fix/perf base64/serial
[nikiroo-utils.git] / src / be / nikiroo / utils / serial / Importer.java
index 0709a96883c77b005f57ddf04de4ac15b04165ac..bca157c501bcb8a512fd5367585b15a06a82771e 100644 (file)
@@ -1,12 +1,11 @@
 package be.nikiroo.utils.serial;
 
+import java.io.IOException;
+import java.io.UnsupportedEncodingException;
+import java.lang.reflect.Field;
 import java.util.HashMap;
 import java.util.Map;
-import java.util.Scanner;
-import java.lang.reflect.InvocationTargetException;
-import java.lang.reflect.Field;
 
-import be.nikiroo.utils.IOUtils;
 import be.nikiroo.utils.StringUtils;
 
 /**
@@ -20,6 +19,9 @@ import be.nikiroo.utils.StringUtils;
  * @author niki
  */
 public class Importer {
+       static private Integer SIZE_ID = null;
+       static private byte[] NEWLINE = null;
+
        private Boolean link;
        private Object me;
        private Importer child;
@@ -27,6 +29,18 @@ public class Importer {
 
        private String currentFieldName;
 
+       static {
+               try {
+                       SIZE_ID = "EXT:".getBytes("UTF-8").length;
+                       NEWLINE = "\n".getBytes("UTF-8");
+               } catch (UnsupportedEncodingException e) {
+                       // UTF-8 is mandated to exist on confirming jre's
+               }
+       }
+
+       /**
+        * Create a new {@link Importer}.
+        */
        public Importer() {
                map = new HashMap<String, Object>();
                map.put("NULL", null);
@@ -36,35 +50,124 @@ public class Importer {
                this.map = map;
        }
 
-       public Importer readLine(String line) {
-               try {
-                       processLine(line);
-               } catch (Exception e) {
-                       throw new IllegalArgumentException(e);
-               }
-               return this;
+       /**
+        * Read some data into this {@link Importer}: it can be the full serialised
+        * content, or a number of lines of it (any given line <b>MUST</b> be
+        * complete though) and accumulate it with the already present data.
+        * 
+        * @param data
+        *            the data to parse
+        * 
+        * @return itself so it can be chained
+        * 
+        * @throws NoSuchFieldException
+        *             if the serialised data contains information about a field
+        *             which does actually not exist in the class we know of
+        * @throws NoSuchMethodException
+        *             if a class described in the serialised data cannot be created
+        *             because it is not compatible with this code
+        * @throws ClassNotFoundException
+        *             if a class described in the serialised data cannot be found
+        * @throws IOException
+        *             if the content cannot be read (for instance, corrupt data)
+        */
+       public Importer read(String data) throws NoSuchFieldException,
+                       NoSuchMethodException, ClassNotFoundException, IOException {
+               return read(data.getBytes("UTF-8"), 0);
        }
 
-       public Importer read(String data) {
-               try {
-                       if (data.startsWith("ZIP:")) {
-                               data = StringUtils.unzip64(data.substring("ZIP:".length()));
+       /**
+        * Read some data into this {@link Importer}: it can be the full serialised
+        * content, or a number of lines of it (any given line <b>MUST</b> be
+        * complete though) and accumulate it with the already present data.
+        * 
+        * @param data
+        *            the data to parse
+        * @param offset
+        *            the offset at which to start reading the data (we ignore
+        *            anything that goes before that offset)
+        * 
+        * @return itself so it can be chained
+        * 
+        * @throws NoSuchFieldException
+        *             if the serialised data contains information about a field
+        *             which does actually not exist in the class we know of
+        * @throws NoSuchMethodException
+        *             if a class described in the serialised data cannot be created
+        *             because it is not compatible with this code
+        * @throws ClassNotFoundException
+        *             if a class described in the serialised data cannot be found
+        * @throws IOException
+        *             if the content cannot be read (for instance, corrupt data)
+        */
+       private Importer read(byte[] data, int offset) throws NoSuchFieldException,
+                       NoSuchMethodException, ClassNotFoundException, IOException {
+
+               int dataStart = offset;
+               while (dataStart < data.length) {
+                       String id = "";
+                       if (data.length - dataStart >= SIZE_ID) {
+                               id = new String(data, dataStart, SIZE_ID);
                        }
-                       Scanner scan = new Scanner(data);
-                       scan.useDelimiter("\n");
-                       while (scan.hasNext()) {
-                               processLine(scan.next());
+
+                       boolean zip = id.equals("ZIP:");
+                       boolean b64 = id.equals("B64:");
+                       if (zip || b64) {
+                               dataStart += SIZE_ID;
                        }
-                       scan.close();
-               } catch (Exception e) {
-                       throw new IllegalArgumentException(e);
+
+                       int count = find(data, dataStart, NEWLINE);
+                       count -= dataStart;
+                       if (count < 0) {
+                               count = data.length - dataStart;
+                       }
+
+                       if (zip || b64) {
+                               boolean unpacked = false;
+                               try {
+                                       byte[] line = StringUtils.unbase64(data, dataStart, count,
+                                                       zip);
+                                       unpacked = true;
+                                       read(line, 0);
+                               } catch (IOException e) {
+                                       throw new IOException("Internal error when decoding "
+                                                       + (unpacked ? "unpacked " : "")
+                                                       + (zip ? "ZIP" : "B64")
+                                                       + " content: input may be corrupt");
+                               }
+                       } else {
+                               String line = new String(data, dataStart, count, "UTF-8");
+                               processLine(line);
+                       }
+
+                       dataStart += count + NEWLINE.length;
                }
+
                return this;
        }
 
-       public boolean processLine(String line) throws IllegalArgumentException,
-                       NoSuchFieldException, SecurityException, IllegalAccessException,
-                       NoSuchMethodException, InstantiationException, ClassNotFoundException, InvocationTargetException {
+       /**
+        * Read a single (whole) line of serialised data into this {@link Importer}
+        * and accumulate it with the already present data.
+        * 
+        * @param line
+        *            the line to parse
+        * 
+        * @return TRUE if we are just done with one object or sub-object
+        * 
+        * @throws NoSuchFieldException
+        *             if the serialised data contains information about a field
+        *             which does actually not exist in the class we know of
+        * @throws NoSuchMethodException
+        *             if a class described in the serialised data cannot be created
+        *             because it is not compatible with this code
+        * @throws ClassNotFoundException
+        *             if a class described in the serialised data cannot be found
+        * @throws IOException
+        *             if the content cannot be read (for instance, corrupt data)
+        */
+       private boolean processLine(String line) throws NoSuchFieldException,
+                       NoSuchMethodException, ClassNotFoundException, IOException {
                // Defer to latest child if any
                if (child != null) {
                        if (child.processLine(line)) {
@@ -85,18 +188,34 @@ public class Importer {
                } else if (line.equals("}")) { // STOP: report self to parent
                        return true;
                } else if (line.startsWith("REF ")) { // REF: create/link self
-                       String ref = line.substring(4).split("@")[1];
+                       String[] tab = line.substring("REF ".length()).split("@");
+                       String type = tab[0];
+                       tab = tab[1].split(":");
+                       String ref = tab[0];
+
                        link = map.containsKey(ref);
                        if (link) {
                                me = map.get(ref);
                        } else {
-                               me = SerialUtils.createObject(line.substring(4).split("@")[0]);
+                               if (line.endsWith(":")) {
+                                       // construct
+                                       me = SerialUtils.createObject(type);
+                               } else {
+                                       // direct value
+                                       int pos = line.indexOf(":");
+                                       String encodedValue = line.substring(pos + 1);
+                                       me = SerialUtils.decode(encodedValue);
+                               }
                                map.put(ref, me);
                        }
-               } else { // FIELD: new field
+               } else { // FIELD: new field *or* direct simple value
                        if (line.endsWith(":")) {
                                // field value is compound
                                currentFieldName = line.substring(0, line.length() - 1);
+                       } else if (line.startsWith(":") || !line.contains(":")
+                                       || line.startsWith("\"") || CustomSerializer.isCustom(line)) {
+                               // not a field value but a direct value
+                               me = SerialUtils.decode(line);
                        } else {
                                // field value is direct
                                int pos = line.indexOf(":");
@@ -118,8 +237,7 @@ public class Importer {
        }
 
        private void setField(String name, Object value)
-                       throws NoSuchFieldException, SecurityException,
-                       IllegalArgumentException, IllegalAccessException {
+                       throws NoSuchFieldException {
 
                try {
                        Field field = me.getClass().getDeclaredField(name);
@@ -130,9 +248,49 @@ public class Importer {
                        throw new NoSuchFieldException(String.format(
                                        "Field \"%s\" was not found in object of type \"%s\".",
                                        name, me.getClass().getCanonicalName()));
+               } catch (Exception e) {
+                       throw new NoSuchFieldException(String.format(
+                                       "Internal error when setting \"%s.%s\": %s", me.getClass()
+                                                       .getCanonicalName(), name, e.getMessage()));
                }
        }
 
+       /**
+        * Find the given needle in the data and return its position (or -1 if not
+        * found).
+        * 
+        * @param data
+        *            the data to look through
+        * @param offset
+        *            the offset at wich to start searching
+        * @param needle
+        *            the needle to find
+        * 
+        * @return the position of the needle if found, -1 if not found
+        */
+       private int find(byte[] data, int offset, byte[] needle) {
+               for (int i = offset; i + needle.length - 1 < data.length; i++) {
+                       boolean same = true;
+                       for (int j = 0; j < needle.length; j++) {
+                               if (data[i + j] != needle[j]) {
+                                       same = false;
+                                       break;
+                               }
+                       }
+
+                       if (same) {
+                               return i;
+                       }
+               }
+
+               return -1;
+       }
+
+       /**
+        * Return the current deserialised value.
+        * 
+        * @return the current value
+        */
        public Object getValue() {
                return me;
        }