S_DO_REPEAT_1,
S_DO_REPEAT_2,
S_DO_REPEAT_3,
+ S_DEFINE_1,
+ S_DEFINE_2,
+ S_DEFINE_3,
+ S_DEFINE_4,
+ S_DEFINE_5,
S_BEGIN_DATA_1,
S_BEGIN_DATA_2,
S_BEGIN_DATA_3,
}
/* fall through */
- case SEG_MACRO_ID:
case SEG_NUMBER:
case SEG_QUOTED_STRING:
case SEG_HEX_STRING:
case SEG_COMMENT_COMMAND:
case SEG_DO_REPEAT_COMMAND:
case SEG_INLINE_DATA:
+ case SEG_MACRO_ID:
+ case SEG_MACRO_BODY:
case SEG_START_DOCUMENT:
case SEG_DOCUMENT:
case SEG_START_COMMAND:
return ofs;
}
}
+ else if (lex_id_match_n (ss_cstr ("DEFINE"), word, 6))
+ {
+ s->state = S_DEFINE_1;
+ return ofs;
+ }
else if (lex_id_match (ss_cstr ("FILE"), word))
{
char id[16];
}
}
+/* We are segmenting a DEFINE command, which consists of:
+
+ - The DEFINE keyword.
+
+ - Anything but "(".
+
+ - "(" followed by a sequence of tokens possibly including balanced parentheses
+ up to a final ")".
+
+ - A newline.
+
+ - A sequence of lines that don't start with "!ENDDEFINE", one string per line,
+ each ending in a newline.
+
+ - "!ENDDEFINE".
+
+ */
+static int
+segmenter_parse_define_1__ (struct segmenter *s,
+ const char *input, size_t n, bool eof,
+ enum segment_type *type)
+{
+ int ofs = segmenter_subparse (s, input, n, eof, type);
+ if (ofs < 0)
+ return -1;
+
+ if (*type == SEG_SEPARATE_COMMANDS
+ || *type == SEG_END_COMMAND
+ || *type == SEG_START_COMMAND)
+ {
+ /* The DEFINE command is malformed because we reached its end without
+ ever hitting a "(" token. Transition back to general parsing. */
+ s->state = S_GENERAL;
+ return ofs;
+ }
+ else if (*type == SEG_PUNCT && input[0] == '(')
+ {
+ s->state = S_DEFINE_2;
+ s->nest = 1;
+ return ofs;
+ }
+
+ return ofs;
+}
+
+static int
+segmenter_parse_define_2__ (struct segmenter *s,
+ const char *input, size_t n, bool eof,
+ enum segment_type *type)
+{
+ int ofs = segmenter_subparse (s, input, n, eof, type);
+ if (ofs < 0)
+ return -1;
+
+ if (*type == SEG_SEPARATE_COMMANDS
+ || *type == SEG_END_COMMAND
+ || *type == SEG_START_COMMAND)
+ {
+ /* The DEFINE command is malformed because we reached its end before
+ closing the set of parentheses. Transition back to general
+ parsing. */
+ s->state = S_GENERAL;
+ return ofs;
+ }
+ else if (*type == SEG_PUNCT && input[0] == '(')
+ {
+ s->nest++;
+ return ofs;
+ }
+ else if (*type == SEG_PUNCT && input[0] == ')')
+ {
+ s->nest--;
+ if (!s->nest)
+ s->state = S_DEFINE_3;
+ return ofs;
+ }
+
+ return ofs;
+}
+
+static int
+segmenter_parse_define_3__ (struct segmenter *s,
+ const char *input, size_t n, bool eof,
+ enum segment_type *type)
+{
+ int ofs = segmenter_subparse (s, input, n, eof, type);
+ if (ofs < 0)
+ return -1;
+
+ if (*type == SEG_NEWLINE)
+ s->state = S_DEFINE_4;
+
+ return ofs;
+}
+
+static bool
+is_enddefine (const char *input, size_t n)
+{
+ int ofs = skip_spaces_and_comments (input, n, true, 0);
+ assert (ofs >= 0);
+
+ const struct substring enddefine = ss_cstr ("!ENDDEFINE");
+ if (n - ofs < enddefine.length)
+ return false;
+
+ if (!ss_equals_case (ss_buffer (input + ofs, enddefine.length), enddefine))
+ return false;
+
+ if (ofs + enddefine.length >= n)
+ return true;
+
+ const uint8_t *u_input = CHAR_CAST (const uint8_t *, input);
+ ucs4_t uc;
+ u8_mbtouc (&uc, u_input + ofs, n - ofs);
+ return uc == '.' || !lex_uc_is_idn (uc);
+}
+
+static int
+segmenter_parse_define_4__ (struct segmenter *s,
+ const char *input, size_t n, bool eof,
+ enum segment_type *type)
+{
+ int ofs;
+
+ ofs = segmenter_parse_full_line__ (input, n, eof, type);
+ if (ofs < 0)
+ return -1;
+ else if (is_enddefine (input, ofs))
+ {
+ s->state = S_GENERAL;
+ s->substate = SS_START_OF_COMMAND | SS_START_OF_LINE;
+ return segmenter_push (s, input, n, eof, type);
+ }
+ else
+ {
+ *type = SEG_MACRO_BODY;
+ s->state = S_DEFINE_5;
+ return input[ofs - 1] == '\n' ? 0 : ofs;
+ }
+}
+
+static int
+segmenter_parse_define_5__ (struct segmenter *s,
+ const char *input, size_t n, bool eof,
+ enum segment_type *type)
+{
+ int ofs;
+
+ ofs = segmenter_parse_newline__ (input, n, eof, type);
+ if (ofs < 0)
+ return -1;
+
+ s->state = S_DEFINE_4;
+ return ofs;
+}
+
static int
segmenter_parse_begin_data_1__ (struct segmenter *s,
const char *input, size_t n, bool eof,
case S_DO_REPEAT_3:
return segmenter_parse_do_repeat_3__ (s, input, n, eof, type);
+ case S_DEFINE_1:
+ return segmenter_parse_define_1__ (s, input, n, eof, type);
+ case S_DEFINE_2:
+ return segmenter_parse_define_2__ (s, input, n, eof, type);
+ case S_DEFINE_3:
+ return segmenter_parse_define_3__ (s, input, n, eof, type);
+ case S_DEFINE_4:
+ return segmenter_parse_define_4__ (s, input, n, eof, type);
+ case S_DEFINE_5:
+ return segmenter_parse_define_5__ (s, input, n, eof, type);
+
case S_BEGIN_DATA_1:
return segmenter_parse_begin_data_1__ (s, input, n, eof, type);
case S_BEGIN_DATA_2:
case S_DO_REPEAT_3:
return PROMPT_DO_REPEAT;
+ case S_DEFINE_1:
+ case S_DEFINE_2:
+ case S_DEFINE_3:
+ return s->substate & SS_START_OF_COMMAND ? PROMPT_FIRST : PROMPT_LATER;
+ case S_DEFINE_4:
+ case S_DEFINE_5:
+ return PROMPT_DEFINE;
+
case S_BEGIN_DATA_1:
return PROMPT_FIRST;
case S_BEGIN_DATA_2:
PSPP_CHECK_SCAN([-b])
AT_CLEANUP
\f
+AT_SETUP([DEFINE command - simple])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1()
+var1 var2 var3
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+RPAREN
+SKIP
+STRING "var1 var2 var3"
+SKIP
+MACRO_ID "!enddefine"
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - empty])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1()
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+RPAREN
+SKIP
+MACRO_ID "!enddefine"
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - arguments])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1(a(), b(), c())
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+ID "a"
+LPAREN
+RPAREN
+COMMA
+SKIP
+ID "b"
+LPAREN
+RPAREN
+COMMA
+SKIP
+ID "c"
+LPAREN
+RPAREN
+RPAREN
+SKIP
+MACRO_ID "!enddefine"
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - multiline arguments])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1(
+ a(), b(
+ ),
+ c()
+)
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+SKIP
+SKIP
+ID "a"
+LPAREN
+RPAREN
+COMMA
+SKIP
+ID "b"
+LPAREN
+SKIP
+SKIP
+RPAREN
+COMMA
+SKIP
+SKIP
+ID "c"
+LPAREN
+RPAREN
+SKIP
+RPAREN
+SKIP
+MACRO_ID "!enddefine"
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - arguments start on second line])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1
+(x,y,z
+)
+content 1
+content 2
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+SKIP
+LPAREN
+ID "x"
+COMMA
+ID "y"
+COMMA
+ID "z"
+SKIP
+RPAREN
+SKIP
+STRING "content 1"
+SKIP
+STRING "content 2"
+SKIP
+MACRO_ID "!enddefine"
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 1])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+ENDCMD
+SKIP
+ID "data"
+SKIP
+ID "list"
+SKIP
+SLASH
+ID "x"
+SKIP
+POS_NUM 1
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 2])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1
+x.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+SKIP
+ID "x"
+ENDCMD
+SKIP
+ID "data"
+SKIP
+ID "list"
+SKIP
+SLASH
+ID "x"
+SKIP
+POS_NUM 1
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 3])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1(.
+x.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+ENDCMD
+SKIP
+ID "x"
+ENDCMD
+SKIP
+ID "data"
+SKIP
+ID "list"
+SKIP
+SLASH
+ID "x"
+SKIP
+POS_NUM 1
+ENDCMD
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - missing !ENDDEFINE])
+AT_KEYWORDS([scan])
+AT_DATA([input], [dnl
+define !macro1().
+content line 1
+content line 2
+])
+AT_DATA([expout-base], [dnl
+ID "define"
+SKIP
+MACRO_ID "!macro1"
+LPAREN
+RPAREN
+ENDCMD
+SKIP
+STRING "content line 1"
+SKIP
+STRING "content line 2"
+-SKIP
+STOP
+])
+PSPP_CHECK_SCAN([-i])
+AT_CLEANUP
+\f
AT_SETUP([batch mode])
AT_KEYWORDS([scan])
AT_DATA([input], [dnl
PSPP_CHECK_SEGMENT([-b])
AT_CLEANUP
\f
+AT_SETUP([DEFINE command - simple])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1()
+var1 var2 var3
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+punct )
+newline \n (DEFINE)
+
+macro_body var1_var2_var3
+newline \n (DEFINE)
+
+macro_id !enddefine
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - empty])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1()
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+punct )
+newline \n (DEFINE)
+
+macro_id !enddefine
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - arguments])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1(a(), b(), c())
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+identifier a
+punct (
+punct )
+punct , space
+identifier b
+punct (
+punct )
+punct , space
+identifier c
+punct (
+punct )
+punct )
+newline \n (DEFINE)
+
+macro_id !enddefine
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - multiline arguments])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1(
+ a(), b(
+ ),
+ c()
+)
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+newline \n (later)
+
+spaces __
+identifier a
+punct (
+punct )
+punct , space
+identifier b
+punct (
+newline \n (later)
+
+spaces __
+punct )
+punct ,
+newline \n (later)
+
+spaces __
+identifier c
+punct (
+punct )
+newline \n (later)
+
+punct )
+newline \n (DEFINE)
+
+macro_id !enddefine
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - arguments start on second line])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1
+(x,y,z
+)
+content 1
+content 2
+!enddefine.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+newline \n (later)
+
+punct (
+identifier x
+punct ,
+identifier y
+punct ,
+identifier z
+newline \n (later)
+
+punct )
+newline \n (DEFINE)
+
+macro_body content_1
+newline \n (DEFINE)
+
+macro_body content_2
+newline \n (DEFINE)
+
+macro_id !enddefine
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 1])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+end_command .
+newline \n (first)
+
+identifier data space
+identifier list space
+punct /
+identifier x space
+number 1
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 2])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1
+x.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+newline \n (later)
+
+identifier x
+end_command .
+newline \n (first)
+
+identifier data space
+identifier list space
+punct /
+identifier x space
+number 1
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - early end of command 3])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1(.
+x.
+data list /x 1.
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+end_command .
+newline \n (first)
+
+identifier x
+end_command .
+newline \n (first)
+
+identifier data space
+identifier list space
+punct /
+identifier x space
+number 1
+end_command .
+-newline \n (first)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
+AT_SETUP([DEFINE command - missing !ENDDEFINE])
+AT_KEYWORDS([segment])
+AT_DATA([input], [dnl
+define !macro1().
+content line 1
+content line 2
+])
+AT_DATA([expout-base], [dnl
+identifier define space
+macro_id !macro1
+punct (
+punct )
+end_command .
+newline \n (DEFINE)
+
+macro_body content_line_1
+newline \n (DEFINE)
+
+macro_body content_line_2
+-newline \n (DEFINE)
+-
+end
+])
+PSPP_CHECK_SEGMENT([-i])
+AT_CLEANUP
+\f
AT_SETUP([batch mode])
AT_KEYWORDS([segment])
AT_DATA([input], [dnl
projects / pspp / commitdiff