ref: 47a288683dbdc55c5df9b9f65db9909de6021af0
parent: ce4a8027322c53b2832f221e2956c90dcd65fd1a
author: sirjofri <[email protected]>
date: Wed Jul 24 11:16:58 EDT 2024
adds namespace support
--- a/README
+++ b/README
@@ -27,6 +27,14 @@
This will remove all the installed files.
+Libxml
+
+This version of libxml adds support for the following features.
+To get a good overview of how these features work, check the man page.
+
+- Namespace support for Elems and Attrs, which supersedes Fstripnamespace
+
+
Libxpath
currently supported rules:
--- a/libxml/mkfile
+++ b/libxml/mkfile
@@ -10,6 +10,7 @@
xmlparse.$O\
xmlprint.$O\
xmlnew.$O\
+ xmlns.$O\
xmllook.$O\
xmlvalue.$O\
heap.$O\
--- a/libxml/xmlattr.c
+++ b/libxml/xmlattr.c
@@ -6,8 +6,9 @@
xmlattr(Xml *xp, Attr **root, Elem *parent, char *name, char *value)
{Attr *ap, *t;
+ char *s;
+ Ns *n;
- USED(xp);
if((ap = xmlcalloc(xp, sizeof(Attr), 1)) == nil)
sysfatal("no memory - %r\n"); if(*root == nil){@@ -20,9 +21,28 @@
}
ap->parent = parent;
- if(name)
- if((ap->name = xmlstrdup(xp, name, 1)) == nil)
+ if (name) {+ n = nil;
+ if (strncmp(name, "xmlns:", 6) == 0) {+ n = xmlfindns(xp, name+6);
+ } else
+ if (strcmp(name, "xmlns") == 0) {+ n = xmlfindns(xp, nil);
+ }
+ if (n && value)
+ n->decl = xmlstrdup(xp, value, 0);
+ }
+
+ if(name){+ s = strchr(name, ':');
+ if (s) {+ ap->ns = xmlfindns(xp, name);
+ s++;
+ } else
+ s = name;
+ if((ap->name = xmlstrdup(xp, s, 1)) == nil)
sysfatal("no memory - %r\n");+ }
if(value)
if((ap->value = xmlstrdup(xp, value, 0)) == nil)
--- a/libxml/xmlelem.c
+++ b/libxml/xmlelem.c
@@ -6,6 +6,8 @@
xmlelem(Xml *xp, Elem **root, Elem *parent, char *name)
{Elem *ep, *t;
+ Ns *ns;
+ char *s;
USED(xp);
if((ep = xmlcalloc(xp, sizeof(Elem), 1)) == nil)
@@ -18,10 +20,18 @@
continue;
t->next = ep;
}
+ ns = strchr(name, ':') ? xmlfindns(xp, name) : xmlfindns(xp, nil);
+ ep->ns = ns;
ep->parent = parent;
- if(name)
- if((ep->name = xmlstrdup(xp, name, 1)) == nil)
+ if(name){+ s = strchr(name, ':');
+ if (s)
+ s++;
+ else
+ s = name;
+ if((ep->name = xmlstrdup(xp, s, 1)) == nil)
sysfatal("no memory - %r\n");+ }
return ep;
}
--- /dev/null
+++ b/libxml/xmlns.c
@@ -1,0 +1,75 @@
+#include <u.h>
+#include <libc.h>
+#include "xml.h"
+
+Ns*
+xmladdns(Xml *xml, char *name, char *decl)
+{+ Ns *n;
+
+ if (!xml->ns) {+ xml->ns = xmlcalloc(xml, sizeof(Ns), 1);
+ n = xml->ns;
+ if (name)
+ n->name = xmlstrdup(xml, name, 1);
+ if (decl)
+ n->decl = xmlstrdup(xml, decl, 0);
+ goto Out;
+ }
+
+ n = xml->ns;
+ while (n->next)
+ n = n->next;
+
+ n->next = xmlcalloc(xml, sizeof(Ns), 1);
+ n = n->next;
+ if (name)
+ n->name = xmlstrdup(xml, name, 1);
+ if (decl)
+ n->decl = xmlstrdup(xml, decl, 0);
+Out:
+ if (xmldebug == 1)
+ fprint(2, "addns: %s = %s\n", name, decl);
+ return n;
+}
+
+Ns*
+xmlfindns(Xml *xml, char *name)
+{+ Ns *n;
+ int i;
+ char *s;
+
+ if (!name) {+ i = 0;
+ goto Search;
+ }
+
+ s = strchr(name, ':');
+ if (!s) {+ i = strlen(name);
+ goto Search;
+ }
+ i = s - name;
+
+Search:
+ if (xmldebug) {+ if (i)
+ fprint(2, "search for xmlns %.*s (%d)\n", i, name, i);
+ else
+ fprint(2, "search for root xmlns\n");
+ }
+ for (n = xml->ns; n; n = n->next) {+ if (n->name == nil && !i)
+ return n;
+ if (n->name && i && strncmp(n->name, name, i) == 0)
+ return n;
+ }
+ if (!i)
+ return xmladdns(xml, nil, nil);
+ s = mallocz(i+1, 1);
+ strncpy(s, name, i);
+ n = xmladdns(xml, s, nil);
+ free(s);
+ return n;
+}
--- a/libxml/xmlparse.c
+++ b/libxml/xmlparse.c
@@ -443,6 +443,8 @@
Lexbuf pcdata, *pc;
Elem *root, *ep;
int os, s, t, a;
+ char *str;
+ Ns *ns;
ap = nil;
ep = nil;
@@ -465,8 +467,6 @@
fprint(2, "%-3d %*.selem name='%s'\n", st->line, depth, "", lb->buf);
if(!isname1(lb->buf[0]))
failed(st, "'%s' is an illegal element name", lb->buf);
- if(st->flags & Fstripnamespace)
- stripns(lb->buf);
assert((ep = xmlelem(st->xml, &root, parent, lb->buf)) != nil);
ep->line = st->line;
break;
@@ -480,8 +480,6 @@
fprint(2, "%-3d %*.sattr name='%s'\n", st->line, depth, "", lb->buf);
if(!isname1(lb->buf[0]))
failed(st, "'%s' is an illegal attribute name", lb->buf);
- if(st->flags & Fstripnamespace)
- stripns(lb->buf);
assert((ap = xmlattr(st->xml, &(ep->attrs), ep, lb->buf, nil)) != nil);
break;
case Avalue:
@@ -488,6 +486,11 @@
assert(ep != nil);
assert(ap != nil);
ap->value = xmlstrdup(st->xml, lb->buf, 0);
+ if(ap->ns && ap->ns->name && strcmp(ap->ns->name, "xmlns") == 0){+ ns = xmlfindns(st->xml, ap->name);
+ if(ns)
+ ns->decl = xmlstrdup(st->xml, lb->buf, 0);
+ }
ap = nil;
if(xmldebug == 1)
fprint(2, "%*.sattr value=%s\n", depth, "", lb->buf);
@@ -514,9 +517,17 @@
break;
case Acheck:
assert(ep != nil);
- if(st->flags & Fstripnamespace)
- stripns(lb->buf);
- if(ep->name && strcmp(lb->buf, ep->name) != 0)
+ if(str = strchr(lb->buf, ':')){+ ns = xmlfindns(st->xml, lb->buf);
+ if(ep->ns != ns){+ failed(st, "</%s> found, but does not match namespace for <%s:%s> (re: line %d) - nesting error",
+ lb->buf, ep->ns->name, ep->name, ep->line);
+ break;
+ }
+ str++;
+ } else
+ str = lb->buf;
+ if(ep->name && strcmp(str, ep->name) != 0)
failed(st, "</%s> found, expecting match for <%s> (re: line %d) - nesting error",
lb->buf, ep->name, ep->line);
break;
--- a/libxml/xmlprint.c
+++ b/libxml/xmlprint.c
@@ -33,13 +33,22 @@
_xmlprint(Biobuf *bp, Elem *ep, int in)
{Attr *ap;
+ char *ns;
enum {indent = 4}; for(; ep; ep = ep->next){- Bprint(bp, "%*s<%s", in, "", ep->name);
-
+ ns = ep->ns ? ep->ns->name : nil;
+ Bprint(bp, "%*s<", in, "");
+ if (ns)
+ Bprint(bp, "%s:", ns);
+ Bprint(bp, "%s", ep->name);
+
for (ap = ep->attrs; ap; ap = ap->next){- Bprint(bp, " %s=\'", ap->name);
+ ns = ap->ns ? ap->ns->name : nil;
+ Bprint(bp, " ");
+ if (ns)
+ Bprint(bp, "%s:", ns);
+ Bprint(bp, "%s=\'", ap->name);
prval(bp, ap->value);
Bprint(bp, "\'");
}
--- a/xml
+++ b/xml
@@ -32,6 +32,7 @@
struct Xml{Elem *root; /* root of tree */
char *doctype; /* DOCTYPE structured comment, or nil */
+ Ns *ns; /* list of all namespaces */
...
};
.PB
@@ -42,6 +43,7 @@
Attr *attrs; /* linked list of atributes */
char *name; /* element name */
char *pcdata; /* pcdata following this element */
+ Ns *ns; /* namespace */
int line; /* Line number (for errors) */
};
.PB
@@ -48,10 +50,17 @@
struct Attr {Attr *next; /* next atribute */
Elem *parent; /* parent element */
- char *name; /* atributes name */
+ char *name; /* attributes name */
char *value; /* atributes value */
+ Ns *ns; /* namespace */
};
.PB
+struct Ns {+ char *name; /* name of namespace */
+ char *decl; /* namespace declaration */
+ Ns *next; /* next namespace */
+};
+.PB
.PD 0
.ta +\w'\fL 'u +\w'\fL 'u +6n +4n
Attr* xmlattr(Xml *xp, Attr **root, Elem *parent,
@@ -78,6 +87,10 @@
void xmlfree(Xml *xp)
.PB
void xmlprint(Xml *xp, int fd)
+.PB
+Ns* xmlfindns(Xml *xp, char *s)
+.PB
+Ns* xmladdns(Xml *xp, char *s, char *v)
.SH DESCRIPTION
.PP
.I Libxml
@@ -158,6 +171,11 @@
writes the XML hierarchy rooted at \fIep\fR as text to the given
file descriptor.
.PP
+.I Xmlfindns
+and
+.I xmladdns
+find, add or adjust a namespace for the given XML document.
+.PP
.IR Xmlmalloc ,
.IR xmlcalloc ,
and
@@ -174,7 +192,9 @@
.SH "SEE ALSO"
.IR xb (1).
.SH BUGS
-Namespaces should be handled properly.
+The current namespace implementation supersedes
+.I Fstripnamespace
+and is probably buggy.
.PP
A SAX model parser will probably be needed sometime (e.g. for Ebooks).
.PP
--- a/xml.h
+++ b/xml.h
@@ -4,6 +4,7 @@
typedef struct Xml Xml;
typedef struct Attr Attr;
typedef struct Elem Elem;
+typedef struct Ns Ns;
typedef struct Xtree Xtree;
typedef struct Xblock Xblock;
@@ -19,6 +20,7 @@
struct Xml {Elem *root; /* root of tree */
char *doctype; /* DOCTYPE structured comment, or nil */
+ Ns *ns;
struct {Xtree *root;
Xblock *active;
@@ -33,6 +35,7 @@
Attr *attrs; /* linked list of atributes */
char *name; /* element name */
char *pcdata; /* pcdata following this element */
+ Ns *ns;
int line; /* Line number (for errors) */
};
@@ -41,8 +44,15 @@
Elem *parent; /* parent element */
char *name; /* atributes name (nil for coments) */
char *value; /* atributes value */
+ Ns *ns;
};
+struct Ns {+ char *name;
+ char *decl;
+ Ns *next;
+};
+
extern int xmldebug;
Attr* xmlattr(Xml *, Attr **, Elem *, char *, char *);
@@ -59,3 +69,5 @@
Xml* xmlparse(int, int, int);
void xmlprint(Xml *, int);
char* xmlvalue(Elem *, char *);
+Ns* xmlfindns(Xml *, char *);
+Ns* xmladdns(Xml *, char *, char *);
--- a/xpath
+++ b/xpath
@@ -15,7 +15,7 @@
#include <xpath.h>
.PB
enum {- Xelems = 1,
+ Xelem = 1,
Xstring = 2,
Xnum = 3,
}