Thanks Ksharp To explain further: I know all the keywords.... The way the code is made, is that the second part of the code simply extracts the series of numbers in textas they occure. So in the following example, the first part of the code identifies the three keywords which I am interested in, which is very adequate: pid=prxparse('/\b(length|height|width|thickness)\b/oi'); on text text text text textdimentionstext text text where length 22.3mm, height 12mm, thickness 2.1mm, width 53mm text text text will return length, height, width ( I am not interested in thickness and I dont want it to be extracted) BUT The second part of the code will pull out 4 series of numbers:   pid=prxparse('/\d+(\.\d+)?mm/oi'); will return    22.3mm, 12mm, 2.1mm, 53mm Now this will cause a missmatch when the data is merged. That is why I am trying to link the number directly to the keyword before it, if that can be done then 2.1mm will not be extracted Kind regards   ... View more

Thank you everyone @Ksharp while the code ran perfectly well i ran into a problems: 1. The code extracted that last number only, i correct this by assing s=1 after the first do-end loop     another question, if the text file lióoks like this:   text text text text textdimentionstext text text where length 22.3mm, height 12mm, width 53mm text text text   is there a way to specifically link the number to the keyword just befre it? The reason I am asking is that there might be another keyword and a number after it in the middle, and I am not interested in these other keywords :   text text text text textdimentionstext text text where length 22.3mm, height 12mm, thickness 2.1mm, width 53mm text text text In the above example i am not interested in thickness Kind regards       ... View more

Thank you again @RW9 I really appreciate the time and effort you have put into your reply and answering this thread   I will explain things further - for your kind information- and i think you would afterwards say that I am definitely "not using the right tool for the right task" The documents themselves have been created over many years, they were not intended for data collection, but to serve as reports for human readers, in such, there data was entered as free text. Even worse, as more than one person entered these data, there was no standardized system in writing the report, and different people described the results in different way (through there were some distinct patterns that I have been trying to exploit using Perl expressions) The documents are stored as .DOC which is even worse. When I used SAS to extract data directly, I had to write a code that would re-save them as rtf files that would be easier to read with SAS. Even after doing this the result was a complete mess. SO as you can see, the whole project is a bit of a mess that I am trying to salvage. One might think that a better alternative would be for a human reader to extract data manually from the results. Believe me i have been there and have done that many years ago with other projects, but the efficiency and reliability of humans significantly drops for such tasks after only doing a relatively small part of the project. So my approach now is to combine both, to try to use SAS / other tools to extract as much as possible and then to use the human part to 1. verify output and tweak the code further 2. extract data that could not be extracted through step 1. I guess another aspect of this is for my own learning. Through this process I learned a few things about using SAS Perl Regular expression, string manipulation, some new aspects in VBA etc.... So at the end of the day, is there a "right tool" to do what I am trying to do? The answer is definitely not, but sometimes the goal is the journey and the challenge and the learning process more than the final result (if I want to be philosophical here!) As for using VBA in word, I havent thought about that, thanks for the suggesion, though iI would imagine that I would still need to export the data to an intermediary storage like excel or access... Anyhow. I thank you again for your answer. Your and other people's answers clearly illustrate the important role that SAS Communities play to make life easier for people like me....     ... View more

Thank you @RW9 I agree DDE is ancient and X command shouldn't be used, but as others previously commented, sometimes it is better to use the right took for the right task, or VBA 8s better equipped (I think) to handle word documents ... View more

Well, the story started with 3000 word documents that contained data in tables. I started using SAS to directly extract data from these documents. That proved to be pain stacking, slow and cumbersome. Approach B was to write a VBA in an excel file that would extract the data from the word documents into the excel sheet, which would then be imported to SAS for further analysis. The way the solution is done now is by SAS executing the Macro stored in the excel document and then importing final results. This means I will need to maintain 2 documents, the SAS code and the excel file. I was just hoping to make it easier by maintaining one document only. I hope this makes it easier to understand why I was looking for this solution ... View more

Thank you @Reeza Do you have the name of that paper as I have been searching in vain Thanks again Kind regards ... View more

Thanks, it shoudnt matter as I have  \s* I did try your code thoiugh and it is the same problem Any other suggesions please   ... View more