Create NLTK corpus reader customized for structured .txt file from Proquest

I need a customized corpus reader that will take a plain .txt file that I have downloaded from Proquest and process it so that I can use in further analysis. The .txt file is structured in a consistent way with standardized tags identifying data fields. These tags are either words or phrases that start with a capitalized letter and that end with ': '. I need two types of data extracted from the plain .txt file. First, I need the 'Full text: ' field (i.e., everything that comes after this tag and before the next tag, which is "Subject: ") extracted with a unique identifier (which could be the existing unique identifier i the file, which is indicated by the tag "ProQuest document ID: ") so that it can be parsed at word, phrase, sentence and paragraph level by NLTK. Second, I need all of the fields EXCEPT the 'Full text: " field extracted and placed into a Pandas dataframe.