[#4907] Add DSLContext.fetchFromHTML() to convert an HTML table into a jOOQ Result

jOOQ/src/main/java/org/jooq/DSLContext.java

@@ -2560,6 +2560,47 @@ public interface DSLContext extends Scope , AutoCloseable  {
     @Support
     Result<Record> fetchFromTXT(String string, String nullLiteral) throws DataAccessException;
 
+    /**
+     * Convert an HTML table into a jOOQ {@link Result}.
+     * <p>
+     * This is the inverse operation of {@link Result#formatHTML()}. It works
+     * according to the following parsing rules:
+     * <ul>
+     * <li>The input is expected to be well-formed XML. XHTML conformance is not
+     * required - i.e. unknown elements / attributes, or elements / attributes
+     * not specified here, such as <code>&lt;caption></code>,
+     * <code>&lt;thead></code>, <code>&lt;tbody></code> are simply ignored.</li>
+     * <li>The surrounding <code>&lt;table></code> element is optional, but it
+     * may appear only once</li>
+     * <li>A single row containing table headings <code>&lt;th></code> is
+     * allowed. Further rows containing table headings are ignored. Table
+     * headings define field names. In the absence of table headings, field
+     * names are generated.</li>
+     * <li>The first row <code>&lt;tr></code> specifies the number of columns in
+     * the table (regardless if it contains table headings or not). Subsequent
+     * rows containing less columns will be padded. Subsequent rows containing
+     * more columns will be truncated.</li>
+     * <li>Comments are ignored</li>
+     * <li>Nested tables are not supported</li>
+     * </ul>
+     * <p>
+     * Ideal input looks like this: <code><pre>
+     * &lt;table>
+     * &lt;tr>&lt;th>COL1&lt;/th>&lt;th>COL2&lt;/th>&lt;/tr>
+     * &lt;tr>&lt;td>1&lt;/td>&lt;td>a&lt;/td>&lt;/tr>
+     * &lt;tr>&lt;td>2&lt;/td>&lt;td>b&lt;/td>&lt;/tr>
+     * &lt;/table>
+     * </code>
+     * </pre>
+     *
+     * @param string The HTML-formatted string.
+     * @return The transformed result
+     * @throws DataAccessException If the supplied string does not adhere to the
+     *             above format rules.
+     */
+    @Support
+    Result<Record> fetchFromHTML(String string) throws DataAccessException;
+
     /**
      * Fetch all data from a CSV string.
      * <p>

jOOQ/src/main/java/org/jooq/impl/DefaultDSLContext.java

@@ -990,6 +990,11 @@ public class DefaultDSLContext extends AbstractScope implements DSLContext, Seri
         return fetchFromStringData(Utils.parseTXT(string, nullLiteral));
     }
 
+    @Override
+    public Result<Record> fetchFromHTML(String string) {
+        return fetchFromStringData(Utils.parseHTML(string));
+    }
+
     @Override
     public Result<Record> fetchFromCSV(String string) {
         return fetchFromCSV(string, true, ',');

jOOQ/src/main/java/org/jooq/impl/Utils.java

@@ -2734,6 +2734,45 @@ final class Utils {
         }
     }
 
+    private static final Pattern P_PARSE_HTML_ROW      = Pattern.compile("<tr>(.*?)</tr>");
+    private static final Pattern P_PARSE_HTML_COL_HEAD = Pattern.compile("<th>(.*?)</th>");
+    private static final Pattern P_PARSE_HTML_COL_BODY = Pattern.compile("<td>(.*?)</td>");
+
+    static List<String[]> parseHTML(String string) {
+
+        List<String[]> result = new ArrayList<String[]>();
+
+        Matcher mRow = P_PARSE_HTML_ROW.matcher(string);
+        while (mRow.find()) {
+            String row = mRow.group(1);
+            List<String> col = new ArrayList<String>();
+
+            // Header was not yet emitted
+            if (result.isEmpty()) {
+                Matcher mColHead = P_PARSE_HTML_COL_HEAD.matcher(row);
+
+                while (mColHead.find()) {
+                    col.add(mColHead.group(1));
+                }
+            }
+
+            if (col.isEmpty()) {
+                Matcher mColBody = P_PARSE_HTML_COL_BODY.matcher(row);
+
+                while (mColBody.find()) {
+                    col.add(mColBody.group(1));
+                }
+
+                if (result.isEmpty())
+                    result.add(fieldNames(col.size()));
+            }
+
+            result.add(col.toArray(new String[col.size()]));
+        }
+
+        return result;
+    }
+
     /**
      * Wrap a <code>DROP .. IF EXISTS</code> statement with
      * <code>BEGIN EXECUTE IMMEDIATE '...' EXCEPTION WHEN ... END;</code>, if