first commit

src/main/java/org/openautonomousconnection/htmlparser/DocumentBuilder.java

@@ -0,0 +1,157 @@
+// Author: maple
+// date: 9/24/25
+
+package org.openautonomousconnection.htmlparser;
+
+import dev.unlegitdqrk.unlegitlibrary.string.StringUtils;
+import org.openautonomousconnection.htmlparser.html.body.misc.HTMLComment;
+import lombok.Getter;
+import lombok.Setter;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+public class DocumentBuilder {
+    @Getter @Setter
+    protected String content;
+
+    @Getter
+    protected List<HTMLComment> comments;
+
+    @Getter
+    protected List<String> attributes, texts, tags;
+
+    public DocumentBuilder(String content) {
+        this.content = content; //content.replace("\n", "");
+        this.comments = new ArrayList<>();
+        this.attributes = new ArrayList<>();
+        this.texts = new ArrayList<>();
+    }
+
+    /**
+     * Extracts all comments and strings into lists
+     */
+    public void extract() {
+        this.extractComments();
+        this.extractStringsAndAttributes();
+        this.extractTexts();
+
+    }
+
+    /**
+     * inserts the extracts back into the content string
+     */
+    public void insert() {
+        this.insertTexts();
+        this.insertStringsAndAttributes();
+        this.insertComments();
+    }
+
+    protected void extractComments() {
+        Pattern pattern = Pattern.compile("<!--(.*?)-->", Pattern.DOTALL);
+
+        Matcher matcher = pattern.matcher(content);
+
+
+        int index = 0;
+
+        while (matcher.find()) {
+            this.content = this.content.replace("<!--" + matcher.group(1) + "-->", "<!--C" + index + "-->");
+
+            this.comments.add(new HTMLComment(matcher.group(1)));
+
+            index++;
+        }
+    }
+
+    protected void insertComments() {
+        if(this.comments.isEmpty())
+            return;
+
+        int i = 0;
+        for(; i < this.comments.size(); i++)
+            this.content = this.content.replace("<!--C" + i + "-->", this.comments.get(i).toString());
+
+        for(; i > 0; i--)
+            this.comments.removeFirst();
+    }
+
+
+
+    protected void extractStringsAndAttributes() {
+        Pattern pattern = Pattern.compile("\"(.*?)\"|'(.*?)'", Pattern.DOTALL);
+
+        Matcher matcher = pattern.matcher(this.content);
+
+
+        int index = 0;
+
+        while (matcher.find()) {
+
+            if(matcher.group(1) != null) {
+                this.content = this.content.replace("\"" + matcher.group(1) + "\"", "\"S" + index + "\"");
+
+                this.attributes.add(matcher.group(1));
+            }
+
+            else {
+                this.content = this.content.replace("'" + matcher.group(2) + "'", "'S" + index + "'");
+
+                this.attributes.add(matcher.group(2));
+            }
+
+            index++;
+        }
+    }
+
+    protected void insertStringsAndAttributes() {
+        if(this.attributes.isEmpty())
+            return;
+        int i = 0;
+        for(; i < this.attributes.size(); i++) {
+            this.content = this.content.replace("\"S" + i + "\"", "\"" + attributes.get(i) + "\"");
+            this.content = this.content.replace("'S" + i + "'", "'" + attributes.get(i) + "'");
+        }
+
+        for(; i > 0; i--)
+            this.attributes.removeFirst();
+    }
+
+    protected void extractTexts() {
+        Pattern pattern = Pattern.compile(">([^<]+)(?=<)", Pattern.DOTALL);
+
+        Matcher matcher = pattern.matcher(content);
+
+
+        int index = 0;
+
+        while (matcher.find()) {
+            if(StringUtils.isEmptyString(matcher.group(1)))
+                continue;
+
+            this.content = this.content.replace(">" + matcher.group(1) + "<", ">T" + index + "<");
+
+            this.texts.add(matcher.group(1));
+
+            index++;
+        }
+    }
+
+    protected void insertTexts() {
+        if(this.texts.isEmpty())
+            return;
+
+        int i = 0;
+        for(; i < this.texts.size(); i++)
+            this.content = this.content.replace(">T" + i + "<", ">" + this.texts.get(i) + "<");
+
+        for(; i > 0; i--)
+            this.texts.removeFirst();
+    }
+
+    protected void extractTags() {
+
+    }
+}