Create a Lexer class for BBC Basic files.

The Lexer class for BBC Basic handles both the numbered lines, and
unnumbered lines, of the detokenised (text) format of BBC BASIC.
The tokeniser copes, in a naive manner, with the orignal versions,
and BASIC V. It does not handle other extensions at this time, nor
does it handle inline assembler. This should be sufficient for most
cases where code needs to be presented in a colourful manner.

3 files changed, 254 insertions, 2 deletions

diff --git a/pygments/lexers/_mapping.py b/pygments/lexers/_mapping.py
index f8207e70..e57124db 100644
--- a/pygments/lexers/_mapping.py
+++ b/pygments/lexers/_mapping.py
@@ -47,6 +47,7 @@ LEXERS = {
     'AutoItLexer': ('pygments.lexers.automation', 'AutoIt', ('autoit',), ('*.au3',), ('text/x-autoit',)),
     'AutohotkeyLexer': ('pygments.lexers.automation', 'autohotkey', ('ahk', 'autohotkey'), ('*.ahk', '*.ahkl'), ('text/x-autohotkey',)),
     'AwkLexer': ('pygments.lexers.textedit', 'Awk', ('awk', 'gawk', 'mawk', 'nawk'), ('*.awk',), ('application/x-awk',)),
+    'BBCBasicLexer': ('pygments.lexers.basic', 'BBC Basic', ('bbcbasic',), ('*.bbc',), ()),
     'BBCodeLexer': ('pygments.lexers.markup', 'BBCode', ('bbcode',), (), ('text/x-bbcode',)),
     'BCLexer': ('pygments.lexers.algebra', 'BC', ('bc',), ('*.bc',), ()),
     'BSTLexer': ('pygments.lexers.bibtex', 'BST', ('bst', 'bst-pybtex'), ('*.bst',), ()),
diff --git a/pygments/lexers/basic.py b/pygments/lexers/basic.py
index b0409386..4d957c2b 100644
--- a/pygments/lexers/basic.py
+++ b/pygments/lexers/basic.py
@@ -18,7 +18,8 @@ from pygments.lexers import _vbscript_builtins
 
 
 __all__ = ['BlitzBasicLexer', 'BlitzMaxLexer', 'MonkeyLexer', 'CbmBasicV2Lexer',
-           'QBasicLexer', 'VBScriptLexer']
+           'QBasicLexer', 'VBScriptLexer', 'BBCBasicLexer']
+
 
 
 class BlitzMaxLexer(RegexLexer):
@@ -561,4 +562,98 @@ class VBScriptLexer(RegexLexer):
             (r'"', String.Double, '#pop'),
             (r'\n', Error, '#pop'),  # Unterminated string
         ],
-    }
\ No newline at end of file
+    }
+
+
+class BBCBasicLexer(RegexLexer):
+    """
+    BBC Basic was supplied on the BBC Micro, and later Acorn RISC OS.
+    It is also used by BBC Basic For Windows.
+
+    .. versionadded:: 2.4????
+    """
+    base_keywords = ['OTHERWISE', 'AND', 'DIV', 'EOR', 'MOD', 'OR', 'ERROR',
+                     'LINE', 'OFF', 'STEP', 'SPC', 'TAB', 'ELSE', 'THEN',
+                     'OPENIN', 'PTR', 'PAGE', 'TIME', 'LOMEM', 'HIMEM', 'ABS',
+                     'ACS', 'ADVAL', 'ASC', 'ASN', 'ATN', 'BGET', 'COS', 'COUNT',
+                     'DEG', 'ERL', 'ERR', 'EVAL', 'EXP', 'EXT', 'FALSE', 'FN',
+                     'GET', 'INKEY', 'INSTR', 'INT', 'LEN', 'LN', 'LOG', 'NOT',
+                     'OPENUP', 'OPENOUT', 'PI', 'POINT', 'POS', 'RAD', 'RND',
+                     'SGN', 'SIN', 'SQR', 'TAN', 'TO', 'TRUE', 'USR', 'VAL',
+                     'VPOS', 'CHR$', 'GET$', 'INKEY$', 'LEFT$', 'MID$',
+                     'RIGHT$', 'STR$', 'STRING$', 'EOF', 'PTR', 'PAGE', 'TIME',
+                     'LOMEM', 'HIMEM', 'SOUND', 'BPUT', 'CALL', 'CHAIN', 'CLEAR',
+                     'CLOSE', 'CLG', 'CLS', 'DATA', 'DEF', 'DIM', 'DRAW', 'END',
+                     'ENDPROC', 'ENVELOPE', 'FOR', 'GOSUB', 'GOTO', 'GCOL', 'IF',
+                     'INPUT', 'LET', 'LOCAL', 'MODE', 'MOVE', 'NEXT', 'ON',
+                     'VDU', 'PLOT', 'PRINT', 'PROC', 'READ', 'REM', 'REPEAT',
+                     'REPORT', 'RESTORE', 'RETURN', 'RUN', 'STOP', 'COLOUR',
+                     'TRACE', 'UNTIL', 'WIDTH', 'OSCLI']
+
+    basic5_keywords = ['WHEN', 'OF', 'ENDCASE', 'ENDIF', 'ENDWHILE', 'CASE',
+                       'CIRCLE', 'FILL', 'ORIGIN', 'POINT', 'RECTANGLE', 'SWAP',
+                       'WHILE', 'WAIT', 'MOUSE', 'QUIT', 'SYS', 'INSTALL',
+                       'LIBRARY', 'TINT', 'ELLIPSE', 'BEATS', 'TEMPO', 'VOICES',
+                       'VOICE', 'STEREO', 'OVERLAY', 'APPEND', 'AUTO', 'CRUNCH',
+                       'DELETE', 'EDIT', 'HELP', 'LIST', 'LOAD', 'LVAR', 'NEW',
+                       'OLD', 'RENUMBER', 'SAVE', 'TEXTLOAD', 'TEXTSAVE',
+                       'TWIN', 'TWINO', 'INSTALL', 'SUM', 'BEAT']
+
+
+    name = 'BBC Basic'
+    aliases = ['bbcbasic']
+    filenames = ['*.bbc']
+
+    tokens = {
+        'root': [
+            (r"[0-9]+", Name.Label),
+            (r"(\*)([^\n]*)",
+             bygroups(Keyword.Pseudo, Comment.Special)),
+            (r"", Whitespace, 'code'),
+        ],
+
+        'code': [
+            (r"(REM)([^\n]*)",
+             bygroups(Keyword.Declaration, Comment.Single)),
+            (r'\n', Whitespace, 'root'),
+            (r'\s+', Whitespace),
+            (r':', Comment.Preproc),
+
+            # Some special cases to make functions come out nicer
+            (r'(DEF)(\s*)(FN|PROC)([A-Za-z_@][A-Za-z0-9_@]*)',
+             bygroups(Keyword.Declaration, Whitespace, Keyword.Declaration, Name.Function)),
+            (r'(FN|PROC)([A-Za-z_@][A-Za-z0-9_@]*)',
+             bygroups(Keyword, Name.Function)),
+
+            (r'(GOTO|GOSUB|THEN|RESTORE)(\s*)(\d+)',
+             bygroups(Keyword, Whitespace, Name.Label)),
+
+            (r'(TRUE|FALSE)', Keyword.Constant),
+            (r'(PAGE|LOMEM|HIMEM|TIME|WIDTH|ERL|ERR|REPORT\$|POS|VPOS|VOICES)', Keyword.Pseudo),
+
+            (words(base_keywords), Keyword),
+            (words(basic5_keywords), Keyword),
+
+            ('"', String.Double, 'string'),
+
+            ('%[01]{1,32}', Number.Bin),
+            ('&[0-9a-f]{1,8}', Number.Hex),
+
+            (r'[+-]?[0-9]+\.[0-9]*(E[+-]?[0-9]+)?', Number.Float),
+            (r'[+-]?\.[0-9]+(E[+-]?[0-9]+)?', Number.Float),
+            (r'[+-]?[0-9]+E[+-]?[0-9]+', Number.Float),
+            (r'[+-]?\d+', Number.Integer),
+
+            (r'([A-Za-z_@][A-Za-z0-9_@]*[%$]?)', Name.Variable),
+            (r'([+\-]=|[$!|?+\-*/%^=><();]|>=|<=|<>|<<|>>|>>>|,)', Operator),
+        ],
+        'string': [
+            (r'[^"\n]+', String.Double),
+            (r'"', String.Double, '#pop'),
+            (r'\n', Error, 'root'),  # Unterminated string
+        ],
+    }
+
+    def analyse_text(text):
+        if text.startswith('10REM >') or text.startswith('REM >'):
+            return 0.9
diff --git a/tests/examplefiles/example.bbc b/tests/examplefiles/example.bbc
new file mode 100644
index 00000000..ebdb8537
--- /dev/null
+++ b/tests/examplefiles/example.bbc
@@ -0,0 +1,156 @@
+10REM >EIRC
+20REM The simplest IRC client you can write. Maybe.
+30REM (C) Justin Fletcher, 1998
+40:
+50END=PAGE+1024*16
+60REM Change these if you wish
+70host$="irc.stealth.net"
+80port=6667
+90nick$="eirc"
+100ourchan$="#acorn"
+110:
+120REM Start connecting to a host
+130SYS "ESocket_ConnectToHost",host$,port TO handle
+140REPEAT
+150 SYS "ESocket_CheckState",handle TO state
+160 IF state<-1 THENSYS "ESocket_Forget",handle:SYS "ESocket_DecodeState",state TO a$:ERROR 1,"Failed ("+a$+")"
+170UNTIL state=4
+180:
+190REM We are now connected
+200PRINT"Connected"
+210:
+220REM Log on to the server
+230SYS "ESocket_SendLine",handle,"USER "+nick$+" x x :"+nick$
+240SYS "ESocket_SendLine",handle,"NICK "+nick$
+250SYS "ESocket_SendLine",handle,"JOIN "+ourchan$
+260REM Install a monitor so that we don't waste time
+270SYS "ESocket_Monitor",0,handle TO monitor
+280SYS "ESocket_ResetMonitor",monitor,0 TO polladdr%
+290:
+300REM If we crash, we should tidy up after ourselves
+310ON ERROR SYS "XESocket_Forget",handle:SYS "XESocket_Forget",monitor:ERROR EXT ERR,REPORT$+" at line "+STR$ERL
+320:
+330REM Memory buffer for our data
+340bufsize%=1024
+350DIM buf% bufsize%
+360:
+370input$="":REM The input line
+380REPEAT
+390 REM In a taskwindow we should yield until there is data
+400 SYS "OS_UpCall",6,polladdr%
+410 IF !polladdr%<>0 THEN
+420  REM Reset the monitor for the time being
+430  SYS "ESocket_ResetMonitor",monitor,0 TO polladdr%
+440  REPEAT
+450   REM Read lines from the connection until this buffer is empty
+460   SYS "ESocket_ReadLine",handle,buf%,bufsize%,%100 TO ,str,len
+470   IF str<>0 AND $str<>"" THEN
+480    line$=$str
+490    IF LEFT$(line$,4)="PING" THEN
+500     REM Ping's must be replied to immediately
+510     SYS "ESocket_SendLine",handle,"PONG "+MID$(line$,6)
+520    ELSE
+530     REM Extract source info
+540     from$=MID$(LEFT$(line$,INSTR(line$+" "," ")-1),2)
+550     line$=MID$(line$,INSTR(line$+" "," ")+1)
+560     uid$=LEFT$(from$,INSTR(from$+"!","!")-1)
+570     com$=LEFT$(line$,INSTR(line$+" "," ")-1)
+580     line$=MID$(line$,INSTR(line$+" "," ")+1)
+590     REM remove the input line
+600     IF input$<>"" THENFORI=1TOLEN(input$):VDU127:NEXT
+610     CASE FNupper(com$) OF
+620      WHEN "PRIVMSG"
+630       REM Extract the destination
+640       chan$=LEFT$(line$,INSTR(line$+" "," ")-1)
+650       line$=MID$(line$,INSTR(line$+" "," ")+2):REM Skip :
+660       IF LEFT$(line$,1)=CHR$1 THEN
+670        REM CTCP, so respond to it
+680        line$=MID$(line$,2,LEN(line$)-2)
+690        com$=LEFT$(line$,INSTR(line$+" "," ")-1)
+700        line$=MID$(line$,INSTR(line$+" "," ")+1)
+710        CASE FNupper(com$) OF
+720         WHEN "PING"
+730          REM Ping lag timing
+740          line$="PONG "+line$
+750          PRINTuid$;" pinged us"
+760         WHEN "VERSION"
+770          REM Version checking
+780          line$="VERSION EIRC 1.00 (c) Justin Fletcher"
+790          PRINTuid$;" wanted our version"
+800         WHEN "ACTION"
+810          PRINT"* ";uid$;" ";line$
+820          line$=""
+830         OTHERWISE
+840          REM everything else is an error
+850          line$="ERRMSG "+com$+" not understood"
+860          PRINT"CTCP '";com$;"' from ";uid$;" (";line$;")"
+870        ENDCASE
+880        IF line$<>"" THEN
+890         SYS "ESocket_SendLine",handle,"NOTICE "+uid$+" :"+CHR$1+line$+CHR$1
+900        ENDIF
+910       ELSE
+920        REM Somebody said something...
+930        PRINT"<";uid$;"> ";FNsafe(line$)
+940       ENDIF
+950      WHEN "JOIN"
+960       REM We (or someone else) has joined the channel
+970       chan$=LEFT$(line$,INSTR(line$+" "," ")):REM Skip :
+980       IF LEFT$(chan$,1)=":" THENchan$=MID$(chan$,2)
+990       PRINTuid$;" has joined ";chan$
+1000      WHEN "PART"
+1010       REM Someone else has left the channel
+1020       chan$=LEFT$(line$,INSTR(line$+" "," ")-1)
+1030       IF LEFT$(chan$,1)=":" THENchan$=MID$(chan$,2)
+1040       PRINTuid$;" has left ";chan$
+1050      WHEN "QUIT"
+1060       REM Someone else has quit IRC
+1070       PRINTuid$;" quit IRC"
+1080      OTHERWISE
+1090       REM Some unknown command
+1100       PRINTuid$;":";com$;":";FNsafe(line$)
+1110     ENDCASE
+1120     REM Re-display our input line
+1130     PRINTinput$;
+1140    ENDIF
+1150   ENDIF
+1160  UNTIL str=0
+1170 ENDIF
+1180 b$=INKEY$(0)
+1190 IF b$<>"" THEN
+1200  CASE b$ OF
+1210   WHEN CHR$13
+1220    SYS "ESocket_SendLine",handle,"PRIVMSG "+ourchan$+" :"+input$
+1230    REM Remove the line
+1240    IF input$<>"" THENFORI=1TOLEN(input$):VDU127:NEXT
+1250    REM We said it...
+1260    PRINT"<"+nick$+"> ";input$
+1270    input$=""
+1280   WHEN CHR$127,CHR$8
+1290    REM Backspace
+1300    IF input$<>"" THENVDU127
+1310    input$=LEFT$(input$)
+1320   OTHERWISE
+1330    REM Ad to current input
+1340    input$+=b$
+1350    PRINTb$;
+1360  ENDCASE
+1370 ENDIF
+1380 REM Has the socket closed
+1390 SYS "ESocket_Closed",handle,%0 TO closed
+1400UNTIL closed
+1410SYS "ESocket_Forget",handle
+1420SYS "ESocket_Forget",monitor
+1430END
+1440:
+1450DEFFNupper(a$):LOCAL c$,b$,I
+1460FORI=1TOLEN(a$)
+1470c$=MID$(a$,I,1):IF c$>="a"ANDc$<="z"THENc$=CHR$(ASC(c$)-32)
+1480b$+=c$:NEXT:=b$
+1490
+1500REM Remove control codes
+1510DEFFNsafe(line$)
+1520LOCAL I
+1530FORI=1TOLEN(line$)
+1540 IF MID$(line$,I,1)<" " THENMID$(line$,I,1)="*"
+1550NEXT
+1560=line$