Add a log_print() function

[magnumripper]

Something like:

    log_print(int verbosity, bool main_only, char *format, ...)

Goals: Any code that now prints to stderr and/or log with a bunch of conditions (with 3-10 lines of code each time) could be replaced with a single call to a shared function in logger.c with following options:

• Send to screen, or log, or both.
• Line-wrap at word boundary (at whatever we deem to be the output width).
• Only print to parent process, or any/all. Only print node number when applicable.
• Optionally colorize serious notes (eg. yellow) and warnings (eg. red).
• Minimum verbosity for sending to screen and a separate minimum "log level" for sending to log. BTW I think I want specific future options like --log-no-rule-details and --log-no-mode-details while --log-none would deprecate the current -nolog. Perhaps this should be considered.

I have initial code for this, but it really needs some thought to end up really good.

[magnumripper]

For screen output, while at it we could easily ensure words don't break at line wrap. We either can know the terminal size or just default to 80.

[magnumripper]

4086

• Also, this could add optional color to the [screen] output, eg. yellow for important notes, red for more serious warnings.
• And, easily done, we could line-wrap at word boundaries as opposed to in the middle of a word. Without the caller needing to care at all.

[magnumripper]

When using -show, color could be great too.

[magnumripper]

Some test cases:

• Input has a word longer than width (disable wrapping for it, set col = 0 appropriately for next line).
• Indentation wider than width (disable indentation, or decrease it to 1).
• Lines containing some or lots of multi-byte UTF-8. Lorem Ipsum transliterated to Greek is Λορεμ ιπσθμ δολορ σιτ αμετ, πλατονεμ vολθπτατιβθσ ιν σιτ, εθ ερατ ιντερπρεταρισ μεα. Vιξ φεθγιατ δεσερθντ ελοqθεντιαμ θτ, μεα εα αθδιαμ φαβελλασ, περ πριμισ τιμεαμ δισcερε cθ. Cθ δολορ ρεcτεqθε δισπθτατιονι μει, μοvετ vοcεντ εθμ αν. Vενιαμ περσεcθτι θσθ νο, σαλθτανδι ιτθπερατα ρεπθδιανδαε αν vιμ. Γραεcι αδιπισcι cονσεcτετθερ νο σεα. Αδ qθο προβο ρατιονιβθσ, qθοδ σιντ cονσεcτετθερ πρι θτ.

Some notes and ideas that hit me:

• john.c sets john_terminal_locale from LC_*. We need to look at it. For example, our output might contain a file name encoded in multi-byte UTF-8.
• We should add strlen_utf8() and/or strlen_enc() (to unicode.[ch]) for counting characters, not bytes. We already have strlen16() that returns number of characters (number of 16-bit words) for a string encoded in UTF-16.
• We should consider hyphen - something we can break right after (but not remove like we'd do with a whitespace).
• We might want to consider / and \ same as - above, for when breaking a long file path: We can break right after but do not remove it.
• We'll automagically honor non-breaking space and non-breaking hyphen. Regardless of encoding they already aren't whitespace nor a regular -.
• We could implement crude hyphenation rules that could be used for otherwise nasty wrapping cases (like a very long word that doesn't fit). Maybe it's overkill!
• We might want blockers like "never break a word starting with -- even though it later has a -" (so we'd not break a word such as --single-wordlist) but that too may be overkill, or even not wanted.
• Likewise, if we do implement hyphenation rules, we could want to add blockers like "never break a word containing digits" and things like that.
• While on the subject of overkill, let's also consider soft hyphen (hint for wrapping but will be invisible/removed unless we actually wrapped there).

[magnumripper]

• john.c sets john_terminal_locale from LC_*. We need to look at it. For example, our output might contain a file name encoded in multi-byte UTF-8.
• We should add strlen_utf8() and/or strlen_enc() (to unicode.[ch]) for counting characters, not bytes. We already have strlen16() that returns number of characters (number of 16-bit words) for a string encoded in UTF-16.

4406 adds this:

• We can test options.terminal_enc for eg. being UTF_8 or not.
• An added best-effort strlen_any() will count number of characters, not bytes (in particular, our messages may include file paths encoded in UTF-8 or not). This function can be safely used even when you're not sure input is UTF-8 or some legacy encoding, or a mix...