Having worked in content and knowledge management, the first approach is valid only if this were a larger, very well funded project, i.e. identifying a COTS product that will do much of what you need right out of the box. There is often additional work required to go from meeting say 75% of your requirements to closer to 100%. Many times that additional work is far from trivial. Potential vendors include those that have invested in parsing PDF, word and possibly OCR files, and then storing the contents in a database. Often, tagging and applying a structure to the contents using XML is offered as well. Astera and Kapow [a Lexmark company] are two vendors that offer such a solution aimed at very large [Enterprise] companies.
More commonly, one will find tools and DIY solutions that will import the files without parsing them into a SQL database - but from what I understand here, that is not what you are [were] looking for.
A second approach is the DIY route, which may not cost much or any money out of pocket, but which will certainly require an investment of time and energy. PHP, for example, has PDFlib which can be used to do what you need to PDF files. This link http://www.php.net/manual/en/ref.pdf.php may help in exploring that particular approach. I reckon you could find PHP scripts that have already done the work of implementing PDFlib that you might be able to use as is or with some minor modifications.
When it comes to working with the contents of .doc and .docx files there appear to be loads of solutions using various tools. This link has many recommendations https://stackoverflow.com/questions/5671988/how-to-extract-just-plain-text-from-doc-docx-files-unix and this is a thread that focuses on a PHP approach https://stackoverflow.com/questions/5540886/extract-text-from-doc-and-docx. I haven't looked into this for years, but I would bet that there are lots of tools for a Microsoft-centric approach to working with the the contents and metadata from .doc and .docx files [and .xls .xlsx etc.] as well.
With OCR your situation is complicated by the fact the files are handwritten. Were they all typed, I think there would more readily available solutions that could extract the contents. I could well be wrong, but I think with hand written and/or drawn images you would be looking at applying metadata to the files and having to call it good at that. The work involved in this approach would likely be quite time consuming even with an ideal setup as all the background work of setting up a taxonomy and such would need to pave the way for the grunt work of reading the handcrafted files and deciding how to categorize each, what tags to apply and so on. Setting aside the complication of handwritten files, there seem to be numerous open source and other projects along the DIY route for an OCR type approach. Nuance seems to be the leading, or one of the leading, vendors in this area.
