Fill pdf form with javascript (client-side only)

I need to fill a pdf form automatically in my angularjs webapp. The pdf form is generated outside the app so I can configure it as I want. In my app, I just need to load the pdf, modify the form fields and flatten the file so it doesn't look like a form anymore. Do you know any way to do it? Edit: I've found iText but it's a java library which won't work for my project (the app runs on tablets so I'm looking for something 100% HTML5)

4 Answers 4

There is now another solution that worked for me.
It's this JS package: https://github.com/phihag/pdfform.js
Also available on NPM: https://www.npmjs.com/package/pdfform.js

If you have troubles opening your PDF files, the best bet is to use the mozilla's PDF.js library which works with every PDF files I tried.
The only drawback as of today is that it's not working as expected with webpack. You have to import the JS files with a script tag.

I've found a solution. not perfect but it should fit most requirements. It doesn't use any server (perfect for privacy requirements) or library! First of all, the PDF must be version 1.5 (Acrobat 6.0 or later). The original pdf can be another version but when you create the fields you must save it as compatible for Acrobat 6.0 or later. If you want to make sure the format is right, you can check there

So let's say I have 'myform.pdf' file (with no form fields); I open it with Acrobat Pro (I have Acrobat Pro 11 but it should work with other versions). I add fields and pre-fill the field's value (not the field name!) with a 'code' (unique text string). This code will be find/replace with the string you want by the javascript function below so let say '%address%' (you can add multiple fields but use a different code to differenciate fields). If you want to have a flatten look of the field, set the field to read-only. To save it, go to File -> Save as other. -> Optimized PDF and choose "Acrobat 6.0 and later" under Make compatible with (top right of the pop-up).

Once you have your file saved, you can check that the format is right by opening it in a text editor and looking for your codes (in my case '%address%'). Count number of occurences, it should appear three times.

The following function does three things: - Changes the fields content - Recalculate the content's length - Fix the cross reference tables

So now the function (look at the end for the final pdf blob):

@param certificate: your pdf form (format of this variable must be compatible with FileReader)

// Works only for PDF 1.5 (Acrobat 6.0 and later) var fillCertificate = function (certificate, changes) < // replace a a substring at a specific position String.prototype.replaceBetween = function(start, end, what) < return this.substring(0, start) + what + this.substring(end); >; // format number with zeros at the beginning (n is the number and length is the total length) var addLeadingZeros = function (n, length) < var str = (n >0 ? n : -n) + ""; var zeros = ""; for (var i = length - str.length; i > 0; i--) zeros += "0"; zeros += str; return n >= 0 ? zeros : "-" + zeros; > // Create the reader first and read the file (call after the onload method) var reader = new FileReader(); // To change the content of a field, three things must be done; - change the text of the field, - change the length of the content field, - change the cross table reference reader.onload = function(aEvent) < var string = aEvent.target.result; // Let's first change the content and the content's length var arrayDiff = []; var char; for(var foo = 0; foo < changes.length; foo++) < // Divide the string into a table of character for finding indices char = new Array(string.length); for (var int = 0; int < string.length; int++) < char[int] = string.charAt(int); >// Let's find the content's field to change and change it everywhere var find = changes[foo].find; var replace = changes[foo].replace; var lengthDiff = replace.length - find.length; var search = new RegExp(find, "g"); var match; var lastElements = []; var int = 0; var objectLenPos; var objectLenEnd; // Each time you change the content, compute the offset difference (number of characters). We'll add it later for the cross tables while (match = search.exec(string)) < arrayDiff.push(); lastElements.push(); // Find length object if(int == 0) < var length = 0; var index; while(char[match.index - length] != '\r')< index = match.index - length; length++; >objectLenPos = index + 10; length = 0; while(char[objectLenPos + length] != ' ') < length++; objectLenEnd = objectLenPos + length; >> int++; > var lengthObject = string.slice(objectLenPos, objectLenEnd) + ' 0 obj'; var objectPositionStart = string.search(new RegExp('\\D' + lengthObject, 'g')) + lengthObject.toString().length + 2; var length = 0; var objectPositionEnd; while(char[objectPositionStart + length] != '\r') < length++; objectPositionEnd = objectPositionStart + length; >// Change the length of the content's field var lengthString = new RegExp('Length ', "g"); var fieldLength; var newLength; string = string.replace(lengthString, function (match, int) < // The length is between the two positions calculated above if (int >objectPositionStart && int < objectPositionEnd) < var length = 0; var end; while (char[int + 7 + length] != '/') < length++; end = int + 7 + length; >fieldLength = string.slice(end - length, end); newLength = parseInt(fieldLength) + lengthDiff; if (fieldLength.length != newLength.toString().length) < arrayDiff.push(); > // Let's modify the length so it's easy to find and replace what interests us; the length number itself return "Length%"; > return match; >); // Replace the length with the new one based on the length difference string = string.replace('Length%' + fieldLength, 'Length ' + (newLength).toString()); string = string.replace(new RegExp(find, 'g'), replace); > // FIND xref and repair cross tables // Rebuild the table of character var char = new Array(string.length); for (var int = 0; int < string.length; int++) < char[int] = string.charAt(int); >; // Find XRefStm (cross reference streams) var regex = /XRefStm/g, result, indices = []; while ( (result = regex.exec(string)) ) < indices.push(result.index); >// Get the position of the stream var xrefstmPositions = []; for(var int = 0; int < indices.length; int++)< var start; var length = 0; while(char[indices[int] - 2 - length] != ' ')< start = indices[int] - 2 - length; length++; >var index = parseInt(string.slice(start, start + length)); var tempIndex = parseInt(string.slice(start, start + length)); // Add the offset (consequence of the content changes) to the index for(var num = 0; num < arrayDiff.length; num++)< if(index >arrayDiff[num].index) < index = index + arrayDiff[num].diff; >> string = string.replaceBetween(start, start + length, index); // If there is a difference in the string length then update what needs to be updated if(tempIndex.toString().length != index.toString().length)< arrayDiff.push(); char = new Array(string.length); for (var int = 0; int < string.length; int++) < char[int] = string.charAt(int); >; > xrefstmPositions.push(index); > // Do the same for non-stream var regex = /startxref/g, result, indices = []; while ( (result = regex.exec(string)) ) < indices.push(result.index); >for(var int = 0; int < indices.length; int++)< var end; var length = 0; while(char[indices[int] + 11 + length] != '\r')< length++; end = indices[int] + 11 + length; >var index = parseInt(string.slice(end - length, end)); var tempIndex = parseInt(string.slice(end - length, end)); for(var num = 0; num < arrayDiff.length; num++)< if(index >arrayDiff[num].index) < index = index + arrayDiff[num].diff; >> string = string.replaceBetween(end - length, end, index); if(tempIndex.toString().length != index.toString().length)< arrayDiff.push(); char = new Array(string.length); for (var int = 0; int < string.length; int++) < char[int] = string.charAt(int); >; > xrefstmPositions.push(index); > xrefstmPositions.reverse(); var firstObject, objectLength, end; var offset; // Updated the cross tables for(var int = 0; int < xrefstmPositions.length; int++) < var length = 0; var end; if(char[xrefstmPositions[int]] == 'x')< offset = 6; >else < offset = 0; >// Get first object index (read pdf documentation) while(char[xrefstmPositions[int] + offset + length] != ' ') < length++; end = xrefstmPositions[int] + offset + length; >firstObject = string.slice(end - length, end); // Get length of objects (read pdf documentation) length = 0; while(char[xrefstmPositions[int] + offset + 1 + firstObject.length + length] != '\r') < length++; end = xrefstmPositions[int] + offset + 1 + firstObject.length + length; >objectLength = string.slice(end - length, end); // Replace the offset by adding the differences from the content's field for(var num = 0; num < objectLength; num++)< if(char[xrefstmPositions[int]] == 'x')< offset = 9; >else < offset = 3; >// Check if it's an available object if (char[xrefstmPositions[int] + 17 + offset + firstObject.length + objectLength.length + (num * 20)] == 'n') < var objectCall = (parseInt(firstObject) + num).toString() + " 0 obj"; var regexp = new RegExp('\\D' + objectCall, "g"); var m; var lastIndexOf; // Get the last index in case an object is created more than once. (not very accurate and can be improved) while (m = regexp.exec(string)) < lastIndexOf = m.index; >string = string.replaceBetween(xrefstmPositions[int] + offset + firstObject.length + objectLength.length + (num * 20), xrefstmPositions[int] + 10 + offset + firstObject.length + objectLength.length + (num * 20), addLeadingZeros(lastIndexOf + 1, 10)); > if(num == objectLength - 1) < if (char[xrefstmPositions[int] + offset + firstObject.length + objectLength.length + ((num + 1) * 20)] != 't')< xrefstmPositions.push(xrefstmPositions[int] + offset + firstObject.length + objectLength.length + ((num + 1) * 20)); >> > > // create a blob from the string var byteNumbers = new Array(string.length); for (var int = 0; int < string.length; int++) < byteNumbers[int] = string.charCodeAt(int); >var byteArray = new Uint8Array(byteNumbers); var blob = new Blob([byteArray], ); // Do whatever you want with the blob here >; reader.readAsBinaryString(certificate); > 

So the code is not clean at all but it works :)

Let me know if you have any question