Extracting tabular data is a typical use case for importing "digitally born" PDF documents such as invoices. Upon reading an invoice, the recognized data will be matched with the corresponding purchase order data in an ERP system.

Import PDF Documents

Like any other supported file type, TX Text Control can import digitally born Adobe PDF documents. Typically, you can use this approach to search PDF documents for strings in document pages and extract the contents of form fields.

Learn More

This article shows different ways to extract data from existing PDF documents.

Extract Text and Data from PDF Documents in C#

Unlike in the above article, the document is actually loaded into the TX Text Control to loop through recognized tables. Consider the following invoice PDF document.

Extracting Tables from PDFs

In this case, the TX Text Control will recognize the table that is highlighted in red.

Extracting Tables from PDFs

Looping through Tables

We use the PDFImportSettings ╰ TX Text Control .NET Server for ASP.NET
╰ TXTextControl Namespace
╰ TXTextControl Enumerations Enumerations
╰ PDFImportSettings Enumeration Enumeration
Specifies how the document structure is generated when a PDF document is imported. to specify that the TX Text Control should import the content and format it in such a way that tables are recognized as well. The following code loads the PDF document and loops through all of the tables it finds.

// create a new load settings object
TXTextControl.LoadSettings ls = new TXTextControl.LoadSettings() {
// set the PDF import settings
PDFImportSettings = PDFImportSettings.GenerateTextFrames
};
// load the PDF document
textControl1.Load("pdf_template.pdf", TXTextControl.StreamType.AdobePDF, ls);
// iterate through all text parts
foreach (IFormattedText formattedText in textControl1.TextParts) {
// iterate through all tables
foreach (Table table in formattedText.Tables) {
Debug.WriteLine(Table2Data(table, true));
}
}
view raw test.cs hosted with ❤ by GitHub

Generate JSON

The Table2Data method loops through all rows and creates a Dictionary for each row. If the containsTableHeader parameter is set to true, the column name is used from the first row. When finished, the list of dictionaries is returned as a JSON string.

// convert a flat (non-nested) table to a JSON string
private string Table2Data(Table table, bool containsTableHeader) {
// create a list of dictionaries
var tableList = new List<Dictionary<string, string>>();
// get the start row
int startRow = containsTableHeader ? 2 : 1;
// iterate through all rows
for (int row = startRow; row < table.Rows.Count; row++) {
// create a dictionary for each row
var rowData = new Dictionary<string, string>();
// iterate through all columns
for (int col = 1; col < table.Columns.Count; col++) {
// get the column title from the first row
var columnTitle = containsTableHeader ?
table.Cells[1, col].Text.Trim('\t') : $"Column {col}";
// add the cell text to the dictionary
rowData[columnTitle] = table.Cells[row, col].Text.Trim('\t');
}
// add the dictionary to the list
tableList.Add(rowData);
}
// return the JSON string
return System.Text.Json.JsonSerializer.Serialize(tableList, new JsonSerializerOptions { WriteIndented = true });
}
view raw test.cs hosted with ❤ by GitHub

For the example document above, the following JSON is returned.

[
{
"ID": "1",
"Name": "Name 1",
"Description": "Description 1",
"Price": "$100",
"Qty": "1"
},
{
"ID": "2",
"Name": "Name 2",
"Description": "Description 2",
"Price": "$200",
"Qty": "1"
}
]
view raw test.json hosted with ❤ by GitHub

If the containsTableHeader parameter is set to false, the following JSON will be returned.

[
{
"Column 1": "ID",
"Column 2": "Name",
"Column 3": "Description",
"Column 4": "Price",
"Column 5": "Qty"
},
{
"Column 1": "1",
"Column 2": "Name 1",
"Column 3": "Description 1",
"Column 4": "$100",
"Column 5": "1"
},
{
"Column 1": "2",
"Column 2": "Name 2",
"Column 3": "Description 2",
"Column 4": "$200",
"Column 5": "1"
}
]
view raw test.json hosted with ❤ by GitHub