XPathEvaluator.m

gistfile1.m

#import "Foundation/Foundation.h"
#include <libxml/HTMLparser.h>
#include <libxml/HTMLtree.h>
#include <libxml/xpath.h>

@interface XPathEvaluator : NSObject
+(NSArray*)arrayWithXPathQuery:(NSString*)query fromHTML:(NSString*)html;
@end

@implementation XPathEvaluator
+(NSArray*)arrayWithXPathQuery:(NSString*)query fromHTML:(NSString*)html
{
  id pool = [NSAutoreleasePool new];

  const char *chtml  = [html  cStringUsingEncoding:NSUTF8StringEncoding];
  const char *cxpath = [query cStringUsingEncoding:NSUTF8StringEncoding];

  NSMutableArray *results = [NSMutableArray new];

  // XPath
  htmlDocPtr         doc   = htmlParseDoc((xmlChar*)chtml, "UTF-8");
  xmlXPathContextPtr xctx  = xmlXPathNewContext(doc);
  xmlXPathObjectPtr  xobj  = xmlXPathEval((xmlChar*)cxpath, xctx);
  xmlNodeSetPtr      nodes = xobj->nodesetval;

  if(doc == NULL)  NSLog(@"parse fail");
  if(xobj == NULL) NSLog(@"xpath fail");

  for(int n = 0; n < xmlXPathNodeSetGetLength(nodes); n++)
  {
      xmlNodePtr           node = nodes->nodeTab[n];
      xmlOutputBufferPtr outbuf = xmlAllocOutputBuffer(NULL);

      if(outbuf == NULL) NSLog(@"outbuf fail");

//      NSLog(@"==== div.body:%d %s %d ====", n, node->name, node->type);

      // inner html
      xmlNodePtr child = node->children;
      while(child)
      {
        htmlNodeDumpFormatOutput(outbuf, doc, child, "UTF-8", 0);
        child = child->next;
      }
      xmlOutputBufferFlush(outbuf);

      // attributes
      NSMutableDictionary *attributes = [NSMutableDictionary dictionary];
      xmlAttr *attr = node->properties;
      while(attr)
      {
        [attributes
          setObject:[NSString stringWithCString:(char*)attr->children->content encoding:NSUTF8StringEncoding]
          forKey:   [NSString stringWithCString:(char*)attr->name encoding:NSUTF8StringEncoding]
        ];
        attr = attr->next;
      }

      // Dictionary
      NSString *name     = [NSString stringWithCString:(char*)node->name encoding:NSUTF8StringEncoding];
      NSString *content  = [[NSString
        stringWithCString:(char*)outbuf->buffer->content encoding:NSUTF8StringEncoding
      ] stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]];
      NSDictionary *nodeinfo = [NSDictionary dictionaryWithObjectsAndKeys:
        name,       @"name",
        attributes, @"attributes",
        content,    @"content",
        nil
      ];

      [results addObject:nodeinfo];

      xmlOutputBufferClose(outbuf);
  }

  xmlXPathFreeObject(xobj);
  xmlXPathFreeContext(xctx);
  xmlFreeDoc(doc);

  [pool release];

  return [results autorelease];
}
@end

int main(int argc, char **argv)
{
    id pool = [NSAutoreleasePool new];

    char *cxpath = (argc > 1) ? argv[1] : "//span[@class='msg']";
    char *chtml  = (argc > 3) ? argv[2] : "<span class=\"msg\">oppai</span><span>futomomo</span>";

    NSString *query = [NSString stringWithCString:cxpath encoding:NSUTF8StringEncoding];
    NSString *html  = [NSString stringWithCString:chtml  encoding:NSUTF8StringEncoding];
    NSArray  *nodes = [XPathEvaluator arrayWithXPathQuery:query fromHTML:html];

    NSLog(@"%@", nodes);
    NSLog(@"%d", [nodes count]);

    [pool release];


    return 0;
}