version 1.30, 2019/04/10 14:22:37 |
version 1.35, 2019/04/12 06:46:45 |
Line 322 warn_msg(struct parse *p, const char *fmt, ...) |
|
Line 322 warn_msg(struct parse *p, const char *fmt, ...) |
|
* Otherwise, create a new one as a child of the current node. |
* Otherwise, create a new one as a child of the current node. |
*/ |
*/ |
static void |
static void |
xml_char(struct parse *p, const char *word, int sz) |
xml_text(struct parse *p, const char *word, int sz) |
{ |
{ |
struct pnode *n; |
struct pnode *n, *np; |
size_t newsz; |
size_t oldsz, newsz; |
|
int i; |
|
|
|
assert(sz > 0); |
if (p->del > 0) |
if (p->del > 0) |
return; |
return; |
|
|
if (p->cur == NULL) { |
if ((n = p->cur) == NULL) { |
error_msg(p, "discarding text before document: %.*s", sz, word); |
error_msg(p, "discarding text before document: %.*s", |
|
sz, word); |
return; |
return; |
} |
} |
|
|
if (p->cur->node != NODE_TEXT) { |
/* Append to the current text node, if one is open. */ |
if ((n = calloc(1, sizeof(*n))) == NULL) |
|
|
if (n->node == NODE_TEXT) { |
|
oldsz = strlen(n->b); |
|
newsz = oldsz + sz; |
|
if (oldsz && (p->flags & PFLAG_SPC)) |
|
newsz++; |
|
if ((n->b = realloc(n->b, newsz + 1)) == NULL) |
fatal(p); |
fatal(p); |
n->node = NODE_TEXT; |
if (oldsz && (p->flags & PFLAG_SPC)) |
n->spc = (p->flags & PFLAG_SPC) != 0; |
n->b[oldsz++] = ' '; |
n->parent = p->cur; |
memcpy(n->b + oldsz, word, sz); |
TAILQ_INIT(&n->childq); |
n->b[newsz] = '\0'; |
TAILQ_INIT(&n->attrq); |
p->flags &= ~PFLAG_SPC; |
TAILQ_INSERT_TAIL(&p->cur->childq, n, child); |
return; |
p->cur = n; |
|
} |
} |
|
|
if (p->tree->flags & TREE_CLOSED && |
if (p->tree->flags & TREE_CLOSED && n == p->tree->root) |
p->cur->parent == p->tree->root) |
|
warn_msg(p, "text after end of document: %.*s", sz, word); |
warn_msg(p, "text after end of document: %.*s", sz, word); |
|
|
/* Append to the current text node. */ |
/* Create a new text node. */ |
|
|
assert(sz >= 0); |
if ((n = pnode_alloc(p->cur)) == NULL) |
newsz = p->cur->bsz + (p->cur->bsz && (p->flags & PFLAG_SPC)) + sz; |
|
if ((p->cur->b = realloc(p->cur->b, newsz + 1)) == NULL) |
|
fatal(p); |
fatal(p); |
if (p->cur->bsz && (p->flags & PFLAG_SPC)) |
n->node = NODE_TEXT; |
p->cur->b[p->cur->bsz++] = ' '; |
n->spc = (p->flags & PFLAG_SPC) != 0; |
memcpy(p->cur->b + p->cur->bsz, word, sz); |
|
p->cur->b[p->cur->bsz = newsz] = '\0'; |
|
p->cur->real = p->cur->b; |
|
p->flags &= ~PFLAG_SPC; |
p->flags &= ~PFLAG_SPC; |
|
|
|
/* |
|
* If this node follows a non-text node without intervening |
|
* whitespace, keep the text in it as short as possible, |
|
* and do not keep it open. |
|
*/ |
|
|
|
if (n->spc == 0 && |
|
(np = TAILQ_PREV(n, pnodeq, child)) != NULL && |
|
np->node != NODE_TEXT && np->node != NODE_ESCAPE) { |
|
i = 0; |
|
while (i < sz && !isspace((unsigned char)word[i])) |
|
i++; |
|
if ((n->b = strndup(word, i)) == NULL) |
|
fatal(p); |
|
if (i == sz) |
|
return; |
|
while (i < sz && isspace((unsigned char)word[i])) |
|
i++; |
|
if (i == sz) { |
|
p->flags |= PFLAG_SPC; |
|
return; |
|
} |
|
|
|
/* Put any remaining text into a second node. */ |
|
|
|
if ((n = pnode_alloc(p->cur)) == NULL) |
|
fatal(p); |
|
n->node = NODE_TEXT; |
|
n->spc = 1; |
|
word += i; |
|
sz -= i; |
|
} |
|
if ((n->b = strndup(word, sz)) == NULL) |
|
fatal(p); |
|
|
|
/* The new node remains open for later pnode_closetext(). */ |
|
|
|
p->cur = n; |
} |
} |
|
|
/* |
/* |
|
|
pnode_closetext(struct parse *p) |
pnode_closetext(struct parse *p) |
{ |
{ |
struct pnode *n; |
struct pnode *n; |
|
char *cp; |
|
|
if ((n = p->cur) == NULL || n->node != NODE_TEXT) |
if ((n = p->cur) == NULL || n->node != NODE_TEXT) |
return; |
return; |
p->cur = n->parent; |
p->cur = n->parent; |
while (n->bsz > 0 && isspace((unsigned char)n->b[n->bsz - 1])) { |
for (cp = strchr(n->b, '\0'); |
n->b[--n->bsz] = '\0'; |
cp > n->b && isspace((unsigned char)cp[-1]); |
|
*--cp = '\0') |
p->flags |= PFLAG_SPC; |
p->flags |= PFLAG_SPC; |
} |
|
} |
} |
|
|
static void |
static void |
Line 438 xml_entity(struct parse *p, const char *name) |
|
Line 480 xml_entity(struct parse *p, const char *name) |
|
} |
} |
|
|
/* Create, append, and close out an entity node. */ |
/* Create, append, and close out an entity node. */ |
if ((n = calloc(1, sizeof(*n))) == NULL || |
if ((n = pnode_alloc(p->cur)) == NULL || |
(n->b = n->real = strdup(entity->roff)) == NULL) |
(n->b = strdup(entity->roff)) == NULL) |
fatal(p); |
fatal(p); |
n->node = NODE_ESCAPE; |
n->node = NODE_ESCAPE; |
n->bsz = strlen(n->b); |
|
n->spc = (p->flags & PFLAG_SPC) != 0; |
n->spc = (p->flags & PFLAG_SPC) != 0; |
n->parent = p->cur; |
|
TAILQ_INIT(&n->childq); |
|
TAILQ_INIT(&n->attrq); |
|
TAILQ_INSERT_TAIL(&p->cur->childq, n, child); |
|
p->flags &= ~PFLAG_SPC; |
p->flags &= ~PFLAG_SPC; |
} |
} |
|
|
Line 503 xml_elem_start(struct parse *p, const char *name) |
|
Line 540 xml_elem_start(struct parse *p, const char *name) |
|
if (p->tree->flags & TREE_CLOSED && p->cur->parent == NULL) |
if (p->tree->flags & TREE_CLOSED && p->cur->parent == NULL) |
warn_msg(p, "element after end of document: <%s>", name); |
warn_msg(p, "element after end of document: <%s>", name); |
|
|
if ((n = calloc(1, sizeof(*n))) == NULL) |
if ((n = pnode_alloc(p->cur)) == NULL) |
fatal(p); |
fatal(p); |
|
|
/* |
/* |
Line 555 xml_elem_start(struct parse *p, const char *name) |
|
Line 592 xml_elem_start(struct parse *p, const char *name) |
|
n->spc = (p->flags & PFLAG_SPC) != 0; |
n->spc = (p->flags & PFLAG_SPC) != 0; |
break; |
break; |
} |
} |
n->parent = p->cur; |
|
TAILQ_INIT(&n->childq); |
|
TAILQ_INIT(&n->attrq); |
|
|
|
if (p->cur != NULL) |
|
TAILQ_INSERT_TAIL(&p->cur->childq, n, child); |
|
|
|
p->cur = n; |
p->cur = n; |
if (n->node == NODE_DOCTYPE) { |
if (n->node == NODE_DOCTYPE) { |
if (p->doctype == NULL) |
if (p->doctype == NULL) |
Line 632 xml_attrval(struct parse *p, const char *name) |
|
Line 662 xml_attrval(struct parse *p, const char *name) |
|
* If we're at a text node, roll that one up first. |
* If we're at a text node, roll that one up first. |
*/ |
*/ |
static void |
static void |
xml_elem_end(struct parse *ps, const char *name) |
xml_elem_end(struct parse *p, const char *name) |
{ |
{ |
const struct element *elem; |
const struct element *elem; |
struct pnode *n; |
struct pnode *n; |
Line 643 xml_elem_end(struct parse *ps, const char *name) |
|
Line 673 xml_elem_end(struct parse *ps, const char *name) |
|
* An ancestor is excluded from the tree; |
* An ancestor is excluded from the tree; |
* keep track of the number of levels excluded. |
* keep track of the number of levels excluded. |
*/ |
*/ |
if (ps->del > 1) { |
if (p->del > 1) { |
ps->del--; |
p->del--; |
return; |
return; |
} |
} |
|
|
if (ps->del == 0) |
if (p->del == 0) |
pnode_closetext(ps); |
pnode_closetext(p); |
|
|
if (name != NULL) { |
if (name != NULL) { |
for (elem = elements; elem->name != NULL; elem++) |
for (elem = elements; elem->name != NULL; elem++) |
Line 657 xml_elem_end(struct parse *ps, const char *name) |
|
Line 687 xml_elem_end(struct parse *ps, const char *name) |
|
break; |
break; |
node = elem->node; |
node = elem->node; |
} else |
} else |
node = ps->ncur; |
node = p->ncur; |
|
|
switch (node) { |
switch (node) { |
case NODE_DELETE_WARN: |
case NODE_DELETE_WARN: |
case NODE_DELETE: |
case NODE_DELETE: |
if (ps->del > 0) |
if (p->del > 0) |
ps->del--; |
p->del--; |
break; |
break; |
case NODE_IGNORE: |
case NODE_IGNORE: |
break; |
break; |
case NODE_INCLUDE: |
case NODE_INCLUDE: |
n = ps->cur; |
n = p->cur; |
ps->cur = ps->cur->parent; |
p->cur = p->cur->parent; |
cp = pnode_getattr_raw(n, ATTRKEY_HREF, NULL); |
cp = pnode_getattr_raw(n, ATTRKEY_HREF, NULL); |
if (cp == NULL) |
if (cp == NULL) |
error_msg(ps, "<xi:include> element " |
error_msg(p, "<xi:include> element " |
"without href attribute"); |
"without href attribute"); |
else |
else |
parse_file(ps, -1, cp); |
parse_file(p, -1, cp); |
pnode_unlink(n); |
pnode_unlink(n); |
ps->flags &= ~PFLAG_SPC; |
p->flags &= ~PFLAG_SPC; |
break; |
break; |
case NODE_DOCTYPE: |
case NODE_DOCTYPE: |
ps->flags &= ~PFLAG_EEND; |
case NODE_SBR: |
|
p->flags &= ~PFLAG_EEND; |
/* FALLTHROUGH */ |
/* FALLTHROUGH */ |
default: |
default: |
if (ps->cur == NULL || node != ps->cur->node) { |
if (p->cur == NULL || node != p->cur->node) { |
warn_msg(ps, "element not open: </%s>", name); |
warn_msg(p, "element not open: </%s>", name); |
break; |
break; |
} |
} |
|
|
Line 695 xml_elem_end(struct parse *ps, const char *name) |
|
Line 726 xml_elem_end(struct parse *ps, const char *name) |
|
* obviously better than discarding it or crashing. |
* obviously better than discarding it or crashing. |
*/ |
*/ |
|
|
if (ps->cur->parent != NULL || node == NODE_DOCTYPE) { |
if (p->cur->parent != NULL || node == NODE_DOCTYPE) { |
ps->cur = ps->cur->parent; |
p->cur = p->cur->parent; |
if (ps->cur != NULL) |
if (p->cur != NULL) |
ps->ncur = ps->cur->node; |
p->ncur = p->cur->node; |
} else |
} else |
ps->tree->flags |= TREE_CLOSED; |
p->tree->flags |= TREE_CLOSED; |
ps->flags &= ~PFLAG_SPC; |
p->flags &= ~PFLAG_SPC; |
break; |
break; |
} |
} |
assert(ps->del == 0); |
assert(p->del == 0); |
} |
} |
|
|
struct parse * |
struct parse * |
Line 970 parse_string(struct parse *p, char *b, size_t rlen, |
|
Line 1001 parse_string(struct parse *p, char *b, size_t rlen, |
|
|
|
} else { |
} else { |
advance(p, b, rlen, &pend, |
advance(p, b, rlen, &pend, |
p->ncur == NODE_DOCTYPE ? "<&]" : "<&", |
p->ncur == NODE_DOCTYPE ? "<&]\n" : "<&\n", |
refill); |
refill); |
xml_char(p, b + poff, pend - poff); |
xml_text(p, b + poff, pend - poff); |
|
if (b[pend] == '\n') |
|
pnode_closetext(p); |
} |
} |
} |
} |
return poff; |
return poff; |