@CodyGray While it would be possible to turn it into a bookmarklet, it's finding the hits that are caught by one entry, but not the other entry, so it needs two, separate, pieces of information. Thus, it wouldn't be just select the text and click the bookmarklet. The bookmarklet would need an input dialog to get the two separate pieces of information. I'd need to think on what would be both easy to implement and intuitive to use. A couple of window.prompt()s would be the easiest to implement.
@CodyGray In retrospect, I realize that both the way that I showed the two differences there and the fact that I didn't provide complete information is confusing. The first difference I linked to was looking at the hits removed by the change, while the second difference was looking at the hits added by the second minor change (for the second, that's "mostly", see below).
⠀⠀1) the hits caught by the first (A);
⠀⠀2) the hits caught by the second (B);
⠀⠀5) the intersection of hits caught by both;
⠀⠀6) the union of hits caught by both.
Effectively, that's filling in the hits that would be in all areas of a basic Venn diagram of the two sets.
(5) and (6) are, usually, not that interesting when considering changing a single existing watchlist/blacklist entry, but are interesting for other operations (e.g., adding/removing entries).
For both of the changes I made, I only linked to searches showing (1), (2), and either (3) or (4), which isn't sufficient to get the entire picture of the change, unless the unshown of either (3) or (4) is null, which dramatically isn't the case for the first change, due to my adding the possibility of an intervening word (again, see below for comments on the second change, because I didn't really provide either (3) or (4) for that change).
For the first change, the additional difference regex (4, above) that shows the added results is (?#in intermediate new, not old)(?s)^(?=[\W\w]*?(?:^|\b)(?:support|service)(?!/)(?!(?:[^<]++|<(?!\/?code>))*+<\/code>)(?:\W++(?:\w(?<!\d))++)?[\W_]*+(?!\d\d(?:\d\d)?[:-]\d\d[:-]\d\d(?:\d\d)?\b)(?:2[\W_]*+4[\W_]*+7|(?:\d[\W_]*+){6,12})(?:\b|$))(?![\W\w]*?(?:^|\b)(?:support|service)[\W_]*+(?<!└─)\d++(?:\b|$))
@Makyen
1482 Total / 1437 TP (96.96% TP) / 40 FP / 5 NAA [Note: numbers won't match, as there's a new FP hit. The new FP hit indicates I should adjust the regex a bit. I also added the second difference and the link in this message to the string of messages to which you were replying, so it's less confusing if someone is reading the transcript.]
For the second change, making the "7" optional, I used a "difference" regex that was less complex and focused only on the specific change I made, rather than the effect on the entirety of the results.
The overall regex has two primary things it's looking for after the (?:support|service). It's looking for any numbers that start with "24", a potential "7", and followed by a \b. The second option of what it's looking for is a telephone-like sequence of digits and non-word characters (loosely described as 6 to 12 digits, potentially separated by non-word characters).
The regex I supplied as a "difference" for the second change completely ignores the second option of what the overall regex looking for (the phone numbers) and only considers the change to the "24/7" portion of making the "7" optional. This ignores the overlap of that portion with the telephone numbers, so the numbers provided as the "difference" are an approximation of the actual effect, but highlight the change in that portion of the regex.
That second "difference" regex is less complex than the generic difference regex I provided because I was focusing specifically on the change I knew I had made and was ignoring the overlap with the other part of the full regex (i.e., the part looking for phone numbers).
That particular change, making the "7" optional, may or may not be all that beneficial and might add too many FP. We'll have to see how it does, but without the change we wouldn't get the data.
I got an email inviting me to a study as it appears I set the settings for it. However I am skeptical to its realness as:
All links do not point to a SO/SE site, but instead to a cryptic link for a site called "pstmrk" and "rallyuxr".
There is a monetary incentive given (with an asterisk going n...
