opentype.php

Bug tracker, and reporter.

     * thickness for an underline, which is used by our layout code.
     *
     * @throws Zend_Pdf_Exception
     */
    protected function _parsePostTable()
    {
        $this->_jumpToTable('post');

        /* We can read versions 1-4 tables.
         */
        $tableVersion = $this->_readTableVersion(1, 4);

        $this->italicAngle = $this->readFixed(16, 16);

        $this->underlinePosition = $this->readInt(2);
        $this->underlineThickness = $this->readInt(2);

        $fixedPitch = $this->readUInt(4);
        $this->isMonospaced = ($fixedPitch !== 0);

        /* Skip over PostScript virtual memory usage.
         */
        $this->skipBytes(16);

        /* The format of the remainder of this table is dependent on the table
         * version. However, since it contains glyph ordering information and
         * PostScript names which we don't use, move on. (This may change at
         * some point in the future though...)
         */
    }


    /**
     * Parses the OpenType hhea (Horizontal Header) table.
     *
     * The hhea table contains information used for horizontal layout. It also
     * contains some vertical layout information for Apple systems. The vertical
     * layout information for the PDF file is usually taken from the OS/2 table.
     *
     * @throws Zend_Pdf_Exception
     */
    protected function _parseHheaTable()
    {
        $this->_jumpToTable('hhea');

        /* We can read any version 1 table.
         */
        $tableVersion = $this->_readTableVersion(1, 1);

        /* The typographic ascent, descent, and line gap values are Apple-
         * specific. Similar values exist in the OS/2 table. We'll use these
         * values unless better values are found in OS/2.
         */
        $this->ascent = $this->readInt(2);
        $this->descent = $this->readInt(2);
        $this->lineGap = $this->readInt(2);

        /* The descent value is supposed to be negative--it's the distance
         * relative to the baseline. However, some fonts improperly store a
         * positive value in this field. If a positive value is found, flip the
         * sign and record a warning in the debug log that we did this.
         */
        if ($this->descent > 0) {
            $this->_debugLog('Warning: Font should specify negative descent. Actual: %d; Using %d',
                             $this->descent, -$this->descent);
            $this->descent = -$this->descent;
        }

        /* Skip over advance width, left and right sidebearing, max x extent,
         * caret slope rise, run, and offset, and the four reserved fields.
         */
        $this->skipBytes(22);

        /* These values are needed to read the hmtx table.
         */
        $this->metricDataFormat = $this->readInt(2);
        $this->numberHMetrics = $this->readUInt(2);
        $this->_debugLog('hmtx data format: %d; number of metrics: %d',
                         $this->metricDataFormat, $this->numberHMetrics);
    }


    /**
     * Parses the OpenType hhea (Horizontal Header) table.
     *
     * The hhea table contains information used for horizontal layout. It also
     * contains some vertical layout information for Apple systems. The vertical
     * layout information for the PDF file is usually taken from the OS/2 table.
     *
     * @throws Zend_Pdf_Exception
     */
    protected function _parseMaxpTable()
    {
        $this->_jumpToTable('maxp');

        /* We don't care about table version.
         */
        $this->_readTableVersion(0, 1);

        /* The number of glyphs in the font.
         */
        $this->numGlyphs = $this->readUInt(2);
        $this->_debugLog('number of glyphs: %d', $this->numGlyphs);
        
        // Skip other maxp table entries (if presented with table version 1.0)...
    }


    /**
     * Parses the OpenType OS/2 (OS/2 and Windows Metrics) table.
     *
     * The OS/2 table contains additional metrics data that is required to use
     * the font on the OS/2 or Microsoft Windows platforms. It is not required
     * for Macintosh fonts, so may not always be present. When available, we use
     * this table to determine most of the vertical layout and stylistic
     * information and for the font.
     *
     * @throws Zend_Pdf_Exception
     */
    protected function _parseOs2Table()
    {
        if (! $this->numberHMetrics) {
            throw new Zend_Pdf_Exception("hhea table must be parsed prior to calling this method",
                                         Zend_Pdf_Exception::PARSED_OUT_OF_ORDER);
        }

        try {
            $this->_jumpToTable('OS/2');
        } catch (Zend_Pdf_Exception $exception) {
            /* This table is not always present. If missing, use default values.
             */
            if ($exception->getCode() == Zend_Pdf_Exception::REQUIRED_TABLE_NOT_FOUND) {
                $this->_debugLog('No OS/2 table found. Using default values');
                $this->fontWeight = Zend_Pdf_Font::WEIGHT_NORMAL;
                $this->fontWidth = Zend_Pdf_Font::WIDTH_NORMAL;
                $this->isEmbeddable = true;
                $this->isSubsettable = true;
                $this->strikeThickness = $this->unitsPerEm * 0.05;
                $this->strikePosition  = $this->unitsPerEm * 0.225;
                $this->isSerifFont      = false;    // the style of the font is unknown
                $this->isSansSerifFont  = false;
                $this->isOrnamentalFont = false;
                $this->isScriptFont     = false;
                $this->isSymbolicFont   = false;
                $this->isAdobeLatinSubset = false;
                $this->vendorID = '';
                $this->xHeight = 0;
                $this->capitalHeight = 0;
                return;
            } else {
                /* Something else went wrong. Throw this exception higher up the chain.
                 */
                throw $exception;
            }
        }

        /* Version 0 tables are becoming rarer these days. They are only found
         * in older fonts.
         *
         * Version 1 formally defines the Unicode character range bits and adds
         * two new fields to the end of the table.
         *
         * Version 2 defines several additional flags to the embedding bits
         * (fsType field) and five new fields to the end of the table.
         *
         * Versions 2 and 3 are structurally identical. There are only two
         * significant differences between the two: First, in version 3, the
         * average character width (xAvgCharWidth field) is calculated using all
         * non-zero width glyphs in the font instead of just the Latin lower-
         * case alphabetic characters; this doesn't affect us. Second, in
         * version 3, the embedding bits (fsType field) have been made mutually
         * exclusive; see additional discusson on this below.
         *
         * We can understand all four of these table versions.
         */
        $tableVersion = $this->readUInt(2);
        if (($tableVersion < 0) || ($tableVersion > 3)) {
            throw new Zend_Pdf_Exception("Unable to read version $tableVersion table",
                                         Zend_Pdf_Exception::DONT_UNDERSTAND_TABLE_VERSION);
        }
        $this->_debugLog('Version %d table', $tableVersion);

        $this->averageCharWidth = $this->readInt(2);

        /* Indicates the visual weight and aspect ratio of the characters. Used
         * primarily to logically sort fonts in lists. Also used to help choose
         * a more appropriate substitute font when necessary. See the WEIGHT_
         * and WIDTH_ constants defined in Zend_Pdf_Font.
         */
        $this->fontWeight = $this->readUInt(2);
        $this->fontWidth  = $this->readUInt(2);

        /* Describes the font embedding licensing rights. We can only embed and
         * subset a font when given explicit permission.
         *
         * NOTE: We always interpret these bits according to the rules defined
         * in version 3 of this table, regardless of the actual version. This
         * means we will perform our checks in order from the most-restrictive
         * to the least.
         */
        $embeddingFlags = $this->readUInt(2);
        $this->_debugLog('Embedding flags: %d', $embeddingFlags);
        if ($this->isBitSet(9, $embeddingFlags)) {
            /* Only bitmaps may be embedded. We don't have the ability to strip
             * outlines from fonts yet, so this means no embed.
             */
            $this->isEmbeddable = false;
        } else if ($this->isBitSet(1, $embeddingFlags)) {
            /* Restricted license embedding. We currently don't have any way to
             * enforce this, so interpret this as no embed. This may be revised
             * in the future...
             */
            $this->isEmbeddable = false;
        } else {
            /* The remainder of the bit settings grant us permission to embed
             * the font. There may be additional usage rights granted or denied
             * but those only affect the PDF viewer application, not our code.
             */
            $this->isEmbeddable = true;
        }
        $this->_debugLog('Font ' . ($this->isEmbeddable ? 'may' : 'may not') . ' be embedded');
        $isSubsettable = $this->isBitSet($embeddingFlags, 8);

        /* Recommended size and offset for synthesized subscript characters.
         */
        $this->subscriptXSize = $this->readInt(2);
        $this->subscriptYSize = $this->readInt(2);
        $this->subscriptXOffset = $this->readInt(2);
        $this->subscriptYOffset = $this->readInt(2);

        /* Recommended size and offset for synthesized superscript characters.
         */
        $this->superscriptXSize = $this->readInt(2);
        $this->superscriptYSize = $this->readInt(2);
        $this->superscriptXOffset = $this->readInt(2);
        $this->superscriptYOffset = $this->readInt(2);

        /* Size and vertical offset for the strikethrough.
         */
        $this->strikeThickness = $this->readInt(2);
        $this->strikePosition  = $this->readInt(2);

        /* Describes the class of font: serif, sans serif, script. etc. These
         * values are defined here:
         *   http://www.microsoft.com/OpenType/OTSpec/ibmfc.htm
         */
        $familyClass = ($this->readUInt(2) >> 8);    // don't care about subclass
        $this->_debugLog('Font family class: %d', $familyClass);
        $this->isSerifFont      = ((($familyClass >= 1) && ($familyClass <= 5)) ||
                                   ($familyClass == 7));
        $this->isSansSerifFont  = ($familyClass == 8);
        $this->isOrnamentalFont = ($familyClass == 9);
        $this->isScriptFont     = ($familyClass == 10);
        $this->isSymbolicFont   = ($familyClass == 12);

        /* Skip over the PANOSE number. The interesting values for us overlap
         * with the font family class defined above.
         */
        $this->skipBytes(10);

        /* The Unicode range is made up of four 4-byte unsigned long integers
         * which are used as bitfields covering a 128-bit range. Each bit
         * represents a Unicode code block. If the bit is set, this font at
         * least partially covers the characters in that block.
         */
        $unicodeRange1 = $this->readUInt(4);
        $unicodeRange2 = $this->readUInt(4);
        $unicodeRange3 = $this->readUInt(4);
        $unicodeRange4 = $this->readUInt(4);
        $this->_debugLog('Unicode ranges: 0x%x 0x%x 0x%x 0x%x',
                        $unicodeRange1, $unicodeRange2, $unicodeRange3, $unicodeRange4);

        /* The Unicode range is currently only used to decide if the character
         * set covered by the font is a subset of the Adobe Latin set, meaning
         * it only has the basic latin set. If it covers any other characters,
         * even any of the extended latin characters, it is considered symbolic
         * to PDF and must be described differently in the Font Descriptor.
         */
        /** 
         * @todo Font is recognized as Adobe Latin subset font if it only contains
         * Basic Latin characters (only bit 0 of Unicode range bits is set).
         * Actually, other Unicode subranges like General Punctuation (bit 31) also 
         * fall into Adobe Latin characters. So this code has to be modified.
         */
        $this->isAdobeLatinSubset = (($unicodeRange1 == 1) && ($unicodeRange2 == 0) &&
                                      ($unicodeRange3 == 0) && ($unicodeRange4 == 0));
        $this->_debugLog(($this->isAdobeLatinSubset ? 'Is' : 'Is not') . ' a subset of Adobe Latin');

        $this->vendorID = $this->readBytes(4);

        /* Skip the font style bits. We use the values found in the 'head' table.
         * Also skip the first Unicode and last Unicode character indicies. Our
         * cmap implementation does not need these values.
         */
        $this->skipBytes(6);

        /* Typographic ascender, descender, and line gap. These values are
         * preferred to those in the 'hhea' table.
         */
        $this->ascent = $this->readInt(2);
        $this->descent = $this->readInt(2);
        $this->lineGap = $this->readInt(2);

        /* The descent value is supposed to be negative--it's the distance
         * relative to the baseline. However, some fonts improperly store a
         * positive value in this field. If a positive value is found, flip the
         * sign and record a warning in the debug log that we did this.
         */
        if ($this->descent > 0) {
            $this->_debugLog('Warning: Font should specify negative descent. Actual: %d; Using %d',
                             $this->descent, -$this->descent);
            $this->descent = -$this->descent;
        }

        /* Skip over Windows-specific ascent and descent.
         */
        $this->skipBytes(4);

        /* Versions 0 and 1 tables do not contain the x or capital height
         * fields. Record zero for unknown.
         */
        if ($tableVersion < 2) {
            $this->xHeight = 0;
            $this->capitalHeight = 0;
        } else {

            /* Skip over the Windows code page coverages. We are only concerned
             * with Unicode coverage.
             */
            $this->skipBytes(8);

            $this->xHeight = $this->readInt(2);
            $this->capitalHeight = $this->readInt(2);

            /* Ignore the remaining fields in this table. They are Windows-specific.
             */
        }
        /**
         * @todo Obtain the x and capital heights from the 'glyf' table if they
         *   haven't been supplied here instead of storing zero.
         */
    }


    /**
     * Parses the OpenType hmtx (Horizontal Metrics) table.
     *
     * The hmtx table contains the horizontal metrics for every glyph contained
     * within the font. These are the critical values for horizontal layout of
     * text.
     *
     * @throws Zend_Pdf_Exception
     */
    protected function _parseHmtxTable()
    {
        $this->_jumpToTable('hmtx');

        if (! $this->numberHMetrics) {
            throw new Zend_Pdf_Exception("hhea table must be parsed prior to calling this method",
                                         Zend_Pdf_Exception::PARSED_OUT_OF_ORDER);
        }

        /* We only understand version 0 tables.
         */
        if ($this->metricDataFormat != 0) {
            throw new Zend_Pdf_Exception("Unable to read format $this->metricDataFormat table.",
                                         Zend_Pdf_Exception::DONT_UNDERSTAND_TABLE_VERSION);
        }

        /* The hmtx table has no header. For each glpyh in the font, it contains