Introduction
Originally, I wrote a C++ parser which was used to parse given MS Word documents and put them into some form of a structure that was more useful for data processing. After I wrote the parser, I started working with .NET and C# to re-create the parser. In the process, I also wrote my first article for Code Project, Automating MS Word Using Visual Studio .NET. Several people have requested to see the C++ version of the application, hence, I finally got some time to put something together. I have written this article with the intention of making it easier for someone who is looking for quick answers. I hope that people can benefit from the information provided and help them get started faster.
Background
No special background is necessary. Just have some hands on experience with C++.
Using the code
I think the best way to present the code would be to first give you the critical sections which you need to get an instance of MS Word, and then give you snapshots of code that perform specific functions. I believe this way will help you get started faster in developing your own programs.
The following block is the header portion of the CPP file.
Note: The most important include files are
// Vahe Karamian - 04-20-2004 - For Code Project
//---------------------------------------------------------------------------
#include
// We need this for the OLE object
#include
#pragma package(smart_init) #pragma resource "*.dfm" TForm1 *Form1; The following block creates MS Word COM Object. This is the object which will be used to access MS Word application functions. To see what functions are available, you can do within MS Word. Refer to the first article, Automating MS Word Using Visual Studio .NET.
As before, you can either make a Windows Forms Application or a Command Line application, the process is the same. The code below is based on a Windows Forms application, that has a button to start the process. When the user clicks the button, the Button1Click(TObject *Sender) event will be called and the code executed.
Note: To better understand the code, ignore everything in the code except the portions that are in bold.
TForm1 *Form1; //--------------------------------------------------------------------------- __fastcall TForm1::TForm1(TComponent* Owner)
: TForm(Owner) { } //---------------------------------------------------------------------------
void __fastcall TForm1::Button1Click(TObject *Sender) {
.
.
.
// used for the file name
OleVariant fileName;
fileName = openDialog->FileName;
Variant my_word;
Variant my_docs;
// create word object
my_word = Variant::CreateObject( "word.application" );
// make word visible, to make invisible put false
my_word.OlePropertySet( "Visible", (Variant) true );
// get document object
my_docs = my_word.OlePropertyGet( "documents" );
Variant wordActiveDocument = my_docs.OleFunction( "open", fileName );
.
.
. So a brief explanation, we define a OleVariant data type called fileName, we assign a file path to our fileName variable. In the code above, this is done using a OpenDialog object. Of course, you can just assign a whole path for testing if you like, i.e., c:\test.doc.
Next, we define two Variant data types called my_word, and my_docs. my_word will be used to create a word.application object and my_docs will be used to create a documents object.
Next, we define another Variant data type called myActiveDocument. Using this referenced object, we can now do what we want! In this case, we are going to open the given MS Word document.
Notice that most of the variables are of type Variant.
At this point, we have a Word document that we can start performing functions on. At first, it might take a while for you to see how it works, but once you get a hang of it, anything in MS Word domain is possible.
Let's take a look at the following code, it is going to be dealing with tables within a MS Word document.
.
.
Variant wordTables = wordActiveDocument.OlePropertyGet( "Tables" );
long table_count = wordTables.OlePropertyGet( "count" );
.
. As I mentioned before, all your data types are going to be of Variant. So we declare a Variant data type called wordTables to represent Tables object in our Document object.
Variant wordTables = wordActiveDocument.OlePropertyGet( "Tables" ); The line above will return all Table objects that are within our active Document object. Since Tables is a property of a Document object, we have to use the OlePropertyGet( "Tables" ); to get the value.
long table_count = wordTables.OlePropertyGet( "count" ); The line above will return the number of tables in out Tables object. This is done by calling the OlePropertyGet( "count" ); to return us the value.
You might be wondering where do I get this information from? The answer to that question is in the first article: Automating MS Word Using Visual Studio .NET.
The next block of code will demonstrate how to extract content from the Tables object.
. . . int t, r, c;
try {
for( t=1; t<=table_count; t++ )
{
Variant wordTable1 = wordTables.OleFunction( "Item", (Variant) t );
Variant tableRows = wordTable1.OlePropertyGet( "Rows" );
Variant tableCols = wordTable1.OlePropertyGet( "Columns" );
long row_count, col_count;
row_count = tableRows.OlePropertyGet( "count" );
col_count = tableCols.OlePropertyGet( "count" );
// LET'S GET THE CONTENT FROM THE TABLES
// THIS IS GOING TO BE FUN!!!
for( r=1; r<=row_count; r++ )
{
Variant tableRow = tableRows.OleFunction( "Item", (Variant) r );
tableRow.OleProcedure( "Select" );
Variant rowSelection = my_word.OlePropertyGet( "Selection" );
Variant rowColumns = rowSelection.OlePropertyGet( "Columns" );
Variant selectionRows = rowSelection.OlePropertyGet( "Rows" );
long rowColumn = rowColumns.OlePropertyGet( "count" );
for( c=1; c<=rowColumn; c++ ) //col_count; c++ )
{
Variant rowCells = tableRow.OlePropertyGet( "cells" );
Variant wordCell = wordTable1.OleFunction( "Cell",
(Variant) r, (Variant) c );
Variant cellRange = wordCell.OlePropertyGet( "Range" );
Variant rangeWords = cellRange.OlePropertyGet( "Words" );
long words_count = rangeWords.OlePropertyGet( "count" );
AnsiString test = '"';
for( int v=1; v<=words_count; v++ )
{
test = test + rangeWords.OleFunction( "Item",
(Variant) v ) + " ";
}
test = test + '"';
}
}
}
my_word.OleFunction( "Quit" ); } catch( Exception &e ) {
ShowMessage( e.Message + "nType: " + __ThrowExceptionName() +
"nFile: "+ __ThrowFileName() +
"nLine: " + AnsiString(__ThrowLineNumber()) ); } . . . Okay, so above we have the code that actually will go through all of the tables in the Document object and extract the content from them. So we have tables, and tables have rows and columns. To go through all of the Tables object in a document, we do a count and get the number of tables within a document.
So we have three nested for loops. The first one is used for the actual Table object, and the 2nd and 3rd are used for the rows and columns of the current Table object. We create three new Variant data types called wordTable1, tableRows, and tableCols.
Note: Notice that wordTable1 comes from the wordTables object. We get out table by calling wordTables.OleFunction( "Item", (Variant) t );. This returns us a unique Table object from the Tables object.
Next, we get the Rows and Columns object of the given Table object. And this is done by calling OlePropertyGet( "Rows" ); and OlePropertyGet( "Columns" ); of the wordTable1 object!
Next, we get a count of rows and columns in the given Rows and Columns objects which belong to the wordTable1 object. We are ready to step through them and get the content.
Now, we will have to define four new Variant data types called tableRow, rowSelection, rowColumsn, and selectionRows. Now, we can start going from column to column in the selected row to get the content.
In the most inner for loop, the final one, we again define four new Variant data types called rowCells, wordCell, cellRange, and rangeWords. Yes, it is tedious, but we have to do it.
Let's sum what we did so far:
We got a collection of Tables object within the current Document object. We got a collection of Rows and Columns in the current Table object. We went through each row and got the number of columns it has. We get the column and the cells, and step through the cells to get to the content of the table. Note: Yes, some steps are repeated, but the reason behind it is because not all tables in a given document are uniform! I.e., it does not necessarily mean that if row 1 has 3 columns, then row 2 must have 3 columns as well. More than likely, it will have different number of columns. You can thank the document authors/owners.
So then the final step will just step through the cells and get the content and concatenate it for a single string output.
And finally, we want to quit Word and close all documents.
...
my_word.OleFunction( "Quit" );
... That is pretty much it. The code does sometimes get pretty tedious and messy. The best way to approach automating/using Word is by first knowing what it is that you exactly want to do. Once you know what you want to achieve, then you will need to find out what objects or properties you need to use to perform what you want. That's the tricky part, you will have to read the documentation: Automating MS Word Using Visual Studio .NET.
In the next code block, I will show you how to open an existing document, create a new document, select content from the existing document and paste it in the new document using Paste Special function, then do clean up, i.e., Find and Replace function.
Before you look at the block of code, the following list will identify which variable is used to identify what object and the function that can be applied to them.
Variables and representations: vk_filename: existing document name vk_converted_filename: new document name vk_this_doc: existing document object vk_converted_document: new document object vk_this_doc_select: existing document selected object vk_this_doc_selection: existing document selection vk_converted_document_select: new document selected object vk_converted_document_selection: new document selection wordSelectionFind: Find and Replace object // Get the filename from the list of files in the OpenDialog vk_filename = openDialog->Files->Strings[i]; vk_converted_filename = openDialog->Files->Strings[i] + "_c.doc";
// Open the given Word file vk_this_doc = vk_word_doc.OleFunction( "Open", vk_filename );
statusBar->Panels->Items[2]->Text = "READING";
// ------------------------------------------------------------------- // Vahe Karamian - 10-10-2003 // This portion of the code will convert the word document into // unformatted text, and do extensive clean up statusBar->Panels->Items[0]->Text = "Converting to text..."; vk_timerTimer( Sender );
// Create a new document Variant vk_converted_document = vk_word_doc.OleFunction( "Add" );
// Select text from the original document Variant vk_this_doc_select = vk_this_doc.OleFunction( "Select" ); Variant vk_this_doc_selection = vk_word_app.OlePropertyGet( "Selection" );
// Copy the selected text vk_this_doc_selection.OleFunction( "Copy" );
// Paste selected text into the new document Variant vk_converted_document_select =
vk_converted_document.OleFunction( "Select" ); Variant vk_converted_document_selection =
vk_word_app.OlePropertyGet( "Selection" ); vk_converted_document_selection.OleFunction( "PasteSpecial",
0, false, 0, false, 2 );
// Re-Select the text in the new document vk_converted_document_select =
vk_converted_document.OleFunction( "Select" ); vk_converted_document_selection =
vk_word_app.OlePropertyGet( "Selection" );
// Close the original document vk_this_doc.OleProcedure( "Close" );
// Let's do out clean-up here ... Variant wordSelectionFind =
vk_converted_document_selection.OlePropertyGet( "Find" );
statusBar->Panels->Items[0]->Text = "Find & Replace..."; vk_timerTimer( Sender );
wordSelectionFind.OleFunction( "Execute", "^l",
false, false, false, false, false, true, 1, false,
" ", 2, false, false, false, false ); wordSelectionFind.OleFunction( "Execute", "^p", false,
false, false, false, false, true, 1, false,
" ", 2, false, false, false, false );
// Save the new document vk_converted_document.OleFunction( "SaveAs", vk_converted_filename );
// Close the new document vk_converted_document.OleProcedure( "Close" ); // ------------------------------------------------------------------- So what we are doing in the code above, we are opening an existing document with vk_this_doc = vk_word_doc.OleFunction( "Open", vk_filename );. Next we add a new document with Variant vk_converted_document = vk_word_doc.OleFunction( "Add" );. Then we want to select the content from the existing document and paste them in our new document. This portion is done by Variant vk_this_doc_select = vk_this_doc.OleFunction( "Select" ); to get a select object and Variant vk_this_doc_selection = vk_word_app.OlePropertyGet( "Selection" ); to get a reference to the actual selection. Then we have to copy the selection using vk_this_doc_selection.OleFunction( "Copy" );. Next, we perform the same task for the new document with Variant vk_converted_document_select = vk_converted_document.OleFunction( "Select" ); and Variant vk_converted_document_selection = vk_word_app.OlePropertyGet( "Selection" );. At this time, we have a selection object for the existing document and the new document. Now, we are going to be using them both to do our special paste using vk_converted_document_selection.OleFunction( "PasteSpecial", 0, false, 0, false, 2 );. Now, we have our original content pasted in a special format in the newly created document. We have to do a new select call in the new document before we do our find and replace. To do so, we simply use the same calls vk_converted_document_select = vk_converted_document.OleFunction( "Select" ); and vk_converted_document_selection = vk_word_app.OlePropertyGet( "Selection" );. Next, we create a Find object with Variant wordSelectionFind = vk_converted_document_selection.OlePropertyGet( "Find" ); and finally, we can use our find object to perform our find and replace with wordSelectionFind.OleFunction( "Execute", "^l", false, false, false, false, false, true, 1, false, " ", 2, false, false, false, false );.
That's all there is to it!
Points of Interest
Putting structure to a Word document is a challenging task, given that many people have different ways of authoring documents. Nevertheless, it would help for organizations to start modeling their documents. This will allow them to apply XML schema to their documents and make extracting content from them much easier. This is a challenging task for most companies; usually, either they are lacking the expertise or the resources. And such projects are huge in scale due to the fact that they will affect more than one functional business area. But on the long run, it will be beneficial to the organization as a whole. The fact that your documents are driven by structured data and not by formatting and lose documents has a lot of value added to your business.