edu.harvard.hul.ois.jhove.module.html.HtmlDocDesc Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of html-hul Show documentation
HTML module developed by Harvard University Library
The newest version!
/**********************************************************************
 * Jhove - JSTOR/Harvard Object Validation Environment
 * Copyright 2004 by JSTOR and the President and Fellows of Harvard College
 **********************************************************************/

package edu.harvard.hul.ois.jhove.module.html;

import java.util.ArrayList;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;

import edu.harvard.hul.ois.jhove.ErrorMessage;
import edu.harvard.hul.ois.jhove.RepInfo;
import edu.harvard.hul.ois.jhove.module.HtmlModule;
import edu.harvard.hul.ois.jhove.module.utf8.Utf8BlockMarker;
import edu.harvard.hul.ois.jhove.module.xml.HtmlMetadata;

/**
 * This is an abstract class for processing an HTML document that has
 * been parsed into a List of HtmlElements. It defines common behavior
 * for all supported versions of HTML except XHTML. Subclasses
 * modify this base as needed.
 *
 * @author Gary McGath
 *
 */
public abstract class HtmlDocDesc {

	/** Metadata for this document. */
	private HtmlMetadata metadata;

	/**
	 * Generic list of supported tags. For efficiency, this is
	 * generated only once. Subclasses will need to get a copy
	 * of this list and make additions or deletions as necessary.
	 * They must not modify any of the existing
	 * members of the list.
	 */
	protected static HashMap commonTags;

	/**
	 * List of supported tags for this version of HTML. The subclass
	 * is responsible for generating this, typically using commonTags
	 * as a starting point.
	 */
	protected Map supportedElements;

	/** A representation of the HTML element. */
	protected HtmlTagDesc htmlElement;
	/** A representation of the HEAD element. */
	protected HtmlTagDesc headElement;
	/** A representation of the BODY element. */
	protected HtmlTagDesc bodyElement;
	/** A representation of the FRAMESET element. */
	protected HtmlTagDesc framesetElement;

	private HtmlStack elementStack;

	/** Header tags, which are invariant for all HTML versions. */
	protected static String[] headings = { "h1", "h2", "h3", "h4", "h5", "h6" };

	/**
	 * Validates the document and puts interesting properties into the
	 * RepInfo.
	 *
	 * @param elements
	 *            The element list constructed by the parser
	 * @param info
	 *            The RepInfo object which will be populated
	 *            with properties
	 */
	public boolean validate(List elements, RepInfo info) {
		// As we get to each open tag, we
		// check it against the corresponding HtmlTagDesc. If there isn't one,
		// we
		// mark the document as invalid but continue anyway; we create a
		// temporary
		// HtmlTagDesc object for the tag that we find, with the closing tag
		// indicated
		// as optional.
		// For each open tag, we push the HtmlTagDesc object onto the stack. We
		// check
		// if it's in the allowed content of the enclosing element. If not, we
		// report it
		// as an error but continue with it anyway.
		//
		// We special-case HTML, HEAD and BODY, which can be implied.
		// If a tag is found which requires the content model for one of
		// these, and it isn't on the stack, we just push it.

		metadata = new HtmlMetadata();
		elementStack = new HtmlStack();
		elementStack.setHeadElement(headElement);
		elementStack.setBodyElement(bodyElement);
		elementStack.setFramesetElement(framesetElement);
		Iterator iter = elements.iterator();
		while (iter.hasNext()) {
			JHElement elem = (JHElement) iter.next();
			if (elem instanceof JHDoctype) {
				// Doctype requires no further processing; grammar
				// will have already caught it if it's not at the top
				continue;
			} else if (elem instanceof JHOpenTag) {
				doOpenTag((JHOpenTag) elem, info);
			} else if (elem instanceof JHCloseTag) {
				doCloseTag((JHCloseTag) elem, info);
			} else if (elem instanceof JHErrorElement) {
				doErrorElement((JHErrorElement) elem, info);
			} else if (elem instanceof JHPCData) {
				doPCData((JHPCData) elem, info, metadata);
			}
		}
		// It's a requirement that there be at least a TITLE,
		// and thus an implicit or explicit HEAD element.
		if (!elementStack.isHeadSeen()) {
			info.setMessage(new ErrorMessage(
					MessageConstants.HTML_HUL_11));
			info.setValid(false);
		}
		return true;
	}

	/** Returns the metadata for this document. */
	public HtmlMetadata getMetadata() {
		return metadata;
	}

	/**
	 * Initialization called by subclass constructors after supportedElements
	 * has been assigned.
	 */
	protected void init() {
		htmlElement = (HtmlTagDesc) supportedElements.get("html");
		headElement = (HtmlTagDesc) supportedElements.get("head");
		bodyElement = (HtmlTagDesc) supportedElements.get("body");
	}

	/* Break out open tag code */
	private void doOpenTag(JHOpenTag tag, RepInfo info) {
		String name = tag.getName().toLowerCase();
		boolean unknownTag = false;

		ErrorMessage msg = tag.getErrorMessage();
		if (msg != null) {
			info.setMessage(msg);
			info.setWellFormed(false);
			// But keep going anyway!
		}

		/*
		 * If it's anything but an HTML tag, and the stack is empty,
		 * push an "HTML" element.
		 */
		if (elementStack.isEmpty() && !"html".equals(name)) {
                    JHOpenTag fakeTag = new JHOpenTag("html");
                    fakeTag.setElement(htmlElement);
                    elementStack.push(fakeTag);
			
		}
		HtmlTagDesc tagDesc = (HtmlTagDesc) supportedElements.get(name);
		if (tagDesc == null) {
			unknownTag = true;
		}
		// Check the context only if it's a known tag;
		// otherwise we'll issue a redundant error message.
		if (!unknownTag && !checkElementContext(tag, info)) {
			String toptag = null;
			if (!elementStack.isEmpty()) {
				JHOpenTag top = elementStack.top();
				toptag = top.getName();
			}
			info.setMessage(new ErrorMessage(
					MessageConstants.HTML_HUL_4,
					"Name = " + name + ", "
							+ (toptag != null ? "Container = " + toptag + ", "
									: "")
							+ "Line = " + tag.getLine() + ", Column = "
							+ tag.getColumn()));
			info.setValid(false);
		}
		if (unknownTag) {
			info.setMessage(
					new ErrorMessage(MessageConstants.HTML_HUL_6,
							"Name = " + name + ", Line = " + tag.getLine()
									+ ", Column = " + tag.getColumn()));
			info.setValid(false);
			// Make a temporary tag descriptor
			tagDesc = new HtmlTempTagDesc(name);
		}
		if (!unknownTag && info.getWellFormed() == RepInfo.TRUE) {
			/* Check if the attributes are valid */
			List atts = tag.getAttributes();
			Iterator iter = atts.iterator();
			// Create a list to accumulate all attribute names.
			List attNames = new ArrayList(atts.size());
			while (iter.hasNext()) {
				JHAttribute att = (JHAttribute) iter.next();
				String attName = att.getName();
				attNames.add(attName);
				String attVal = att.getValue();
				HtmlAttributeDesc attDesc = tagDesc.namedAttDesc(attName);
				if (attDesc == null) {
					info.setMessage(new ErrorMessage(
							MessageConstants.HTML_HUL_7,
							"Name = " + name + ", Attribute = " + attName
									+ ", Line = " + att.getLine()
									+ ", Column = " + att.getColumn()));
					info.setValid(false);
				} else {
					/* Check if value is legit */
					if (!attDesc.valueOK(attName, attVal)) {
						info.setMessage(new ErrorMessage(
								MessageConstants.HTML_HUL_8,
								"Element = " + name + ", Attribute = " + attName
										+ ", Value = " + attVal + ", Line = "
										+ att.getLine() + ", Column = "
										+ att.getColumn()));
						info.setValid(false);
					}
				}
				// Extract entities from attribute value
				if (attVal != null) {
					Iterator entIter = tag.getEntities(attVal).iterator();
					Utf8BlockMarker utf8BM = metadata.getUtf8BlockMarker();
					while (entIter.hasNext()) {
						String ent = (String) entIter.next();
						metadata.addEntity(ent);
						// If it's a numerical entity, note which UTF8 block
						// it's in
						try {
							if (ent.charAt(1) == '#') {
								int entval = Integer.parseInt(
										ent.substring(2, ent.length() - 1));
								utf8BM.markBlock(entval);
							}
						} catch (Exception e) {
							// Any exception means it's the wrong kind of entity
						}
					}
				}
			}
			// Check if all required attributes were found.
			List missingAtts = tagDesc.missingRequiredAttributes(attNames);
			if (!missingAtts.isEmpty()) {
				info.setValid(false);
				Iterator miter = missingAtts.iterator();
				while (miter.hasNext()) {
					String matt = (String) miter.next();
					info.setMessage(new ErrorMessage(
							MessageConstants.HTML_HUL_9,
							"Tag = " + name + ", Attribute = " + matt
									+ ", Line = " + tag.getLine()
									+ ", Column = " + tag.getColumn()));
				}
			}
		}
		tag.processElement(metadata);
		// If the content is empty, then a closing tag isn't permitted
		// (SGML handbook 7.3), so we don't push the open tag.
		// But if it's a temporary tag descriptor, we don't know
		// anything about it, so all guesses are wild. Push it anyway.
		if (tagDesc.isTemp() || !tagDesc.isContentEmpty()) {
			tag.setElement(tagDesc);
			elementStack.push(tag);
		}
	}

	private void doCloseTag(JHCloseTag tag, RepInfo info) {
		String name = tag.getName();
		// Dig down into the stack till we find an element which
		// matches this. If there's none, report the document
		// as not well formed. Also allow for the special case
		// of an empty body. (An empty head is illegal.)
		int idx = elementStack.search(name);
		if (idx == -1) {
			info.setMessage(new ErrorMessage(
					MessageConstants.HTML_HUL_10,
					"Name = " + name + ", Line = " + tag.getLine()
							+ ", Column = " + tag.getColumn()));
			info.setValid(false);
		} else {
			// Pop the stack down to the level of the matching tag.
			elementStack.popTo(idx);
		}

	}

	private static void doErrorElement(JHErrorElement elem, RepInfo info) {
		elem.reportError(info);
	}

	private void doPCData(JHPCData elem, RepInfo info, HtmlMetadata metadata) {
		// Pop any elements that have optional close tags and do not
		// allow PCDATA.
		if (elementStack.isEmpty()) {
			// PCData before any content. This generates an implicit
			// html and body if they haven't already been seen.
			// It also means the document isn't valid, since the title
			// should precede any PCData.
			info.setMessage(new ErrorMessage(
					MessageConstants.HTML_HUL_12));
			info.setValid(false);
			return;
		}
		HtmlTagDesc top = elementStack.top().getElement();
		if (top.isTemp() || top.allowsPCData()) {
			// We assume that PCData is allowed with unknown tags.
			elem.processPCData(elementStack, metadata);
			return;
		}
		// If we can pop elements with optional closing tags till we find
		// one that allows PCData, we should do that. But popping the
		// stack empty, as could happen if we're in a HEAD element, is
		// wrong. So we always allow two elements to remain on the stack.
		while (!top.isCloseTagRequired()) {
			if (elementStack.size() <= 2) {
				break;
			}
			elementStack.popp();
			top = elementStack.top().getElement();
			if (top.allowsPCData()) {
				elem.processPCData(elementStack, metadata);
				return;
			}
		}
		info.setMessage(new ErrorMessage(
				MessageConstants.HTML_HUL_13,
				"Line = " + elem.getLine() + ", Column = " + elem.getColumn()));
		info.setValid(false);
	}

	/*
	 * Returns true if the element is permissible at this point.
	 * This may pop elements off the stack and push implied tags.
	 */
	private boolean checkElementContext(JHOpenTag elem, RepInfo info) {
		/*
		 * We are guaranteed there's something on the stack
		 * unless the tag is "html", but Paranoia Is A Virtue
		 */
		String name = elem.getName();
		if (elementStack.isEmpty()) {                    
			return "html".equals(name);
		}
		if (elementStack.excludesTag(name)) {
			return false;
		}
		JHOpenTag top = elementStack.top();
		for (;;) {
			if (top.canGetMore() && top.allowsTag(name, this)) {
				top.countComponent();
				return true;
			}
			if (!top.canAdvance()) {
				/* Can't advance, can't stay put. */
				break;
			}
			top.advanceIndex();
		}

		/* Kludgy special-case code for optional tags */
		HtmlTagDesc topElem = top.getElement();
		if (topElem == htmlElement) {
			if (!elementStack.isHeadSeen()
					&& headElement.allowsTag(name, this)) {
				JHOpenTag fakeTag = new JHOpenTag("head");
				fakeTag.setElement(headElement);
				elementStack.push(fakeTag);
				return true;
			}
			if (!elementStack.isBodySeen() && bodyElement != null
					&& bodyElement.allowsTag(name, this)) {
				JHOpenTag fakeTag = new JHOpenTag("body");
				fakeTag.setElement(bodyElement);
				elementStack.push(fakeTag);
				return true;
			}
			return false;
		} else if (topElem == headElement) {
			if ("body".equals(name) || "frameset".equals(name)) {
				// Pop implied head end tag. Is this too much
				// special-casing?
				elementStack.popp();
				elementStack.push(elem);
				return true;
			} else if (!elementStack.isBodySeen() && bodyElement != null
					&& bodyElement.allowsTag(name, this)) {
				// Similar to above case except that the head is
				// implicitly terminated.
				elementStack.popp();
				JHOpenTag fakeTag = new JHOpenTag("body");
				fakeTag.setElement(bodyElement);
				elementStack.push(fakeTag);
				return true;
			} else {
				return false;
			}
		}

		// Pop elements till we find a valid context. If
		// the enclosing element doesn't have an optional close
		// tag, report an error but pop it anyway. But first
		// check if there even is a context to which we can pop things.
		boolean complained = false;
		boolean searchStack = false;
		if (elementStack.size() > 2) {
			Iterator iter = elementStack.iterator();
			// Discard html element
			iter.next();
			while (iter.hasNext()) {
				JHOpenTag otag = (JHOpenTag) iter.next();
				if (otag.allowsTag(name, this)) {
					searchStack = true;
					break;
				}
			}
		}
		if (searchStack) {
			// We've established we can pop down to something.
			while (elementStack.size() > 2) {
				if (!complained) {
					top = elementStack.top();
					topElem = top.getElement();
					if (topElem.isCloseTagRequired()) {
						info.setValid(false);
						info.setMessage(new ErrorMessage(
								MessageConstants.HTML_HUL_5,
								"Name = " + name + ", " + "Container = "
										+ top.getName() + ", " + "Line = "
										+ elem.getLine() + ", Column = "
										+ elem.getColumn()));
					}
				}
				elementStack.popp();
				top = elementStack.top();
				// topElem = top.getElement ();
				if (top.allowsTag(name, this)) {
					return true;
				}
				if (elementStack.isEmpty()) {
					break;
				}
			}
		}
		return false;
	}

	/** Adds all the Strings in an array to the end of a List. */
	protected static void addStringsToList(String[] names, List lst) {
		for (int i = 0; i < names.length; i++) {
			lst.add(names[i]);
		}
	}

	/**
	 * Adds an attribute to a List, with unrestricted values and
	 * type IMPLIED.
	 */
	protected static void addSimpleAttribute(List atts, String name) {
		atts.add(new HtmlAttributeDesc(name));
	}

	/**
	 * Adds an attribute to a List, with unrestricted values and
	 * type REQUIRED.
	 */
	protected static void addRequiredAttribute(List atts, String name) {
		atts.add(new HtmlAttributeDesc(name, null, HtmlAttributeDesc.REQUIRED));
	}

	/**
	 * Adds an attribute to a List, with the only permitted value being
	 * the name of the attribute. This kind of attribute is normally
	 * represented in HTML without an explicit value; in fact, some (most?)
	 * readers won't permit an explicit value.
	 */
	protected static void addSelfAttribute(List atts, String name) {
		atts.add(new HtmlAttributeDesc(name, new String[] { name },
				HtmlAttributeDesc.IMPLIED));
	}

	/** Removes excluded strings from a List. */
	protected static void removeStringsFromList(List lst, String[] strs) {
		for (int i = 0; i < strs.length; i++) {
			lst.remove(strs[i]);
		}
	}

	/** Pushes an element onto the element stack. */
	protected void pushElementStack(JHOpenTag tag) {
		elementStack.push(tag);
	}

}