The sbbs_t::wordwrap[] buffer was 81 characters (80 printable characters, plus
NUL terminator) and sbbs_t.getstr(), when used with the K_WRAP mode flag could
potentially write more than 81 characters to this buffer (e.g. when using a
wider than 80 column terminal and writing a message with the internal line
editor which calls sbbs_t::getstr(... K_WRAP)) - would corrupt sbbs_t members
after wordwrap[], which included pointers that would be freed in the sbbs_t
destructor (~sbbs_t) and subsequently page/segfault as seen in issue #545.

This change increases the wordwrap buffer to likely twice the same needed
(maximum columns + NUL terminator) and adds wordwrap bounds checking to
sbbs_t::getstr().

There were comments indicating crash sightings in the sbsb_t destructor going
back to 2002, so this commit removes those comments.

Thanks to Nelgin for providing the gdb dump details ('print *this') that was
the clue needed to reach the root-cause determination.

This fixes issue #545.

At one time, Synchronet was a 16-bit DOS project, plagued by the 16-bit [u]int, so long's were used everywhere > 16-bits were known to be needed/wanted (This is before the days of the standard sized types from stdint.h), and they've persisted.

But '[u]long int' is 64-bits on *nix 64-bit builds, 32-bits everywhere else (even 64-bit Windows builds if/when we ever get around to that), so this could lead to insidious bugs that would only show up on one flavor or the other. Since [u]int is 32-bits on everything we currently support, we'll use that instead of [u]long.

This "part 1" because I'm sure there's going to be warnings and errors from the GCC/Clang builds as a result, which I'll get to next.

The internal line editor's quoting feature add some hard-coded strings ("Done" and "All") and the (L)ist key was hard-coded. Use the text.dat string (new and pre-existing) for these now.

Also, use the new sbbs_t *_key() methods for referencing the configured key bindings (via text.dat) for these common key-stroke commands.
Convert the text.dat strings for keys (e.g. YNQP) to uppercase always as well.

Hopefully not introducing any bugs in the process.

Hopefully not introducing any bugs in the process.

I'm fed-up with MSVC assertions in ctype functions (e.g. isdigit, isprint, isspace, etc.) when called with out-of-range (e.g. negative) values.

This problem only affects MSVC debug builds, but if you run them (like I do), these things are like little time bombs that can drive you crazy (knocking your board out of service).

The new macros names are bit more descriptive as well.

- User Defaults menu mousified
- When all hot spots have scrolled off the screen, clear the hot spot list
- Insert new hot spots in the front of the list so they will take precedence
  over any previously defined duplicate hot spots
- Numeric hot spot commands end in a carriage-return
- No auto-acceptence of numbers when there are keys in the keyboard buffer
- Baja UNGETKEY function now inserts keys into the "front" of the keyboard
  buffer (next to be consumed)
- Hot key @-code command text supports C-style escape sequences
  (allows encoding of ctrl chars, spaces, etc.)

getstr's insert mode (Ctrl-V) indicator and edit/line modes. Global Hot Keys
(e.g. Ctrl-T) would do weird stuff like return the cursor to the end of the
line or redraw the input string multiple times.
So as part of this fix:
- backspasce() now removed the last count characters from the save-line buffer
  (it is a destructive backspace afterall).
- outchar('\b') (a non-destructive backspace) adds the \b to the save-line buf
- restoreline() now always redraws the current insert/overwrite indicator in
  the upper right-most corner fo the screen
- insert_indicator() now uses functions that aren't line-saved.

An unrelated change included in this comment: center() now sends a CR before
the cursor movement and centered-text, for times when the current column is
*not* already at 0. Very useful for use with @-codes (e.g. display some pattern
or "graphics" and then center some text over it).

when echo is disabled or when we're echoing X's (for password input).
Ctrl-R key would redraw the raw password at a password prompt, even when the
console is in Echo-X mode, pass a non-zero mode flag to redrwstr() to fix that.
Also, some experimental UTF-8 input stuff that doesn't work right yet.

- Export all FIDOCTRL (other FTN kludge lines) to the QWK HEADERS.DAT file.
  These should already be imported if they exist, but were never added during
  export, so untested/new behavior. The control paragraph (kludge line) of
  specific interest here is the "CHRS" (charset) kludge we need for UTF-8.

- Don't use the QWK "newline" character (0xE3) when the message is UTF-8.
  Use bare-LF's instead. This is pretty untested at this point as I will need
  another QWKnet board to post or receive UTF-8 encoded messages to test,
  getting the code into CVS is the first step. At least for now, there's no
  opt-in/out for this behavior. If your BBS has UTF-8 encoded messages, some
  QWK nodes or offline readers may have trouble with packets which include
  those messages. Or they may work fine (but likely display garbage CP437
  chars in-place of the proper Unicode codepoint glyph).

- The beginning of UTF-8 input support in getstr() - which needs more work,
  particularly around character and word deletion and insertion.
- The internal message editor now supports UTF-8 messages and kind of somewhat
  supports inputting UTF-8 characters in message text.

New put/print text flag: P_AUTO_UTF8 which can auto-detect UTF8 strings and
do the "right thing" for the user's terminal. New associated sbbs_t method:
auto_utf8() which automatically sets P_UTF8 for any stirng that begins with
a UTF-8 BOM (ZWNBSP). Else, if the P_AUTO_UTF8 mode flag is set, then
it checks to see if the string contains invalid US-ASCII chars but valid UTF-8
sequences and then sets P_UTF8 accordingly. Used by putmsg() and bputs().

There's a new permuation of bprintf() which accepts a mode argument
(i.e. for P_UTF8) and passes it on to the new mode-capable bputs().

- for PETSCII, this is automatic and you don't lose blink, but you do lose
  colored foreground when enabling a bright background ("reverse video")
- for ANSI, this mode is typically referred to as "iCE colors" and disables
  blinking-text support
- Ctrl-AE (^AE) is the new attribute code to enable bright-background
  (will have no effect on ANSI terminals that are not iCE color enabled)
  "E" is now valid in the ctrl/attr.cfg file and string values for JS
  console.attributes assignments, as well
- Ctrl-AI (^AI) - blink - now does nothing for ANSI/ICE color terminals
  (blinking is not supported in combination with bright-background)
- Using a new/non-standard CGA attribute bit-flag to indicate the selection of
  bright-background colors (BG_BRIGHT, bit 10), separate from BLINK.
  This change required all/most char/uchar attribute representations to be
  converted to int/uint.

New text.dat strings:
- PetTerminalDetected (renamed from PetTermDetected)
- PetTerminalQ
- TerminalAutoDetect
- TerminalColumns
- TerminalRows
- TerminalMonochrome
- TerminalColor
- TerminalIceColor
- IceColorTerminalQ
This also moved the MsgCarbonCopyList definition to the end of the file
for now.

PETSCII reverse-video attribute fix:
When a CR is sent to the terminal, the reverse-video attibute is auto-disabled
so update our "current attribute" (curatr) value to match the remote.

Support new printfile/putmsg mode flag: P_WRAP to force an ungraceful
line-wrap (splitting) to the specified column width. If no column width is
specified (0), then this mode will force an ungraceful wrap before the last
terminal column where some terminals *may* auto-wrap.

JS console.printfile() and printtail() methods now support an optional
"orig_columns" argument, similar to console.putmsg(). Must specify P_WORDWRAP
or P_WRAP for this argument to have any effect.

Much improved terminal-type selection/configuration in the user defaults
menu and abort (^C) at any of the yes/no prompts is now detected/handled much
better (to answers to the prompted questions are not saved to the user
settings).

So Omegix recently asked in the Synchronet Discussion group whether or not
a PETSCII (Commodore) terminal could be used to access his Synchronet BBS.
Now, the answer is "Yes". :-)
The major issues addressed:

- detecting a PETSCII terminal, solved by assigning specific (configurable):
  TCP ports to be used for incoming PETSCII connections, by default:
  port 64 is for 40-column PETSCII and port 128 is for 80-column PETSCII,
  but if the terminal sends a Telnet Window Size reply (e.g. SyncTERM), then
  either size terminal should fine on either port.
  The port numbers are configurable in the [BBS] section of your sbbs.ini
  file using the new keys: PET40Port (default value: 64) and PET80Port
  (default value: 128). Having these keys set doesn't make make the terminal
  server listen on that additional port - you'll need to add more
  IP:port combinations to one of Interfaces values, example:
    TelnetInterface=71.95.196.34,71.95.196.34:64,71.95.196.34:128
  And you don't have to use Telnet for the PETSCII connections - you could use
  RLogin or SSH instead (or in addition).

- support for terminal widths < 80 columns:
  This was achieved through a combination of text.dat changes (numerous),
  new Ctrl-A and @-codes and new optional terminal-width-specific menu files
  (e.g. text/menu/main.40col.asc)
  A side effect of these changes is actually better support for terminals
  *wider* than 80 columns as well!

- support for terminals that don't expand tabs to spaces (e.g. PETSCII):
  The terminal server now handles tab expansion with a run-time settable
  tab-size (default size: 8)

- conditional access based on PETSCII (or small) terminal use (or not):
  + New PETSCII ARS keyword (boolean)
  + New COLS and ROWS ARS keywords (for terminal width and height requirements)
  + New TERM (string) ARS keyword

New @-codes:
  - WORDWRAP, when placed at the top of a file, enables auto-wordwrap for
              lines longer than the terminal width
  - CENTER, the text following before an end of line will be displayed centered
            on the terminal (whatever the width, in columns)
  - CLEAR, like CLS, except it ignores (doesn't display) a CRLF that follows
  - COLS, current number of terminal columns (width)
  - ROWS, current number of terminal rows (height)
  - TERM, the auto-detected or reported terminal type (e.g. ANSI, TTY, etc.)
  - SYSONLY, toggles "echo" (display) off/back-on for non-sysops
            similar to the Ctrl-A( and ) codes, but more convenient to use
            (and PabloDraw won't strip the @-code from the file like it does
             with Ctrl-A codes it doesn't support)

New Ctrl-A codes:
  - \ conditional new-line/continuation when the terminal width is < 80 cols
      prints the new text.dat string LongLineContinuationPrefix

yesno() will now return true if passed a blank string.
noyes() will now return false if passed a blank string.
getstr()'s input length limiting based on terminal width is more broadly
           applied now (not just when using the K_LINE mode flag).

New JS bbs object method: menu_exists(<base_filename>) returns Boolean
New JS console object property: tabstop (Number)
New JS console object methods: getbyte() and putbyte() to recv/send raw byte
    value with (very little) interpretation/intervention by the terminal server
New JS console object method: creturn() - performs a carriage return
    (or equivalent)
New JS (and C) printfile() mode flag: P_TRUNCATE, causes long lines to be
    truncated, rather than displaying causing a line-wrap.

New text.dat strings:
  - NoAccessTerminal (for ARS check failures)
  - LongLineContinuationPrefix (for breaking long lines for 40col terminals)
  - Scanning (replaces a previously hard-coded "Scanning" string)
  - Done (replaces a previusly hard-coded "Done")
  - Scanned  (when finished scannning, clears the progress bar)

from both ends of the input string automatically and disallow (ignore)
more than one consecutive space character mid-string.
The returned value is still the non-trimmed string length - perhaps that
should be changed.

NULL. When history is NULL, treat TAB the same as before (in weird ways).
The TAB-completion algo is very simple: just find the first (case
insensitive) match in the history list and if there is one, use it as the
current string content.

additional, optional, "history" argument - an array of strings that can
be recalled/scrolled using the up/down arrow keys. Currently, I just plan
to use this for netmail destination address recall, but it could be used
other places easily enough.

fix the resulting warnings.

than sprinkling the mapping around the source files.
Also, disable the message thread nav keys (arrow keys, home, end) when not
in Thread View Mode - too confusing.

was never used. This was intended to be a method for internationalization,
letting the sysop change which key is used as the uiniversal "quit" key.
This commit replaces most of the uses of the hard-coded 'Q' for quit wtih the
3rd charcter in the text.dat YNQP string. Some hard-coded menus still have the
'Q' key hard-coded and will need to be addressed later. The text.h YN index
macro was changed to YNQP and the JS text index variable name will change
as well.

Fix Ctrl-Z/Undo so it only redisplays from the correct column/offset.
Note: K_NOECHO mode appears to still have some bugs if/when K_EDIT is also
used (which would be a weird combination of modes).

for non-ANSI terminals (that have a non-destructive backspace) in getstr().

Simplify redrwstr() by combining ANSI/non-ANSI terminal support.

with ansi_gotoxy(), ansi_save(), and ansi_restore() functions.

If default value is not zero, it is the value returned if enter is
pressed with no input.  In this case, zero is allowed to be entered.

- returns true/false if the user's terminal supports all the passed flags
- returns the supported flags if no flags are passed for comparison
- automatically uses the auto-terminal-type flags if no user is logged-on

CON_R_ECHOX and CON_L_ECHOX, then sends "\b \b"

Replace all instances of [br]puts("\b \b") with call to backspace()

This fixes once and for all the "spaces entered at password prompt are
echoed as spaces" problem.

Har.

PR:42->closed

(I always wanted to do that commit!)

editor).

indicator.
Created CON_DELETELINE status to indicate getstr() returned due to line
deletion (versus backspace).

Removed CRLFs even when K_NOCRLF mode bit is set under some circumstances.
Ctrl-Z (if not handled by inkey) will undo any changes made to the line.
Insert indicator (flashing block) is now placed in (normally unused) cell 1,80
(fixes wrap problem when using insert during input on bottom line of terminal).
Insert mode is no longer automatically turned off at end of string.
Ctrl-Y will now delete the entire line (ala Ctrl-X) if used at EOL.
Optimized Ctrl-Y/X clearing of line for ANSI terminals.

for full-screen editor functionality.
Added support for CON_DOWNARROW, CON_LEFTARROW, and CON_BACKSPACE console
status bits, used in combination with getstr(K_LEFTEXIT)  for full-screen
editor functionality.
Insert mode (^V) still needs work.