i = 0; i < blocks.Count; i++)
{
for (int j = i + 1; j < blocks.Count; j++)
{
if (blocks[i] == null)
{
continue;
}
if (blocks[j] == null)
{
continue;
}
if (Block.HasOverlap(blocks[i], blocks[j]))
{
// precheck: contained block?
bool blockContainsA = BlockContains(blocks[i], blocks[j]);
bool blockContainsB = BlockContains(blocks[j], blocks[i]);
if (blockContainsA || blockContainsB)
{
}
int k = SelectBlock(splitted, blocks, i, j);
bool breakInTheMiddle = false;
if ((k == -1) && (blockContainsA || blockContainsB))
{
k = (blockContainsA) ? i : k;
k = (blockContainsB) ? j : k;
breakInTheMiddle = true;
}
if (k == -1)
{
PdfReaderException.Warning("BreakColumnsLight:k == -1");
continue;
//// the blocks can merge?
//float wdiff = Math.Abs(blocks[i].GetWidth() - blocks[j].GetWidth());
//float xdiff = Math.Abs( blocks[i].GetX() - blocks[j].GetX() );
//// ignore?
//if (wdiff < 10f && xdiff < 10f)
// continue;
//throw new NotImplementedException("merge blockLines");
//// cannot break the blocks ?!?!?!?!
//throw new InvalidOperationException("should be handled previously in precheck");
////continue;
}
var selected_block = blocks[k];
var selected_block_split = splitted[k];
IBlock otherBlock = (selected_block == blocks[i]) ? blocks[j] : blocks[i];
float middle = otherBlock.GetH() + otherBlock.GetHeight() / 2;
int size = -1;
if (breakInTheMiddle)
{
size = SelectSize(selected_block, middle);
}
else
{
size = SelectSize(blocks[i], blocks[j], selected_block_split);
if (size == -1)
{
size = SelectSize(selected_block, middle);
}
}
if (size == 0)
{
PdfReaderException.AlwaysThrow("size == 0");
}
if (size == -1)
{
PdfReaderException.AlwaysThrow("size == -1");
}
if (size == ((BlockSet <IBlock>)selected_block).Count())
{
PdfReaderException.AlwaysThrow("size > total_blocks");
}
var newblocks = CreateNewBlocks((BlockSet <IBlock>)selected_block, size);
if (breakInTheMiddle)
{
// Check if newblocks has collision
bool checkOverlap = CheckOverlapCrossIntersection(newblocks, otherBlock);
if (checkOverlap)
{
PdfReaderException.Warning("BreakColumnsLight:checkOverlap");
}
}
// replace
blocks[k] = null;
blocks.Add(newblocks[0]);
blocks.Add(newblocks[1]);
splitted[k] = null;
splitted.Add(SplitBlock(newblocks[0]));
splitted.Add(SplitBlock(newblocks[1]));
}
}
}
result.AddRange(blocks.Where(b => b != null));
return(result);
}
public BlockPage BreakPage(BlockPage page)
{
var blocks = page.AllBlocks.ToList();
var result = new BlockPage();
for (int i = 0; i < blocks.Count; i++)
{
var current = blocks[i] as BlockSet <IBlock>;
if (current == null)
{
continue;
}
for (int j = 0; j < blocks.Count; j++)
{
if (i == j)
{
continue;
}
if (blocks[j] == null)
{
continue;
}
if (blocks[i] == null)
{
break;
}
if (Block.HasOverlap(blocks[i], blocks[j]))
{
float otherH_bottom = blocks[j].GetH();
float otherH_top = blocks[j].GetH() + blocks[j].GetHeight();
if (otherH_bottom > otherH_top)
{
PdfReaderException.AlwaysThrow("negative height");
}
var blockList = current.ToList();
int idxTop = FindTop(blockList, otherH_top);
int idxBottom = FindBottom(blockList, otherH_bottom);
var topBlock = RewriteBlockTop(blockList, idxTop);
var bodyBlock = RewriteBlockBody(blockList, idxBottom, idxTop);
var bottomBlock = RewriteBlockBottom(blockList, idxBottom);
if (topBlock != null || bottomBlock != null)
{
int total = 0;
if (topBlock != null)
{
blocks.Add(CreateNewBlock(result, topBlock));
total += topBlock.Count;
}
if (bodyBlock != null)
{
blocks.Add(CreateNewBlock(result, bodyBlock));
total += bodyBlock.Count;
}
if (bottomBlock != null)
{
blocks.Add(CreateNewBlock(result, bottomBlock));
total += bottomBlock.Count;
}
if (total != blockList.Count)
{
PdfReaderException.AlwaysThrow("incorrect number of blocks");
}
// replace the blocks
blocks[i] = null;
}
else
{
}
// replace
//blocks.Add(newblocks[0]);
//blocks.Add(newblocks[1]);
}
}
}
result.AddRange(blocks.Where(b => b != null));
return(result);
}
public BlockPage Process(BlockPage page)
{
float error_othercolumn = 2f;
var blocksets = page.AllBlocks.ToList();
if (blocksets.Count == 0)
{
return(page);
}
float x1 = page.AllBlocks.GetX();
float x2 = page.AllBlocks.GetX() + page.AllBlocks.GetWidth();
float dx = page.AllBlocks.GetWidth() + 2;
float h1 = page.AllBlocks.GetH();
float h2 = page.AllBlocks.GetH() + page.AllBlocks.GetHeight();
float dh = page.AllBlocks.GetHeight() + 2;
float pageSize = page.AllBlocks.Max(b => b.GetX() + b.GetWidth());
// Prepare the values order by X
int id = 0;
var values = page.AllBlocks.Select(b => new Data
{
ID = id++,
X = (int)(6.0 * ((b.GetX() - x1) / dx) + 0.5),
X2 = (int)(6.0 * ((b.GetX() + b.GetWidth() - x1) / dx) + 0.5),
Y = (int)(1000 * (b.GetH() - h1) / (dh)),
Y1 = (int)(1000 * (b.GetH() + b.GetHeight() - h1) / (dh)),
W = (int)(6.0 * (b.GetWidth() / dx) + 0.5),
RW = b.GetWidth(),
B = b
})
.OrderByDescending(p => p.W)
.ToList();
var columnW = (from v in values
group v by v.W into g
select new { g.Key, size = g.Max(ta => ta.RW) }).ToDictionary(t => t.Key);
foreach (var blsearch in values)
{
if (blsearch.B is TableSet)
{
continue;
}
if (blsearch.B is ImageBlock)
{
continue;
}
// we could have used predefined blocks (w=6, w=3, etc)
var predefinedBlocks = values;
var over = predefinedBlocks
.Where(v => v != blsearch && v.X <= blsearch.X && v.X2 >= blsearch.X2)
.Where(v => v.RW > blsearch.RW)
.Where(v => Math.Abs(v.RW - blsearch.RW) > error_othercolumn)
.Select(v => v.B)
.ToList();
var curblocks = values.Select(v => v.B).ToList();
List <IBlock> repls = new List <IBlock>();
foreach (var bl in over)
{
var compareBlocks = curblocks.Except(new IBlock[] { bl, blsearch.B });
var block = new Block()
{
X = bl.GetX(),
Width = bl.GetWidth(),
H = blsearch.B.GetH(),
Height = blsearch.B.GetHeight()
};
// ensure it will increase
float diff = block.GetWidth() - blsearch.B.GetWidth();
if (diff < 0)
{
PdfReaderException.AlwaysThrow("should never decrease the block size");
}
if (CheckBoundary(compareBlocks, block))
{
// may receive multiples - confusing...
var original = (IEnumerable <IBlock>)blsearch.B;
if ((original is TableSet) || (original is ImageBlock))
{
PdfReaderException.AlwaysThrow("Block should not be resized");
}
var replace = new BlockSet2 <IBlock>(original, block.GetX(), block.GetH(), block.GetX() + block.GetWidth(), block.GetH() + block.GetHeight());
bool isStillContained = Block.Contains(replace, blsearch.B);
if (!isStillContained)
{
bool hasOverlap = Block.HasOverlap(replace, blsearch.B);
// TODO: review this issue
if (!hasOverlap)
{
PdfReaderException.Warning("Block was moved to another place -- ignore");
continue;
}
}
repls.Add(replace);
}
}
if (repls.Count > 0)
{
// this is important because repls.Count can be > 1
if (repls.Count > 1)
{
// add a breakpoint to monitor if needed
}
var largest_replace = repls.OrderByDescending(t => t.GetWidth()).First();
blsearch.B = largest_replace;
}
}
var result = new BlockPage();
result.AddRange(values.Select(p => (IBlock)p.B));
//result.AddRange(OrderedBlocks);
return(result);
}