DEFINT a-z
DECLARE FUNCTION trimhtml$(text$)

FUNCTION PBMAIN()

   File$ = TRIM$(COMMAND$)

   IF File$ = "" OR DIR$(File$)="" THEN
      PRINT "strip <filemask>"
      PRINT "removes HTML code from a text file"
      EXIT FUNCTION
   END IF

   x$ = DIR$(File$)
   DIM filedata$(1 TO 1000)   'max 1000 lines of text

   DO
      IF x$ = "" THEN EXIT DO
      PRINT x$ 'disp filename
      'suck in file
      OPEN x$ FOR INPUT AS #1
      lines=0
      DO UNTIL EOF(1)
         lines=lines+1
         LINE INPUT #1, filedata$(lines)
      LOOP
      CLOSE #1
      'pump it out and hope it doesn't crash
      OPEN x$ FOR OUTPUT AS #1
      FOR r=1 TO lines
          PRINT #1, trimhtml$(filedata$(r))
      NEXT
      CLOSE #1
      x$ = DIR$
   LOOP


END FUNCTION

FUNCTION trimhtml$(text$)
    'removes html code from text$
    'NOTE: routine is brute force and will remove more than it needs
    '      to in many cases where improper tags are used
    hstart=INSTR(text$,"<")
    hend=INSTR(text$,">")
    IF hend<>0 AND hend<hstart THEN hstart=0
    IF hstart=0 AND hend=0 THEN
        temp$=text$
    ELSE
        'A few possibilities:
        IF hstart=1 THEN  'starts with HTML
            IF hend=0 THEN  'all HTML tag!
                temp$ = ""
            ELSE            'starts with tag, text follows
                temp$ = MID$(text$+" ", hend+1)
            END IF
        ELSEIF hstart=0 THEN    'starts with interior of tag, trim to end
            temp$ = MID$(text$+" ", hend+1)
        ELSE    'embedded
            IF hend=0 THEN  'trim to end, tag continues on next line
                temp$ = MID$(text$,1,hstart-1)
            ELSE    'trim full tag
                temp$ = MID$(text$,1,hstart-1)+MID$(text$+" ", hend+1)
            END IF
        END IF

    END IF
    IF (INSTR(temp$,"<") + INSTR(temp$,">"))>1 THEN
        temp$=trimhtml$(temp$)  'wu hoo recursive
    END IF
    trimhtml$=temp$
END FUNCTION