csharp/mutil/XmlExtractor.cs

   1 using System;
   2 using System.Collections.Generic;
   3 using System.Xml;
   4
   5 namespace mutil {
   6
   7 public class XmlExtractor {
   8         public const int EC_BEG = 1; // begin of subtree (current node descends) - in
   9         public const int EC_CTN = 2; // continue with next sibling element - in / out: data of found element
  10         public const int EC_END = 3; // subtree completion (current node ascends); no data - out
  11
  12
  13         private XmlDocument docNode;
  14         private bool textMode;
  15         public XmlNode CurrentNode;
  16
  17         public XmlExtractor(object BaseNode = null, bool TextMode = false) {
  18                 this.CurrentNode = (XmlNode)BaseNode;
  19                 this.textMode = TextMode;
  20         }
  21
  22         public void openInput(string input){
  23                 this.docNode = new XmlDocument();
  24                 this.docNode.Load(input);
  25                 this.CurrentNode = this.docNode;
  26         }
  27
  28         public void toTop(){
  29                 this.CurrentNode = this.docNode;
  30         }
  31         public void toParent(){
  32                 this.CurrentNode = this.CurrentNode.ParentNode;
  33         }
  34         public int extractElement(int extr_ctl, out string elem, out string value, out Dictionary<string,string> attrs) {
  35                 elem  = "";
  36                 value = null;
  37                 attrs = new Dictionary<string,string>();
  38
  39                 XmlNode srh_node;
  40
  41                 if (extr_ctl == EC_BEG) {
  42                         srh_node = this.CurrentNode.FirstChild;
  43                 } else if (extr_ctl == EC_CTN) {
  44                         srh_node = this.CurrentNode.NextSibling;
  45                 } else {
  46                         return EC_END;
  47                 }
  48
  49                 // search for next element node
  50                 if (! this.textMode) {
  51                         while (srh_node != null) {
  52                                 if (srh_node.NodeType != XmlNodeType.Element) {
  53                                         srh_node = srh_node.NextSibling;
  54                                 } else {
  55                                         break;
  56                                 }
  57                         }
  58                 }
  59
  60                 if (srh_node != null) {
  61                         // element node found: extract tag name, value and attributes
  62                     this.CurrentNode = srh_node;
  63                     this.getCurrentNodeData(out elem, out value, out attrs);
  64                         return EC_CTN;
  65                 } else {
  66                         // terminate / ascend subtree after child extraction
  67                         if (extr_ctl == EC_CTN) {
  68                                 this.CurrentNode = this.CurrentNode.ParentNode;
  69                         }
  70                         return EC_END;
  71                 }
  72
  73         }
  74         public void getCurrentNodeData(out string elem, out string value, out Dictionary<string,string> attrs) {
  75             elem = this.CurrentNode.Name;
  76                 value = null;
  77                 attrs = new Dictionary<string,string>();
  78                 if (this.textMode) {
  79                         if (this.CurrentNode.NodeType == XmlNodeType.Text) value = this.CurrentNode.Value;
  80                 }
  81                 else {
  82                         if (this.CurrentNode.InnerText != null) value = this.CurrentNode.InnerText;
  83                 }
  84                 if (this.CurrentNode.Attributes != null) {
  85                         for (int att_idx = 0; att_idx < this.CurrentNode.Attributes.Count; att_idx++) {
  86                                 var att_node = this.CurrentNode.Attributes[att_idx];
  87                                 attrs[att_node.Name] = this.CurrentNode.Attributes[att_idx].Value;
  88                         }
  89                 }
  90         }
  91         public XmlNode gotoChild(string name) {
  92            // return True/False, if child node could be found
  93            int cntl = EC_BEG;
  94            string elem;
  95            string value;
  96            Dictionary<string,string> attrs;
  97            while ((cntl = this.extractElement(cntl, out elem, out value, out attrs)) != EC_END) {
  98                if (elem == name) return this.CurrentNode;
  99            }
 100            return null;
 101         }
 102         public XmlNode requireChild(string name) {
 103            // throws exception, if child not present
 104            XmlNode child = this.gotoChild(name);
 105            if (child == null) throw new Exception("Could not find node "+name);
 106            return child;
 107         }
 108 }
 109 }