static struct high_state *find_state(struct high_syntax *syntax, unsigned char *name)
{
struct high_state *state;
/* Find state */
state = htfind(syntax->ht_states, name);
/* It doesn't exist, so create it */
if (!state)
{
int y;
state = joe_malloc(sizeof(struct high_state));
state->name = zdup(name);
state->no = syntax->nstates;
state->color = FG_WHITE;
/* Expand the state table if necessary */
if (syntax->nstates == syntax->szstates)
{
syntax->states = joe_realloc(syntax->states, sizeof(struct high_state *) * (syntax->szstates *= 2));
}
syntax->states[syntax->nstates++] = state;
for (y = 0; y != 256; ++y)
{
state->cmd[y] = &syntax->default_cmd;
}
state->delim = 0;
htadd(syntax->ht_states, state->name, state);
}
return state;
}
unsigned char *my_gettext(unsigned char *s)
{
if (gettext_ht) {
unsigned char *r = htfind(gettext_ht, s);
if (r)
s = r;
}
if (s[0] == '|')
return ignore_prefix(s);
else
return s;
}
const char *my_gettext(const char *s)
{
if (gettext_ht) {
const char *r = (const char *)htfind(gettext_ht, s);
if (r)
s = r;
}
if (s[0] == '|')
return ignore_prefix(s);
else
return s;
}
/* count all 'buf' substrings of length 'len', return count for buf[0..len-1] */
static void do_cnt(const struct buf *seq, unsigned len, char *buf)
{
const char *Match = "GGTATTTTAATTTATAGT";
unsigned long cnt = 0;
struct ht t;
htinit(&t, HT_BINS, MIN(dna_combo(len), MAX_ENTRIES));
freq_build(&t, seq, len);
struct htentry *e = htfind(&t, Match, len, index(Match, len));
if (e)
cnt = e->val.cnt;
htfree(&t);
sprintf(buf, "%lu\t%.*s\n", cnt, len, Match);
}
HIGHLIGHT_STATE parse(struct high_syntax *const syntax, line_desc *const ld, HIGHLIGHT_STATE h_state, const bool utf8)
{
struct high_frame *stack = h_state.stack;
struct high_state *h = (stack ? stack->syntax : syntax)->states[h_state.state];
/* Current state */
unsigned char buf[24]; /* Name buffer (trunc after 23 characters) */
unsigned char lbuf[24]; /* Lower case version of name buffer */
unsigned char lsaved_s[24]; /* Lower case version of delimiter match buffer */
int buf_idx = 0; /* Index into buffer */
int c; /* Current character */
int c_len; /* Character length in bytes */
uint32_t *attr = attr_buf;
uint32_t *attr_end = attr_buf + attr_size;
int buf_en = 0; /* Set for name buffering */
int ofst = 0; /* record offset after we've stopped buffering */
int mark1 = 0; /* offset to mark start from current pos */
int mark2 = 0; /* offset to mark end from current pos */
int mark_en = 0; /* set if marking */
int recolor_delimiter_or_keyword;
const unsigned char *p = (const unsigned char *)ld->line;
unsigned char *q = (unsigned char *)(ld->line + ld->line_len);
buf[0] = 0; /* Forgot this originally... took 5 months to fix! */
/* Get next character */
/* Una iterazione in più: aggiungo '\n' come ultimo carattere. */
while (p <= q) /* On the last itteration, process the virtual '\n' character. */
{
struct high_cmd *cmd;
struct high_cmd *kw_cmd;
int x;
if (p == q)
{
c = '\n';
}
else
{
c = utf8 ? get_char((const char *)p, ENC_UTF8) : *p;
}
c_len = utf8 ? utf8seqlen(c) : 1;
p += c_len;
/* Hack so we can have UTF-8 characters without crashing */
if (c < 0 || c > 255)
{
c = 0x1F;
}
/* Create or expand attribute array if necessary */
if (attr == attr_end)
{
if (!attr_buf)
{
attr_size = 1024;
attr_buf = joe_malloc(sizeof(int) * attr_size);
attr = attr_buf;
}
else
{
attr_buf = joe_realloc(attr_buf, sizeof(int) * (attr_size * 2));
attr = attr_buf + attr_size;
attr_size *= 2;
}
attr_end = attr_buf + attr_size;
}
/* Advance to next attribute position (note attr[-1] below) */
attr++;
/* Loop while noeat */
do
{
/* Color with current state */
attr[-1] = h->color;
/* Get command for this character */
if (h->delim && c == h_state.saved_s[0] && h_state.saved_s[1] == 0)
{
cmd = h->delim;
}
else
{
cmd = h->cmd[c];
}
/* Lowerize strings for case-insensitive matching */
if (cmd->ignore)
{
zcpy(lbuf, buf);
lowerize(lbuf);
if (cmd->delim)
{
zcpy(lsaved_s, h_state.saved_s);
lowerize(lsaved_s);
}
}
/* Check for delimiter or keyword matches */
recolor_delimiter_or_keyword = 0;
if (cmd->delim && (cmd->ignore ? !zcmp(lsaved_s, lbuf) : !zcmp(h_state.saved_s, buf)))
{
cmd = cmd->delim;
recolor_delimiter_or_keyword = 1;
}
else if (cmd->keywords && (cmd->ignore ? (kw_cmd = htfind(cmd->keywords, lbuf)) : (kw_cmd = htfind(cmd->keywords, buf))))
{
cmd = kw_cmd;
recolor_delimiter_or_keyword = 1;
}
/* Determine new state */
if (cmd->call)
{
/* Call */
struct high_frame **frame_ptr = stack ? &stack->child : &syntax->stack_base;
/* Search for an existing stack frame for this call */
while (*frame_ptr && !((*frame_ptr)->syntax == cmd->call && (*frame_ptr)->return_state == cmd->new_state))
{
frame_ptr = &(*frame_ptr)->sibling;
}
if (*frame_ptr)
{
stack = *frame_ptr;
}
else
{
struct high_frame *frame = joe_malloc(sizeof(struct high_frame));
frame->parent = stack;
frame->child = 0;
frame->sibling = 0;
frame->syntax = cmd->call;
frame->return_state = cmd->new_state;
*frame_ptr = frame;
stack = frame;
++stack_count;
}
h = stack->syntax->states[0];
}
else if (cmd->rtn)
{
/* Return */
if (stack)
{
h = stack->return_state;
stack = stack->parent;
}
else
/* Not in a subroutine, so ignore the return */
{
h = cmd->new_state;
}
}
else if (cmd->reset)
{
/* Reset the state and call stack */
h = syntax->states[0];
stack = syntax->stack_base;
}
else
{
/* Normal edge */
h = cmd->new_state;
}
/* Recolor if necessary */
if (recolor_delimiter_or_keyword)
for (x = -(buf_idx + 1); x < -1; ++x)
{
attr[x - ofst] = h->color;
}
for (x = cmd->recolor; x < 0; ++x)
if (attr + x >= attr_buf)
{
attr[x] = h->color;
}
/* Mark recoloring */
if (cmd->recolor_mark)
for (x = -mark1; x < -mark2; ++x)
{
attr[x] = h->color;
}
/* Save string? */
if (cmd->save_s)
{
zcpy(h_state.saved_s, buf);
}
/* Save character? */
if (cmd->save_c)
{
h_state.saved_s[1] = 0;
if (c == '<')
{
h_state.saved_s[0] = '>';
}
else if (c == '(')
{
h_state.saved_s[0] = ')';
}
else if (c == '[')
{
h_state.saved_s[0] = ']';
}
else if (c == '{')
{
h_state.saved_s[0] = '}';
}
else if (c == '`')
{
h_state.saved_s[0] = '\'';
}
else
{
h_state.saved_s[0] = c;
}
}
/* Start buffering? */
if (cmd->start_buffering)
{
buf_idx = 0;
buf_en = 1;
ofst = 0;
}
/* Stop buffering? */
if (cmd->stop_buffering)
{
buf_en = 0;
}
/* Set mark begin? */
if (cmd->start_mark)
{
mark2 = 1;
mark1 = 1;
mark_en = 1;
}
/* Set mark end? */
if (cmd->stop_mark)
{
mark_en = 0;
mark2 = 1;
}
}
while (cmd->noeat);
/* Save character in buffer */
if (buf_idx < 23 && buf_en)
{
buf[buf_idx++] = c;
}
if (!buf_en)
{
++ofst;
}
buf[buf_idx] = 0;
/* Update mark pointers */
++mark1;
if (!mark_en)
{
++mark2;
}
/* if(c=='\n')
break;*/
}
/* Return new state */
h_state.stack = stack;
h_state.state = h->no;
attr_len = attr - attr_buf - 1; /* -1 because of the fake newline. */
return h_state;
}
CMD *findcmd(unsigned char *s)
{
if (!cmdhash)
izcmds();
return (CMD *) htfind(cmdhash, s);
}