Re: I need help understanding a match

Thank you!


On 1/13/25 18:20, William Michels via perl6-users wrote:
> Hi Todd,
> 
> First I should apologize for one of my earlier posts. The first token was a bit of a jumble. I think now you just want the literal string "download" to start your capture.
> 
> As per usual I tried a few different approaches to your regex problem, and posted what I thought was the best one, However an older iteration crept into one of my email posts: it used `^` which is Raku's zero-width "start-of-string" regex token.
> 
> If you use `^` you will capture from the start-of-string onward, in this case through the `.*?` any-character token and up to the \> angle. You may not want this as it actually means the word "download" isn't required for you to capture that sequence of characters.
> 
> I'm not sure where you got the impression that `\...\` actually means anything specific in Raku. If you're asking for a match against alphanumeric characters in Raku you don't have to escape them. Anything else (e.g. punctuation) you'll have to escape. So this means if you're trying to match ">" the "greater-than" sign (angle), you'll have to escape it  via a backslash (e.g. `\>`), or by quoting (e.g. ">").
> 
> For non-alphanumeric characters, an unescaped punctuation characters is reserved for special "metacharacter" purposes: for example an unescaped "." dot means "any-character". You'll also note backslashing used to denote characters that are difficult to represent otherwise. Think for example how `\n` means newline, `\t` means tab. There are others: `\s` means whitespace, `\h` means horizontal-whitespace, and `\v` means vertical whitespace. Also `\S` means non-whitespace, `\H` means non- horizontal-whitespace, and `\V` means non- vertical-whitespace.
> 
> I've also posted direct links to Raku regex forms, such as `<?before ... >` (a positive lookahead) and `<?after ... >` (a positive lookbehind). You can try this in the REPL:
> 
> [0] > my $a = "XYZ"
> XYZ
> [1] > say $a ~~ m/ <?after X > Y <?before Z > /;
> ｢Y｣
> 
> Try reading that out loud in English, "say $a smartmatching against a requested `m` match comprising after-X, Y, before-Z". If you read it that way, you'll understand why only the `｢Y｣` ends up in the match variable. You can also `andthen` the smartmatch, which will put the match in the `$_` topic variable for you, which can help with stringification:
> 
> [1] > $a ~~ m/ <?after X > Y <?before Z > / andthen put $_.Str;
> Y
> 
> I'll try to go through and correct what you wrote below. Best, Bill.
> 
>> On Jan 12, 2025, at 03:11, ToddAndMargo via perl6-users <[email protected]> wrote:
>>
>> Hi Bill,
>>
>> Please correct my notes.
>>
>> Many thanks,
>> -T
>>
>>
>>
>> Explanation:
>> my @y = $x ~~ m:g/ <?before ^ | download > .*? <?before \> | \h+ > /;
>>
>> `m:g`     # match and global
> CORRECT
>> `\...\`   # the constrains (beginning and end) of the match
> NO, backslashes are used to escape non-alphanumeric characters, denote invisible characters (e.g. `\n`), etc.
>> `<...>`   # constraints of instructions inside the match
> NO, `<?after ... >` is a lookbehind and `<?before ... >` is a lookahead.
>>
>>
>>
>> First instruction: `<?before ^ | download >`
> NO, this should just be the literal string `download` (or `"download"`)
>>
>>   `?download ^`   # positive look-behind, match but don`t capture `download `
>>                   # `^` means "look behind"
>>
>>   `|`             # This is logical "OR"
>>
>>   `download `     # positive look-behind, match but don`t capture `download `
>>
>>    summary: capture everything behind `before ` or capture just `download`
>>
>>
>> Second instruction: `.*?`
>>    `.*?`       # any-character, one-or-more, frugal up to the third instruction
> YES, CORRECT
>>
>>
>> Third instruction: `<?before \> | \h+ >`
> NO, SIMPLIFY THIS TO  `<?before \>` and the match will stop when it encounters  ">" the "greater-than" sign (angle). Because you're using a lookahead (match characters and "lookahead" to find a pattern but don't capture, example ), the ">" angle doesn't get captured.
>>
>>    `<?before \>`  # positive look-ahead, match but don`t capture `download \>`
> KINDA, the actual construct is `<?before \> >` or (even more readable), `<?before ">" >`
>>                   # Note that the `\` in `\>` is escaping the `>` and is removing
> KINDA, the `\` backslash in front of a non-alphanumeric is a rule in Raku. If it isn't backslashed Raku will try to interpret the non-alphanumeric as a metacharacter.
>>                   # the `>` from the instructions constraints and making is part
>>                   # of the match
> The unescaped `>` is part of the lookahead/lookbehind construct, either  `<?after ... >` (lookbehind) or  `<?before ... >` (lookahead).
>>
>>   `|`             # This is logical "OR"
> YES
>>
>>   `\h+ `          # one-or-more horizontal whitespace character
> YES
>>
>>   summary: capture everything before `before` or one-or-more whitespace characters
> KINDA. Match the previous tokens, and stop matching when (before) you find one-or-more whitespace characters.
>>
>>
> 
> HTH.