/*****************************************************************************
*
* Int2 AsnLexScan(aip, name)
* scans until name ::= to start reading asn.1 past garbage
*
*****************************************************************************/
static Int2 AsnLexScan (AsnIoPtr aip, CharPtr name)
{
register CharPtr pos;
Int2 token = 0;
register int linepos, len;
int done;
Boolean started = FALSE, matched_ref = FALSE;
if (aip->type_indent) /* only for start */
return ERROR_TOKEN;
if (! aip->bytes) /* no data loaded */
AsnIoGets(aip);
linepos = aip->linepos;
pos = aip->linebuf + linepos;
len = 0;
while (*pos == '\n' || *pos == '\r') /* skip empty lines */
{
pos = AsnIoGets(aip); /* get a line */
if (pos == NULL)
return EOF_TOKEN;
}
while (! started)
{
while ((* pos <= ' ') || ((*pos == '-') && (*(pos+1) == '-'))) /* skip leading white space */
{
if (*pos == '\n' || *pos == '\r')
{
pos = AsnIoGets(aip);
if (pos == NULL)
return EOF_TOKEN;
}
else if ((*pos == '-') && (*(pos+1) == '-')) /* skip comments */
{
pos += 2;
done = 0;
while (! done)
{
if ((*pos == '-') && (*(pos +1) == '-'))
{
pos += 2;
done = 1;
}
else if (*pos == '\n' || *pos == '\r')
done = 1;
else
pos++;
}
}
else
pos++;
}
aip->word = pos;
len = 0;
if (* pos == ':')
{
if ((*(pos + 1) == ':') && (*(pos + 2) == '='))
{
token = ISDEF;
pos += 2;
len = 3;
if (matched_ref)
started = TRUE;
else
{
AsnIoErrorMsg(aip, 59, *pos, aip->linenumber);
return ERROR_TOKEN;
}
}
else
matched_ref = FALSE;
}
else if (IS_UPPER(*pos)) /* a reference or keyword */
{
token = REF;
while ((IS_ALPHANUM(*pos)) || (*pos == '-'))
{
pos++; len++;
}
pos--; /* move back for increment at end */
aip->wordlen = len;
if (StrMatch(name, aip->word, (Int2)len))
matched_ref = TRUE;
else
matched_ref = FALSE;
}
else
matched_ref = FALSE;
pos++; /* move over last symbol */
}
/* found it , do normal return after ::= */
aip->linepos = pos - aip->linebuf;
aip->wordlen = len;
aip->token = token;
return token;
}
/*****************************************************************************
*
* Int2 AsnLexTWord(aip)
* reads words, punctuation, and asn keywords with 2 parts
* returns tokens defined at top
*
*****************************************************************************/
NLM_EXTERN Int2 AsnLexTWord (AsnIoPtr aip)
{
register CharPtr pos;
register int len;
Int1 state;
Int2 token, asntype, linepos;
int done;
Boolean first = FALSE, hitnewline = FALSE;
CharPtr commentptr;
if (! aip->bytes) /* no data loaded */
{
hitnewline = TRUE;
first = TRUE;
AsnIoGets(aip);
}
linepos = aip->linepos;
pos = aip->linebuf + linepos;
state = aip->state;
len = 0;
token = -1;
while (*pos == '\n' || *pos == '\r') /* skip empty lines */
{
hitnewline = TRUE;
pos = AsnIoGets(aip);
if (pos == NULL)
return EOF_TOKEN;
}
if (state == IN_STRING_STATE)
{
aip->word = pos;
if (* pos == '\"') /* end of string */
{
token = END_STRING;
pos++;
state = 0; /* reset state */
}
else
{
token = IN_STRING;
while ((* pos != '\"') && (* pos != '\n') && (* pos != '\r'))
{
pos++; len++;
}
if ((*pos != '\n') && (*pos != '\r') && (* (pos + 1) == '\"')) /* internal quote */
{
len++; /* include in previous string */
pos += 2; /* point to rest of string */
}
}
}
else if (state == IN_BITHEX_STATE)
{
aip->word = pos;
if (*pos == '\'') /* end of binhex */
{
state = 0; /* set to normal */
pos++; /* move past quote */
while (IS_WHITESP(*pos))
pos++;
if (* pos == 'H')
token = OCTETS;
else if (* pos == 'B')
token = ASNBITS;
else
{
AsnIoErrorMsg(aip, 58, aip->linenumber);
token = ERROR_TOKEN;
}
}
else
{
token = IN_BITHEX;
while ((* pos != '\'') && (* pos != '\n') && (* pos != '\r'))
{
pos++; len++;
}
}
}
else /* normal scanning */
{
done = 0;
while (! done)
{
while (* pos <= ' ') /* skip leading white space */
{
if (*pos == '\n' || *pos == '\r')
{
hitnewline = TRUE;
pos = AsnIoGets(aip);
if (pos == NULL)
return EOF_TOKEN;
}
else
pos++;
}
done = 1;
while (done && (*pos == '-') && (*(pos+1) == '-')) /* skip comments */
{
pos += 2;
if (first) /* could be revision */
{
first = FALSE;
if (StrMatch(asnwords[57], pos, 10)) /* $Revision: */
{
token = REVISION_TOKEN;
pos += 10;
while (IS_WHITESP(*pos))
pos++;
aip->word = pos;
while (IS_DIGIT(*pos)) /* eg. 1.2 */
{
len++;
pos++;
}
if (*pos == '.') /* take after . if present */
{
pos++;
len++;
while (IS_DIGIT(*pos))
{
len++;
pos++;
}
}
}
}
commentptr = pos;
done = 0;
while (! done) /* skip to end of comment */
{
if ((*pos == '-') && (*(pos +1) == '-'))
{
if (token != REVISION_TOKEN)
{
AsnLexTAddComment(commentptr, pos, aip);
if ((! hitnewline) && (aip->token != COMMENT_TOKEN))
token = COMMENT_TOKEN;
}
pos += 2;
done = 1;
}
else if (*pos == '\n' || *pos == '\r')
{
if (token != REVISION_TOKEN)
{
AsnLexTAddComment(commentptr, pos, aip);
if ((! hitnewline) && (aip->token != COMMENT_TOKEN))
token = COMMENT_TOKEN;
}
done = 1;
}
else
pos++;
}
if ((token == REVISION_TOKEN) || (token == COMMENT_TOKEN))
{
aip->linepos = pos - aip->linebuf;
aip->state = state;
aip->wordlen = len;
aip->token = token;
return token;
}
if (*pos <= ' ')
done = 0;
else
done = 1;
}
}
aip->word = pos;
if (* pos == '\"')
{
token = START_STRING;
state = IN_STRING_STATE;
}
else if (* pos == '\'')
{
token = START_BITHEX;
state = IN_BITHEX_STATE;
}
else if (* pos == ',')
token = COMMA;
else if (* pos == '{')
token = START_STRUCT;
else if (* pos == '}')
token = END_STRUCT;
else if (* pos == '[')
token = START_TAG;
else if (* pos == ']')
token = END_TAG;
else if (* pos == '(')
token = OPEN_PAREN;
else if (* pos == ')')
token = CLOSE_PAREN;
else if (* pos == ';')
token = SEMI_COLON;
else if (* pos == ':')
{
if ((*(pos + 1) == ':') && (*(pos + 2) == '='))
{
token = ISDEF;
pos += 2;
len = 3;
}
else
{
AsnIoErrorMsg(aip, 59, *pos, aip->linenumber);
token = ERROR_TOKEN;
}
}
else if (IS_UPPER(*pos)) /* a reference or keyword */
{
token = REF;
while ((IS_ALPHANUM(*pos)) || (*pos == '-'))
{
pos++; len++;
}
aip->wordlen = len;
asntype = AsnLexTMatchToken(aip); /* check types */
if (asntype) /* did it match ? */
{
if ((asntype > 27) && (asntype < 57)) /* not a primitive type */
{
token = asntype + 400; /* make a keyword type */
if (asntype == COMPS_TOKEN) /* COMPONENTS OF */
{
if ((*(pos + 1) == 'O') &&
(*(pos + 2) == 'F') &&
(IS_WHITESP(*(pos+3))))
{
pos += 3; /* move past OF */
len += 3;
}
else
AsnIoErrorMsg(aip, 89, aip->linenumber);
}
}
else if (asntype == 57) /* StringStore */
token = STRSTORE_TYPE;
else if (asntype == 59) /* BitInt */
token = BIGINT_TYPE;
else
{
switch (asntype)
{
case 3: /* BIT */
case 4: /* OCTET */
if (! StrMatch(asnwords[11], (pos+1), 6))
AsnIoErrorMsg(aip, 90, aip->linenumber);
pos += 7; /* move past STRING */
len += 7;
break;
case 11: /* SEQUENCE */
case 13: /* SET */
if ((*(pos + 1) == 'O') &&
(*(pos + 2) == 'F'))
{
asntype++; /* SET or SEQ OF */
pos += 3;
len += 3;
if (! IS_WHITESP(*pos))
AsnIoErrorMsg(aip, 91, aip->linenumber);
}
break;
case 6: /* OBJECT */
if ((! StrMatch(asnwords[55], (pos+1), 10))) /* IDENTIFIER */
AsnIoErrorMsg(aip, 92, aip->linenumber);
pos += 11;
len += 11;
break;
default:
break;
}
token = asntype + 300; /* change to point at type */
}
}
pos--; /* move back for increment at end */
len--;
}
else if (IS_LOWER(*pos)) /* an identifier or valuereference */
{
token = IDENT;
while ((IS_ALPHANUM(*pos)) || (*pos == '-'))
{
pos++; len++;
}
pos--; /* move back for increment at end */
len--;
}
else if ((IS_DIGIT(*pos)) || ((*pos == '-') && (IS_DIGIT(*(pos+1)))))
{
token = NUMBER;
if (*pos == '-')
{
pos++; len++;
}
while (IS_DIGIT(*pos))
{
pos++; len++;
}
pos--; /* move back for increment at end */
len--;
}
else
{
AsnIoErrorMsg(aip, 59, *pos, aip->linenumber);
token = ERROR_TOKEN;
}
len++; pos++; /* move over last symbol */
}
aip->linepos = pos - aip->linebuf;
aip->state = state;
aip->wordlen = len;
aip->token = token;
return token;
}
static Int2 AsnLexWordEx (AsnIoPtr aip, Uint1 fix_non_print)
{
register CharPtr pos;
Int1 state;
Int2 token;
register int linepos, len;
int done;
if (aip->tagsaved) /* had to read ahead */
{
aip->tagsaved = FALSE;
return aip->token;
}
if (! aip->bytes) /* no data loaded */
AsnIoGets(aip);
linepos = aip->linepos;
pos = aip->linebuf + linepos;
state = aip->state;
len = 0;
while (*pos == '\n' || *pos == '\r') /* skip empty lines */
{
pos = AsnIoGets(aip); /* get a line */
if (pos == NULL)
return EOF_TOKEN;
}
if (state == IN_STRING_STATE)
{
aip->word = pos;
if ((* pos == '\"') && (*(pos + 1) != '\"')) /* end of string */
{
token = END_STRING;
pos++;
state = 0; /* reset state */
}
else
{
token = IN_STRING;
while ((* pos != '\"') && (* pos != '\n') && (* pos != '\r'))
{
if ((fix_non_print != 2) && ((*pos < ' ') || (*pos > '~')))
{
done = (int)(*pos);
*pos = '\0';
if ((fix_non_print == 0) || (fix_non_print == 3))
{
AsnIoErrorMsg(aip, 106, done, aip->word);
}
done = 0;
*pos = '#'; /* replace with # */
}
pos++; len++;
}
if ((*pos != '\n') && (*pos != '\r') && (* (pos + 1) == '\"')) /* internal quote */
{
len++; /* include in previous string */
pos += 2; /* point to rest of string */
}
}
}
else if (state == IN_BITHEX_STATE)
{
aip->word = pos;
if (*pos == '\'') /* end of binhex */
{
state = 0; /* set to normal */
pos++; /* move past quote */
while (IS_WHITESP(*pos))
{
if (*pos == '\n' || *pos == '\r') /* skip empty lines */
{
pos = AsnIoGets(aip); /* get a line */
if (pos == NULL)
return EOF_TOKEN;
}
else
pos++;
}
if (* pos == 'H')
token = OCTETS;
else if (* pos == 'B')
token = ASNBITS;
else
{
AsnIoErrorMsg(aip, 58, aip->linenumber);
token = ERROR_TOKEN;
}
pos++; /* move past H or B */
}
else
{
token = IN_BITHEX;
while ((* pos != '\'') && (* pos != '\n') && (* pos != '\r'))
{
pos++; len++;
}
}
}
else /* normal scanning */
{
while ((* pos <= ' ') || ((*pos == '-') && (*(pos+1) == '-'))) /* skip leading white space */
{
if (*pos == '\n' || *pos == '\r')
{
pos = AsnIoGets(aip);
if (pos == NULL)
return EOF_TOKEN;
}
else if ((*pos == '-') && (*(pos+1) == '-')) /* skip comments */
{
pos += 2;
done = 0;
while (! done)
{
if ((*pos == '-') && (*(pos +1) == '-'))
{
pos += 2;
done = 1;
}
else if (*pos == '\n' || *pos == '\r')
done = 1;
else
pos++;
}
}
else
pos++;
}
aip->word = pos;
if (* pos == '\"')
{
token = START_STRING;
state = IN_STRING_STATE;
}
else if (* pos == '\'')
{
token = START_BITHEX;
state = IN_BITHEX_STATE;
}
else if (* pos == ',')
token = COMMA;
else if (* pos == '{')
token = START_STRUCT;
else if (* pos == '}')
token = END_STRUCT;
else if (* pos == '[')
token = START_TAG;
else if (* pos == ']')
token = END_TAG;
else if (* pos == '(')
token = OPEN_PAREN;
else if (* pos == ')')
token = CLOSE_PAREN;
else if (* pos == ';')
token = SEMI_COLON;
else if (* pos == ':')
{
if ((*(pos + 1) == ':') && (*(pos + 2) == '='))
{
token = ISDEF;
pos += 2;
len = 3;
}
else
{
AsnIoErrorMsg(aip, 59, *pos, aip->linenumber);
token = ERROR_TOKEN;
}
}
else if (IS_UPPER(*pos)) /* a reference or keyword */
{
token = REF;
while ((IS_ALPHANUM(*pos)) || (*pos == '-'))
{
pos++; len++;
}
pos--; /* move back for increment at end */
len--;
}
else if (IS_LOWER(*pos)) /* an identifier or valuereference */
{
token = IDENT;
while ((IS_ALPHANUM(*pos)) || (*pos == '-'))
{
pos++; len++;
}
pos--; /* move back for increment at end */
len--;
}
else if ((IS_DIGIT(*pos)) || ((*pos == '-') && (IS_DIGIT(*(pos+1)))))
{
token = NUMBER;
if (*pos == '-')
{
pos++; len++;
}
while (IS_DIGIT(*pos))
{
pos++; len++;
}
pos--; /* move back for increment at end */
len--;
}
else
{
AsnIoErrorMsg(aip, 59, *pos, aip->linenumber);
token = ERROR_TOKEN;
}
len++; pos++; /* move over last symbol */
}
aip->linepos = pos - aip->linebuf;
/******************** check on MSWIN
linepos = 0;
while (pos != linebuf)
{
linepos++;
pos++;
}
aip->linepos = linepos;
**********************/
aip->state = state;
aip->wordlen = len;
aip->token = token;
return token;
}
