dictd-wiktionary: Let dictd serve an offline copy of wiktionary.

@vcunat: add -O to python, as it takes lots of time to process

2 files changed, 810 insertions, 0 deletions

diff --git a/pkgs/servers/dict/dictd-wiktionary.nix b/pkgs/servers/dict/dictd-wiktionary.nix
new file mode 100644
index 00000000000..48aaf187605
--- /dev/null
+++ b/pkgs/servers/dict/dictd-wiktionary.nix
@@ -0,0 +1,32 @@
+{stdenv, fetchurl, python, dict, glibcLocales, writeScript}:
+
+stdenv.mkDerivation rec {
+  version = "20121021";
+  name = "dict-db-wiktionary-${version}";
+  data = fetchurl {
+    url = "http://dumps.wikimedia.org/enwiktionary/${version}/enwiktionary-${version}-pages-articles.xml.bz2";
+    sha256 = "1i4xwdpc2bx58495iy62iz0kn50c3qmnh4qribi82f2rd4qkfjd2";
+  };
+
+  convert = ./wiktionary2dict.py;
+  buildInputs = [python dict glibcLocales];
+
+  builder = writeScript "wiktionary-builder.sh" ''
+    source $stdenv/setup
+
+    ensureDir $out/share/dictd/
+    cd $out/share/dictd
+
+    export LOCALE_ARCHIVE=${glibcLocales}/lib/locale/locale-archive
+    python -O ${convert} ${data}
+    dictzip wiktionary-en.dict
+    echo en_US.UTF-8 > locale
+  '';
+
+  meta = {
+    description = "DICT version of English Wiktionary";
+    homepage = http://en.wiktionary.org/;
+    maintainers = [ stdenv.lib.maintainers.mornfall ];
+    platforms = stdenv.lib.platforms.all;
+  };
+}
diff --git a/pkgs/servers/dict/wiktionary2dict.py b/pkgs/servers/dict/wiktionary2dict.py
new file mode 100644
index 00000000000..8a0210e7254
--- /dev/null
+++ b/pkgs/servers/dict/wiktionary2dict.py
@@ -0,0 +1,778 @@
+# Adapted to produce DICT-compatible files by Petr Rockai in 2012
+# Based on code from wiktiondict by Greg Hewgill
+import re
+import sys
+import codecs
+import os
+import textwrap
+import time
+import xml.sax
+
+class Text:
+    def __init__(self, s):
+        self.s = s
+    def process(self):
+        return s
+
+class TemplateCall:
+    def __init__(self):
+        pass
+    def process(self):
+        pass
+
+class Template:
+    def __init__(self):
+        self.parts = []
+    def append(self, part):
+        self.parts.append(part)
+    def process(self):
+        return ''.join(x.process() for x in self.parts)
+
+class Whitespace:
+    def __init__(self, s):
+        self.s = s
+
+class OpenDouble: pass
+class OpenTriple: pass
+class CloseDouble: pass
+class CloseTriple: pass
+
+class Equals:
+    def __str__(self):
+        return "="
+
+class Delimiter:
+    def __init__(self, c):
+        self.c = c
+    def __str__(self):
+        return self.c
+
+def Tokenise(s):
+    s = unicode(s)
+    stack = []
+    last = 0
+    i = 0
+    while i < len(s):
+        if s[i] == '{' and i+1 < len(s) and s[i+1] == '{':
+            if i > last:
+                yield s[last:i]
+            if i+2 < len(s) and s[i+2] == '{':
+                yield OpenTriple()
+                stack.append(3)
+                i += 3
+            else:
+                yield OpenDouble()
+                stack.append(2)
+                i += 2
+            last = i
+        elif s[i] == '}' and i+1 < len(s) and s[i+1] == '}':
+            if i > last:
+                yield s[last:i]
+            if len(stack) == 0:
+                yield "}}"
+                i += 2
+            elif stack[-1] == 2:
+                yield CloseDouble()
+                i += 2
+                stack.pop()
+            elif i+2 < len(s) and s[i+2] == '}':
+                yield CloseTriple()
+                i += 3
+                stack.pop()
+            else:
+                raise SyntaxError()
+            last = i
+        elif s[i] == ':' or s[i] == '|':
+            if i > last:
+                yield s[last:i]
+            yield Delimiter(s[i])
+            i += 1
+            last = i
+        elif s[i] == '=':
+            if i > last:
+                yield s[last:i]
+            yield Equals()
+            i += 1
+            last = i
+        #elif s[i] == ' ' or s[i] == '\t' or s[i] == '\n':
+        #    if i > last:
+        #        yield s[last:i]
+        #    last = i
+        #    m = re.match(r"\s+", s[i:])
+        #    assert m
+        #    yield Whitespace(m.group(0))
+        #    i += len(m.group(0))
+        #    last = i
+        else:
+            i += 1
+    if i > last:
+        yield s[last:i]
+
+def processSub(templates, tokens, args):
+    t = tokens.next()
+    if not isinstance(t, unicode):
+        raise SyntaxError
+    name = t
+    t = tokens.next()
+    default = None
+    if isinstance(t, Delimiter) and t.c == '|':
+        default = ""
+        while True:
+            t = tokens.next()
+            if isinstance(t, unicode):
+                default += t
+            elif isinstance(t, OpenDouble):
+                default += processTemplateCall(templates, tokens, args)
+            elif isinstance(t, OpenTriple):
+                default += processSub(templates, tokens, args)
+            elif isinstance(t, CloseTriple):
+                break
+            else:
+                print "Unexpected:", t
+                raise SyntaxError()
+    if name in args:
+        return args[name]
+    if default is not None:
+        return default
+    if name == "lang":
+        return "en"
+    return "{{{%s}}}" % name
+
+def processTemplateCall(templates, tokens, args):
+    template = tokens.next().strip().lower()
+    args = {}
+    a = 1
+    t = tokens.next()
+    while True:
+        if isinstance(t, Delimiter):
+            name = unicode(a)
+            arg = ""
+            while True:
+                t = tokens.next()
+                if isinstance(t, unicode):
+                    arg += t
+                elif isinstance(t, OpenDouble):
+                    arg += processTemplateCall(templates, tokens, args)
+                elif isinstance(t, OpenTriple):
+                    arg += processSub(templates, tokens, args)
+                elif isinstance(t, Delimiter) and t.c != '|':
+                    arg += str(t)
+                else:
+                    break
+            if isinstance(t, Equals):
+                name = arg.strip()
+                arg = ""
+                while True:
+                    t = tokens.next()
+                    if isinstance(t, (unicode, Equals)):
+                        arg += unicode(t)
+                    elif isinstance(t, OpenDouble):
+                        arg += processTemplateCall(templates, tokens, args)
+                    elif isinstance(t, OpenTriple):
+                        arg += processSub(templates, tokens, args)
+                    elif isinstance(t, Delimiter) and t.c != '|':
+                        arg += str(t)
+                    else:
+                        break
+                arg = arg.strip()
+            else:
+                a += 1
+            args[name] = arg
+        elif isinstance(t, CloseDouble):
+            break
+        else:
+            print "Unexpected:", t
+            raise SyntaxError
+    #print template, args
+    if template[0] == '#':
+        if template == "#if":
+            if args['1'].strip():
+                return args['2']
+            elif '3' in args:
+                return args['3']
+            else:
+                return ""
+        elif template == "#ifeq":
+            if args['1'].strip() == args['2'].strip():
+                return args['3']
+            elif '4' in args:
+                return args['4']
+            else:
+                return ""
+        elif template == "#ifexist":
+            return ""
+        elif template == "#switch":
+            sw = args['1'].strip()
+            if sw in args:
+                return args[sw]
+            else:
+                return ""
+        else:
+            print "Unknown ParserFunction:", template
+            sys.exit(1)
+    if template not in templates:
+        return "{{%s}}" % template
+    return process(templates, templates[template], args)
+
+def process(templates, s, args = {}):
+    s = re.compile(r"<!--.*?-->", re.DOTALL).sub("", s)
+    s = re.compile(r"<noinclude>.*?</noinclude>", re.DOTALL).sub("", s)
+    assert "<onlyinclude>" not in s
+    #s = re.sub(r"(.*?)<onlyinclude>(.*?)</onlyinclude>(.*)", r"\1", s)
+    s = re.compile(r"<includeonly>(.*?)</includeonly>", re.DOTALL).sub(r"\1", s)
+    r = ""
+    #print list(Tokenise(s))
+    tokens = Tokenise(s)
+    try:
+        while True:
+            t = tokens.next()
+            if isinstance(t, OpenDouble):
+                r += processTemplateCall(templates, tokens, args)
+            elif isinstance(t, OpenTriple):
+                r += processSub(templates, tokens, args)
+            else:
+                r += unicode(t)
+    except StopIteration:
+        pass
+    return r
+
+def test():
+    templates = {
+        'lb': "{{",
+        'name-example': "I am a template example, my first name is '''{{{firstName}}}''' and my last name is '''{{{lastName}}}'''. You can reference my page at [[{{{lastName}}}, {{{firstName}}}]].",
+        't': "start-{{{1|pqr}}}-end",
+        't0': "start-{{{1}}}-end",
+        't1': "start{{{1}}}end<noinclude>moo</noinclude>",
+        't2a1': "{{t2demo|a|{{{1}}}}}",
+        't2a2': "{{t2demo|a|2={{{1}}}}}",
+        't2demo': "start-{{{1}}}-middle-{{{2}}}-end",
+        't5': "{{t2demo|{{{a}}}=b}}",
+        't6': "t2demo|a",
+    }
+    def t(text, expected):
+        print "text:", text
+        s = process(templates, text)
+        if s != expected:
+            print "got:", s
+            print "expected:", expected
+            sys.exit(1)
+    t("{{Name-example}}", "I am a template example, my first name is '''{{{firstName}}}''' and my last name is '''{{{lastName}}}'''. You can reference my page at [[{{{lastName}}}, {{{firstName}}}]].")
+    t("{{Name-example | firstName=John | lastName=Smith }}", "I am a template example, my first name is '''John''' and my last name is '''Smith'''. You can reference my page at [[Smith, John]].")
+    t("{{t0|a}}", "start-a-end")
+    t("{{t0| }}", "start- -end")
+    t("{{t0|}}", "start--end")
+    t("{{t0}}", "start-{{{1}}}-end")
+    t("{{t0|     }}", "start-     -end")
+    t("{{t0|\n}}", "start-\n-end")
+    t("{{t0|1=     }}", "start--end")
+    t("{{t0|1=\n}}", "start--end")
+    t("{{T}}", "start-pqr-end")
+    t("{{T|}}", "start--end")
+    t("{{T|abc}}", "start-abc-end")
+    t("{{T|abc|def}}", "start-abc-end")
+    t("{{T|1=abc|1=def}}", "start-def-end")
+    t("{{T|abc|1=def}}", "start-def-end")
+    t("{{T|1=abc|def}}", "start-def-end")
+    t("{{T|{{T}}}}", "start-start-pqr-end-end")
+    t("{{T|{{T|{{T}}}}}}", "start-start-start-pqr-end-end-end")
+    t("{{T|{{T|{{T|{{T}}}}}}}}", "start-start-start-start-pqr-end-end-end-end")
+    t("{{T|a{{t|b}}}}", "start-astart-b-end-end")
+    t("{{T|{{T|a=b}}}}", "start-start-pqr-end-end")
+    t("{{T|a=b}}", "start-pqr-end")
+    t("{{T|1=a=b}}", "start-a=b-end")
+    #t("{{t1|{{lb}}tc}}}}", "start{{tcend}}")
+    #t("{{t2a1|1=x=y}}", "start-a-middle-{{{2}}}-end")
+    #t("{{t2a2|1=x=y}}", "start-a-middle-x=y-end")
+    #t("{{t5|a=2=d}}", "start-{{{1}}}-middle-d=b-end")
+    #t("{{ {{t6}} }}", "{{ t2demo|a }}")
+    t("{{t|[[a|b]]}}", "start-b-end")
+    t("{{t|[[a|b]] }}", "start-b -end")
+
+Parts = {
+    # Standard POS headers
+    'noun': "n.",
+    'Noun': "n.",
+    'Noun 1': "n.",
+    'Noun 2': "n.",
+    'Verb': "v.",
+    'Adjective': "adj.",
+    'Adverb': "adv.",
+    'Pronoun': "pron.",
+    'Conjunction': "conj.",
+    'Interjection': "interj.",
+    'Preposition': "prep.",
+    'Proper noun': "n.p.",
+    'Proper Noun': "n.p.",
+    'Article': "art.",
+
+    # Standard non-POS level 3 headers
+    '{{acronym}}': "acr.",
+    'Acronym': "acr.",
+    '{{abbreviation}}': "abbr.",
+    '[[Abbreviation]]': "abbr.",
+    'Abbreviation': "abbr.",
+    '[[initialism]]': "init.",
+    '{{initialism}}': "init.",
+    'Initialism': "init.",
+    'Contraction': "cont.",
+    'Prefix': "prefix",
+    'Suffix': "suffix",
+    'Symbol': "sym.",
+    'Letter': "letter",
+    'Idiom': "idiom",
+    'Idioms': "idiom",
+    'Phrase': "phrase",
+
+    # Debated POS level 3 headers
+    'Number': "num.",
+    'Numeral': "num.",
+    'Cardinal number': "num.",
+    'Ordinal number': "num.",
+    'Cardinal numeral': "num.",
+    'Ordinal numeral': "num.",
+
+    # Other headers in use
+    'Personal pronoun': "pers.pron.",
+    'Adjective/Adverb': "adj./adv.",
+    'Proper adjective': "prop.adj.",
+    'Determiner': "det.",
+    'Demonstrative determiner': "dem.det.",
+    'Clitic': "clitic",
+    'Infix': "infix",
+    'Counter': "counter",
+    'Kanji': None,
+    'Kanji reading': None,
+    'Hiragana letter': None,
+    'Katakana letter': None,
+    'Pinyin': None,
+    'Han character': None,
+    'Hanzi': None,
+    'Hanja': None,
+    'Proverb': "prov.",
+    'Expression': None,
+    'Adjectival noun': None,
+    'Quasi-adjective': None,
+    'Particle': "part.",
+    'Infinitive particle': "part.",
+    'Possessive adjective': "poss.adj.",
+    'Verbal prefix': "v.p.",
+    'Postposition': "post.",
+    'Prepositional article': "prep.art.",
+    'Phrasal verb': "phr.v.",
+    'Participle': "participle",
+    'Interrogative auxiliary verb': "int.aux.v.",
+    'Pronominal adverb': "pron.adv.",
+    'Adnominal': "adn.",
+    'Abstract pronoun': "abs.pron.",
+    'Conjunction particle': None,
+    'Root': "root",
+
+    # Non-standard, deprecated headers
+    'Noun form': "n.",
+    'Verb form': "v.",
+    'Adjective form': "adj.form.",
+    'Nominal phrase': "nom.phr.",
+    'Noun phrase': "n. phrase",
+    'Verb phrase': "v. phrase",
+    'Transitive verb': "v.t.",
+    'Intransitive verb': "v.i.",
+    'Reflexive verb': "v.r.",
+    'Cmavo': None,
+    'Romaji': "rom.",
+    'Hiragana': None,
+    'Furigana': None,
+    'Compounds': None,
+
+    # Other headers seen
+    'Alternative forms': None,
+    'Alternative spellings': None,
+    'Anagrams': None,
+    'Antonym': None,
+    'Antonyms': None,
+    'Conjugation': None,
+    'Declension': None,
+    'Declension and pronunciations': None,
+    'Definite Article': "def.art.",
+    'Definite article': "def.art.",
+    'Demonstrative pronoun': "dem.pron.",
+    'Derivation': None,
+    'Derived expression': None,
+    'Derived expressions': None,
+    'Derived forms': None,
+    'Derived phrases': None,
+    'Derived terms': None,
+    'Derived, Related terms': None,
+    'Descendants': None,
+    #'Etymology': None,
+    #'Etymology 1': None,
+    #'Etymology 2': None,
+    #'Etymology 3': None,
+    #'Etymology 4': None,
+    #'Etymology 5': None,
+    'Examples': None,
+    'External links': None,
+    '[[Gismu]]': None,
+    'Gismu': None,
+    'Homonyms': None,
+    'Homophones': None,
+    'Hyphenation': None,
+    'Indefinite article': "art.",
+    'Indefinite pronoun': "ind.pron.",
+    'Indefinite Pronoun': "ind.pron.",
+    'Indetermined pronoun': "ind.pron.",
+    'Interrogative conjunction': "int.conj.",
+    'Interrogative determiner': "int.det.",
+    'Interrogative particle': "int.part.",
+    'Interrogative pronoun': "int.pron.",
+    'Legal expression': "legal",
+    'Mass noun': "n.",
+    'Miscellaneous': None,
+    'Mutations': None,
+    'Noun and verb': "n/v.",
+    'Other language': None,
+    'Pinyin syllable': None,
+    'Possessive determiner': "poss.det.",
+    'Possessive pronoun': "poss.pron.",
+    'Prepositional phrase': "prep.phr.",
+    'Prepositional Pronoun': "prep.pron.",
+    'Pronunciation': None,
+    'Pronunciation 1': None,
+    'Pronunciation 2': None,
+    'Quotations': None,
+    'References': None,
+    'Reflexive pronoun': "refl.pron.",
+    'Related expressions': None,
+    'Related terms': None,
+    'Related words': None,
+    'Relative pronoun': "rel.pron.",
+    'Saying': "saying",
+    'See also': None,
+    'Shorthand': None,
+    '[http://en.wikipedia.org/wiki/Shorthand Shorthand]': None,
+    'Sister projects': None,
+    'Spelling note': None,
+    'Synonyms': None,
+    'Translation': None,
+    'Translations': None,
+    'Translations to be checked': None,
+    'Transliteration': None,
+    'Trivia': None,
+    'Usage': None,
+    'Usage in English': None,
+    'Usage notes': None,
+    'Verbal noun': "v.n.",
+}
+PartsUsed = {}
+for p in Parts.keys():
+    PartsUsed[p] = 0
+
+def encode(s):
+    r = e(s)
+    assert r[1] == len(s)
+    return r[0]
+
+def dowikilink(m):
+    a = m.group(1).split("|")
+    if len(a) > 1:
+        link = a[1]
+    else:
+        link = a[0]
+    if ':' in link:
+        link = ""
+    return link
+
+seentemplates = {}
+def dotemplate(m):
+    aa = m.group(1).split("|")
+    args = {}
+    n = 0
+    for a in aa:
+        am = re.match(r"(.*?)(=(.*))?", a)
+        if am:
+            args[am.group(1)] = am.group(3)
+        else:
+            n += 1
+            args[n] = am.group(1)
+
+    #if aa[0] in seentemplates:
+    #    seentemplates[aa[0]] += 1
+    #else:
+    #    seentemplates[aa[0]] = 1
+    #    print len(seentemplates), aa[0]
+    #print aa[0]
+
+    #if aa[0] not in Templates:
+    #    return "(unknown template %s)" % aa[0]
+    #body = Templates[aa[0]]
+    #body = re.sub(r"<noinclude>.*?</noinclude>", "", body)
+    #assert "<onlyinclude>" not in body
+    ##body = re.sub(r"(.*?)<onlyinclude>(.*?)</onlyinclude>(.*)", r"\1", body)
+    #body = re.sub(r"<includeonly>(.*?)</includeonly>", r"\1", body)
+    #def dotemplatearg(m):
+    #    ta = m.group(1).split("|")
+    #    if ta[0] in args:
+    #        return args[ta[0]]
+    #    elif len(ta) > 1:
+    #        return ta[1]
+    #    else:
+    #        return "{{{%s}}}" % ta[0]
+    #body = re.sub(r"{{{(.*?)}}}", dotemplatearg, body)
+    #return dewiki(body)
+
+def doparserfunction(m):
+    a = m.group(2).split("|")
+    if m.group(1) == "ifeq":
+        if a[0] == a[1]:
+            return a[2]
+        elif len(a) >= 4:
+            return a[3]
+    return ""
+
+def dewiki(body, indent = 0):
+    # process in this order:
+    #   {{{ }}}
+    #   <> <>
+    #   [[ ]]
+    #   {{ }}
+    #   ''' '''
+    #   '' ''
+    #body = wikimediatemplate.process(Templates, body)
+    body = re.sub(r"\[\[(.*?)\]\]", dowikilink, body)
+    #body = re.sub(r"{{(.*?)}}", dotemplate, body)
+    #body = re.sub(r"{{#(.*?):(.*?)}}", doparserfunction, body)
+    body = re.sub(r"'''(.*?)'''", r"\1", body)
+    body = re.sub(r"''(.*?)''", r"\1", body)
+    lines = body.split("\n")
+    n = 0
+    i = 0
+    while i < len(lines):
+        if len(lines[i]) > 0 and lines[i][0] == "#":
+            if len(lines[i]) > 1 and lines[i][1] == '*':
+                wlines = textwrap.wrap(lines[i][2:].strip(),
+                    initial_indent = "    * ",
+                    subsequent_indent = "      ")
+            elif len(lines[i]) > 1 and lines[i][1] == ':':
+                wlines = textwrap.wrap(lines[i][2:].strip(),
+                    initial_indent = "        ",
+                    subsequent_indent = "        ")
+            else:
+                n += 1
+                wlines = textwrap.wrap(str(n) + ". " + lines[i][1:].strip(),
+                    subsequent_indent = "   ")
+        elif len(lines[i]) > 0 and lines[i][0] == "*":
+            n = 0
+            wlines = textwrap.wrap(lines[i][1:].strip(),
+                initial_indent = "* ",
+                subsequent_indent = "  ")
+        else:
+            n = 0
+            wlines = textwrap.wrap(lines[i].strip())
+            if len(wlines) == 0:
+                wlines = ['']
+        lines[i:i+1] = wlines
+        i += len(wlines)
+    return ''.join("  "*(indent-1)+x+"\n" for x in lines)
+
+class WikiSection:
+    def __init__(self, heading, body):
+        self.heading = heading
+        self.body = body
+        #self.lines = re.split("\n+", body.strip())
+        #if len(self.lines) == 1 and len(self.lines[0]) == 0:
+        #    self.lines = []
+        self.children = []
+    def __str__(self):
+        return "<%s:%i:%s>" % (self.heading, len(self.body or ""), ','.join([str(x) for x in self.children]))
+    def add(self, section):
+        self.children.append(section)
+
+def parse(word, text):
+    headings = list(re.finditer("^(=+)\s*(.*?)\s*=+\n", text, re.MULTILINE))
+    #print [x.group(1) for x in headings]
+    doc = WikiSection(word, "")
+    stack = [doc]
+    for i, m in enumerate(headings):
+        depth = len(m.group(1))
+        if depth < len(stack):
+            stack = stack[:depth]
+        else:
+            while depth > len(stack):
+                s = WikiSection(None, "")
+                stack[-1].add(s)
+                stack.append(s)
+        if i+1 < len(headings):
+            s = WikiSection(m.group(2), text[m.end(0):headings[i+1].start(0)].strip())
+        else:
+            s = WikiSection(m.group(2), text[m.end(0):].strip())
+        assert len(stack) == depth
+        stack[-1].add(s)
+        stack.append(s)
+    #while doc.heading is None and len(doc.lines) == 0 and len(doc.children) == 1:
+    #    doc = doc.children[0]
+    return doc
+
+def formatFull(word, doc):
+    def f(depth, section):
+        if section.heading:
+            r = "  "*(depth-1) + section.heading + "\n\n"
+        else:
+            r = ""
+        if section.body:
+            r += dewiki(section.body, depth+1)+"\n"
+        #r += "".join("  "*depth + x + "\n" for x in dewiki(section.body))
+        #if len(section.lines) > 0:
+        #    r += "\n"
+        for c in section.children:
+            r += f(depth+1, c)
+        return r
+    s = f(0, doc)
+    s += "Ref: http://en.wiktionary.org/wiki/%s\n" % word
+    return s
+
+def formatNormal(word, doc):
+    def f(depth, posdepth, section):
+        r = ""
+        if depth == posdepth:
+            if not section.heading or section.heading.startswith("Etymology"):
+                posdepth += 1
+            elif section.heading in Parts:
+                #p = Parts[section.heading]
+                #if p:
+                #    r += "  "*(depth-1) + word + " (" + p + ")\n\n"
+                r += "  "*(depth-1) + section.heading + "\n\n"
+            else:
+                print >>errors, "Unknown part: (%s) %s" % (word, section.heading)
+                return ""
+        elif depth > posdepth:
+            return ""
+        elif section.heading:
+            r += "  "*(depth-1) + section.heading + "\n\n"
+        if section.body:
+            r += dewiki(section.body, depth+1)+"\n"
+        #r += "".join("  "*depth + x + "\n" for x in dewiki(section.lines))
+        #if len(section.lines) > 0:
+        #    r += "\n"
+        for c in section.children:
+            r += f(depth+1, posdepth, c)
+        return r
+    s = f(0, 3, doc)
+    s += "Ref: http://en.wiktionary.org/wiki/%s\n" % word
+    return s
+
+def formatBrief(word, doc):
+    def f(depth, posdepth, section):
+        if depth == posdepth:
+            h = section.heading
+            if not section.heading or section.heading.startswith("Etymology"):
+                posdepth += 1
+            elif section.heading in Parts:
+                #h = Parts[section.heading]
+                #if h:
+                #    h = "%s (%s)" % (word, h)
+                pass
+            stack.append([h, False])
+        elif depth > 0:
+            stack.append([section.heading, False])
+        else:
+            stack.append(["%h " + section.heading, False])
+        r = ""
+        #if section.heading:
+        #    r += "  "*(depth-1) + section.heading + "\n"
+        body = ''.join(x+"\n" for x in section.body.split("\n") if len(x) > 0 and x[0] == '#')
+        if len(body) > 0:
+            for i in range(len(stack)):
+                if not stack[i][1]:
+                    if stack[i][0]:
+                        r += "  "*(i-1) + stack[i][0] + "\n"
+                    stack[i][1] = True
+            r += dewiki(body, depth+1)
+        for c in section.children:
+            r += f(depth+1, posdepth, c)
+        stack.pop()
+        return r
+    stack = []
+    s = f(0, 3, doc)
+    s += "Ref: http://en.wiktionary.org/wiki/%s\n" % word
+    return s
+
+class WikiHandler(xml.sax.ContentHandler):
+    def __init__(self):
+        self.element = None
+        self.page = None
+        self.text = ""
+        self.long = {}
+    def startElement(self, name, attrs):
+        #print "start", name, attrs
+        self.element = name
+    def endElement(self, name):
+        #print "end", name
+        if self.element == "text":
+            if self.page:
+                if self.page in self.long:
+                    print self.page, len(self.text)
+                    print
+                self.doPage(self.page, self.text)
+                self.page = None
+            self.text = ""
+        self.element = None
+    def characters(self, content):
+        #print "characters", content
+        if self.element == "title":
+            if self.checkPage(content):
+                self.page = content
+        elif self.element == "text":
+            if self.page:
+                self.text += content
+                if len(self.text) > 100000 and self.page not in self.long:
+                    self.long[self.page] = 1
+    def checkPage(self, page):
+        return False
+    def doPage(self, page, text):
+        pass
+
+class TemplateHandler(WikiHandler):
+    def checkPage(self, page):
+        return page.startswith("Template:")
+    def doPage(self, page, text):
+        Templates[page[page.find(':')+1:].lower()] = text
+
+class WordHandler(WikiHandler):
+    def checkPage(self, page):
+        return ':' not in page
+    def doPage(self, page, text):
+        m = re.match(r"#redirect\s*\[\[(.*?)\]\]", text, re.IGNORECASE)
+        if m:
+            out.write("  See <%s>" % page)
+            return
+        doc = parse(page, text)
+        out.write(formatBrief(page, doc))
+        #print formatBrief(page, doc)
+
+fn = sys.argv[1]
+info = """   This file was converted from the original database on:
+             %s
+
+   The original data is available from:
+             http://en.wiktionary.org
+   The version from which this file was generated was:
+             %s
+
+  Wiktionary is available under the GNU Free Documentation License.
+""" % (time.ctime(), os.path.basename(fn))
+
+errors = codecs.open("mkdict.err", "w", "utf_8")
+e = codecs.getencoder("utf_8")
+
+Templates = {}
+f = os.popen("bunzip2 -c %s" % fn, "r")
+xml.sax.parse(f, TemplateHandler())
+f.close()
+
+f = os.popen("bunzip2 -c %s" % fn, "r")
+out = codecs.getwriter("utf_8")(
+        os.popen("dictfmt -p wiktionary-en --locale en_US.UTF-8 --columns 0 -u http://en.wiktionary.org", "w"))
+
+out.write(("%%h English Wiktionary\n%s" % info).encode('utf-8'))
+xml.sax.parse(f, WordHandler())
+f.close()
+out.close()