Text heavy data analysis

General information for the business: Personal project
Description of requirements/functionality: I need someone to:
Visit http://www.ombudsman-decisions.org.uk/

Write a script to download three years’ worth of ombudsman decisions except those containing the following phrases:
“payment protection insurance”
“packaged bank account”
“closed his/her/its account”

Each decision has a unique File ID, which may help in writing the script. A live example below:
http://www.ombudsman-decisions.org.uk/viewPDF.aspx?FileID=143331

Having downloaded the necessary files (likely to be in the low tens of thousands) isolate the following attributes for each file and paste as columns into a Microsoft Excel compatible table.

Unique reference number
e.g DRN0795971 (top left hand corner of first page)

Decision date
Commonly in the final paragraph in the format “accept my decision by Day/Month/Year”.

Date of sale
Where given, it will be in the first paragraph. “In Month/Year..”

Name of firm complained about.
Usually in the first two lines, ending Limited, Ltd or Plc or similar. A non-exhaustive list of firm names is available here: http://www.ombudsman-complaints-data.org.uk/

Product complained about
Also usually in the first paragraph. The text may, or may not, be a phrase contained in this list:
http://www.financial-ombudsman.org.uk/publications/ombudsman-news/139/chart_issue139.pdf
Please capture linked products, like this:
“self-invested personal pension (SIPP) so he could make an investment in Harlequin property”
http://www.ombudsman-decisions.org.uk/viewPDF.aspx?FileID=143331

URL
As above.

Decision
Infer from the paragraph following “my final decision” whether the firm does, or does not need to pay compensation. “I don’t uphold Mr. C’s complaint” means the firm complained of will not pay compensation.

Compensation
Standard compensation looks like this: “refund the £41,700”. Please capture the amount.
Formulaic compensation looks like this: “should calculate fair compensation by comparing the value of”. Example: http://www.ombudsman-decisions.org.uk/viewPDF.aspx?FileID=117578. Reflect this in the table by writing “Formulaic”

Name of ombudsman
Contained in the penultimate line of the document.

Perform analysis
Using the table, or the underlying data, identify any products with similar properties to Stirling Mortimer, Harlequin, Brandeaux, and explain what these properties are.

There's no rush on the job, next few weeks is fine.
Specific technologies required: "R"
OS requirements: Windows
Extra notes: