X-Git-Url: http://www2.svjatoslav.eu/gitweb/?p=sixth.git;a=blobdiff_plain;f=src%2Fmain%2Fjava%2Feu%2Fsvjatoslav%2Fsixth%2Fcore%2Fdocument%2FDocument.java;h=c61d613842249402978b07ba39afc865df307542;hp=23852845cf06134c1fe57ef58826982c62368e1e;hb=9b683d223e06fc5ec9e2561c659de5a79d3a8194;hpb=f9ba1a34f75fc8bfe9e6507762947566c8e22bce

diff --git a/src/main/java/eu/svjatoslav/sixth/core/document/Document.java b/src/main/java/eu/svjatoslav/sixth/core/document/Document.java
index 2385284..c61d613 100644
--- a/src/main/java/eu/svjatoslav/sixth/core/document/Document.java
+++ b/src/main/java/eu/svjatoslav/sixth/core/document/Document.java
@@ -1,12 +1,16 @@
 package eu.svjatoslav.sixth.core.document;
 
+import eu.svjatoslav.commons.string.tokenizer.InvalidSyntaxException;
+import eu.svjatoslav.commons.string.tokenizer.Tokenizer;
+import eu.svjatoslav.commons.string.tokenizer.TokenizerMatch;
 import eu.svjatoslav.sixth.core.document.text.FormattedText;
 
+import static eu.svjatoslav.commons.string.tokenizer.Terminator.TerminationStrategy.PRESERVE;
+import static eu.svjatoslav.sixth.core.document.Helper.*;
 import static eu.svjatoslav.sixth.core.document.text.FormattedText.fromOrg;
 
 public class Document {
-    public final Heading rootHeading = new Heading( fromOrg("<root>") , 0, null);
-
+    public final Heading rootHeading = new Heading( null , 0, null);
     private Heading currentHeading = rootHeading;
 
     public Heading createHeading(FormattedText name, int targetLevel){
@@ -27,4 +31,80 @@ public class Document {
         currentHeading = missingIntermediate;
         return createHeading(name, targetLevel);
     }
+
+    public Heading getCurrentHeading(){
+        return currentHeading;
+    }
+
+    private void parseHeading(TokenizerMatch token) throws InvalidSyntaxException {
+        String[] headingSections = token.getRegExpGroups();
+        int level = headingSections[0].length();
+        String title = headingSections[1];
+        createHeading(fromOrg(title), level);
+    }
+
+    public void parse(String fileContentsAsString) throws InvalidSyntaxException {
+        final Tokenizer tokenizer = new Tokenizer(fileContentsAsString);
+
+        // Org heading:
+        // "*** Example Heading 1234"
+        tokenizer.addTerminator(PRESERVE, "(\\*+)[ \\t](.*)\\r?\\n", TG_HEADING);
+
+        // Org list. Examples:
+        // "   + my list title"
+        // "   - my list title"
+        tokenizer.addTerminator(PRESERVE, "([ \\t]*)(\\+|-)[ \\t]+(.*)?\\r?\\n", TG_LIST);
+
+        // "   * my list title"
+        tokenizer.addTerminator(PRESERVE, "([ \\t]+)(\\*)[ \\t]+(.*)?\\r?\\n", TG_LIST);
+
+        // TODO: add numbered list
+
+        // DocumentProperty:
+        // "#+OPTIONS: H:20 num:20"
+        tokenizer.addTerminator(PRESERVE, "#\\+([^\\s]+):(.*)\\r?\\n", TG_DOCUMENT_PROPERTY);
+
+        // Drawer property:
+        //  " :ID:       533734b9-0456-4448-9830-a43646345615"
+        tokenizer.addTerminator(PRESERVE, "([ \\t]*):([^\\s]+):(.*)\\r?\\n", TG_DRAWER_PROPERTY);
+
+
+        // multiline code block
+        tokenizer.addTerminator(PRESERVE,
+                "([ \\t]*)#\\+BEGIN_SRC" +  // source begin identifier
+                        "(([ \\t]+)(.*))?(\\r?\\n)" + // source block parameters
+                        "((?:.|\\n|\\r)*?)" + // source content
+                        "(\\r?\\n)([ \\t]*)#\\+END_SRC(.*)\\r?\\n" // source end identifier
+                , TG_MULTILINE_CODE);
+
+        // verse
+        tokenizer.addTerminator(PRESERVE,
+                "([ \\t]*)#\\+BEGIN_VERSE" + // verse begin identifier
+                        "(([ \\t]+)(.*))?(\\r?\\n)" + // verse block parameters
+                        "((?:.|\\n|\\r)*?)" + // verse
+                        "(\\r?\\n)([ \\t]*)#\\+END_VERSE(.*)\\r?\\n" // verse end identifier
+                , TG_VERSE);
+
+
+        // TODO: add support for export blocks:
+        //        #+begin_export latex
+        //  \clearpage
+        //#+end_export
+
+
+        // normal text
+        tokenizer.addTerminator(PRESERVE,".*\\r?\\n", TG_NORMAL_TEXT);
+
+        while (tokenizer.hasMoreContent()) {
+            final TokenizerMatch tm = tokenizer.getNextToken();
+
+            if (tm.isGroup(TG_HEADING)){
+                parseHeading(tm);
+                continue;
+            }
+
+            currentHeading.parse(tm);
+        }
+
+    }
 }