Fred Drake | 361ee65 | 1998-03-06 21:29:00 +0000 | [diff] [blame^] | 1 | #! /usr/bin/env python |
| 2 | |
| 3 | """Convert a LaTeX .toc file to some PDFTeX magic to create that neat outline. |
| 4 | |
| 5 | The output file has an extension of '.bkm' instead of '.out', since hyperref |
| 6 | already uses that extension. Let's avoid clashing. |
| 7 | """ |
| 8 | |
| 9 | import os |
| 10 | import re |
| 11 | import string |
| 12 | import sys |
| 13 | |
| 14 | |
| 15 | # Ench item in an entry is a tuple of: |
| 16 | # |
| 17 | # Section #, Title String, Page #, List of Sub-entries |
| 18 | |
| 19 | cline_re = r"""^ |
| 20 | \\contentsline\ \{([a-z]*)} # type of section in $1 |
| 21 | \{(?:\\numberline\ \{([0-9.A-Z]+)})? # section number |
| 22 | (.*)} # title string |
| 23 | \{(\d+)}$""" # page number |
| 24 | |
| 25 | cline_rx = re.compile(cline_re, re.VERBOSE) |
| 26 | |
| 27 | OUTER_TO_INNER = -1 |
| 28 | |
| 29 | _transition_map = { |
| 30 | ('chapter', 'section'): OUTER_TO_INNER, |
| 31 | ('section', 'subsection'): OUTER_TO_INNER, |
| 32 | ('subsection', 'subsubsection'): OUTER_TO_INNER, |
| 33 | ('subsubsection', 'subsection'): 1, |
| 34 | ('subsection', 'section'): 1, |
| 35 | ('section', 'chapter'): 1, |
| 36 | ('subsection', 'chapter'): 2, |
| 37 | ('subsubsection', 'section'): 2, |
| 38 | ('subsubsection', 'chapter'): 3, |
| 39 | } |
| 40 | |
| 41 | def parse_toc(fp): |
| 42 | toc = top = [] |
| 43 | stack = [toc] |
| 44 | level = 'chapter' |
| 45 | lineno = 0 |
| 46 | while 1: |
| 47 | line = fp.readline() |
| 48 | if not line: |
| 49 | break |
| 50 | lineno = lineno + 1 |
| 51 | m = cline_rx.match(line) |
| 52 | if m: |
| 53 | stype, snum, title, pageno = m.group(1, 2, 3, 4) |
| 54 | title = clean_title(title) |
| 55 | entry = (stype, snum, title, string.atoi(pageno), []) |
| 56 | if stype == level: |
| 57 | toc.append(entry) |
| 58 | else: |
| 59 | direction = _transition_map[(level, stype)] |
| 60 | if direction == OUTER_TO_INNER: |
| 61 | toc = toc[-1][-1] |
| 62 | stack.insert(0, toc) |
| 63 | toc.append(entry) |
| 64 | else: |
| 65 | for i in range(direction): |
| 66 | del stack[0] |
| 67 | toc = stack[0] |
| 68 | toc.append(entry) |
| 69 | level = stype |
| 70 | else: |
| 71 | sys.stderr.write("l.%s: " + line) |
| 72 | return top |
| 73 | |
| 74 | |
| 75 | hackscore_rx = re.compile(r"\\(hackscore|raisebox)\s*{[^}]*}") |
| 76 | title_rx = re.compile(r"\\[a-zA-Z]+\s*") |
| 77 | title_trans = string.maketrans("", "") |
| 78 | |
| 79 | def clean_title(title): |
| 80 | title = hackscore_rx.sub("_", title) |
| 81 | while 1: |
| 82 | m = title_rx.search(title) |
| 83 | if m: |
| 84 | title = title[:m.start()] + title[m.end():] |
| 85 | else: |
| 86 | break |
| 87 | return string.translate(title, title_trans, "{}") |
| 88 | |
| 89 | |
| 90 | def write_toc(toc, fp): |
| 91 | for entry in toc: |
| 92 | write_toc_entry(entry, fp, 0) |
| 93 | |
| 94 | def write_toc_entry(entry, fp, layer): |
| 95 | stype, snum, title, pageno, toc = entry |
| 96 | s = "\\pdfoutline goto name{page.%d}" % pageno |
| 97 | if toc: |
| 98 | s = "%s count -%d" % (s, len(toc)) |
| 99 | if snum: |
| 100 | title = "%s %s" % (snum, title) |
| 101 | s = "%s {%s}\n" % (s, title) |
| 102 | fp.write(s) |
| 103 | for entry in toc: |
| 104 | write_toc_entry(entry, fp, layer + 1) |
| 105 | |
| 106 | |
| 107 | def main(): |
| 108 | base, ext = os.path.splitext(sys.argv[1]) |
| 109 | ext = ext or ".toc" |
| 110 | toc = parse_toc(open(base + ext)) |
| 111 | write_toc(toc, open(base + ".bkm", "w")) |
| 112 | |
| 113 | |
| 114 | if __name__ == "__main__": |
| 115 | main() |