Table clipper functionality (norma-0.7.0-alpha) and supporting analysis code

pom.xml

@@ -5,22 +5,27 @@
     <modelVersion>4.0.0</modelVersion>
 
     <properties>
-       <svg2xml.version>2.1.0</svg2xml.version>
+		   <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <!--  upstream -->
-       <pdf2svg.version>2.1.0</pdf2svg.version>
+       <!-- <svghtml.version>1.0.0-SNAPSHOT</svghtml.version> --> <!-- not used -->
+
+       <pdf2svg.version>2.4.0-SNAPSHOT</pdf2svg.version>
+
     </properties>
 
-    <groupId>org.contentmine</groupId>
+    <groupId>org.xmlcml</groupId>
     <artifactId>svg2xml</artifactId>
-    <version>${svg2xml.version}</version>
+    <version>2.4.0-SNAPSHOT</version>
     <description>Takes SVG from PDF2SVG and analyzes the domain-independent semantics</description>
 
     <dependencies>
+		<!-- uses new combined library through pdf2svg -->
         <dependency>
-            <groupId>org.contentmine</groupId>
+            <groupId>org.xmlcml</groupId>
             <artifactId>pdf2svg</artifactId>
             <version>${pdf2svg.version}</version>
         </dependency>
+        <!-- org.xmlcml:pdf2svg:jar:2.4.0-SNAPSHOT -->
     </dependencies>
 
 
@@ -33,11 +38,12 @@
         </resources>
 
         <plugins>
+        <!-- 
             <plugin>
                 <groupId>org.apache.maven.plugins</groupId>
                 <artifactId>maven-surefire-plugin</artifactId>
             </plugin>
-
+        -->
         			<!-- giant jar -->
 			<plugin>
 				<groupId>org.apache.maven.plugins</groupId>
@@ -76,6 +82,10 @@
 							<mainClass>org.xmlcml.svg2xml.pdf.PDFAnalayzer</mainClass>
 							<id>svg2xml</id>
 						</program>
+						<program>
+							<mainClass>org.xmlcml.svg2xml.page.PageCropper</mainClass>
+							<id>cropper</id>
+						</program>
 					</programs>
 				</configuration>
 			</plugin>

src/main/java/org/xmlcml/svg2xml/PDF2SVGConverterWrapper.java

@@ -0,0 +1,104 @@
+package org.xmlcml.svg2xml;
+
+import java.io.File;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+
+import org.apache.commons.io.FileExistsException;
+import org.apache.commons.io.FileUtils;
+import org.apache.log4j.Level;
+import org.apache.log4j.Logger;
+import org.xmlcml.graphics.svg.SVGSVG;
+import org.xmlcml.pdf2svg.PDF2SVGConverter;
+
+/** dummy class to enable new PDF2SVG to be slotted in.
+ * 
+ * wrapper for PDF2SVG
+ * 
+ * @author pm286
+ *
+ */
+public class PDF2SVGConverterWrapper {
+	private static final Logger LOG = Logger.getLogger(PDF2SVGConverterWrapper.class);
+	private PDF2SVGConverter converter;
+
+	static {
+		LOG.setLevel(Level.DEBUG);
+	}
+
+	public PDF2SVGConverterWrapper() {
+		converter = new PDF2SVGConverter();
+	}
+
+	public void run(String ...pdf2svgArgs) {
+		converter.run(pdf2svgArgs);
+		return;
+	}
+
+	public List<SVGSVG> getPageList() {
+		return converter.getPageList();
+	}
+
+	public void createSVGFromPDF(File pdfOrigDir, File projectDir) {
+		projectDir.mkdirs();
+		File[] pdfDirs = pdfOrigDir.listFiles();
+		for (File pdfDir : pdfDirs) {
+			File projectPdfDir = new File(projectDir, pdfDir.getName());
+			projectPdfDir.mkdirs();
+			File pdfFile = new File(pdfDir, "fulltext.pdf");
+			File svgDir = new File(projectPdfDir, "svg/");
+			svgDir.mkdirs();
+
+			String[] args0 = {"-logger", "-infofiles", "-logglyphs", "-outdir", svgDir.toString(), pdfFile.toString()};
+			List<String> argList = new ArrayList<String>(Arrays.asList(args0));
+			argList.add("-compact");
+			run(argList.toArray(new String[0]));
+
+			File pngDir = new File(projectPdfDir, "png/");
+			List<File> pngFiles = new ArrayList<File>(FileUtils.listFiles(svgDir, new String[] {"png"}, false));
+			for (File pngFile : pngFiles) {
+				movePngFileToDirectory(pngDir, pngFile);
+			}
+		}
+	}
+
+	private void movePngFileToDirectory(File pngDir, File pngFile) {
+		if (FileUtils.sizeOf(pngFile) == 0) {
+			pngFile.delete();
+		} else {
+			try {
+				FileUtils.moveToDirectory(pngFile, pngDir, true);
+			} catch (FileExistsException fee) {
+				File oldFile = new File(pngDir, pngFile.getName());
+				try {
+					oldFile.delete();
+					FileUtils.moveToDirectory(pngFile, pngDir, true);
+				} catch (IOException e) {
+					// TODO Auto-generated catch block
+					e.printStackTrace();
+				}
+			} catch (Exception e) {
+				LOG.error("cannot move file: "+e);
+			}
+		}
+	}
+
+	/** 
+	 * Page numbered from ONE
+	 * 
+	 * @param file
+	 * @param page
+	 * @return
+	 */
+	public static SVGSVG getSVGPageFromPDF(File file, int page) {
+		PDF2SVGConverterWrapper converter = new PDF2SVGConverterWrapper();
+		converter.run("-outdir target "+file);
+		LOG.warn("PDF2SVGConverter shorted out");
+		SVGSVG svgPage = (page < 1 || page > converter.getPageList().size() ? null : converter.getPageList().get(page - 1));
+		return svgPage;
+	}
+
+
+}