2 # -*- coding: utf-8 -*-
4 # file lyxpreview2bitmap.py
5 # This file is part of LyX, the document processor.
6 # Licence details can be found in the file COPYING.
9 # with much advice from members of the preview-latex project:
10 # David Kastrup, dak@gnu.org and
11 # Jan-Åke Larsson, jalar@mai.liu.se.
13 # Full author contact details are available in file CREDITS
15 # This script takes a LaTeX file and generates a collection of
16 # png or ppm image files, one per previewed snippet.
19 # * A latex executable;
23 # * pngtoppm (if outputing ppm format images).
25 # preview.sty and dvipng are part of the preview-latex project
26 # http://preview-latex.sourceforge.net/
28 # preview.sty can alternatively be obtained from
29 # CTAN/support/preview-latex/
32 # lyxpreview2bitmap.py --bg=faf0e6 0lyxpreview.tex
34 # This script takes one obligatory argument:
36 # <input file>: The name of the .tex file to be converted.
38 # and these optional arguments:
40 # --png, --ppm: The desired output format. Either 'png' or 'ppm'.
41 # --dpi=<res>: A scale factor, used to ascertain the resolution of the
42 # generated image which is then passed to gs.
43 # --fg=<color>: The foreground color as a hexadecimal string, eg '000000'.
44 # --bg=<color>: The background color as a hexadecimal string, eg 'faf0e6'.
45 # --latex=<exe>: The converter for latex files. Default is latex.
47 # Decomposing TEXFILE's name as DIR/BASE.tex, this script will,
48 # if executed successfully, leave in DIR:
49 # * a (possibly large) number of image files with names
51 # * a file BASE.metrics, containing info needed by LyX to position
52 # the images correctly on the screen.
54 # What does this script do?
55 # 1) Call latex/pdflatex/xelatex/whatever (CONVERTER parameter)
56 # 2) If the output is a PDF fallback to legacy
57 # 3) Otherwise check each page of the DVI (with dv2dt) looking for
58 # PostScript literals, not well supported by dvipng. Pages
59 # containing them are passed to the legacy method in a new LaTeX file.
60 # 4) Call dvipng on the pages without PS literals
61 # 5) Join metrics info coming from both methods (legacy and dvipng)
62 # and write them to file
64 # dvipng is fast but gives problem in several cases, like with
65 # PSTricks, TikZ and other packages using PostScript literals
66 # for all these cases the legacy route is taken (step 3).
67 # Moreover dvipng can't work with PDF files, so, if the CONVERTER
68 # paramter is pdflatex we have to fallback to legacy route (step 2).
70 import getopt, glob, os, re, string, sys
72 from legacy_lyxpreview2ppm import legacy_conversion, \
73 legacy_conversion_step2, legacy_extract_metrics_info
75 from lyxpreview_tools import copyfileobj, error, filter_pages, find_exe, \
76 find_exe_or_terminate, join_metrics_and_rename, latex_commands, \
77 latex_file_re, make_texcolor, mkstemp, run_command, warning, \
83 Usage: %s <options> <input file>
86 -h, --help: Show this help and exit
87 --dpi=<res>: Resolution per inch (default: 128)
88 --png, --ppm: Select the output format (default: png)
89 --fg=<color>: Foreground color (default: black, ie '000000')
90 --bg=<color>: Background color (default: white, ie 'ffffff')
91 --latex=<exe>: Specify the executable for latex (default: latex)
93 The colors are hexadecimal strings, eg 'faf0e6'."""
94 return msg % prog_name
96 # Returns a list of tuples containing page number and ascent fraction
97 # extracted from dvipng output.
98 # Use write_metrics_info to create the .metrics file with this info
99 def extract_metrics_info(dvipng_stdout):
100 # "\[[0-9]+" can match two kinds of numbers: page numbers from dvipng
101 # and glyph numbers from mktexpk. The glyph numbers always match
102 # "\[[0-9]+\]" while the page number never is followed by "\]". Thus:
103 page_re = re.compile("\[([0-9]+)[^]]");
104 metrics_re = re.compile("depth=(-?[0-9]+) height=(-?[0-9]+)")
111 match = page_re.search(dvipng_stdout, pos)
114 page = match.group(1)
116 match = metrics_re.search(dvipng_stdout, pos)
121 # Calculate the 'ascent fraction'.
122 descent = string.atof(match.group(1))
123 ascent = string.atof(match.group(2))
126 if ascent >= 0 or descent >= 0:
127 if abs(ascent + descent) > 0.1:
128 frac = ascent / (ascent + descent)
134 results.append((int(page), frac))
135 pos = match.end() + 2
138 error("Failed to extract metrics info from dvipng")
143 def color_pdf(latex_file, bg_color, fg_color):
144 use_preview_pdf_re = re.compile("(\s*\\\\usepackage\[[^]]+)((pdftex|xetex)\]{preview})")
149 if fg_color != "0.000000,0.000000,0.000000":
150 fg = ' \\AtBeginDocument{\\let\\oldpreview\\preview\\renewcommand\\preview{\\oldpreview\\color[rgb]{%s}}}\n' % (fg_color)
154 for line in open(latex_file, 'r').readlines():
155 match = use_preview_pdf_re.match(line)
160 tmp.write(" \\usepackage{color}\n" \
161 " \\pagecolor[rgb]{%s}\n" \
164 % (bg_color, fg, match.group()))
168 # Unable to open the file, but do nothing here because
169 # the calling function will act on the value of 'success'.
170 warning('Warning in color_pdf! Unable to open "%s"' % latex_file)
171 warning(`sys.exc_type` + ',' + `sys.exc_value`)
174 copyfileobj(tmp, open(latex_file,"wb"), 1)
179 def fix_latex_file(latex_file):
180 documentclass_re = re.compile("(\\\\documentclass\[)(1[012]pt,?)(.+)")
185 for line in open(latex_file, 'r').readlines():
186 match = documentclass_re.match(line)
192 tmp.write("%s%s\n" % (match.group(1), match.group(3)))
195 copyfileobj(tmp, open(latex_file,"wb"), 1)
200 def convert_to_ppm_format(pngtopnm, basename):
201 png_file_re = re.compile("\.png$")
203 for png_file in glob.glob("%s*.png" % basename):
204 ppm_file = png_file_re.sub(".ppm", png_file)
206 p2p_cmd = '%s "%s"' % (pngtopnm, png_file)
207 p2p_status, p2p_stdout = run_command(p2p_cmd)
208 if p2p_status != None:
209 error("Unable to convert %s to ppm format" % png_file)
211 ppm = open(ppm_file, 'w')
212 ppm.write(p2p_stdout)
215 # Returns a tuple of:
216 # ps_pages: list of page indexes of pages containing PS literals
217 # page_count: total number of pages
218 # pages_parameter: parameter for dvipng to exclude pages with PostScript
219 def find_ps_pages(dvi_file):
221 # FIXME: try with pdflatex
222 if not os.path.isfile(dvi_file):
223 error("No DVI output.")
225 # Check for PostScript specials in the dvi, badly supported by dvipng
226 # This is required for correct rendering of PSTricks and TikZ
227 dv2dt = find_exe_or_terminate(["dv2dt"])
228 dv2dt_call = '%s "%s"' % (dv2dt, dvi_file)
230 # The output from dv2dt goes to stdout
231 dv2dt_status, dv2dt_output = run_command(dv2dt_call)
232 psliteral_re = re.compile("^special[1-4] [0-9]+ '(\"|ps:)")
234 # Parse the dtl file looking for PostScript specials.
235 # Pages using PostScript specials are recorded in ps_pages and then
236 # used to create a different LaTeX file for processing in legacy mode.
241 for line in dv2dt_output.split("\n"):
243 if line.startswith("bop"):
248 if line.startswith("eop") and page_has_ps:
249 # We save in a list all the PostScript pages
250 ps_pages.append(page_index)
252 if psliteral_re.match(line) != None:
253 # Literal PostScript special detected!
256 # Create the -pp parameter for dvipng
258 if len(ps_pages) > 0 and len(ps_pages) < page_index:
259 # Don't process Postscript pages with dvipng by selecting the
260 # wanted pages through the -pp parameter. E.g., dvipng -pp 4-12,14,64
261 pages_parameter = " -pp "
265 # Use page ranges, as a list of pages could exceed command line
266 # maximum length (especially under Win32)
267 for index in xrange(1, page_index + 1):
268 if (not index in ps_pages) and skip:
269 # We were skipping pages but current page shouldn't be skipped.
270 # Add this page to -pp, it could stay alone or become the
272 pages_parameter += str(index)
273 # Save the starting index to avoid things such as "11-11"
275 # We're not skipping anymore
277 elif (index in ps_pages) and (not skip):
278 # We weren't skipping but current page should be skipped
279 if last != index - 1:
280 # If the start index of the range is the previous page
281 # then it's not a range
282 pages_parameter += "-" + str(index - 1)
285 pages_parameter += ","
289 # Remove the trailing separator
290 pages_parameter = pages_parameter.rstrip(",")
291 # We've to manage the case in which the last page is closing a range
292 if (not index in ps_pages) and (not skip) and (last != index):
293 pages_parameter += "-" + str(index)
295 return (ps_pages, page_index, pages_parameter)
303 output_format = "png"
304 script_name = argv[0]
306 # Parse and manipulate the command line arguments.
308 (opts, args) = getopt.gnu_getopt(argv[1:], "h", ["bg=",
309 "dpi=", "fg=", "help", "latex=", "png", "ppm"])
310 except getopt.GetoptError:
311 error(usage(script_name))
314 for opt, val in opts:
315 if opt in ("-h", "--help"):
316 print usage(script_name)
322 dpi = string.atoi(val)
324 error("Cannot convert %s to an integer value" % val)
327 elif opt == "--latex":
329 elif opt in ("--png", "--ppm"):
330 output_format = opt[2:]
333 error(usage(script_name))
336 dir, latex_file = os.path.split(input_path)
340 fg_color_dvipng = make_texcolor(fg_color, False)
341 bg_color_dvipng = make_texcolor(bg_color, False)
343 fg_color_gr = make_texcolor(fg_color, True)
344 bg_color_gr = make_texcolor(bg_color, True)
346 # External programs used by the script.
347 latex = find_exe_or_terminate(latex or latex_commands)
349 # Omit font size specification in latex file.
350 fix_latex_file(latex_file)
352 # This can go once dvipng becomes widespread.
353 dvipng = find_exe(["dvipng"])
355 # The data is input to legacy_conversion in as similar
356 # as possible a manner to that input to the code used in
358 vec = [ script_name, input_path, str(dpi), output_format, fg_color, bg_color, latex ]
359 return legacy_conversion(vec)
362 if output_format == "ppm":
363 pngtopnm = find_exe_or_terminate(["pngtopnm"])
365 # Move color information for PDF into the latex file.
366 if not color_pdf(latex_file, bg_color_gr, fg_color_gr):
367 error("Unable to move color info into the latex file")
369 # Compile the latex file.
370 latex_call = '%s "%s"' % (latex, latex_file)
372 latex_status, latex_stdout = run_command(latex_call)
373 if latex_status != None:
374 warning("%s had problems compiling %s" \
375 % (os.path.basename(latex), latex_file))
377 if latex == "xelatex":
378 warning("Using XeTeX")
379 # FIXME: skip unnecessary dvips trial in legacy_conversion_step2
380 return legacy_conversion_step2(latex_file, dpi, output_format)
382 # The dvi output file name
383 dvi_file = latex_file_re.sub(".dvi", latex_file)
385 # If there's no DVI output, look for PDF and go to legacy or fail
386 if not os.path.isfile(dvi_file):
387 # No DVI, is there a PDF?
388 pdf_file = latex_file_re.sub(".pdf", latex_file)
389 if os.path.isfile(pdf_file):
390 warning("%s produced a PDF output, fallback to legacy." % \
391 (os.path.basename(latex)))
392 return legacy_conversion_step2(latex_file, dpi, output_format)
394 error("No DVI or PDF output. %s failed." \
395 % (os.path.basename(latex)))
397 # Look for PS literals in DVI pages
398 # ps_pages: list of page indexes of pages containing PS literals
399 # page_count: total number of pages
400 # pages_parameter: parameter for dvipng to exclude pages with PostScript
401 (ps_pages, page_count, pages_parameter) = find_ps_pages(dvi_file)
403 # If all pages need PostScript, directly use the legacy method.
404 if len(ps_pages) == page_count:
405 vec = [ script_name, input_path, str(dpi), output_format, fg_color, bg_color, latex ]
406 return legacy_conversion(vec)
408 # Run the dvi file through dvipng.
409 dvipng_call = '%s -Ttight -depth -height -D %d -fg "%s" -bg "%s" %s "%s"' \
410 % (dvipng, dpi, fg_color_dvipng, bg_color_dvipng, pages_parameter, dvi_file)
411 dvipng_status, dvipng_stdout = run_command(dvipng_call)
413 if dvipng_status != None:
414 warning("%s failed to generate images from %s... fallback to legacy method" \
415 % (os.path.basename(dvipng), dvi_file))
416 # FIXME: skip unnecessary dvips trial in legacy_conversion_step2
417 return legacy_conversion_step2(latex_file, dpi, output_format)
419 # Extract metrics info from dvipng_stdout.
420 metrics_file = latex_file_re.sub(".metrics", latex_file)
421 dvipng_metrics = extract_metrics_info(dvipng_stdout)
423 # If some pages require PostScript pass them to legacy method
424 if len(ps_pages) > 0:
425 # Create a new LaTeX file just for the snippets needing
427 legacy_latex_file = latex_file_re.sub("_legacy.tex", latex_file)
428 filter_pages(latex_file, legacy_latex_file, ps_pages)
430 # Pass the new LaTeX file to the legacy method
431 vec = [ script_name, latex_file_re.sub("_legacy.tex", input_path),
432 str(dpi), output_format, fg_color, bg_color, latex ]
433 legacy_metrics = legacy_conversion(vec, True)[1]
435 # Now we need to mix metrics data from dvipng and the legacy method
436 original_bitmap = latex_file_re.sub("%d." + output_format, legacy_latex_file)
437 destination_bitmap = latex_file_re.sub("%d." + output_format, latex_file)
439 # Join metrics from dvipng and legacy, and rename legacy bitmaps
440 join_metrics_and_rename(dvipng_metrics, legacy_metrics, ps_pages,
441 original_bitmap, destination_bitmap)
443 # Convert images to ppm format if necessary.
444 if output_format == "ppm":
445 convert_to_ppm_format(pngtopnm, latex_file_re.sub("", latex_file))
447 # Actually create the .metrics file
448 write_metrics_info(dvipng_metrics, metrics_file)
450 return (0, dvipng_metrics)
452 if __name__ == "__main__":
453 exit(main(sys.argv)[0])