#include <unictype.h>
#include <unistd.h>
#include <unistr.h>
-#include <uniwidth.h>
#include "language/command.h"
#include "language/lexer/macro.h"
location of the token in terms of the lex_source's buffer.
For a token produced through macro expansion, this is the entire macro
- call.
-
- src->tail <= line_pos <= token_pos <= src->head. */
- size_t token_pos; /* Start of token. */
+ call. */
+ size_t token_pos; /* Offset into src->buffer of token start. */
size_t token_len; /* Length of source for token in bytes. */
- size_t line_pos; /* Start of line containing token_pos. */
int first_line; /* Line number at token_pos. */
/* For a token obtained through macro expansion, this is just this token.
bool eof; /* True if T_STOP was read from 'reader'. */
/* Buffer of UTF-8 bytes. */
- char *buffer;
+ char *buffer; /* Source file contents. */
+ size_t length; /* Number of bytes filled. */
size_t allocated; /* Number of bytes allocated. */
- size_t tail; /* &buffer[0] offset into UTF-8 source. */
- size_t head; /* &buffer[head - tail] offset into source. */
- /* Positions in source file, tail <= pos <= head for each member here. */
+ /* Offsets into 'buffer'. */
size_t journal_pos; /* First byte not yet output to journal. */
size_t seg_pos; /* First byte not yet scanned as token. */
- size_t line_pos; /* First byte of line containing seg_pos. */
int n_newlines; /* Number of new-lines up to seg_pos. */
bool suppress_next_newline;
return 0;
else
{
- char *token_str = &src->buffer[token->token_pos - src->tail];
+ char *token_str = &src->buffer[token->token_pos];
return token->first_line + count_newlines (token_str, token->token_len) + 1;
}
}
static int
-count_columns (const char *s_, size_t length)
+lex_token_get_column__ (const struct lex_source *src, size_t offset)
{
- const uint8_t *s = CHAR_CAST (const uint8_t *, s_);
- int columns;
- size_t ofs;
- int mblen;
-
- columns = 0;
- for (ofs = 0; ofs < length; ofs += mblen)
- {
- ucs4_t uc;
-
- mblen = u8_mbtouc (&uc, s + ofs, length - ofs);
- if (uc != '\t')
- {
- int width = uc_width (uc, "UTF-8");
- if (width > 0)
- columns += width;
- }
- else
- columns = ROUND_UP (columns + 1, 8);
- }
-
- return columns + 1;
+ const char *newline = memrchr (src->buffer, '\n', offset);
+ size_t line_ofs = newline ? newline - src->buffer + 1 : 0;
+ return utf8_count_columns (&src->buffer[line_ofs], offset - line_ofs) + 1;
}
static int
lex_token_get_first_column (const struct lex_source *src,
const struct lex_token *token)
{
- return count_columns (&src->buffer[token->line_pos - src->tail],
- token->token_pos - token->line_pos);
+ return lex_token_get_column__ (src, token->token_pos);
}
static int
lex_token_get_last_column (const struct lex_source *src,
const struct lex_token *token)
{
- char *start, *end, *newline;
-
- start = &src->buffer[token->line_pos - src->tail];
- end = &src->buffer[(token->token_pos + token->token_len) - src->tail];
- newline = memrchr (start, '\n', end - start);
- if (newline != NULL)
- start = newline + 1;
- return count_columns (start, end - start);
+ return lex_token_get_column__ (src, token->token_pos + token->token_len);
}
static struct msg_location
struct lex_source *src = lex_source__ (lexer);
if (src != NULL && src->reader->error == LEX_ERROR_TERMINAL)
{
- src->head = src->tail = 0;
- src->journal_pos = src->seg_pos = src->line_pos = 0;
+ src->length = 0;
+ src->journal_pos = src->seg_pos = 0;
src->n_newlines = 0;
src->suppress_next_newline = false;
src->segmenter = segmenter_init (segmenter_get_mode (&src->segmenter),
}
}
\f
-static size_t
-lex_source_max_tail__ (const struct lex_source *src_)
-{
- struct lex_source *src = CONST_CAST (struct lex_source *, src_);
-
- assert (src->seg_pos >= src->line_pos);
- size_t max_tail = MIN (src->journal_pos, src->line_pos);
-
- /* Use the oldest token also. */
- struct lex_stage *stages[] = { &src->lookahead, &src->merge, &src->pp };
- for (size_t i = 0; i < sizeof stages / sizeof *stages; i++)
- if (!lex_stage_is_empty (stages[i]))
- {
- struct lex_token *first = lex_stage_first (stages[i]);
- assert (first->token_pos >= first->line_pos);
- return MIN (max_tail, first->line_pos);
- }
-
- return max_tail;
-}
-
static void
lex_source_expand__ (struct lex_source *src)
{
- if (src->head - src->tail >= src->allocated)
- {
- size_t max_tail = lex_source_max_tail__ (src);
- if (max_tail > src->tail)
- {
- /* Advance the tail, freeing up room at the head. */
- memmove (src->buffer, src->buffer + (max_tail - src->tail),
- src->head - max_tail);
- src->tail = max_tail;
- }
- else
- {
- /* Buffer is completely full. Expand it. */
- src->buffer = x2realloc (src->buffer, &src->allocated);
- }
- }
- else
- {
- /* There's space available at the head of the buffer. Nothing to do. */
- }
+ if (src->length >= src->allocated)
+ src->buffer = x2realloc (src->buffer, &src->allocated);
}
static void
{
lex_source_expand__ (src);
- size_t head_ofs = src->head - src->tail;
- size_t space = src->allocated - head_ofs;
+ size_t space = src->allocated - src->length;
enum prompt_style prompt = segmenter_get_prompt (&src->segmenter);
- size_t n = src->reader->class->read (src->reader, &src->buffer[head_ofs],
+ size_t n = src->reader->class->read (src->reader,
+ &src->buffer[src->length],
space, prompt);
assert (n <= space);
{
/* End of input. */
src->reader->eof = true;
- lex_source_expand__ (src);
return;
}
- src->head += n;
+ src->length += n;
}
- while (!memchr (&src->buffer[src->seg_pos - src->tail], '\n',
- src->head - src->seg_pos));
+ while (!memchr (&src->buffer[src->seg_pos], '\n',
+ src->length - src->seg_pos));
}
static struct lex_source *
{
size_t start = first->token_pos;
size_t end = last->token_pos + last->token_len;
- ds_put_substring (&s, ss_buffer (&src->buffer[start - src->tail],
- end - start));
+ ds_put_substring (&s, ss_buffer (&src->buffer[start], end - start));
}
else
{
size_t start = token0->token_pos;
size_t end = token1->token_pos + token1->token_len;
- return ss_buffer (&src->buffer[start - src->tail], end - start);
+ return ss_buffer (&src->buffer[start], end - start);
}
static void
lex_get_error (struct lex_source *src, const struct lex_token *token)
{
char syntax[64];
- str_ellipsize (ss_buffer (&src->buffer[token->token_pos - src->tail],
- token->token_len),
+ str_ellipsize (ss_buffer (&src->buffer[token->token_pos], token->token_len),
syntax, sizeof syntax);
struct string s = DS_EMPTY_INITIALIZER;
token->token = (struct token) { .type = T_STOP };
token->macro_rep = NULL;
token->ref_cnt = NULL;
- token->line_pos = src->line_pos;
token->token_pos = src->seg_pos;
if (src->reader->line_number > 0)
token->first_line = src->reader->line_number + src->n_newlines;
int seg_len;
for (;;)
{
- segment = &src->buffer[src->seg_pos - src->tail];
+ segment = &src->buffer[src->seg_pos];
seg_len = segmenter_push (&src->segmenter, segment,
- src->head - src->seg_pos,
+ src->length - src->seg_pos,
src->reader->eof, &seg_type);
if (seg_len >= 0)
break;
token->token_len = seg_len;
src->seg_pos += seg_len;
if (seg_type == SEG_NEWLINE)
- {
- src->line_pos = src->seg_pos;
- src->n_newlines++;
- }
+ src->n_newlines++;
/* Get a token from the segment. */
enum tokenize_result result = token_from_segment (
for (int i = 0; i < n_lines; i++)
{
/* Beginning of line. */
- const char *line = &src->buffer[src->journal_pos - src->tail];
+ const char *line = &src->buffer[src->journal_pos];
/* Calculate line length, including \n or \r\n end-of-line if present.
- We use src->head even though that may be beyond what we've actually
- converted to tokens (which is only through line_pos). That's because,
- if we're emitting the line due to SEG_END_COMMAND, we want to take the
- whole line through the newline, not just through the '.'. */
- size_t max_len = src->head - src->journal_pos;
+ We use src->length even though that may be beyond what we've actually
+ converted to tokens. That's because, if we're emitting the line due
+ to SEG_END_COMMAND, we want to take the whole line through the
+ newline, not just through the '.'. */
+ size_t max_len = src->length - src->journal_pos;
const char *newline = memchr (line, '\n', max_len);
size_t line_len = newline ? newline - line + 1 : max_len;
size_t end = t->token_pos + t->token_len;
const struct macro_token mt = {
.token = t->token,
- .syntax = ss_buffer (&src->buffer[start - src->tail], end - start),
+ .syntax = ss_buffer (&src->buffer[start], end - start),
};
const struct msg_location loc = lex_token_location (src, t, t);
n_call = macro_call_add (mc, &mt, &loc);
.token = expansion.mts[i].token,
.token_pos = c0->token_pos,
.token_len = (c1->token_pos + c1->token_len) - c0->token_pos,
- .line_pos = c0->line_pos,
.first_line = c0->first_line,
.macro_rep = macro_rep,
.ofs = ofs[i],
.token = out,
.token_pos = first->token_pos,
.token_len = (last->token_pos - first->token_pos) + last->token_len,
- .line_pos = first->line_pos,
.first_line = first->first_line,
/* This works well if all the tokens were not expanded from macros,
projects / pspp / blobdiff