parsing prefix version (binary operators) of msph
This commit is contained in:
parent
7e5f080282
commit
17be15d7b5
12 changed files with 617 additions and 92 deletions
130
src/msph/token.c
130
src/msph/token.c
|
|
@ -8,7 +8,6 @@
|
|||
#include "msph/err.h"
|
||||
#include "msph/token.h"
|
||||
|
||||
|
||||
struct msph_matcher {
|
||||
size_t off;
|
||||
size_t matchlen;
|
||||
|
|
@ -98,6 +97,8 @@ static int tok_match(struct msph_ctx *, struct msph_token_src *,
|
|||
struct msph_matcher *);
|
||||
static int tok_commit(struct msph_ctx *, struct msph_token_src *,
|
||||
struct msph_matcher *, struct msph_token *);
|
||||
static void tok_update_pos(struct msph_ctx *, struct msph_token_src *,
|
||||
struct msph_matcher *m);
|
||||
static int char_at(struct msph_ctx *, struct msph_token_src *, size_t,
|
||||
char *);
|
||||
static int fromcbuf_charcpy(char *, const char *, size_t, size_t, size_t);
|
||||
|
|
@ -116,8 +117,9 @@ void msph_ctx_init(struct msph_ctx *ctx)
|
|||
}
|
||||
|
||||
struct msph_token_stream *
|
||||
msph_token_stream_file(struct msph_ctx *ctx, FILE *f)
|
||||
msph_token_stream_file(struct msph_ctx *ctx, const char *name, FILE *f)
|
||||
{
|
||||
size_t res;
|
||||
struct msph_token_stream *ret;
|
||||
|
||||
if (ctx == NULL || f == NULL) {
|
||||
|
|
@ -132,7 +134,13 @@ msph_token_stream_file(struct msph_ctx *ctx, FILE *f)
|
|||
}
|
||||
|
||||
ret->ctx = ctx;
|
||||
if ((res = strlcpy(ret->name, name, BUF_LEN(ret->name)))
|
||||
>= BUF_LEN(ret->name)) {
|
||||
MSPH_ERR(ctx, MSPH_ERR_TOOLONG);
|
||||
goto err;
|
||||
}
|
||||
ret->src.type = MSPH_TOKEN_SRC_FILE;
|
||||
ret->src.pos = (struct msph_text_pos) { .line = 1, .col = 1 };
|
||||
ret->src.inner.file.f = f;
|
||||
ret->src.inner.file.pos = 0;
|
||||
ret->src.inner.file.end = 0;
|
||||
|
|
@ -149,8 +157,10 @@ err:
|
|||
}
|
||||
|
||||
struct msph_token_stream *
|
||||
msph_token_stream_frombuf(struct msph_ctx *ctx, const char *buf, size_t len)
|
||||
msph_token_stream_frombuf(struct msph_ctx *ctx, const char *name,
|
||||
const char *buf, size_t len)
|
||||
{
|
||||
size_t res;
|
||||
struct msph_token_stream *ret;
|
||||
|
||||
if ((ret = calloc(1, sizeof(struct msph_token_stream))) == NULL) {
|
||||
|
|
@ -159,12 +169,23 @@ msph_token_stream_frombuf(struct msph_ctx *ctx, const char *buf, size_t len)
|
|||
}
|
||||
|
||||
ret->ctx = ctx;
|
||||
if ((res = strlcpy(ret->name, name, BUF_LEN(ret->name)))
|
||||
>= BUF_LEN(ret->name)) {
|
||||
MSPH_ERR(ctx, MSPH_ERR_TOOLONG);
|
||||
goto err;
|
||||
}
|
||||
ret->src.type = MSPH_TOKEN_SRC_STR;
|
||||
ret->src.pos = (struct msph_text_pos) { .line = 1, .col = 1 };
|
||||
ret->src.inner.str.s = buf;
|
||||
ret->src.inner.str.len = strnlen(buf, len);
|
||||
ret->src.inner.str.pos = 0;
|
||||
|
||||
return (ret);
|
||||
|
||||
err:
|
||||
free(ret);
|
||||
|
||||
return (NULL);
|
||||
}
|
||||
|
||||
ssize_t
|
||||
|
|
@ -244,7 +265,9 @@ msph_token_stream_close(struct msph_token_stream *s)
|
|||
return (ret);
|
||||
}
|
||||
|
||||
/* -1 on error or num tokens read */
|
||||
/* read at most n tokens from s into p.
|
||||
* return -1 on error, or num tokens read
|
||||
*/
|
||||
ssize_t
|
||||
msph_token_stream_read(struct msph_token *ptr, size_t n,
|
||||
struct msph_token_stream *s)
|
||||
|
|
@ -279,6 +302,7 @@ read_single_tok(struct msph_token *ptr, struct msph_token_stream *s)
|
|||
/* Skipping whitespace */
|
||||
if (tok_match(ctx, src, &wspace) == -1)
|
||||
return (-1);
|
||||
SPHO_DEBUG_PRINT("wspace.matchlen=%zu\n", wspace.matchlen);
|
||||
if (wspace.matchlen > 0 &&
|
||||
tok_commit(ctx, src, &wspace, NULL) == -1)
|
||||
return (-1);
|
||||
|
|
@ -329,42 +353,34 @@ msph_token_stream_eof(struct msph_token_stream *s)
|
|||
}
|
||||
|
||||
struct msph_token *
|
||||
msph_token_create(struct msph_ctx *ctx, int type, union msph_token_data *data)
|
||||
msph_token_copy(struct msph_ctx *ctx, struct msph_token *token)
|
||||
{
|
||||
size_t i;
|
||||
struct msph_token *tok;
|
||||
struct msph_token *copy;
|
||||
struct msph_token_info *info;
|
||||
|
||||
info = NULL;
|
||||
|
||||
for (i = 0; token_info[i].type != TOK_END; i++) {
|
||||
if (token_info[i].type == type) {
|
||||
if (token_info[i].type == token->type) {
|
||||
info = &token_info[i];
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (info == NULL) {
|
||||
MSPH_ERR_INFO(ctx, MSPH_ERR_TOKEN_INVAL, type);
|
||||
MSPH_ERR_INFO(ctx, MSPH_ERR_TOKEN_INVAL, token->type);
|
||||
return (NULL);
|
||||
}
|
||||
|
||||
if ((tok = malloc(sizeof(*tok))) == NULL) {
|
||||
if ((copy = malloc(sizeof(*copy))) == NULL) {
|
||||
MSPH_ERR(ctx, MSPH_ERR_SYS);
|
||||
return (NULL);
|
||||
}
|
||||
|
||||
tok->type = type;
|
||||
memcpy(copy, token, sizeof(*copy));
|
||||
|
||||
switch (type) {
|
||||
case TOK_IDENT:
|
||||
memcpy(&tok->data, data, sizeof(*data));
|
||||
break;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
|
||||
return (tok);
|
||||
return (copy);
|
||||
}
|
||||
|
||||
static ssize_t
|
||||
|
|
@ -404,8 +420,6 @@ src_file_fill_buf(struct msph_ctx *ctx, struct msph_token_src_file *file)
|
|||
|
||||
} while (file->end != file->pos);
|
||||
|
||||
SPHO_DEBUG_PRINT("src_file_fill_buf: read %zd\n", ret);
|
||||
|
||||
return (ret);
|
||||
}
|
||||
|
||||
|
|
@ -414,20 +428,16 @@ static int
|
|||
file_char_at(struct msph_ctx *ctx, struct msph_token_src *src, size_t i,
|
||||
char *out)
|
||||
{
|
||||
int ret;
|
||||
ssize_t fill;
|
||||
struct msph_token_src_file *file;
|
||||
|
||||
SPHO_PRECOND(src != NULL);
|
||||
SPHO_PRECOND(src->type == MSPH_TOKEN_SRC_FILE);
|
||||
|
||||
ret = 0;
|
||||
file = &src->inner.file;
|
||||
fill = 0;
|
||||
|
||||
do {
|
||||
SPHO_DEBUG_PRINT("want to read %zu, valid range (%zu, %zu)\n",
|
||||
(file->pos + i) % BUF_LEN(file->buf), file->pos, file->end);
|
||||
/* simplest case */
|
||||
if (file->pos + i < file->end) {
|
||||
*out = file->buf[file->pos + i];
|
||||
|
|
@ -445,9 +455,9 @@ file_char_at(struct msph_ctx *ctx, struct msph_token_src *src, size_t i,
|
|||
if ((fill = src_file_fill_buf(ctx, file)) == -1)
|
||||
return (-1);
|
||||
|
||||
} while (fill > 0 && ret++);
|
||||
} while (fill > 0);
|
||||
|
||||
return (ret);
|
||||
return (0);
|
||||
}
|
||||
|
||||
static int
|
||||
|
|
@ -474,7 +484,34 @@ char_at(struct msph_ctx *ctx, struct msph_token_src *src, size_t i, char *out)
|
|||
break;
|
||||
}
|
||||
|
||||
SPHO_DEBUG_PRINT("char_at: ret=%d, *out=%c\n", ret, *out);
|
||||
#ifdef SPHO_ENABLE_DEBUG_PRINT
|
||||
if (isspace(*out)) {
|
||||
const char *charrep;
|
||||
switch (*out) {
|
||||
case '\n':
|
||||
charrep = "\\n";
|
||||
break;
|
||||
case '\t':
|
||||
charrep = "\\t";
|
||||
break;
|
||||
case '\r':
|
||||
charrep = "\\r";
|
||||
break;
|
||||
case '\v':
|
||||
charrep = "\\v";
|
||||
break;
|
||||
case '\f':
|
||||
charrep = "\\f";
|
||||
break;
|
||||
default:
|
||||
charrep = "WOOOOOOOOOOPS";
|
||||
break;
|
||||
}
|
||||
SPHO_DEBUG_PRINT("char_at: ret=%d, *out=%s\n", ret, charrep);
|
||||
} else {
|
||||
SPHO_DEBUG_PRINT("char_at: ret=%d, *out=%c\n", ret, *out);
|
||||
}
|
||||
#endif
|
||||
return (ret);
|
||||
}
|
||||
|
||||
|
|
@ -563,6 +600,10 @@ tok_match(struct msph_ctx *ctx, struct msph_token_src *src,
|
|||
MATCH_CHAR(')');
|
||||
case TOK_COLON:
|
||||
MATCH_CHAR(':');
|
||||
case TOK_DOT:
|
||||
MATCH_CHAR('.');
|
||||
case TOK_COMMA:
|
||||
MATCH_CHAR(',');
|
||||
case TOK_EQUALS:
|
||||
MATCH_CHAR('=');
|
||||
case TOK_AMP:
|
||||
|
|
@ -626,12 +667,16 @@ tok_commit(struct msph_ctx *ctx, struct msph_token_src *src,
|
|||
struct msph_matcher *m, struct msph_token *ptr)
|
||||
{
|
||||
size_t pos_old;
|
||||
struct msph_text_pos tok_pos;
|
||||
struct msph_token_src_str *str;
|
||||
struct msph_token_src_file *file;
|
||||
|
||||
SPHO_PRECOND(ctx != NULL && m != NULL);
|
||||
SPHO_PRECOND(m->matchlen != 0);
|
||||
|
||||
tok_pos = src->pos;
|
||||
tok_update_pos(ctx, src, m);
|
||||
|
||||
switch (src->type) {
|
||||
case MSPH_TOKEN_SRC_FILE:
|
||||
file = &src->inner.file;
|
||||
|
|
@ -646,6 +691,7 @@ tok_commit(struct msph_ctx *ctx, struct msph_token_src *src,
|
|||
return (0);
|
||||
|
||||
ptr->type = m->type;
|
||||
ptr->pos = tok_pos;
|
||||
if (! TOK_HAS_DATA(ptr->type))
|
||||
return (0);
|
||||
|
||||
|
|
@ -674,6 +720,7 @@ tok_commit(struct msph_ctx *ctx, struct msph_token_src *src,
|
|||
return (0);
|
||||
|
||||
ptr->type = m->type;
|
||||
ptr->pos = tok_pos;
|
||||
if (! TOK_HAS_DATA(ptr->type))
|
||||
return (0);
|
||||
|
||||
|
|
@ -692,7 +739,34 @@ tok_commit(struct msph_ctx *ctx, struct msph_token_src *src,
|
|||
}
|
||||
}
|
||||
|
||||
static void
|
||||
tok_update_pos(struct msph_ctx *ctx, struct msph_token_src *src,
|
||||
struct msph_matcher *m)
|
||||
{
|
||||
int res;
|
||||
char c;
|
||||
size_t i;
|
||||
|
||||
for (i = 0; i < m->matchlen; i++) {
|
||||
res = char_at(ctx, src, i, &c);
|
||||
SPHO_ASSERT(res == 1);
|
||||
|
||||
switch (c) {
|
||||
case '\t':
|
||||
src->pos.col += MSPH_TAB_WIDTH;
|
||||
break;
|
||||
case '\n':
|
||||
src->pos.line++;
|
||||
src->pos.col = 1;
|
||||
break;
|
||||
case '\r':
|
||||
break;
|
||||
default:
|
||||
src->pos.col++;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
static const char *
|
||||
tok_base_str(struct msph_token *tok)
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue