lyx_mirror/lib/lyx2lyx/lyx_2_3.py

# -*- coding: utf-8 -*-
# This file is part of lyx2lyx
# Copyright (C) 2016 The LyX team
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.

""" Convert files to the file format generated by lyx 2.3"""

import re, string
import unicodedata
import sys, os

# Uncomment only what you need to import, please.

from parser_tools import (del_token, del_value, del_complete_lines,
    find_complete_lines, find_end_of, find_end_of_layout, find_end_of_inset,
    find_re, find_substring, find_token, find_token_backwards, find_across_lines,
    get_containing_inset, get_containing_layout, get_bool_value, get_value,
    get_quoted_value, is_in_inset, set_bool_value)
#  find_tokens, find_token_exact, check_token, get_option_value

from lyx2lyx_tools import (add_to_preamble, put_cmd_in_ert, revert_font_attrs,
    insert_to_preamble, latex_length, is_document_option,
    insert_document_option, remove_document_option, revert_language)

####################################################################
# Private helper functions



###############################################################################
###
### Conversion and reversion routines
###
###############################################################################

def convert_microtype(document):
    " Add microtype settings. "
    i = find_token(document.header, "\\font_tt_scale")
    j = find_token(document.preamble, "\\usepackage{microtype}")
    if j == -1:
        document.header.insert(i + 1, "\\use_microtype false")
    else:
        document.header.insert(i + 1, "\\use_microtype true")
        del document.preamble[j]
        if j and document.preamble[j-1] == "% Added by lyx2lyx":
            del document.preamble[j-1]


def revert_microtype(document):
    " Remove microtype settings. "
    use_microtype = get_bool_value(document.header, "\\use_microtype", delete=True)
    if use_microtype:
        add_to_preamble(document, ["\\usepackage{microtype}"])


def convert_dateinset(document):
    ' Convert date external inset to ERT '
    i = 0
    while True:
        i = find_token(document.body, "\\begin_inset External", i+1)
        if i == -1:
            return
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Malformed lyx document: Missing '\\end_inset' in convert_dateinset.")
            continue
        if get_value(document.body, 'template', i, j) == "Date":
            document.body[i : j + 1] = put_cmd_in_ert("\\today ")
        i = j # skip inset


def convert_inputenc(document):
    """Replace no longer supported input encoding setting."""
    i = find_token(document.header, "\\inputencoding pt254")
    if i != -1:
        document.header[i] = "\\inputencoding pt154"


def convert_ibranches(document):
    ' Add "inverted 0" to branch insets'
    i = 0
    while True:
        i = find_token(document.body, "\\begin_inset Branch", i+1)
        if i == -1:
            return
        document.body.insert(i + 1, "inverted 0")


def revert_ibranches(document):
    ' Convert inverted branches to explicit anti-branches'
    # Get list of branches
    ourbranches = {}
    i = 0
    while True:
        i = find_token(document.header, "\\branch", i+1)
        if i == -1:
            break
        branch = document.header[i][8:].strip()
        selected = get_bool_value(document.header, "\\selected", i+1, i+2)
        if selected is None:
            document.warning("Malformed LyX document: No selection indicator "
                             "for branch %s." % branch)
            selected = True
        # the value tells us whether the branch is selected
        ourbranches[branch] = selected

    # Find branch insets, remove "inverted" tag and
    # convert inverted insets to "Anti-OldBranch" insets
    antibranches = {}
    i = 0
    while True:
        i = find_token(document.body, "\\begin_inset Branch", i+1)
        if i == -1:
            break
        inverted = get_bool_value(document.body, "inverted", i+1, i+2, delete=True)
        if inverted is None:
            document.warning("Malformed LyX document: Missing 'inverted' tag in branch inset.")
            continue
        if inverted:
            branch = document.body[i][20:].strip()
            if not branch in antibranches:
                antibranch = "Anti-" + branch
                while antibranch in antibranches:
                    antibranch = "x" + antibranch
                antibranches[branch] = antibranch
            else:
                antibranch = antibranches[branch]
            document.body[i] = "\\begin_inset Branch " + antibranch

    # now we need to add the new branches to the header
    for old, new in antibranches.items():
        i = find_token(document.header, "\\branch " + old, 0)
        if i == -1:
            document.warning("Can't find branch %s even though we found it before!" % (old))
            continue
        j = find_token(document.header, "\\end_branch", i)
        if j == -1:
            document.warning("Malformed LyX document! Can't find end of branch " + old)
            continue
        lines = ["\\branch " + new,
                 "\\selected %d" % (not ourbranches[old])]
        # these are the old lines telling us color, etc.
        lines += document.header[i+2 : j+1]
        document.header[i:i] = lines


beamer_article_styles = [
    "### Inserted by lyx2lyx (more [scr]article styles) ###",
    "Input article.layout",
    "Input beamer.layout",
    "Provides geometry 0",
    "Provides hyperref 0",
    "DefaultFont",
    "     Family                Roman",
    "     Series                Medium",
    "     Shape                 Up",
    "     Size                  Normal",
    "     Color                 None",
    "EndFont",
    "Preamble",
    "     \\usepackage{beamerarticle,pgf}",
    "     % this default might be overridden by plain title style",
    "     \\newcommand\makebeamertitle{\\frame{\\maketitle}}%",
    "     \\AtBeginDocument{",
    "             \\let\\origtableofcontents=\\tableofcontents",
    "             \\def\\tableofcontents{\\@ifnextchar[{\\origtableofcontents}{\\gobbletableofcontents}}",
    "             \\def\\gobbletableofcontents#1{\\origtableofcontents}",
    "     }",
    "EndPreamble",
    "### End of insertion by lyx2lyx (more [scr]article styles) ###"]

def revert_beamer_article_styles(document):
    " Include (scr)article styles in beamer article "

    beamer_articles = ["article-beamer", "scrarticle-beamer"]
    if document.textclass not in beamer_articles:
        return

    if document.textclass == "scrarticle-beamer":
        beamer_article_styles[1] = "Input scrartcl.layout"
    document.append_local_layout(beamer_article_styles)

def convert_beamer_article_styles(document):
    " Remove included (scr)article styles in beamer article "

    beamer_articles = ["article-beamer", "scrarticle-beamer"]
    if document.textclass not in beamer_articles:
        return

    if document.textclass == "scrarticle-beamer":
        beamer_article_styles[1] = "Input scrartcl.layout"
    document.del_local_layout(beamer_article_styles)


def revert_new_babel_languages(document):
    """Revert "bosnian", "friulan", "macedonian", "piedmontese", "romansh".

    Set the document language to English but use correct babel setting.
    """

    nblanguages = ["bosnian", "friulan", "macedonian", "piedmontese", "romansh"]

    for lang in nblanguages:
        if lang == "bosnian" or lang == "macedonian":
            # These are only supported by babel
            revert_language(document, lang, lang, "")
        else:
            # These are supported by babel and polyglossia
            revert_language(document, lang, lang, lang)

# TODO:
# def convert_new_babel_languages(document)
# set to native support if get_value(document.header, "\\options") in
# ["bosnian", "friulan", "macedonian", "piedmontese", "romansh"]
# and Babel is used.


def revert_amharic(document):
    "Set the document language to English but assure Amharic output"

    revert_language(document, "amharic", "", "amharic")


def revert_asturian(document):
    "Set the document language to English but assure Asturian output"

    revert_language(document, "asturian", "", "asturian")


def revert_kannada(document):
    "Set the document language to English but assure Kannada output"

    revert_language(document, "kannada", "", "kannada")


def revert_khmer(document):
    "Set the document language to English but assure Khmer output"

    revert_language(document, "khmer", "", "khmer")


def revert_urdu(document):
    "Set the document language to English but assure Urdu output"

    revert_language(document, "urdu", "", "urdu")


def revert_syriac(document):
    "Set the document language to English but assure Syriac output"

    revert_language(document, "syriac", "", "syriac")


def revert_quotes(document):
    " Revert Quote Insets in verbatim or Hebrew context to plain quotes "

    # First handle verbatim insets
    i = 0
    j = 0
    while i < len(document.body):
        words = document.body[i].split()
        if len(words) > 1 and words[0] == "\\begin_inset" and \
           ( words[1] in ["ERT", "listings"] or ( len(words) > 2 and words[2] in ["URL", "Chunk", "Sweave", "S/R"]) ):
            j = find_end_of_inset(document.body, i)

            if j == -1:
                document.warning("Malformed LyX document: Can't find end of " + words[1] + " inset at line " + str(i))
                i += 1
                continue
            while True:
                k = find_token(document.body, '\\begin_inset Quotes', i, j)
                if k == -1:
                    i += 1
                    break
                l = find_end_of_inset(document.body, k)
                if l == -1:
                    document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
                    i = k
                    continue
                replace = '"'
                if document.body[k].endswith("s"):
                    replace = "'"
                document.body[k:l+2] = [replace]
        else:
            i += 1
            continue

    # Now verbatim layouts
    i = 0
    j = 0
    while i < len(document.body):
        words = document.body[i].split()
        if len(words) > 1 and words[0] == "\\begin_layout" and \
           words[1] in ["Verbatim", "Verbatim*", "Code", "Author_Email", "Author_URL"]:
            j = find_end_of_layout(document.body, i)
            if j == -1:
                document.warning("Malformed LyX document: Can't find end of " + words[1] + " layout at line " + str(i))
                i += 1
                continue
            while True:
                k = find_token(document.body, '\\begin_inset Quotes', i, j)
                if k == -1:
                    i += 1
                    break
                l = find_end_of_inset(document.body, k)
                if l == -1:
                    document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
                    i = k
                    continue
                replace = "\""
                if document.body[k].endswith("s"):
                    replace = "'"
                document.body[k:l+2] = [replace]
        else:
            i += 1
            continue

    # Now handle Hebrew
    if not document.language == "hebrew" and find_token(document.body, '\\lang hebrew', 0) == -1:
        return

    i = 0
    j = 0
    while True:
        k = find_token(document.body, '\\begin_inset Quotes', i)
        if k == -1:
            return
        l = find_end_of_inset(document.body, k)
        if l == -1:
            document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
            i = k
            continue
        hebrew = False
        parent = get_containing_layout(document.body, k)
        ql = find_token_backwards(document.body, "\\lang", k)
        if ql == -1 or ql < parent[1]:
            hebrew = document.language == "hebrew"
        elif document.body[ql] == "\\lang hebrew":
            hebrew = True
        if hebrew:
            replace = "\""
            if document.body[k].endswith("s"):
                replace = "'"
            document.body[k:l+2] = [replace]
        i = l


iopart_local_layout = ["### Inserted by lyx2lyx (stdlayouts) ###",
                       "Input stdlayouts.inc",
                       "### End of insertion by lyx2lyx (stdlayouts) ###"""]

def revert_iopart(document):
    " Input new styles via local layout "
    if document.textclass != "iopart":
        return
    document.append_local_layout(iopart_local_layout)


def convert_iopart(document):
    " Remove local layout we added, if it is there "
    if document.textclass != "iopart":
        return
    document.del_local_layout(iopart_local_layout)


def convert_quotestyle(document):
    " Convert \\quotes_language to \\quotes_style "
    i = find_token(document.header, "\\quotes_language", 0)
    if i == -1:
        document.warning("Malformed LyX document! Can't find \\quotes_language!")
        return
    val = get_value(document.header, "\\quotes_language", i)
    document.header[i] = "\\quotes_style " + val


def revert_quotestyle(document):
    " Revert \\quotes_style to \\quotes_language "
    i = find_token(document.header, "\\quotes_style", 0)
    if i == -1:
        document.warning("Malformed LyX document! Can't find \\quotes_style!")
        return
    val = get_value(document.header, "\\quotes_style", i)
    document.header[i] = "\\quotes_language " + val


def revert_plainquote(document):
    " Revert plain quote insets "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style plain", 0)
    if i != -1:
        document.header[i] = "\\quotes_style english"

    # now the insets
    i = 0
    j = 0
    while True:
        k = find_token(document.body, '\\begin_inset Quotes q', i)
        if k == -1:
            return
        l = find_end_of_inset(document.body, k)
        if l == -1:
            document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
            i = k
            continue
        replace = "\""
        if document.body[k].endswith("s"):
            replace = "'"
        document.body[k:l+2] = [replace]
        i = l


def convert_frenchquotes(document):
    " Convert french quote insets to swiss "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style french", 0)
    if i != -1:
        document.header[i] = "\\quotes_style swiss"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes f', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        newval = val.replace("f", "c", 1)
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_swissquotes(document):
    " Revert swiss quote insets to french "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style swiss", 0)
    if i != -1:
        document.header[i] = "\\quotes_style french"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes c', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        newval = val.replace("c", "f", 1)
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_britishquotes(document):
    " Revert british quote insets to english "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style british", 0)
    if i != -1:
        document.header[i] = "\\quotes_style english"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes b', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        newval = val.replace("b", "e", 1)
        if val[2] == "d":
            # opening mark
            newval = newval.replace("d", "s")
        else:
            # closing mark
            newval = newval.replace("s", "d")
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_swedishgquotes(document):
    " Revert swedish quote insets "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style swedishg", 0)
    if i != -1:
        document.header[i] = "\\quotes_style danish"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes w', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        if val[2] == "d":
            # outer marks
            newval = val.replace("w", "a", 1).replace("r", "l")
        else:
            # inner marks
            newval = val.replace("w", "s", 1)
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_frenchquotes(document):
    " Revert french inner quote insets "

    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes f', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        if val[2] == "s":
            # inner marks
            newval = val.replace("f", "e", 1).replace("s", "d")
            document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_frenchinquotes(document):
    " Revert inner frenchin quote insets "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style frenchin", 0)
    if i != -1:
        document.header[i] = "\\quotes_style french"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes i', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        newval = val.replace("i", "f", 1)
        if val[2] == "s":
            # inner marks
            newval = newval.replace("s", "d")
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_russianquotes(document):
    " Revert russian quote insets "

    # First, revert style setting
    i = find_token(document.header, "\\quotes_style russian", 0)
    if i != -1:
        document.header[i] = "\\quotes_style french"

    # now the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes r', i)
        if i == -1:
            return
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        newval = val
        if val[2] == "s":
            # inner marks
            newval = val.replace("r", "g", 1).replace("s", "d")
        else:
            # outer marks
            newval = val.replace("r", "f", 1)
        document.body[i] = document.body[i].replace(val, newval)
        i += 1


def revert_dynamicquotes(document):
    " Revert dynamic quote insets "

    # First, revert header
    i = find_token(document.header, "\\dynamic_quotes", 0)
    if i != -1:
        del document.header[i]

    # Get global style
    style = "english"
    i = find_token(document.header, "\\quotes_style", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\quotes_style")
    else:
        style = get_value(document.header, "\\quotes_style", i)

    s = "e"
    if style == "english":
        s = "e"
    elif style == "swedish":
        s = "s"
    elif style == "german":
        s = "g"
    elif style == "polish":
        s = "p"
    elif style == "swiss":
        s = "c"
    elif style == "danish":
        s = "a"
    elif style == "plain":
        s = "q"
    elif style == "british":
        s = "b"
    elif style == "swedishg":
        s = "w"
    elif style == "french":
        s = "f"
    elif style == "frenchin":
        s = "i"
    elif style == "russian":
        s = "r"

    # now transform the insets
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset Quotes x', i)
        if i == -1:
            return
        document.body[i] = document.body[i].replace("x", s)
        i += 1


def revert_cjkquotes(document):
    " Revert cjk quote insets "

    # Get global style
    style = "english"
    i = find_token(document.header, "\\quotes_style", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\quotes_style")
    else:
        style = get_value(document.header, "\\quotes_style", i)

    global_cjk = style.find("cjk") != -1

    if global_cjk:
        document.header[i] = "\\quotes_style english"
        # transform dynamic insets
        s = "j"
        if style == "cjkangle":
            s = "k"
        i = 0
        while True:
            i = find_token(document.body, '\\begin_inset Quotes x', i)
            if i == -1:
                break
            document.body[i] = document.body[i].replace("x", s)
            i += 1

    cjk_langs = ["chinese-simplified", "chinese-traditional", "japanese", "japanese-cjk", "korean"]

    i = 0
    j = 0
    while True:
        k = find_token(document.body, '\\begin_inset Quotes j', i)
        if k == -1:
            break
        l = find_end_of_inset(document.body, k)
        if l == -1:
            document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
            i = k
            continue
        cjk = False
        parent = get_containing_layout(document.body, k)
        ql = find_token_backwards(document.body, "\\lang", k)
        if ql == -1 or ql < parent[1]:
            cjk = document.language in cjk_langs
        elif document.body[ql].split()[1] in cjk_langs:
            cjk = True
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        replace = []
        if val[2] == "s":
            # inner marks
            if val[1] == "l":
                # inner opening mark
                if cjk:
                    replace = [u"\u300E"]
                else:
                    replace = ["\\begin_inset Formula $\\llceil$", "\\end_inset"]
            else:
                # inner closing mark
                if cjk:
                    replace = [u"\u300F"]
                else:
                    replace = ["\\begin_inset Formula $\\rrfloor$", "\\end_inset"]
        else:
            # outer marks
            if val[1] == "l":
                # outer opening mark
                if cjk:
                    replace = [u"\u300C"]
                else:
                    replace = ["\\begin_inset Formula $\\lceil$", "\\end_inset"]
            else:
                # outer closing mark
                if cjk:
                    replace = [u"\u300D"]
                else:
                    replace = ["\\begin_inset Formula $\\rfloor$", "\\end_inset"]

        document.body[k:l+1] = replace
        i = l

    i = 0
    j = 0
    while True:
        k = find_token(document.body, '\\begin_inset Quotes k', i)
        if k == -1:
            return
        l = find_end_of_inset(document.body, k)
        if l == -1:
            document.warning("Malformed LyX document: Can't find end of Quote inset at line " + str(k))
            i = k
            continue
        cjk = False
        parent = get_containing_layout(document.body, k)
        ql = find_token_backwards(document.body, "\\lang", k)
        if ql == -1 or ql < parent[1]:
            cjk = document.language in cjk_langs
        elif document.body[ql].split()[1] in cjk_langs:
            cjk = True
        val = get_value(document.body, "\\begin_inset Quotes", i)[7:]
        replace = []
        if val[2] == "s":
            # inner marks
            if val[1] == "l":
                # inner opening mark
                if cjk:
                    replace = [u"\u3008"]
                else:
                    replace = ["\\begin_inset Formula $\\langle$", "\\end_inset"]
            else:
                # inner closing mark
                if cjk:
                    replace = [u"\u3009"]
                else:
                    replace = ["\\begin_inset Formula $\\rangle$", "\\end_inset"]
        else:
            # outer marks
            if val[1] == "l":
                # outer opening mark
                if cjk:
                    replace = [u"\u300A"]
                else:
                    replace = ["\\begin_inset Formula $\\langle\\kern -2.5pt\\langle$", "\\end_inset"]
            else:
                # outer closing mark
                if cjk:
                    replace = [u"\u300B"]
                else:
                    replace = ["\\begin_inset Formula $\\rangle\\kern -2.5pt\\rangle$", "\\end_inset"]

        document.body[k:l+1] = replace
        i = l


def convert_crimson(document):
    """Transform preamble code to native font setting."""
    # Quick-check:
    i = find_substring(document.preamble, "{cochineal}")
    if i == -1:
        return
    # Find and delete user-preamble code:
    if document.preamble[i] == "\\usepackage[proportional,osf]{cochineal}":
        osf = True
    elif document.preamble[i] == "\\usepackage{cochineal}":
        osf = False
    else:
        return
    del document.preamble[i]
    if i and document.preamble[i-1] == "% Added by lyx2lyx":
        del document.preamble[i-1]

    # Convert to native font setting:
    j = find_token(document.header, '\\font_roman')
    if j == -1:
        romanfont = ['\font_roman', '"cochineal"', '"default"']
    else:
        romanfont = document.header[j].split()
        romanfont[1] = '"cochineal"'
    document.header[j] = " ".join(romanfont)
    try:
        set_bool_value(document.header, '\\font_osf', osf)
    except ValueError: # no \\font_osf setting in document.header
        if osf:
            document.header.insert(-1, "\\font_osf true")


def revert_crimson(document):
    " Revert native Cochineal/Crimson font definition to LaTeX "

    i = find_token(document.header, '\\font_roman "cochineal"')
    if i == -1:
        return
    # replace unsupported font setting
    document.header[i] = document.header[i].replace("cochineal", "default")
    # no need for preamble code with system fonts
    if get_bool_value(document.header, "\\use_non_tex_fonts"):
        return
    # transfer old style figures setting to package options
    j = find_token(document.header, "\\font_osf true")
    if j != -1:
        options = "[proportional,osf]"
        document.header[j] = "\\font_osf false"
    else:
        options = ""
    add_to_preamble(document, ["\\usepackage%s{cochineal}"%options])


def revert_cochinealmath(document):
    " Revert cochineal newtxmath definitions to LaTeX "

    if find_token(document.header, "\\use_non_tex_fonts false", 0) != -1:
        i = find_token(document.header, "\\font_math \"cochineal-ntxm\"", 0)
        if i != -1:
            add_to_preamble(document, "\\usepackage[cochineal]{newtxmath}")
            document.header[i] = document.header[i].replace("cochineal-ntxm", "auto")


def revert_labelonly(document):
    " Revert labelonly tag for InsetRef "
    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset ref", i)
        if i == -1:
            return
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of reference inset at line %d!!" %(i))
            i += 1
            continue
        k = find_token(document.body, "LatexCommand labelonly", i, j)
        if k == -1:
            i = j
            continue
        label = get_quoted_value(document.body, "reference", i, j)
        if not label:
            document.warning("Can't find label for reference at line %d!" %(i))
            i = j + 1
            continue
        document.body[i:j+1] = put_cmd_in_ert([label])
        i += 1


def revert_plural_refs(document):
    " Revert plural and capitalized references "
    i = find_token(document.header, "\\use_refstyle 1", 0)
    use_refstyle = (i != 0)

    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset ref", i)
        if i == -1:
            return
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of reference inset at line %d!!" %(i))
            i += 1
            continue

        plural = caps = suffix = False
        k = find_token(document.body, "LaTeXCommand formatted", i, j)
        if k != -1 and use_refstyle:
            plural = get_bool_value(document.body, "plural", i, j, False)
            caps   = get_bool_value(document.body, "caps", i, j, False)
            label  = get_quoted_value(document.body, "reference", i, j)
            if label:
                try:
                    (prefix, suffix) = label.split(":", 1)
                except:
                    document.warning("No `:' separator in formatted reference at line %d!" % (i))
            else:
                document.warning("Can't find label for reference at line %d!" % (i))

        # this effectively tests also for use_refstyle and a formatted reference
        # we do this complicated test because we would otherwise do this erasure
        # over and over and over
        if not ((plural or caps) and suffix):
            del_token(document.body, "plural", i, j)
            del_token(document.body, "caps", i, j - 1) # since we deleted a line
            i = j - 1
            continue

        if caps:
            prefix = prefix[0].title() + prefix[1:]
        cmd = "\\" + prefix + "ref"
        if plural:
            cmd += "[s]"
        cmd += "{" + suffix + "}"
        document.body[i:j+1] = put_cmd_in_ert([cmd])
        i += 1


def revert_noprefix(document):
    " Revert labelonly tags with 'noprefix' set "
    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset ref", i)
        if i == -1:
            return
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of reference inset at line %d!!" %(i))
            i += 1
            continue
        k = find_token(document.body, "LatexCommand labelonly", i, j)
        noprefix = False
        if k != -1:
            noprefix = get_bool_value(document.body, "noprefix", i, j)
        if not noprefix:
            # either it was not a labelonly command, or else noprefix was not set.
            # in that case, we just delete the option.
            del_token(document.body, "noprefix", i, j)
            i = j
            continue
        label = get_quoted_value(document.body, "reference", i, j)
        if not label:
            document.warning("Can't find label for reference at line %d!" %(i))
            i = j + 1
            continue
        try:
            (prefix, suffix) = label.split(":", 1)
        except:
            document.warning("No `:' separator in formatted reference at line %d!" % (i))
            # we'll leave this as an ordinary labelonly reference
            del_token(document.body, "noprefix", i, j)
            i = j
            continue
        document.body[i:j+1] = put_cmd_in_ert([suffix])
        i += 1


def revert_biblatex(document):
    " Revert biblatex support "

    #
    # Header
    #

    # 1. Get cite engine
    engine = "basic"
    i = find_token(document.header, "\\cite_engine", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\cite_engine")
    else:
        engine = get_value(document.header, "\\cite_engine", i)

    # 2. Store biblatex state and revert to natbib
    biblatex = False
    if engine in ["biblatex", "biblatex-natbib"]:
        biblatex = True
        document.header[i] = "\\cite_engine natbib"

    # 3. Store and remove new document headers
    bibstyle = ""
    i = find_token(document.header, "\\biblatex_bibstyle", 0)
    if i != -1:
        bibstyle = get_value(document.header, "\\biblatex_bibstyle", i)
        del document.header[i]

    citestyle = ""
    i = find_token(document.header, "\\biblatex_citestyle", 0)
    if i != -1:
        citestyle = get_value(document.header, "\\biblatex_citestyle", i)
        del document.header[i]

    biblio_options = ""
    i = find_token(document.header, "\\biblio_options", 0)
    if i != -1:
        biblio_options = get_value(document.header, "\\biblio_options", i)
        del document.header[i]

    if biblatex:
        bbxopts = "[natbib=true"
        if bibstyle != "":
            bbxopts += ",bibstyle=" + bibstyle
        if citestyle != "":
            bbxopts += ",citestyle=" + citestyle
        if biblio_options != "":
            bbxopts += "," + biblio_options
        bbxopts += "]"
        add_to_preamble(document, "\\usepackage" + bbxopts + "{biblatex}")

    #
    # Body
    #

    # 1. Bibtex insets
    i = 0
    bibresources = []
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset bibtex", i)
        if i == -1:
            break
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of bibtex inset at line %d!!" %(i))
            i += 1
            continue
        bibs = get_quoted_value(document.body, "bibfiles", i, j)
        opts = get_quoted_value(document.body, "biblatexopts", i, j)
        # store resources
        if bibs:
            bibresources += bibs.split(",")
        else:
            document.warning("Can't find bibfiles for bibtex inset at line %d!" %(i))
        # remove biblatexopts line
        k = find_token(document.body, "biblatexopts", i, j)
        if k != -1:
            del document.body[k]
        # Re-find inset end line
        j = find_end_of_inset(document.body, i)
        # Insert ERT \\printbibliography and wrap bibtex inset to a Note
        if biblatex:
            pcmd = "printbibliography"
            if opts:
                pcmd += "[" + opts + "]"
            repl = ["\\begin_inset ERT", "status open", "", "\\begin_layout Plain Layout",\
                    "", "", "\\backslash", pcmd, "\\end_layout", "", "\\end_inset", "", "",\
                    "\\end_layout", "", "\\begin_layout Standard", "\\begin_inset Note Note",\
                    "status open", "", "\\begin_layout Plain Layout" ]
            repl += document.body[i:j+1]
            repl += ["", "\\end_layout", "", "\\end_inset", "", ""]
            document.body[i:j+1] = repl
            j += 27

        i = j + 1

    if biblatex:
        for b in bibresources:
            add_to_preamble(document, "\\addbibresource{" + b + ".bib}")

    # 2. Citation insets

    # Specific citation insets used in biblatex that need to be reverted to ERT
    new_citations = {
        "Cite" : "Cite",
        "citebyear" : "citeyear",
        "citeyear" : "cite*",
        "Footcite" : "Smartcite",
        "footcite" : "smartcite",
        "Autocite" : "Autocite",
        "autocite" : "autocite",
        "citetitle" : "citetitle",
        "citetitle*" : "citetitle*",
        "fullcite" : "fullcite",
        "footfullcite" : "footfullcite",
        "supercite" : "supercite",
        "citeauthor" : "citeauthor",
        "citeauthor*" : "citeauthor*",
        "Citeauthor" : "Citeauthor",
        "Citeauthor*" : "Citeauthor*"
        }

    # All commands accepted by LyX < 2.3. Everything else throws an error.
    old_citations = ["cite", "nocite", "citet", "citep", "citealt", "citealp",
                     "citeauthor", "citeyear", "citeyearpar", "citet*", "citep*",
                     "citealt*", "citealp*", "citeauthor*", "Citet",  "Citep",
                     "Citealt",  "Citealp",  "Citeauthor", "Citet*", "Citep*",
                     "Citealt*", "Citealp*", "Citeauthor*", "fullcite", "footcite",
                     "footcitet", "footcitep", "footcitealt", "footcitealp",
                     "footciteauthor", "footciteyear", "footciteyearpar",
                     "citefield", "citetitle", "cite*" ]

    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset citation", i)
        if i == -1:
            break
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of citation inset at line %d!!" %(i))
            i += 1
            continue
        k = find_token(document.body, "LatexCommand", i, j)
        if k == -1:
            document.warning("Can't find LatexCommand for citation inset at line %d!" %(i))
            i = j + 1
            continue
        cmd = get_value(document.body, "LatexCommand", k)
        if biblatex and cmd in list(new_citations.keys()):
            pre = get_quoted_value(document.body, "before", i, j)
            post = get_quoted_value(document.body, "after", i, j)
            key = get_quoted_value(document.body, "key", i, j)
            if not key:
                document.warning("Citation inset at line %d does not have a key!" %(i))
                key = "???"
            # Replace known new commands with ERT
            res = "\\" + new_citations[cmd]
            if pre:
                res += "[" + pre + "]"
            if post:
                res += "[" + post + "]"
            elif pre:
                res += "[]"
            res += "{" + key + "}"
            document.body[i:j+1] = put_cmd_in_ert([res])
        elif cmd not in old_citations:
            # Reset unknown commands to cite. This is what LyX does as well
            # (but LyX 2.2 would break on unknown commands)
            document.body[k] = "LatexCommand cite"
            document.warning("Reset unknown cite command '%s' with cite" % cmd)
        i = j + 1

    # Emulate the old biblatex-workaround (pretend natbib in order to use the styles)
    if biblatex:
        biblatex_emulation = [
            "### Inserted by lyx2lyx (biblatex emulation) ###",
            "Provides natbib 1",
            "### End of insertion by lyx2lyx (biblatex emulation) ###"
        ]
        document.append_local_layout(biblatex_emulation)


def revert_citekeyonly(document):
    " Revert keyonly cite command to ERT "

    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset citation", i)
        if i == -1:
            break
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of citation inset at line %d!!" %(i))
            i += 1
            continue
        k = find_token(document.body, "LatexCommand", i, j)
        if k == -1:
            document.warning("Can't find LatexCommand for citation inset at line %d!" %(i))
            i = j + 1
            continue
        cmd = get_value(document.body, "LatexCommand", k)
        if cmd != "keyonly":
            i = j + 1
            continue

        key = get_quoted_value(document.body, "key", i, j)
        if not key:
            document.warning("Citation inset at line %d does not have a key!" %(i))
        # Replace known new commands with ERT
        document.body[i:j+1] = put_cmd_in_ert([key])
        i = j + 1



def revert_bibpackopts(document):
    " Revert support for natbib/jurabib package options "

    engine = "basic"
    i = find_token(document.header, "\\cite_engine", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\cite_engine")
    else:
        engine = get_value(document.header, "\\cite_engine", i)

    biblatex = False
    if engine not in ["natbib", "jurabib"]:
        return

    i = find_token(document.header, "\\biblio_options", 0)
    if i == -1:
        # Nothing to do if we have no options
        return

    biblio_options = get_value(document.header, "\\biblio_options", i)
    del document.header[i]

    if not biblio_options:
        # Nothing to do for empty options
        return

    bibliography_package_options = [
        "### Inserted by lyx2lyx (bibliography package options) ###",
        "PackageOptions " + engine + " " + biblio_options,
        "### End of insertion by lyx2lyx (bibliography package options) ###"
    ]
    document.append_local_layout(bibliography_package_options)

def revert_qualicites(document):
    " Revert qualified citation list commands to ERT "

    # Citation insets that support qualified lists, with their LaTeX code
    ql_citations = {
        "cite" : "cites",
        "Cite" : "Cites",
        "citet" : "textcites",
        "Citet" : "Textcites",
        "citep" : "parencites",
        "Citep" : "Parencites",
        "Footcite" : "Smartcites",
        "footcite" : "smartcites",
        "Autocite" : "Autocites",
        "autocite" : "autocites",
        }

    # Get cite engine
    engine = "basic"
    i = find_token(document.header, "\\cite_engine", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\cite_engine")
    else:
        engine = get_value(document.header, "\\cite_engine", i)

    biblatex = engine in ["biblatex", "biblatex-natbib"]

    i = 0
    while (True):
        i = find_token(document.body, "\\begin_inset CommandInset citation", i)
        if i == -1:
            break
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Can't find end of citation inset at line %d!!" %(i))
            i += 1
            continue
        pres = find_token(document.body, "pretextlist", i, j)
        posts = find_token(document.body, "posttextlist", i, j)
        if pres == -1 and posts == -1:
            # nothing to do.
            i = j + 1
            continue
        pretexts = get_quoted_value(document.body, "pretextlist", pres)
        posttexts = get_quoted_value(document.body, "posttextlist", posts)
        k = find_token(document.body, "LatexCommand", i, j)
        if k == -1:
            document.warning("Can't find LatexCommand for citation inset at line %d!" %(i))
            i = j + 1
            continue
        cmd = get_value(document.body, "LatexCommand", k)
        if biblatex and cmd in list(ql_citations.keys()):
            pre = get_quoted_value(document.body, "before", i, j)
            post = get_quoted_value(document.body, "after", i, j)
            key = get_quoted_value(document.body, "key", i, j)
            if not key:
                document.warning("Citation inset at line %d does not have a key!" %(i))
                key = "???"
            keys = key.split(",")
            prelist = pretexts.split("\t")
            premap = dict()
            for pp in prelist:
                ppp = pp.split(" ", 1)
                premap[ppp[0]] = ppp[1]
            postlist = posttexts.split("\t")
            postmap = dict()
            for pp in postlist:
                ppp = pp.split(" ", 1)
                postmap[ppp[0]] = ppp[1]
            # Replace known new commands with ERT
            if "(" in pre or ")" in pre:
                pre = "{" + pre + "}"
            if "(" in post or ")" in post:
                post = "{" + post + "}"
            res = "\\" + ql_citations[cmd]
            if pre:
                res += "(" + pre + ")"
            if post:
                res += "(" + post + ")"
            elif pre:
                res += "()"
            for kk in keys:
                if premap.get(kk, "") != "":
                    res += "[" + premap[kk] + "]"
                if postmap.get(kk, "") != "":
                    res += "[" + postmap[kk] + "]"
                elif premap.get(kk, "") != "":
                    res += "[]"
                res += "{" + kk + "}"
            document.body[i:j+1] = put_cmd_in_ert([res])
        else:
            # just remove the params
            del document.body[posttexts]
            del document.body[pretexts]
            i += 1


command_insets = ["bibitem", "citation", "href", "index_print", "nomenclature"]
def convert_literalparam(document):
    " Add param literal "

    pos = len("\\begin_inset CommandInset ")
    i = 0
    while True:
        i = find_token(document.body, '\\begin_inset CommandInset', i)
        if i == -1:
            break
        inset = document.body[i][pos:].strip()
        if not inset in command_insets:
            i += 1
            continue
        j = find_end_of_inset(document.body, i)
        if j == -1:
            document.warning("Malformed LyX document: Can't find end of %s inset at line %d" % (inset, i))
            i += 1
            continue
        while i < j and document.body[i].strip() != '':
            i += 1
        # href is already fully latexified. Here we can switch off literal.
        if inset == "href":
            document.body.insert(i, 'literal "false"')
        else:
            document.body.insert(i, 'literal "true"')
        i = j + 1


def revert_literalparam(document):
    " Remove param literal "

    for inset in command_insets:
        i = 0
        while True:
            i = find_token(document.body, '\\begin_inset CommandInset %s' % inset, i+1)
            if i == -1:
                break
            j = find_end_of_inset(document.body, i)
            if j == -1:
                document.warning("Malformed LyX document: Can't find end of %s inset at line %d" % (inset, i))
                continue
            del_token(document.body, 'literal', i, j)


def revert_multibib(document):
    " Revert multibib support "

    # 1. Get cite engine
    engine = "basic"
    i = find_token(document.header, "\\cite_engine", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\cite_engine")
    else:
        engine = get_value(document.header, "\\cite_engine", i)

    # 2. Do we use biblatex?
    biblatex = False
    if engine in ["biblatex", "biblatex-natbib"]:
        biblatex = True

    # 3. Store and remove multibib document header
    multibib = ""
    i = find_token(document.header, "\\multibib", 0)
    if i != -1:
        multibib = get_value(document.header, "\\multibib", i)
        del document.header[i]

    if not multibib:
        return

    # 4. The easy part: Biblatex
    if biblatex:
        i = find_token(document.header, "\\biblio_options", 0)
        if i == -1:
            k = find_token(document.header, "\\use_bibtopic", 0)
            if k == -1:
                # this should not happen
                document.warning("Malformed LyX document! No \\use_bibtopic header found!")
                return
            document.header[k-1 : k-1] = ["\\biblio_options " + "refsection=" + multibib]
        else:
            biblio_options = get_value(document.header, "\\biblio_options", i)
            if biblio_options:
                biblio_options += ","
            biblio_options += "refsection=" + multibib
            document.header[i] = "\\biblio_options " + biblio_options

        # Bibtex insets
        i = 0
        while (True):
            i = find_token(document.body, "\\begin_inset CommandInset bibtex", i)
            if i == -1:
                break
            j = find_end_of_inset(document.body, i)
            if j == -1:
                document.warning("Can't find end of bibtex inset at line %d!!" %(i))
                i += 1
                continue
            btprint = get_quoted_value(document.body, "btprint", i, j)
            if btprint != "bibbysection":
                i += 1
                continue
            opts = get_quoted_value(document.body, "biblatexopts", i, j)
            # change btprint line
            k = find_token(document.body, "btprint", i, j)
            if k != -1:
                document.body[k] = "btprint \"btPrintCited\""
            # Insert ERT \\bibbysection and wrap bibtex inset to a Note
            pcmd = "bibbysection"
            if opts:
                pcmd += "[" + opts + "]"
            repl = ["\\begin_inset ERT", "status open", "", "\\begin_layout Plain Layout",\
                    "", "", "\\backslash", pcmd, "\\end_layout", "", "\\end_inset", "", "",\
                    "\\end_layout", "", "\\begin_layout Standard", "\\begin_inset Note Note",\
                    "status open", "", "\\begin_layout Plain Layout" ]
            repl += document.body[i:j+1]
            repl += ["", "\\end_layout", "", "\\end_inset", "", ""]
            document.body[i:j+1] = repl
            j += 27

            i = j + 1
        return

    # 5. More tricky: Bibtex/Bibtopic
    k = find_token(document.header, "\\use_bibtopic", 0)
    if k == -1:
        # this should not happen
        document.warning("Malformed LyX document! No \\use_bibtopic header found!")
        return
    document.header[k] = "\\use_bibtopic true"

    # Possible units. This assumes that the LyX name follows the std,
    # which might not always be the case. But it's as good as we can get.
    units = {
        "part" : "Part",
        "chapter" : "Chapter",
        "section" : "Section",
        "subsection" : "Subsection",
        }

    if multibib not in units.keys():
        document.warning("Unknown multibib value `%s'!" % nultibib)
        return
    unit = units[multibib]
    btunit = False
    i = 0
    while (True):
        i = find_token(document.body, "\\begin_layout " + unit, i)
        if i == -1:
            break
        if btunit:
            document.body[i-1 : i-1] = ["\\begin_layout Standard",
                                "\\begin_inset ERT", "status open", "",
                                "\\begin_layout Plain Layout", "", "",
                                "\\backslash",
                                "end{btUnit}", "\\end_layout",
                                "\\begin_layout Plain Layout", "",
                                "\\backslash",
                                "begin{btUnit}"
                                "\\end_layout", "", "\\end_inset", "", "",
                                "\\end_layout", ""]
            i += 21
        else:
            document.body[i-1 : i-1] = ["\\begin_layout Standard",
                                "\\begin_inset ERT", "status open", "",
                                "\\begin_layout Plain Layout", "", "",
                                "\\backslash",
                                "begin{btUnit}"
                                "\\end_layout", "", "\\end_inset", "", "",
                                "\\end_layout", ""]
            i += 16
        btunit = True
        i += 1

    if btunit:
        i = find_token(document.body, "\\end_body", i)
        document.body[i-1 : i-1] = ["\\begin_layout Standard",
                                "\\begin_inset ERT", "status open", "",
                                "\\begin_layout Plain Layout", "", "",
                                "\\backslash",
                                "end{btUnit}"
                                "\\end_layout", "", "\\end_inset", "", "",
                                "\\end_layout", ""]


def revert_chapterbib(document):
    " Revert chapterbib support "

    # 1. Get cite engine
    engine = "basic"
    i = find_token(document.header, "\\cite_engine", 0)
    if i == -1:
        document.warning("Malformed document! Missing \\cite_engine")
    else:
        engine = get_value(document.header, "\\cite_engine", i)

    # 2. Do we use biblatex?
    biblatex = False
    if engine in ["biblatex", "biblatex-natbib"]:
        biblatex = True

    # 3. Store multibib document header value
    multibib = ""
    i = find_token(document.header, "\\multibib", 0)
    if i != -1:
        multibib = get_value(document.header, "\\multibib", i)

    if not multibib or multibib != "child":
        # nothing to do
        return

    # 4. remove multibib header
    del document.header[i]

    # 5. Biblatex
    if biblatex:
        # find include insets
        i = 0
        while (True):
            i = find_token(document.body, "\\begin_inset CommandInset include", i)
            if i == -1:
                break
            j = find_end_of_inset(document.body, i)
            if j == -1:
                document.warning("Can't find end of bibtex inset at line %d!!" %(i))
                i += 1
                continue
            parent = get_containing_layout(document.body, i)
            parbeg = parent[1]

            # Insert ERT \\newrefsection before inset
            beg = ["\\begin_layout Standard",
                   "\\begin_inset ERT", "status open", "",
                   "\\begin_layout Plain Layout", "", "",
                   "\\backslash",
                   "newrefsection"
                   "\\end_layout", "", "\\end_inset", "", "",
                   "\\end_layout", ""]
            document.body[parbeg-1:parbeg-1] = beg
            j += len(beg)
            i = j + 1
        return

    # 6. Bibtex/Bibtopic
    i = find_token(document.header, "\\use_bibtopic", 0)
    if i == -1:
        # this should not happen
        document.warning("Malformed LyX document! No \\use_bibtopic header found!")
        return
    if get_value(document.header, "\\use_bibtopic", i) == "true":
        # find include insets
        i = 0
        while (True):
            i = find_token(document.body, "\\begin_inset CommandInset include", i)
            if i == -1:
                break
            j = find_end_of_inset(document.body, i)
            if j == -1:
                document.warning("Can't find end of bibtex inset at line %d!!" %(i))
                i += 1
                continue
            parent = get_containing_layout(document.body, i)
            parbeg = parent[1]
            parend = parent[2]

            # Insert wrap inset into \\begin{btUnit}...\\end{btUnit}
            beg = ["\\begin_layout Standard",
                   "\\begin_inset ERT", "status open", "",
                   "\\begin_layout Plain Layout", "", "",
                   "\\backslash",
                   "begin{btUnit}"
                   "\\end_layout", "", "\\end_inset", "", "",
                   "\\end_layout", ""]
            end = ["\\begin_layout Standard",
                   "\\begin_inset ERT", "status open", "",
                   "\\begin_layout Plain Layout", "", "",
                   "\\backslash",
                   "end{btUnit}"
                   "\\end_layout", "", "\\end_inset", "", "",
                   "\\end_layout", ""]
            document.body[parend+1:parend+1] = end
            document.body[parbeg-1:parbeg-1] = beg
            j += len(beg) + len(end)
            i = j + 1
        return

    # 7. Chapterbib proper
    add_to_preamble(document, ["\\usepackage{chapterbib}"])


def convert_dashligatures(document):
    """Set 'use_dash_ligatures' according to content.
    """
    # Look for and remove dashligatures workaround from 2.3->2.2 reversion,
    # set use_dash_ligatures to True if found, to None else.
    use_dash_ligatures = del_complete_lines(document.preamble,
                                ['% Added by lyx2lyx',
                                 r'\renewcommand{\textendash}{--}',
                                 r'\renewcommand{\textemdash}{---}']) or None

    if use_dash_ligatures is None:
        # Look for dashes (Documents by LyX 2.1 or older have "\twohyphens\n"
        # or "\threehyphens\n" as interim representation for -- an ---.)
        lines = document.body
        has_literal_dashes = has_ligature_dashes = False
        dash_pattern = re.compile(u".*[\u2013\u2014]|\\twohyphens|\\threehyphens")
        i = j = 0
        while True:
            # skip lines without dashes:
            i = find_re(lines, dash_pattern, i+1)
            if i == -1:
                break
            line = lines[i]
            # skip label width string (see bug 10243):
            if line.startswith("\\labelwidthstring"):
                continue
            # do not touch hyphens in some insets (cf. lyx_2_2.convert_dashes):
            try:
                inset_type, start, end = get_containing_inset(lines, i)
            except TypeError: # no containing inset
                inset_type, start, end = "no inset", -1, -1
            if (inset_type.split()[0] in
                ["CommandInset", "ERT", "External", "Formula",
                 "FormulaMacro", "Graphics", "IPA", "listings"]
                or inset_type == "Flex Code"):
                i = end
                continue
            try:
                layoutname, start, end, j = get_containing_layout(lines, i)
            except TypeError: # no (or malformed) containing layout
                document.warning("Malformed LyX document: "
                                "Can't find layout at line %d" % i)
                continue
            if not layoutname:
                document.warning("Malformed LyX document: "
                                 "Missing layout name on line %d"%start)
            if layoutname == "LyX-Code":
                i = end
                continue

            # literal dash followed by a non-white-character or no-break space:
            if re.search(u"[\u2013\u2014]([\S\u00A0\u202F\u2060]|$)",
                         line, flags=re.UNICODE):
                has_literal_dashes = True
            # ligature dash followed by non-white-char or no-break space on next line:
            if (re.search(r"(\\twohyphens|\\threehyphens)", line) and
                re.match(u"[\S\u00A0\u202F\u2060]", lines[i+1], flags=re.UNICODE)):
                has_ligature_dashes = True
            if has_literal_dashes and has_ligature_dashes:
                # TODO: insert a warning note in the document?
                document.warning('This document contained both literal and '
                                 '"ligature" dashes.\n Line breaks may have changed. '
                                 'See UserGuide chapter 3.9.1 for details.')
                break

        if has_literal_dashes and not has_ligature_dashes:
            use_dash_ligatures = False
        elif has_ligature_dashes and not has_literal_dashes:
            use_dash_ligatures = True

    # insert the setting if there is a preferred value
    if use_dash_ligatures is True:
        document.header.insert(-1, "\\use_dash_ligatures true")
    elif use_dash_ligatures is False:
        document.header.insert(-1, "\\use_dash_ligatures false")


def revert_dashligatures(document):
    """Remove font ligature settings for en- and em-dashes.
    Revert conversion of \twodashes or \threedashes to literal dashes.
    """
    use_dash_ligatures = del_value(document.header, "\\use_dash_ligatures")
    if use_dash_ligatures != "true" or document.backend != "latex":
        return
    i = 0
    dash_pattern = re.compile(u".*[\u2013\u2014]")
    while True:
        # skip lines without dashes:
        i = find_re(document.body, dash_pattern, i+1)
        if i == -1:
            break
        line = document.body[i]
        # skip label width string (see bug 10243):
        if line.startswith("\\labelwidthstring"):
            continue
        # do not touch hyphens in some insets (cf. lyx_2_2.convert_dashes):
        try:
            inset_type, start, end = get_containing_inset(document.body, i)
        except TypeError: # no containing inset
            inset_type, start, end = "no inset", -1, -1
        if (inset_type.split()[0] in
            ["CommandInset", "ERT", "External", "Formula",
                "FormulaMacro", "Graphics", "IPA", "listings"]
            or inset_type == "Flex Code"):
            i = end
            continue
        try:
            layoutname, start, end, j = get_containing_layout(document.body, i)
        except TypeError: # no (or malformed) containing layout
            document.warning("Malformed LyX document: "
                            "Can't find layout at body line %d" % i)
            continue
        if layoutname == "LyX-Code":
            i = end
            continue
        # TODO: skip replacement in typewriter fonts
        line = line.replace(u'\u2013', '\\twohyphens\n')
        line = line.replace(u'\u2014', '\\threehyphens\n')
        document.body[i:i+1] = line.split('\n')
    # redefine the dash LICRs to use ligature dashes:
    add_to_preamble(document, [r'\renewcommand{\textendash}{--}',
                               r'\renewcommand{\textemdash}{---}'])


def revert_noto(document):
    " Revert Noto font definitions to LaTeX "

    if find_token(document.header, "\\use_non_tex_fonts false", 0) != -1:
        preamble = ""
        i = find_token(document.header, "\\font_roman \"NotoSerif-TLF\"", 0)
        if i != -1:
            add_to_preamble(document, ["\\renewcommand{\\rmdefault}{NotoSerif-TLF}"])
            document.header[i] = document.header[i].replace("NotoSerif-TLF", "default")
        i = find_token(document.header, "\\font_sans \"NotoSans-TLF\"", 0)
        if i != -1:
            add_to_preamble(document, ["\\renewcommand{\\sfdefault}{NotoSans-TLF}"])
            document.header[i] = document.header[i].replace("NotoSans-TLF", "default")
        i = find_token(document.header, "\\font_typewriter \"NotoMono-TLF\"", 0)
        if i != -1:
            add_to_preamble(document, ["\\renewcommand{\\ttdefault}{NotoMono-TLF}"])
            document.header[i] = document.header[i].replace("NotoMono-TLF", "default")


def revert_xout(document):
  " Reverts \\xout font attribute "
  changed = revert_font_attrs(document.body, "\\xout", "\\xout")
  if changed == True:
    insert_to_preamble(document, \
        ['%  for proper cross-out',
        '\\PassOptionsToPackage{normalem}{ulem}',
        '\\usepackage{ulem}'])


def convert_mathindent(document):
    """Add the \\is_math_indent tag.
    """
    k = find_token(document.header, "\\quotes_style") # where to insert
    # check if the document uses the class option "fleqn"
    options = get_value(document.header, "\\options")
    if 'fleqn' in options:
        document.header.insert(k, "\\is_math_indent 1")
        # delete the fleqn option
        i = find_token(document.header, "\\options")
        options = [option for option in options.split(",")
                   if option.strip() != "fleqn"]
        if options:
            document.header[i] = "\\options " + ",".join(options)
        else:
            del document.header[i]
    else:
        document.header.insert(k, "\\is_math_indent 0")

def revert_mathindent(document):
    " Define mathindent if set in the document "
    # emulate and delete \math_indentation
    value = get_value(document.header, "\\math_indentation",
                      default="default", delete=True)
    if value != "default":
        add_to_preamble(document, [r"\setlength{\mathindent}{%s}"%value])
    # delete \is_math_indent and emulate via document class option
    if not get_bool_value(document.header, "\\is_math_indent", delete=True):
        return
    i = find_token(document.header, "\\options")
    if i != -1:
        document.header[i] = document.header[i].replace("\\options ",
                                                        "\\options fleqn,")
    else:
        l = find_token(document.header, "\\use_default_options")
        document.header.insert(l, "\\options fleqn")


def revert_baselineskip(document):
    " Revert baselineskips to TeX code "
    i = 0
    while True:
        i = find_substring(document.body, "baselineskip%", i+1)
        if i == -1:
            return
        if  document.body[i].startswith("\\begin_inset VSpace"):
            # output VSpace inset as TeX code
            end = find_end_of_inset(document.body, i)
            if end == -1:
                document.warning("Malformed LyX document: "
                        "Can't find end of VSpace inset at line %d." % i)
                continue
            # read out the value
            baselineskip = document.body[i].split()[-1]
            # check if it is the starred version
            star = '*' if '*' in document.body[i] else ''
            # now output TeX code
            cmd = "\\vspace%s{%s}" %(star, latex_length(baselineskip)[1])
            document.body[i:end+1] = put_cmd_in_ert(cmd)
            i += 8
            continue
        begin, end = is_in_inset(document.body, i, "\\begin_inset space \\hspace")
        if  begin != - 1:
            # output space inset as TeX code
            baselineskip = document.body[i].split()[-1]
            star = '*' if '*' in document.body[i-1] else ''
            cmd = "\\hspace%s{%s}" %(star, latex_length(baselineskip)[1])
            document.body[begin:end+1] = put_cmd_in_ert(cmd)


def revert_rotfloat(document):
  " Revert placement options for rotated floats "
  i = 0
  j = 0
  k = 0
  while True:
    i = find_token(document.body, "sideways true", i)
    if i == -1:
      return
    if not document.body[i-2].startswith('placement '):
        i = i + 1
        continue
    # we found a sideways float with placement options
    # at first store the placement
    beg = document.body[i-2].rfind(" ");
    placement = document.body[i-2][beg+1:]
    # check if the option'H' is used
    if placement.find("H") != -1:
      add_to_preamble(document, ["\\usepackage{float}"])
    # now check if it is a starred type
    if document.body[i-1].find("wide true") != -1:
      star = '*'
    else:
      star = ''
    # store the float type
    beg = document.body[i-3].rfind(" ");
    fType = document.body[i-3][beg+1:]
    # now output TeX code
    endInset = find_end_of_inset(document.body, i-3)
    if endInset == -1:
      document.warning("Malformed LyX document: Missing '\\end_inset' of Float inset.")
      return
    else:
      document.body[endInset-2: endInset+1] = put_cmd_in_ert("\\end{sideways" + fType + star + '}')
      document.body[i-3: i+2] = put_cmd_in_ert("\\begin{sideways" + fType + star + "}[" + placement + ']')
      add_to_preamble(document, ["\\usepackage{rotfloat}"])

    i = i + 1


allowbreak_emulation =  [r"\begin_inset space \hspace{}",
                         r"\length 0dd",
                         r"\end_inset",
                         r""]

def convert_allowbreak(document):
    " Zero widths Space-inset -> \SpecialChar allowbreak. "
    lines = document.body
    i = find_complete_lines(lines, allowbreak_emulation, 2)
    while i != -1:
        lines[i-1:i+4] = [lines[i-1] + r"\SpecialChar allowbreak"]
        i = find_complete_lines(lines, allowbreak_emulation, i+3)


def revert_allowbreak(document):
    " \SpecialChar allowbreak -> Zero widths Space-inset. "
    i = 1
    lines = document.body
    while i < len(lines):
        if lines[i].endswith(r"\SpecialChar allowbreak"):
            lines[i:i+1] = [lines[i].replace(r"\SpecialChar allowbreak", "")
                           ] + allowbreak_emulation
            i += 5
        else:
            i += 1


def convert_mathnumberpos(document):
    " add the \\math_number_before tag "
    i = find_token(document.header, "\\quotes_style")
    # check if the document uses the class option "leqno"
    if is_document_option(document, "leqno"):
        remove_document_option(document, "leqno")
        document.header.insert(i, "\\math_number_before 1")
    else:
        document.header.insert(i, "\\math_number_before 0")


def revert_mathnumberpos(document):
    """Remove \\math_number_before tag,
    add the document class option leqno if required.
    """
    math_number_before = get_bool_value(document.header, '\\math_number_before',
                                        delete=True)
    if math_number_before:
        insert_document_option(document, "leqno")


def convert_mathnumberingname(document):
    " rename the \\math_number_before tag to \\math_numbering_side "
    i = find_token(document.header, "\\math_number_before")
    math_number_before = get_bool_value(document.header, '\\math_number_before', i)
    if math_number_before:
        document.header[i] = "\\math_numbering_side left"
        return
    # check if the document uses the class option "reqno"
    k = find_token(document.header, "\\options")
    if 'reqno' in document.header[k]:
        document.header[i] = "\\math_numbering_side right"
        # delete the found option
        document.header[k] = document.header[k].replace(",reqno", "")
        document.header[k] = document.header[k].replace(", reqno", "")
        document.header[k] = document.header[k].replace("reqno,", "")
        if 'reqno' in document.header[k]:
            # then we have reqno as the only option
            del document.header[k]
    else:
        document.header[i] = "\\math_numbering_side default"


def revert_mathnumberingname(document):
    " rename the \\math_numbering_side tag back to \\math_number_before "
    i = find_token(document.header, "\\math_numbering_side")
    math_numbering_side = get_value(document.header, '\\math_numbering_side', i)
    # rename tag and set boolean value:
    if math_numbering_side == "left":
        document.header[i] = "\\math_number_before 1"
    elif math_numbering_side == "right":
        # also add the option reqno:
        document.header[i] = "\\math_number_before 0"
        k = find_token(document.header, "\\options")
        if k != -1  and 'reqno' not in document.header[k]:
            document.header[k] = document.header[k].replace("\\options", "\\options reqno,")
        else:
            l = find_token(document.header, "\\use_default_options", 0)
            document.header.insert(l, "\\options reqno")
    else:
        document.header[i] = "\\math_number_before 0"


def convert_minted(document):
    " add the \\use_minted tag "
    i = find_token(document.header, "\\index ")
    document.header.insert(i, "\\use_minted 0")


def revert_minted(document):
    " remove the \\use_minted tag "
    del_token(document.header, "\\use_minted")


def revert_longtable_lscape(document):
    " revert the longtable landcape mode to ERT "
    i = 0
    regexp = re.compile(r'^<features rotate=\"90\"\s.*islongtable=\"true\"\s.*$', re.IGNORECASE)
    while True:
        i = find_re(document.body, regexp, i)
        if i == -1:
            return

        document.body[i] = document.body[i].replace(" rotate=\"90\"", "")
        lay = get_containing_layout(document.body, i)
        if lay == False:
            document.warning("Longtable has not layout!")
            i += 1
            continue
        begcmd = put_cmd_in_ert("\\begin{landscape}")
        endcmd = put_cmd_in_ert("\\end{landscape}")
        document.body[lay[2] : lay[2]] = endcmd + ["\\end_layout"]
        document.body[lay[1] : lay[1]] = ["\\begin_layout " + lay[0], ""] + begcmd

        add_to_preamble(document, ["\\usepackage{pdflscape}"])
        i = lay[2]


##
# Conversion hub
#

supported_versions = ["2.3.0", "2.3"]
convert = [
           [509, [convert_microtype]],
           [510, [convert_dateinset]],
           [511, [convert_ibranches]],
           [512, [convert_beamer_article_styles]],
           [513, []],
           [514, []],
           [515, []],
           [516, [convert_inputenc]],
           [517, []],
           [518, [convert_iopart]],
           [519, [convert_quotestyle]],
           [520, []],
           [521, [convert_frenchquotes]],
           [522, []],
           [523, []],
           [524, [convert_crimson]],
           [525, []],
           [526, []],
           [527, []],
           [528, []],
           [529, []],
           [530, []],
           [531, []],
           [532, [convert_literalparam]],
           [533, []],
           [534, []],
           [535, [convert_dashligatures]],
           [536, []],
           [537, []],
           [538, [convert_mathindent]],
           [539, []],
           [540, []],
           [541, [convert_allowbreak]],
           [542, [convert_mathnumberpos]],
           [543, [convert_mathnumberingname]],
           [544, [convert_minted]]
          ]

revert =  [
           [543, [revert_minted, revert_longtable_lscape]],
           [542, [revert_mathnumberingname]],
           [541, [revert_mathnumberpos]],
           [540, [revert_allowbreak]],
           [539, [revert_rotfloat]],
           [538, [revert_baselineskip]],
           [537, [revert_mathindent]],
           [536, [revert_xout]],
           [535, [revert_noto]],
           [534, [revert_dashligatures]],
           [533, [revert_chapterbib]],
           [532, [revert_multibib]],
           [531, [revert_literalparam]],
           [530, [revert_qualicites]],
           [529, [revert_bibpackopts]],
           [528, [revert_citekeyonly]],
           [527, [revert_biblatex]],
           [526, [revert_noprefix]],
           [525, [revert_plural_refs]],
           [524, [revert_labelonly]],
           [523, [revert_crimson, revert_cochinealmath]],
           [522, [revert_cjkquotes]],
           [521, [revert_dynamicquotes]],
           [520, [revert_britishquotes, revert_swedishgquotes, revert_frenchquotes, revert_frenchinquotes, revert_russianquotes, revert_swissquotes]],
           [519, [revert_plainquote]],
           [518, [revert_quotestyle]],
           [517, [revert_iopart]],
           [516, [revert_quotes]],
           [515, []],
           [514, [revert_urdu, revert_syriac]],
           [513, [revert_amharic, revert_asturian, revert_kannada, revert_khmer]],
           [512, [revert_new_babel_languages]],
           [511, [revert_beamer_article_styles]],
           [510, [revert_ibranches]],
           [509, []],
           [508, [revert_microtype]]
          ]


if __name__ == "__main__":
    pass