| 1 | package org.docx4j.fonts; |
|---|
| 2 | |
|---|
| 3 | import java.io.File; |
|---|
| 4 | import java.net.URL; |
|---|
| 5 | import java.util.HashMap; |
|---|
| 6 | import java.util.Iterator; |
|---|
| 7 | import java.util.List; |
|---|
| 8 | import java.util.Map; |
|---|
| 9 | |
|---|
| 10 | import org.apache.log4j.Logger; |
|---|
| 11 | import org.docx4j.fonts.fop.fonts.EmbedFontInfo; |
|---|
| 12 | import org.docx4j.fonts.fop.fonts.FontCache; |
|---|
| 13 | import org.docx4j.fonts.fop.fonts.FontResolver; |
|---|
| 14 | import org.docx4j.fonts.fop.fonts.FontSetup; |
|---|
| 15 | import org.docx4j.fonts.fop.fonts.FontTriplet; |
|---|
| 16 | import org.docx4j.fonts.fop.fonts.autodetect.FontFileFinder; |
|---|
| 17 | import org.docx4j.fonts.fop.fonts.autodetect.FontInfoFinder; |
|---|
| 18 | import org.docx4j.fonts.microsoft.MicrosoftFonts; |
|---|
| 19 | import org.docx4j.fonts.microsoft.MicrosoftFontsRegistry; |
|---|
| 20 | import org.docx4j.openpackaging.parts.WordprocessingML.ObfuscatedFontPart; |
|---|
| 21 | |
|---|
| 22 | //import com.lowagie.text.pdf.BaseFont; |
|---|
| 23 | |
|---|
| 24 | /** |
|---|
| 25 | * The fonts which are physically installed on the system. |
|---|
| 26 | * |
|---|
| 27 | * They can be discovered automatically, or you can |
|---|
| 28 | * just add specific fonts. |
|---|
| 29 | * |
|---|
| 30 | * @author dev |
|---|
| 31 | * |
|---|
| 32 | */ |
|---|
| 33 | public class PhysicalFonts { |
|---|
| 34 | |
|---|
| 35 | protected static Logger log = Logger.getLogger(PhysicalFonts.class); |
|---|
| 36 | |
|---|
| 37 | protected static FontCache fontCache; |
|---|
| 38 | |
|---|
| 39 | |
|---|
| 40 | /** These are the physical fonts on the system which we have discovered. */ |
|---|
| 41 | private final static Map<String, PhysicalFont> physicalFontMap; |
|---|
| 42 | public static Map<String, PhysicalFont> getPhysicalFonts() { |
|---|
| 43 | return physicalFontMap; |
|---|
| 44 | } |
|---|
| 45 | |
|---|
| 46 | private final static Map<String, PhysicalFont> physicalFontMapByFilenameLowercase; |
|---|
| 47 | |
|---|
| 48 | |
|---|
| 49 | // private final static Map<String, PhysicalFontFamily> physicalFontFamiliesMap; |
|---|
| 50 | // int lastSeenNumberOfPhysicalFonts = 0; |
|---|
| 51 | // |
|---|
| 52 | // |
|---|
| 53 | // /** Max difference for it to be considered an acceptable match. |
|---|
| 54 | // * Note that this value will depend on the weights in the |
|---|
| 55 | // * difference function. |
|---|
| 56 | // */ |
|---|
| 57 | // public static final int MATCH_THRESHOLD = 30; |
|---|
| 58 | |
|---|
| 59 | private static FontResolver fontResolver; |
|---|
| 60 | |
|---|
| 61 | // parse font to ascertain font info |
|---|
| 62 | private static FontInfoFinder fontInfoFinder; |
|---|
| 63 | |
|---|
| 64 | static { |
|---|
| 65 | |
|---|
| 66 | try { |
|---|
| 67 | |
|---|
| 68 | fontCache = FontCache.load(); |
|---|
| 69 | if (fontCache == null) { |
|---|
| 70 | fontCache = new FontCache(); |
|---|
| 71 | } |
|---|
| 72 | |
|---|
| 73 | physicalFontMap = new HashMap<String, PhysicalFont>(); |
|---|
| 74 | physicalFontMapByFilenameLowercase |
|---|
| 75 | = new HashMap<String, PhysicalFont>(); |
|---|
| 76 | |
|---|
| 77 | // physicalFontFamiliesMap = new HashMap<String, PhysicalFontFamily>(); |
|---|
| 78 | |
|---|
| 79 | fontResolver = FontSetup.createMinimalFontResolver(); |
|---|
| 80 | |
|---|
| 81 | // parse font to ascertain font info |
|---|
| 82 | fontInfoFinder = new FontInfoFinder(); |
|---|
| 83 | |
|---|
| 84 | // setupPhysicalFonts(); |
|---|
| 85 | |
|---|
| 86 | } catch (Exception exc) { |
|---|
| 87 | throw new RuntimeException(exc); |
|---|
| 88 | } |
|---|
| 89 | } |
|---|
| 90 | |
|---|
| 91 | /** |
|---|
| 92 | * Autodetect fonts available on the system. |
|---|
| 93 | * |
|---|
| 94 | */ |
|---|
| 95 | public final static void discoverPhysicalFonts() throws Exception { |
|---|
| 96 | |
|---|
| 97 | // Currently we use FOP - inspired by org.apache.fop.render.PrintRendererConfigurator |
|---|
| 98 | // iText also has a font discoverer (which we could use |
|---|
| 99 | // instead, but don't). |
|---|
| 100 | // It remains to be seen which of XSL FO or iText |
|---|
| 101 | // PDF generation becomes the most popular. |
|---|
| 102 | // (If it is iText, then there _may_ be something |
|---|
| 103 | // to be said for using their font discovery instead) |
|---|
| 104 | |
|---|
| 105 | |
|---|
| 106 | FontFileFinder fontFileFinder = new FontFileFinder(); |
|---|
| 107 | |
|---|
| 108 | // Automagically finds a list of font files on local system |
|---|
| 109 | // based on os.name |
|---|
| 110 | List fontFileList = fontFileFinder.find(); |
|---|
| 111 | for (Iterator iter = fontFileList.iterator(); iter.hasNext();) { |
|---|
| 112 | |
|---|
| 113 | URL fontUrl = getURL(iter.next()); |
|---|
| 114 | |
|---|
| 115 | // parse font to ascertain font info |
|---|
| 116 | FontInfoFinder finder = new FontInfoFinder(); |
|---|
| 117 | addPhysicalFont( fontUrl); |
|---|
| 118 | } |
|---|
| 119 | |
|---|
| 120 | // Add fonts from our Temporary Embedded Fonts dir |
|---|
| 121 | fontFileList = fontFileFinder.find( ObfuscatedFontPart.getTemporaryEmbeddedFontsDir() ); |
|---|
| 122 | for (Iterator iter = fontFileList.iterator(); iter.hasNext();) { |
|---|
| 123 | URL fontUrl = getURL(iter.next()); |
|---|
| 124 | addPhysicalFont( fontUrl); |
|---|
| 125 | } |
|---|
| 126 | |
|---|
| 127 | fontCache.save(); |
|---|
| 128 | |
|---|
| 129 | } |
|---|
| 130 | |
|---|
| 131 | private static URL getURL(Object o) throws Exception { |
|---|
| 132 | |
|---|
| 133 | if (o instanceof java.io.File) { |
|---|
| 134 | // Running in Tomcat |
|---|
| 135 | java.io.File f = (java.io.File)o; |
|---|
| 136 | return f.toURL(); |
|---|
| 137 | } else if (o instanceof java.net.URL) { |
|---|
| 138 | return (URL)o; |
|---|
| 139 | } else { |
|---|
| 140 | throw new Exception("Unexpected object:" + o.getClass().getName() ); |
|---|
| 141 | } |
|---|
| 142 | } |
|---|
| 143 | |
|---|
| 144 | private static boolean loggedWarningAlready = false; |
|---|
| 145 | |
|---|
| 146 | /** |
|---|
| 147 | * Add a physical font's EmbedFontInfo object. |
|---|
| 148 | * |
|---|
| 149 | * @param fontUrl eg new java.net.URL("file:" + path) |
|---|
| 150 | */ |
|---|
| 151 | public static void addPhysicalFont(URL fontUrl) { |
|---|
| 152 | |
|---|
| 153 | //List<EmbedFontInfo> embedFontInfoList = fontInfoFinder.find(fontUrl, fontResolver, fontCache); |
|---|
| 154 | EmbedFontInfo[] embedFontInfoList = fontInfoFinder.find(fontUrl, fontResolver, fontCache); |
|---|
| 155 | /* FOP r644208 (Bugzilla #44737) 3/04/08 made this an array, |
|---|
| 156 | // so if you are using non-patched FOP, it needs to be at least this revision |
|---|
| 157 | // (but doesn't seem to be in FOP 0.95 binary?!) */ |
|---|
| 158 | |
|---|
| 159 | if (embedFontInfoList==null) { |
|---|
| 160 | // Quite a few fonts exist that we can't seem to get |
|---|
| 161 | // EmbedFontInfo for. To be investigated. |
|---|
| 162 | log.warn("Aborting: " + fontUrl.toString() + " (can't get EmbedFontInfo[] .. try deleting fop-fonts.cache?)"); |
|---|
| 163 | return; |
|---|
| 164 | } |
|---|
| 165 | |
|---|
| 166 | StringBuffer debug = new StringBuffer(); |
|---|
| 167 | |
|---|
| 168 | for ( EmbedFontInfo fontInfo : embedFontInfoList ) { |
|---|
| 169 | |
|---|
| 170 | /* EmbedFontInfo has: |
|---|
| 171 | * - subFontName (if the underlying CustomFont is a TTC) |
|---|
| 172 | * - PostScriptName = CustomFont.getFontName() |
|---|
| 173 | * - FontTriplets named: |
|---|
| 174 | * - CustomFont.getFullName() with quotes stripped |
|---|
| 175 | * - CustomFont.getFontName() with whitespace stripped |
|---|
| 176 | * - each family name (with quotes stripped) |
|---|
| 177 | * |
|---|
| 178 | * By creating one PhysicalFont object |
|---|
| 179 | * per triplet, each referring to the same |
|---|
| 180 | * EmbedFontInfo, we increase the chances |
|---|
| 181 | * of a match |
|---|
| 182 | * |
|---|
| 183 | ComicSansMS |
|---|
| 184 | .. triplet Comic Sans MS (priority + 0 |
|---|
| 185 | .. triplet ComicSansMS (priority + 0 |
|---|
| 186 | |
|---|
| 187 | ComicSansMS-Bold |
|---|
| 188 | .. triplet Comic Sans MS Bold (priority + 0 |
|---|
| 189 | .. triplet ComicSansMS-Bold (priority + 0 |
|---|
| 190 | .. triplet Comic Sans MS (priority + 5 |
|---|
| 191 | * |
|---|
| 192 | * but the second triplet is what FOP creates where its |
|---|
| 193 | * getPostScriptName() |
|---|
| 194 | * does FontUtil.stripWhiteSpace(getFullName());. |
|---|
| 195 | * |
|---|
| 196 | * and the third is just the family name. |
|---|
| 197 | * |
|---|
| 198 | * So we only get the first. |
|---|
| 199 | * |
|---|
| 200 | */ |
|---|
| 201 | |
|---|
| 202 | |
|---|
| 203 | if (fontInfo == null) { |
|---|
| 204 | // return; |
|---|
| 205 | continue; |
|---|
| 206 | } |
|---|
| 207 | |
|---|
| 208 | debug.append("------- \n"); |
|---|
| 209 | |
|---|
| 210 | try { |
|---|
| 211 | debug.append(fontInfo.getPostScriptName() + "\n" ); |
|---|
| 212 | if (!fontInfo.isEmbeddable() ) { |
|---|
| 213 | // log.info(tokens[x] + " is not embeddable; skipping."); |
|---|
| 214 | |
|---|
| 215 | /* |
|---|
| 216 | * No point looking at this font, since if we tried to use it, |
|---|
| 217 | * later, we'd get: |
|---|
| 218 | * |
|---|
| 219 | * com.lowagie.text.DocumentException: file:/usr/share/fonts/truetype/ttf-tamil-fonts/lohit_ta.ttf cannot be embedded due to licensing restrictions. |
|---|
| 220 | at com.lowagie.text.pdf.TrueTypeFont.<init>(TrueTypeFont.java:364) |
|---|
| 221 | at com.lowagie.text.pdf.TrueTypeFont.<init>(TrueTypeFont.java:335) |
|---|
| 222 | at com.lowagie.text.pdf.BaseFont.createFont(BaseFont.java:399) |
|---|
| 223 | at com.lowagie.text.pdf.BaseFont.createFont(BaseFont.java:345) |
|---|
| 224 | at org.xhtmlrenderer.pdf.ITextFontResolver.addFont(ITextFontResolver.java:164) |
|---|
| 225 | |
|---|
| 226 | will be thrown if os_2.fsType == 2 |
|---|
| 227 | |
|---|
| 228 | */ |
|---|
| 229 | log.warn(fontInfo.getEmbedFile() + " is not embeddable; ignoring this font."); |
|---|
| 230 | |
|---|
| 231 | //return; |
|---|
| 232 | continue; |
|---|
| 233 | } |
|---|
| 234 | } catch (Exception e1) { |
|---|
| 235 | // NB isEmbeddable() only exists in our patched FOP |
|---|
| 236 | if (!loggedWarningAlready) { |
|---|
| 237 | log.warn("Not using patched FOP; isEmbeddable() method missing."); |
|---|
| 238 | loggedWarningAlready = true; |
|---|
| 239 | } |
|---|
| 240 | } |
|---|
| 241 | |
|---|
| 242 | PhysicalFont pf; |
|---|
| 243 | |
|---|
| 244 | // for (Iterator iterIn = fontInfo.getFontTriplets().iterator() ; iterIn.hasNext();) { |
|---|
| 245 | // FontTriplet triplet = (FontTriplet)iterIn.next(); |
|---|
| 246 | |
|---|
| 247 | FontTriplet triplet = (FontTriplet)fontInfo.getFontTriplets().get(0); |
|---|
| 248 | // There is one triplet for each of the font family names |
|---|
| 249 | // this font has, and we create a PhysicalFont object |
|---|
| 250 | // for each of them. For our purposes though, each of |
|---|
| 251 | // these physical font objects contains the same info |
|---|
| 252 | |
|---|
| 253 | String lower = fontInfo.getEmbedFile().toLowerCase(); |
|---|
| 254 | log.debug("Processing physical font: " + lower); |
|---|
| 255 | debug.append(".. triplet " + triplet.getName() |
|---|
| 256 | + " (priority " + triplet.getPriority() +"\n" ); |
|---|
| 257 | |
|---|
| 258 | pf = null; |
|---|
| 259 | // xhtmlrenderer's org.xhtmlrenderer.pdf.ITextFontResolver.addFont |
|---|
| 260 | // can handle |
|---|
| 261 | // .otf, .ttf, .ttc, .pfb |
|---|
| 262 | if (lower.endsWith(".otf") || lower.endsWith(".ttf") || lower.endsWith(".ttc") ) { |
|---|
| 263 | pf = new PhysicalFont(triplet.getName(), fontInfo, fontResolver); |
|---|
| 264 | } else if (lower.endsWith(".pfb") ) { |
|---|
| 265 | // See whether we have everything org.xhtmlrenderer.pdf.ITextFontResolver.addFont |
|---|
| 266 | // will need - for a .pfb file, it needs a corresponding .afm or .pfm |
|---|
| 267 | String afm = FontUtils.pathFromURL(lower); |
|---|
| 268 | afm = afm.substring(0, afm.length()-4 ) + ".afm"; // drop the 'file:' |
|---|
| 269 | log.debug("Looking for: " + afm); |
|---|
| 270 | File f = new File(afm); |
|---|
| 271 | if (f.exists()) { |
|---|
| 272 | |
|---|
| 273 | log.debug(".. found"); |
|---|
| 274 | |
|---|
| 275 | // Uncomment if you want to use the iText stuff in docx4j-extras |
|---|
| 276 | // // We're only interested if this font supports UTF-8 encoding |
|---|
| 277 | // // since otherwise iText can't use it (at least on a |
|---|
| 278 | // // UTF8 encoded XHTML document) |
|---|
| 279 | // try { |
|---|
| 280 | // BaseFont bf = BaseFont.createFont(afm, |
|---|
| 281 | // BaseFont.IDENTITY_H, |
|---|
| 282 | // BaseFont.NOT_EMBEDDED); |
|---|
| 283 | // } catch (java.io.UnsupportedEncodingException uee) { |
|---|
| 284 | // log.error(afm + " does not support UTF encoding, so ignoring"); |
|---|
| 285 | // continue; |
|---|
| 286 | // } catch (Exception e) { |
|---|
| 287 | // log.error(e); |
|---|
| 288 | // continue; |
|---|
| 289 | // } |
|---|
| 290 | pf = new PhysicalFont(triplet.getName(),fontInfo, fontResolver); |
|---|
| 291 | |
|---|
| 292 | |
|---|
| 293 | } else { |
|---|
| 294 | // Should we be doing afm first, or pfm? |
|---|
| 295 | String pfm = FontUtils.pathFromURL(lower); |
|---|
| 296 | pfm = pfm.substring(0, pfm.length()-4 ) + ".pfm"; // drop the 'file:' |
|---|
| 297 | log.debug("Looking for: " + pfm); |
|---|
| 298 | f = new File(pfm); |
|---|
| 299 | if (f.exists()) { |
|---|
| 300 | log.debug(".. found"); |
|---|
| 301 | |
|---|
| 302 | // Uncomment if you want to use the iText stuff in docx4j-extras |
|---|
| 303 | // // We're only interested if this font supports UTF-8 encoding |
|---|
| 304 | // try { |
|---|
| 305 | // BaseFont bf = BaseFont.createFont(pfm, |
|---|
| 306 | // BaseFont.IDENTITY_H, |
|---|
| 307 | // BaseFont.NOT_EMBEDDED); |
|---|
| 308 | // } catch (java.io.UnsupportedEncodingException uee) { |
|---|
| 309 | // log.error(pfm + " does not support UTF encoding, so ignoring"); |
|---|
| 310 | // continue; |
|---|
| 311 | // } catch (Exception e) { |
|---|
| 312 | // log.error(e); |
|---|
| 313 | // continue; |
|---|
| 314 | // } |
|---|
| 315 | pf = new PhysicalFont(triplet.getName(), fontInfo, fontResolver); |
|---|
| 316 | } else { |
|---|
| 317 | log.warn("Skipping " + triplet.getName() + "; couldn't find .afm or .pfm for : " + fontInfo.getEmbedFile()); |
|---|
| 318 | } |
|---|
| 319 | } |
|---|
| 320 | } else { |
|---|
| 321 | log.warn("Skipping " + triplet.getName() + "; unsupported type: " + fontInfo.getEmbedFile()); |
|---|
| 322 | } |
|---|
| 323 | |
|---|
| 324 | |
|---|
| 325 | if (pf!=null) { |
|---|
| 326 | |
|---|
| 327 | // Add it to the map |
|---|
| 328 | physicalFontMap.put(pf.getName(), pf); |
|---|
| 329 | log.debug("Added " + pf.getName() + " -> " + pf.getEmbeddedFile()); |
|---|
| 330 | |
|---|
| 331 | // We also need to add it to map by filename |
|---|
| 332 | String filename = pf.getEmbeddedFile(); |
|---|
| 333 | filename = filename.substring( filename.lastIndexOf("/")+1).toLowerCase(); |
|---|
| 334 | physicalFontMapByFilenameLowercase.put(filename, pf); |
|---|
| 335 | log.debug("added to filename map: " + filename); |
|---|
| 336 | |
|---|
| 337 | // String familyName = triplet.getName(); |
|---|
| 338 | // pf.setFamilyName(familyName); |
|---|
| 339 | // |
|---|
| 340 | // PhysicalFontFamily pff; |
|---|
| 341 | // if (physicalFontFamiliesMap.get(familyName)==null) { |
|---|
| 342 | // pff = new PhysicalFontFamily(familyName); |
|---|
| 343 | // physicalFontFamiliesMap.put(familyName, pff); |
|---|
| 344 | // } else { |
|---|
| 345 | // pff = physicalFontFamiliesMap.get(familyName); |
|---|
| 346 | // } |
|---|
| 347 | // pff.addFont(pf); |
|---|
| 348 | |
|---|
| 349 | } |
|---|
| 350 | } |
|---|
| 351 | |
|---|
| 352 | log.debug(debug.toString() ); |
|---|
| 353 | } |
|---|
| 354 | |
|---|
| 355 | public static PhysicalFont getBoldForm( PhysicalFont pf) { |
|---|
| 356 | |
|---|
| 357 | // look up the font in MicrosoftFontsRegistry |
|---|
| 358 | MicrosoftFonts.Font msFont = MicrosoftFontsRegistry.getMsFonts().get(pf.getName() ); |
|---|
| 359 | |
|---|
| 360 | if (msFont==null) { |
|---|
| 361 | log.warn("No entry in MicrosoftFontsRegistry for: " + pf.getName()); |
|---|
| 362 | return null; |
|---|
| 363 | } |
|---|
| 364 | |
|---|
| 365 | if (msFont.getBold()==null) { |
|---|
| 366 | log.debug("No bold form for: " + pf.getName()); |
|---|
| 367 | return null; |
|---|
| 368 | } else { |
|---|
| 369 | |
|---|
| 370 | // We have to go via the file name, grrr.. |
|---|
| 371 | // since MicrosoftFonts.xml doesn't give the associate font name |
|---|
| 372 | String filename = msFont.getBold().getFilename().toLowerCase(); |
|---|
| 373 | log.debug("Fetching: " + filename); |
|---|
| 374 | return physicalFontMapByFilenameLowercase.get(filename); |
|---|
| 375 | } |
|---|
| 376 | } |
|---|
| 377 | |
|---|
| 378 | public static PhysicalFont getBoldItalicForm( PhysicalFont pf) { |
|---|
| 379 | |
|---|
| 380 | // look up the font in MicrosoftFontsRegistry |
|---|
| 381 | MicrosoftFonts.Font msFont = MicrosoftFontsRegistry.getMsFonts().get(pf.getName() ); |
|---|
| 382 | |
|---|
| 383 | if (msFont==null) { |
|---|
| 384 | log.warn("No entry in MicrosoftFontsRegistry for: " + pf.getName()); |
|---|
| 385 | return null; |
|---|
| 386 | } |
|---|
| 387 | |
|---|
| 388 | if (msFont.getBolditalic()==null) { |
|---|
| 389 | log.debug("No Bolditalic form for: " + pf.getName()); |
|---|
| 390 | return null; |
|---|
| 391 | } else { |
|---|
| 392 | |
|---|
| 393 | // We have to go via the file name, grrr.. |
|---|
| 394 | // since MicrosoftFonts.xml doesn't give the associate font name |
|---|
| 395 | String filename = msFont.getBolditalic().getFilename().toLowerCase(); |
|---|
| 396 | log.debug("Fetching: " + filename); |
|---|
| 397 | return physicalFontMapByFilenameLowercase.get(filename); |
|---|
| 398 | } |
|---|
| 399 | } |
|---|
| 400 | |
|---|
| 401 | public static PhysicalFont getItalicForm( PhysicalFont pf) { |
|---|
| 402 | |
|---|
| 403 | // look up the font in MicrosoftFontsRegistry |
|---|
| 404 | MicrosoftFonts.Font msFont = MicrosoftFontsRegistry.getMsFonts().get(pf.getName() ); |
|---|
| 405 | |
|---|
| 406 | if (msFont==null) { |
|---|
| 407 | log.debug("No entry in MicrosoftFontsRegistry for: " + pf.getName()); |
|---|
| 408 | return null; |
|---|
| 409 | } |
|---|
| 410 | |
|---|
| 411 | if (msFont.getItalic()==null) { |
|---|
| 412 | log.info("No italic form for: " + pf.getName()); |
|---|
| 413 | return null; |
|---|
| 414 | } else { |
|---|
| 415 | |
|---|
| 416 | // We have to go via the file name, grrr.. |
|---|
| 417 | // since MicrosoftFonts.xml doesn't give the associate font name |
|---|
| 418 | String filename = msFont.getItalic().getFilename().toLowerCase(); |
|---|
| 419 | log.debug("Fetching: " + filename); |
|---|
| 420 | return physicalFontMapByFilenameLowercase.get(filename); |
|---|
| 421 | } |
|---|
| 422 | |
|---|
| 423 | } |
|---|
| 424 | |
|---|
| 425 | public static void main(String[] args) throws Exception { |
|---|
| 426 | |
|---|
| 427 | discoverPhysicalFonts(); |
|---|
| 428 | System.out.println("That should have listed your physical fonts (provided you have logging enabled)."); |
|---|
| 429 | } |
|---|
| 430 | |
|---|
| 431 | |
|---|
| 432 | } |
|---|
