| 1 |
|
| 2 |
Requirements for XML parsing in neon |
| 3 |
------------------------------------ |
| 4 |
|
| 5 |
Before describing the interface given in neon for parsing XML, here |
| 6 |
are the requirements which it must satisfy: |
| 7 |
|
| 8 |
1. to support using either libxml or expat as the underlying parser |
| 9 |
2. to allow "independent" sections to handle parsing one XML |
| 10 |
document |
| 11 |
3. to map element namespaces/names to an integer for easier |
| 12 |
comparison. |
| 13 |
|
| 14 |
A description of requirement (2) is useful since it is the "hard" |
| 15 |
requirement, and adds most of the complexity of interface: WebDAV |
| 16 |
PROPFIND responses are made up of a large boilerplate XML |
| 17 |
|
| 18 |
<multistatus><response><propstat><prop>...</prop></propstat> etc. |
| 19 |
|
| 20 |
neon should handle the parsing of these standard elements, and expose |
| 21 |
the meaning of the response using a convenient interface. But, within |
| 22 |
the <prop> elements, there may also be fragments of XML: neon can |
| 23 |
never know how to parse these, since they are property- and hence |
| 24 |
application-specific. The simplest example of this is the |
| 25 |
DAV:resourcetype property. |
| 26 |
|
| 27 |
So there is requirement (2) that two "independent" sections of code |
| 28 |
can handle the parsing of one XML document. |
| 29 |
|
| 30 |
Callback-based XML parsing |
| 31 |
-------------------------- |
| 32 |
|
| 33 |
There are two ways of parsing XML documents commonly used: |
| 34 |
|
| 35 |
1. Build an in-memory tree of the document |
| 36 |
2. Use callbacks |
| 37 |
|
| 38 |
Where practical, using callbacks is more efficient than building a |
| 39 |
tree, so this is what neon uses. The standard interface for |
| 40 |
callback-based XML parsing is called SAX, so understanding the SAX |
| 41 |
interface is useful to understanding the XML parsing interface |
| 42 |
provided by neon. |
| 43 |
|
| 44 |
The SAX interface works by registering callbacks which are called *as |
| 45 |
the XML is parsed*. The most important callbacks are for 'start |
| 46 |
element' and 'end element'. For instance, if the XML document below is |
| 47 |
parsed by a SAX-like interface: |
| 48 |
|
| 49 |
<hello> |
| 50 |
<foobar></foobar> |
| 51 |
</hello> |
| 52 |
|
| 53 |
Say we have registered callbacks "startelm" for 'start element' and |
| 54 |
"endelm" for 'end element'. Simplified somewhat, the callbacks will |
| 55 |
be called in this order, with these arguments: |
| 56 |
|
| 57 |
1. startelm("hello") |
| 58 |
2. startelm("foobar") |
| 59 |
3. endelm("foobar") |
| 60 |
4. endelm("hello") |
| 61 |
|
| 62 |
See the expat 'xmlparse.h' header for a more complete definition of a |
| 63 |
SAX-like interface. |
| 64 |
|
| 65 |
The hip_xml interface |
| 66 |
--------------------- |
| 67 |
|
| 68 |
The hip_xml interface satisfies requirement (2) by introducing the |
| 69 |
"handler" concept. A handler is made up of these things: |
| 70 |
|
| 71 |
- a set of XML elements |
| 72 |
- a callback to validate an element |
| 73 |
- a callback which is called when an element is opened |
| 74 |
- a callback which is called when an element is closed |
| 75 |
- (optionally, a callback which is called for CDATA) |
| 76 |
|
| 77 |
Registering a handler essentially says: |
| 78 |
|
| 79 |
"If you encounter any of this set of elements, I want these |
| 80 |
callbacks to be called." |
| 81 |
|
| 82 |
Handlers are kept in a STACK inside hip_xml. The first handler |
| 83 |
registered becomes the BASE of the stack, subsequent handlers are |
| 84 |
PUSHed on top. |
| 85 |
|
| 86 |
During XML parsing, the handler which is used for an XML element is |
| 87 |
recorded. When a new element is started, the search for a handler for |
| 88 |
this element begins at the handler used for the parent element, and |
| 89 |
carries on up the stack. For the root element, the search always |
| 90 |
starts at the BASE of the stack. |
| 91 |
|
| 92 |
A user's guide to hip_xml |
| 93 |
------------------------- |
| 94 |
|
| 95 |
The first thing to do when using hip_xml is to know what set of XML |
| 96 |
elements you are going to be parsing. This can usually be done by |
| 97 |
looking at the DTD provided for the documents you are going to be |
| 98 |
parsing. The DTD is also very useful in writing the 'validate' |
| 99 |
callback function, since it can tell you what parent/child pairs are |
| 100 |
valid, and which aren't. |
| 101 |
|
| 102 |
In this example, we'll parse XML documents which look like: |
| 103 |
|
| 104 |
<T:list-of-things xmlns:T="http://things.org/"> |
| 105 |
<T:a-thing>foo</T:a-thing> |
| 106 |
<T:a-thing>bar</T:a-thing> |
| 107 |
</T:list-of-things> |
| 108 |
|
| 109 |
So, given the set of elements, declare the element id's and the |
| 110 |
element array: |
| 111 |
|
| 112 |
#define ELM_listofthings (HIP_ELM_UNUSED) |
| 113 |
#define ELM_a_thing (HIP_ELM_UNUSED + 1) |
| 114 |
|
| 115 |
const static struct my_elms[] = { |
| 116 |
{ "http://things.org/", "list-of-things", ELM_listofthings, 0 }, |
| 117 |
{ "http://things.org/", "a-thing", ELM_a_thing, HIP_XML_CDATA }, |
| 118 |
{ NULL } |
| 119 |
}; |
| 120 |
|
| 121 |
This declares we know about two elements: list-of-things, and a-thing, |
| 122 |
and that the 'a-thing' element contains character data. |
| 123 |
|
| 124 |
The definition of the validation callback is very simple: |
| 125 |
|
| 126 |
static int validate(hip_xml_elmid parent, hip_xml_elmid child) |
| 127 |
{ |
| 128 |
/* Only allow 'list-of-things' as the root element. */ |
| 129 |
if (parent == HIP_ELM_root && child == ELM_listofthings || |
| 130 |
parent = ELM_listofthings && child == ELM_a_thing) { |
| 131 |
return HIP_XML_VALID; |
| 132 |
} else { |
| 133 |
return HIP_XML_INVALID; |
| 134 |
} |
| 135 |
} |
| 136 |
|
| 137 |
For this example, we can ignore the start-element callback, and just |
| 138 |
use the end-element callback: |
| 139 |
|
| 140 |
static int endelm(void *userdata, const struct hip_xml_elm *s, |
| 141 |
const char *cdata) |
| 142 |
{ |
| 143 |
printf("Got a thing: %s\n", cdata); |
| 144 |
return 0; |
| 145 |
} |
| 146 |
|
| 147 |
This endelm callback just prints the cdata which was contained in the |
| 148 |
"a-thing" element. |
| 149 |
|
| 150 |
Now, on to parsing. A new parser object is created for parsing each |
| 151 |
XML document. Creating a new parser object is as simple as: |
| 152 |
|
| 153 |
hip_xml_parser *parser; |
| 154 |
|
| 155 |
parser = hip_xml_create(); |
| 156 |
|
| 157 |
Next register the handler, passing NULL as the start-element callback, |
| 158 |
and also as userdata, which we don't use here. |
| 159 |
|
| 160 |
hip_xml_push_handler(parser, my_elms, |
| 161 |
validate, NULL, endelm, |
| 162 |
NULL); |
| 163 |
|
| 164 |
Finally, call hip_xml_parse, passing the chunks of XML document to the |
| 165 |
hip_xml as you get them. The output should be: |
| 166 |
|
| 167 |
Got a thing: foo |
| 168 |
Got a thing: bar |
| 169 |
|
| 170 |
for the XML document. |