paul@0 | 1 | # -*- coding: iso-8859-1 -*- |
paul@0 | 2 | """ |
paul@0 | 3 | MoinMoin - Export as PDF action |
paul@0 | 4 | |
paul@0 | 5 | @copyright: 2013 by Paul Boddie <paul@boddie.org.uk> |
paul@0 | 6 | @license: GNU GPL (v2 or later), see COPYING.txt for details. |
paul@0 | 7 | """ |
paul@0 | 8 | |
paul@3 | 9 | from MoinMoin import caching |
paul@3 | 10 | from MoinMoin.action import ActionBase, cache |
paul@0 | 11 | from MoinMoin.wikiutil import escape |
paul@4 | 12 | from MoinMoin.Page import Page |
paul@0 | 13 | from MoinSupport import ActionSupport, escattr, getFormatterClass, formatText, get_send_headers |
paul@0 | 14 | from os.path import join |
paul@1 | 15 | import subprocess, os, codecs |
paul@0 | 16 | |
paul@0 | 17 | __version__ = "0.1" |
paul@0 | 18 | |
paul@0 | 19 | Dependencies = ['page'] |
paul@0 | 20 | |
paul@0 | 21 | # Configuration settings. |
paul@0 | 22 | |
paul@1 | 23 | # Choose one value for the export mode. |
paul@1 | 24 | |
paul@1 | 25 | PDF_EXPORT_MODE = "docbook" |
paul@2 | 26 | #PDF_EXPORT_MODE = "wkhtmltopdf" |
paul@1 | 27 | #PDF_EXPORT_MODE = "htmldoc" |
paul@1 | 28 | |
paul@1 | 29 | # Settings for "docbook" mode. |
paul@1 | 30 | |
paul@0 | 31 | XSLT_PROCESSOR = "/usr/bin/xsltproc" |
paul@0 | 32 | FO_PROCESSOR = "/usr/bin/fop" |
paul@0 | 33 | DOCBOOK_STYLESHEET_BASE = "/usr/share/xml/docbook/stylesheet" |
paul@0 | 34 | |
paul@1 | 35 | # Tool settings for "docbook" mode. |
paul@0 | 36 | |
paul@0 | 37 | DOCBOOK_TO_FO_STYLESHEET = "docbook-xsl/fo/docbook.xsl" |
paul@0 | 38 | |
paul@2 | 39 | # Settings for "wkhtmltopdf" mode. |
paul@2 | 40 | |
paul@2 | 41 | XVFB_WRAPPER = "/usr/bin/xvfb-run" |
paul@2 | 42 | WKHTMLTOPDF_PROCESSOR = "/usr/bin/wkhtmltopdf" |
paul@2 | 43 | |
paul@1 | 44 | # Settings for "htmldoc" mode. |
paul@1 | 45 | |
paul@1 | 46 | HTMLDOC_PROCESSOR = "/usr/bin/htmldoc" |
paul@1 | 47 | |
paul@0 | 48 | # NOTE: From docbook-xsl/fo/param.xsl. |
paul@0 | 49 | |
paul@1 | 50 | docbook_paper_sizes = [ |
paul@0 | 51 | "A0", "A1", "A2", "A3", "A4", "A5", "A6", "A7", "A8", "A9", "A10", |
paul@0 | 52 | "B0", "B1", "B2", "B3", "B4", "B5", "B6", "B7", "B8", "B9", "B10", |
paul@0 | 53 | "C0", "C1", "C2", "C3", "C4", "C5", "C6", "C7", "C8", "C9", "C10", |
paul@0 | 54 | "A4landscape", "USletter", "USlandscape", "4A0", "2A0", |
paul@0 | 55 | ] |
paul@0 | 56 | |
paul@1 | 57 | docbook_paper_size_labels = { |
paul@0 | 58 | "A4landscape" : "A4 landscape", |
paul@0 | 59 | "USletter" : "US letter", |
paul@0 | 60 | "USlandscape" : "US landscape", |
paul@0 | 61 | "4A0" : "Quadruple A0", |
paul@0 | 62 | "2A0" : "Double A0" |
paul@0 | 63 | } |
paul@0 | 64 | |
paul@2 | 65 | wkhtmltopdf_paper_sizes = [ |
paul@2 | 66 | "A4", "Letter" |
paul@2 | 67 | ] |
paul@2 | 68 | |
paul@2 | 69 | wkhtmltopdf_paper_size_labels = {} |
paul@2 | 70 | |
paul@1 | 71 | # NOTE: From the htmldoc man page. |
paul@1 | 72 | |
paul@1 | 73 | htmldoc_paper_sizes = [ |
paul@1 | 74 | "a4", "legal", "letter", "universal" |
paul@1 | 75 | ] |
paul@1 | 76 | |
paul@1 | 77 | htmldoc_paper_size_labels = { |
paul@1 | 78 | "a4" : "A4", |
paul@1 | 79 | "legal" : "US legal", |
paul@1 | 80 | "letter" : "US letter", |
paul@1 | 81 | "universal" : "US universal" |
paul@1 | 82 | } |
paul@1 | 83 | |
paul@2 | 84 | paper_sizes = { |
paul@2 | 85 | "docbook" : docbook_paper_sizes, |
paul@2 | 86 | "wkhtmltopdf" : wkhtmltopdf_paper_sizes, |
paul@2 | 87 | "htmldoc" : htmldoc_paper_sizes |
paul@2 | 88 | } |
paul@2 | 89 | |
paul@2 | 90 | paper_size_labels = { |
paul@2 | 91 | "docbook" : docbook_paper_size_labels, |
paul@2 | 92 | "wkhtmltopdf" : wkhtmltopdf_paper_size_labels, |
paul@2 | 93 | "htmldoc" : htmldoc_paper_size_labels |
paul@2 | 94 | } |
paul@2 | 95 | |
paul@0 | 96 | class ExportPDF(ActionBase, ActionSupport): |
paul@0 | 97 | |
paul@0 | 98 | "Export the current page as PDF." |
paul@0 | 99 | |
paul@1 | 100 | mode = PDF_EXPORT_MODE |
paul@1 | 101 | |
paul@1 | 102 | def _get_paper_sizes(self): |
paul@2 | 103 | return paper_sizes.get(self.mode) |
paul@1 | 104 | |
paul@1 | 105 | def _get_paper_size_labels(self): |
paul@2 | 106 | return paper_size_labels.get(self.mode) |
paul@1 | 107 | |
paul@0 | 108 | def get_form_html(self, buttons_html): |
paul@0 | 109 | |
paul@0 | 110 | "Return the action's form incorporating the 'buttons_html'." |
paul@0 | 111 | |
paul@0 | 112 | _ = self._ |
paul@0 | 113 | request = self.request |
paul@0 | 114 | form = self.get_form() |
paul@0 | 115 | |
paul@0 | 116 | paper_size = form.get("paper-size", ["A4"])[0] |
paul@0 | 117 | |
paul@0 | 118 | paper_size_options = [] |
paul@2 | 119 | paper_size_labels = self._get_paper_size_labels() or {} |
paul@0 | 120 | |
paul@2 | 121 | for size in self._get_paper_sizes() or []: |
paul@0 | 122 | paper_size_options.append('<option value="%s" %s>%s</option>' % ( |
paul@0 | 123 | escattr(size), self._get_selected(size, paper_size), |
paul@0 | 124 | escape(_(paper_size_labels.get(size) or size)) |
paul@0 | 125 | )) |
paul@0 | 126 | |
paul@0 | 127 | d = { |
paul@0 | 128 | "paper_size_label" : escape(_("Paper size")), |
paul@0 | 129 | "paper_size_options" : u"".join(paper_size_options), |
paul@0 | 130 | "buttons_html" : buttons_html, |
paul@4 | 131 | "rev" : escattr(form.get("rev", ["0"])[0]), |
paul@0 | 132 | } |
paul@0 | 133 | |
paul@0 | 134 | return u"""\ |
paul@4 | 135 | <input name="rev" type="hidden" value="%(rev)s" /> |
paul@0 | 136 | <table> |
paul@0 | 137 | <tr> |
paul@0 | 138 | <td class="label"><label>%(paper_size_label)s</label></td> |
paul@0 | 139 | <td><select name="paper-size">%(paper_size_options)s</select></td> |
paul@0 | 140 | </tr> |
paul@0 | 141 | <tr> |
paul@0 | 142 | <td></td> |
paul@0 | 143 | <td class="buttons">%(buttons_html)s</td> |
paul@0 | 144 | </tr> |
paul@0 | 145 | </table> |
paul@0 | 146 | """ % d |
paul@0 | 147 | |
paul@0 | 148 | def do_action(self): |
paul@0 | 149 | |
paul@0 | 150 | "Attempt to post a comment." |
paul@0 | 151 | |
paul@0 | 152 | _ = self._ |
paul@0 | 153 | form = self.get_form() |
paul@3 | 154 | request = self.request |
paul@0 | 155 | |
paul@4 | 156 | # Permit other revisions, but only if the current revision is readable. |
paul@4 | 157 | |
paul@4 | 158 | if not request.user.may.read(self.page.page_name): |
paul@4 | 159 | return 0, _("This page no longer allows read access.") |
paul@4 | 160 | |
paul@4 | 161 | self.page = Page(request, self.page.page_name, rev=int(form.get("rev", ["0"])[0])) |
paul@4 | 162 | |
paul@4 | 163 | # Check the paper size. |
paul@4 | 164 | |
paul@0 | 165 | paper_size = form.get("paper-size", [""])[0] |
paul@0 | 166 | |
paul@2 | 167 | if not paper_size in self._get_paper_sizes() or []: |
paul@0 | 168 | return 0, _("A paper size must be chosen.") |
paul@0 | 169 | |
paul@3 | 170 | # See if the revision is cached. |
paul@3 | 171 | |
paul@4 | 172 | cache_key = cache.key(request, content="%s-%s" % (self.page.get_real_rev(), paper_size)) |
paul@3 | 173 | cache_entry = caching.CacheEntry(request, self.page, cache_key, scope="item") |
paul@3 | 174 | |
paul@3 | 175 | # Open any available cache entry and read it. |
paul@3 | 176 | |
paul@3 | 177 | if cache_entry.exists(): |
paul@3 | 178 | cache_entry.open() |
paul@3 | 179 | try: |
paul@3 | 180 | self._write_pdf(cache_entry.read()) |
paul@3 | 181 | return 1, None |
paul@3 | 182 | finally: |
paul@3 | 183 | cache_entry.close() |
paul@3 | 184 | |
paul@3 | 185 | # Otherwise, prepare the PDF. |
paul@3 | 186 | |
paul@1 | 187 | if self.mode == "docbook": |
paul@3 | 188 | return self._export_using_docbook(paper_size, cache_entry) |
paul@2 | 189 | elif self.mode == "wkhtmltopdf": |
paul@3 | 190 | return self._export_using_wkhtmltopdf(paper_size, cache_entry) |
paul@1 | 191 | elif self.mode == "htmldoc": |
paul@3 | 192 | return self._export_using_htmldoc(paper_size, cache_entry) |
paul@1 | 193 | else: |
paul@1 | 194 | return 0, _("The action must be configured to use a particular PDF generation tool.") |
paul@1 | 195 | |
paul@2 | 196 | def _get_page_as_html(self): |
paul@2 | 197 | |
paul@2 | 198 | "Get the page in HTML format." |
paul@1 | 199 | |
paul@1 | 200 | request = self.request |
paul@1 | 201 | page = self.page |
paul@1 | 202 | |
paul@1 | 203 | fmt = getFormatterClass(request, "text_html")(request) |
paul@1 | 204 | fmt.setPage(page) |
paul@1 | 205 | |
paul@1 | 206 | page_as_html = [] |
paul@1 | 207 | append = page_as_html.append |
paul@1 | 208 | |
paul@1 | 209 | append("""\ |
paul@1 | 210 | <html> |
paul@1 | 211 | <head> |
paul@1 | 212 | <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> |
paul@1 | 213 | </head> |
paul@1 | 214 | <body> |
paul@1 | 215 | """) |
paul@1 | 216 | append(formatText(page.get_raw_body(), request, fmt, inhibit_p=False)) |
paul@1 | 217 | append("""\ |
paul@1 | 218 | </body> |
paul@1 | 219 | </html> |
paul@1 | 220 | """) |
paul@1 | 221 | |
paul@2 | 222 | return u"".join(page_as_html) |
paul@1 | 223 | |
paul@2 | 224 | def _get_page_as_docbook(self): |
paul@1 | 225 | |
paul@2 | 226 | "Get the page in DocBook format." |
paul@1 | 227 | |
paul@1 | 228 | request = self.request |
paul@1 | 229 | page = self.page |
paul@1 | 230 | |
paul@0 | 231 | fmt = getFormatterClass(request, "text_docbook")(request) |
paul@0 | 232 | fmt.setPage(page) |
paul@0 | 233 | |
paul@0 | 234 | # The DocBook formatter needs to pretend a full document is being made. |
paul@0 | 235 | |
paul@0 | 236 | page_as_docbook = [] |
paul@0 | 237 | append = page_as_docbook.append |
paul@0 | 238 | |
paul@0 | 239 | append(fmt.startDocument(page.page_name)) |
paul@0 | 240 | append(fmt.startContent()) |
paul@0 | 241 | append(formatText(page.get_raw_body(), request, fmt, inhibit_p=False).encode("utf-8")) |
paul@0 | 242 | append(fmt.endContent()) |
paul@0 | 243 | append(fmt.endDocument()) |
paul@0 | 244 | |
paul@2 | 245 | return "".join(page_as_docbook) |
paul@2 | 246 | |
paul@2 | 247 | def _write_pdf_for_html(self, p, page_as_html): |
paul@2 | 248 | |
paul@2 | 249 | """ |
paul@2 | 250 | Write to the process 'p', the HTML for the page, reading the PDF output |
paul@2 | 251 | from the process and writing it to the browser. |
paul@2 | 252 | """ |
paul@2 | 253 | |
paul@2 | 254 | writer = codecs.getwriter("utf-8")(p.stdin) |
paul@2 | 255 | writer.write(page_as_html) |
paul@2 | 256 | |
paul@2 | 257 | out, err = p.communicate() |
paul@2 | 258 | |
paul@2 | 259 | retcode = p.wait() |
paul@2 | 260 | |
paul@2 | 261 | if retcode != 0: |
paul@2 | 262 | return 0, err |
paul@2 | 263 | |
paul@2 | 264 | self._write_pdf(out) |
paul@2 | 265 | return 1, None |
paul@2 | 266 | |
paul@3 | 267 | def _export_using_wkhtmltopdf(self, paper_size, cache_entry): |
paul@2 | 268 | |
paul@2 | 269 | """ |
paul@2 | 270 | Send the page HTML to the processor, indicating the given 'paper_size'. |
paul@2 | 271 | """ |
paul@2 | 272 | |
paul@2 | 273 | p = subprocess.Popen([ |
paul@2 | 274 | XVFB_WRAPPER, "--", |
paul@2 | 275 | WKHTMLTOPDF_PROCESSOR, |
paul@2 | 276 | "--page-size", paper_size, |
paul@2 | 277 | "-", |
paul@2 | 278 | "-" |
paul@2 | 279 | ], |
paul@2 | 280 | shell=False, |
paul@2 | 281 | stdin=subprocess.PIPE, |
paul@2 | 282 | stdout=subprocess.PIPE, |
paul@2 | 283 | stderr=subprocess.PIPE) |
paul@2 | 284 | |
paul@3 | 285 | return self._write_pdf_for_html(p, self._get_page_as_html(), cache_entry) |
paul@2 | 286 | |
paul@3 | 287 | def _export_using_htmldoc(self, paper_size, cache_entry): |
paul@2 | 288 | |
paul@2 | 289 | """ |
paul@2 | 290 | Send the page HTML to the processor, indicating the given 'paper_size'. |
paul@2 | 291 | """ |
paul@2 | 292 | |
paul@2 | 293 | os.environ["HTMLDOC_NOCGI"] = "1" |
paul@2 | 294 | |
paul@2 | 295 | p = subprocess.Popen([ |
paul@2 | 296 | HTMLDOC_PROCESSOR, |
paul@2 | 297 | "-t", "pdf", "--quiet", "--webpage", |
paul@2 | 298 | "--size", paper_size, |
paul@2 | 299 | "-" |
paul@2 | 300 | ], |
paul@2 | 301 | shell=False, |
paul@2 | 302 | stdin=subprocess.PIPE, |
paul@2 | 303 | stdout=subprocess.PIPE, |
paul@2 | 304 | stderr=subprocess.PIPE) |
paul@2 | 305 | |
paul@3 | 306 | return self._write_pdf_for_html(p, self._get_page_as_html(), cache_entry) |
paul@2 | 307 | |
paul@3 | 308 | def _export_using_docbook(self, paper_size, cache_entry): |
paul@2 | 309 | |
paul@2 | 310 | """ |
paul@2 | 311 | Send the page DocBook XML to the processor, indicating the given |
paul@2 | 312 | 'paper_size'. |
paul@2 | 313 | """ |
paul@0 | 314 | |
paul@0 | 315 | p1 = subprocess.Popen([ |
paul@0 | 316 | XSLT_PROCESSOR, |
paul@0 | 317 | "-stringparam", "fop1.extensions", "1", |
paul@0 | 318 | "--stringparam", "paper.type", paper_size, |
paul@0 | 319 | join(DOCBOOK_STYLESHEET_BASE, DOCBOOK_TO_FO_STYLESHEET), |
paul@0 | 320 | "-" |
paul@0 | 321 | ], |
paul@0 | 322 | shell=False, |
paul@0 | 323 | stdin=subprocess.PIPE, |
paul@0 | 324 | stdout=subprocess.PIPE, |
paul@0 | 325 | stderr=subprocess.PIPE) |
paul@0 | 326 | |
paul@2 | 327 | p1.stdin.write(self._get_page_as_docbook()) |
paul@0 | 328 | p1.stdin.close() |
paul@0 | 329 | |
paul@0 | 330 | # Pipe the XML-FO output to the FO processor. |
paul@0 | 331 | |
paul@0 | 332 | p2 = subprocess.Popen([ |
paul@0 | 333 | FO_PROCESSOR, |
paul@0 | 334 | "-fo", "-", |
paul@0 | 335 | "-pdf", "-", |
paul@0 | 336 | ], |
paul@0 | 337 | shell=False, |
paul@0 | 338 | stdin=p1.stdout, |
paul@0 | 339 | stdout=subprocess.PIPE, |
paul@0 | 340 | stderr=subprocess.PIPE) |
paul@0 | 341 | |
paul@0 | 342 | out, err = p2.communicate() |
paul@0 | 343 | |
paul@0 | 344 | retcode = p1.wait() |
paul@0 | 345 | |
paul@0 | 346 | if retcode != 0: |
paul@0 | 347 | return 0, err |
paul@0 | 348 | |
paul@0 | 349 | retcode = p2.wait() |
paul@0 | 350 | |
paul@0 | 351 | if retcode != 0: |
paul@0 | 352 | return 0, err |
paul@0 | 353 | |
paul@3 | 354 | self._write_to_cache(out, cache_entry) |
paul@1 | 355 | self._write_pdf(out) |
paul@1 | 356 | return 1, None |
paul@1 | 357 | |
paul@3 | 358 | def _write_to_cache(self, out, cache_entry): |
paul@3 | 359 | |
paul@3 | 360 | "Write the output 'out' to the given 'cache_entry'." |
paul@3 | 361 | |
paul@3 | 362 | cache_entry.open(mode="w") |
paul@3 | 363 | try: |
paul@3 | 364 | try: |
paul@3 | 365 | cache_entry.write(out) |
paul@3 | 366 | finally: |
paul@3 | 367 | cache_entry.close() |
paul@3 | 368 | except IOError: |
paul@3 | 369 | if cache_entry.exists(): |
paul@3 | 370 | cache_entry.remove() |
paul@3 | 371 | |
paul@1 | 372 | def _write_pdf(self, out): |
paul@3 | 373 | |
paul@3 | 374 | "Write the output 'out' to the request/response." |
paul@3 | 375 | |
paul@1 | 376 | request = self.request |
paul@1 | 377 | |
paul@0 | 378 | send_headers = get_send_headers(request) |
paul@0 | 379 | headers = ["Content-Type: application/pdf"] |
paul@0 | 380 | send_headers(headers) |
paul@0 | 381 | request.write(out) |
paul@0 | 382 | |
paul@0 | 383 | def render_success(self, msg, msgtype=None): |
paul@0 | 384 | |
paul@0 | 385 | """ |
paul@0 | 386 | Render neither 'msg' nor 'msgtype' since a resource has already been |
paul@0 | 387 | produced. |
paul@0 | 388 | NOTE: msgtype is optional because MoinMoin 1.5.x does not support it. |
paul@0 | 389 | """ |
paul@0 | 390 | |
paul@0 | 391 | pass |
paul@0 | 392 | |
paul@0 | 393 | # Action invocation function. |
paul@0 | 394 | |
paul@0 | 395 | def execute(pagename, request): |
paul@0 | 396 | ExportPDF(pagename, request).render() |
paul@0 | 397 | |
paul@0 | 398 | # vim: tabstop=4 expandtab shiftwidth=4 |