예제 #1
0
파일: main.py 프로젝트: Yuukin256/plain-ref
def from_citeweb(wikicode: Wikicode) -> Wikicode:
    count = 0
    for t in wikicode.filter_templates():
        if t.name == 'Cite web' and not t.has(format) and not t.has('author'):
            title = getvalue(t, 'title')
            url = getvalue(t, 'url')
            date = format_date(getvalue(t, 'date'))
            website = getvalue(t, 'website')
            publisher = getvalue(t, 'publisher')
            accessdate = format_date(getvalue(t, 'accessdate')) if getvalue(
                t, 'accessdate') else format_date(getvalue(t, 'access-date'))

            if all(title, url, accessdate):
                new = f'“[{url} {title}]”'

                if publisher and website:
                    new += f'. \'\'{website}\'\'. {publisher} ({date}). ' if date else f'. \'\'{website}\'\'. {publisher}. '
                elif publisher and not website:
                    new += f'. {publisher} ({date}). ' if date else f'. {publisher}. '
                elif not publisher and website:
                    new += f'. \'\'{website}\'\' ({date}). ' if date else f'. \'\'{website}\'\'. '
                else:
                    new += f' ({date}) .' if date else f'. '
                new += f'{accessdate}閲覧。'

                wikicode.replace(t, new)
                count += 1

    pywikibot.output(f'{{{{Cite web}}}} を {count} 回置換しました')
    return wikicode
예제 #2
0
 def repl_conditional(self, arg: Template, code: Wikicode,
                      index: Union[str, int]):
     if arg.has(index):
         param = arg.get(index)
         self.apply_wikitext(param.value)
         code.replace(
             arg,
             str(param.value).strip() if param.showkey else param.value)
     else:
         code.remove(arg)
예제 #3
0
def merge_etyl_templates(wc: Wikicode) -> Wikicode:
    """
    Given a chunk of wikicode, finds instances where the deprecated `etyl` template is immediately followed by
    either a word in free text, a linked word, or a generic `mention`/`link`/`langname-mention` template.
    It replaces this pattern with a new `derived-parsed` template -- meaning the same thing as the `derived` template
    but namespaced to differentiate. For cases where the `mention` language is different from the `etyl` language,
    we use the former. The template is removed if we can't parse it effectively.
    """
    etyl_indices = [
        i for i, node in enumerate(wc.nodes) if isinstance(node, Template)
        and node.name == "etyl" and i < len(wc.nodes) - 1
    ]

    nodes_to_remove = []
    for i in etyl_indices:
        make_new_template = False
        etyl: Template = wc.nodes[i]
        related_language = etyl.params[0]
        if len(etyl.params) == 1:
            language = "en"
        else:
            language = etyl.params[1]
        node = wc.nodes[i + 1]
        if isinstance(node, Text):
            val = re.split(",| |", node.value.strip())[0]
            if val:
                make_new_template = True
        elif isinstance(node, Wikilink):
            val = node.text or node.title
            val = re.split(",| |", val.strip())[0]
            if val:
                make_new_template = True
        elif isinstance(node, Template):
            if node.name in ("m", "mention", "m+", "langname-mention", "l",
                             "link"):
                related_language = node.params[0]
                if len(node.params) > 1:
                    val = node.params[1].value
                    make_new_template = True
                    nodes_to_remove.append(node)

        if make_new_template:
            params = [
                Parameter(str(i + 1), str(param), showkey=False)
                for i, param in enumerate([language, related_language, val])
            ]
            new_template = Template("derived-parsed", params=params)
            wc.replace(etyl, new_template, recursive=False)
        else:
            nodes_to_remove.append(etyl)

    for node in nodes_to_remove:
        wc.remove(node, recursive=False)
    return wc
예제 #4
0
파일: common.py 프로젝트: nyurik/lexicator
def flag_template(self: TemplateParser,
                  code: Wikicode,
                  template: Template,
                  flag,
                  index=None):
    if index and template.has(index):
        param = template.get(index)
        self.apply_wikitext(param.value)
        code.replace(template, param)
    else:
        code.remove(template)
    self.state.flags.add(flag)
예제 #5
0
def remove_links(wc: Wikicode) -> None:
    """
    Given a chunk of wikicode, replaces all inner links with their text representation
    """
    for link in wc.filter_wikilinks():
        wc.replace(link, link.text)
예제 #6
0
 def apply_value(self, code: Wikicode, arg: Union[Node, Template]):
     typ = type(arg)
     if typ == Text:
         return
     elif typ == Argument:
         self.apply_wikitext(arg.name)
         arg_name = str(arg.name)
         if arg_name in self.arguments:
             code.replace(arg, self.arguments[arg_name])
         elif arg.default is not None:
             self.apply_wikitext(arg.default)
             code.replace(arg, str(arg.default).strip())
     elif typ == Template:
         self.apply_wikitext(arg.name)
         name = self.to_template_name(str(arg.name).strip())
         if name == '':
             self.warn(f"Template name is blank in {arg}")
             code.remove(arg)
             return
         if name.startswith('safesubst:'):
             name = name[len('safesubst:'):].strip()
         if name.startswith('#'):
             if name.startswith('#if:'):
                 self.repl_conditional(
                     arg, code, 2 if len(arg.name.nodes) == 1
                     or arg.name.get(1).strip() == '' else 1)
             elif name.startswith('#ifeq:'):
                 if not arg.has('1'):
                     code.remove(arg)
                 else:
                     val1 = name[len('#ifeq:'):].strip()
                     val2 = arg.get('1')
                     self.apply_wikitext(val2.value)
                     val2 = str(val2.value).strip()
                     self.repl_conditional(arg, code,
                                           3 if val1 == val2 else 2)
             elif name.startswith('#switch:'):
                 key = name[len('#switch:'):].strip()
                 if not arg.has(key):
                     key = '#default'
                     if not arg.has(key):
                         key = '1'
                         # if not arg.has(key):
                         #     self.warn(f'switch value "{key}" not found in {arg}')
                 self.repl_conditional(arg, code, key)
             elif name.startswith('#ifexist:'):
                 key = name[len('#ifexist:'):].strip().replace(
                     self.state.page_parser.template_ns, '').strip()
                 self.repl_conditional(
                     arg, code, 1 if key
                     in self.state.page_parser.templates_no_ns else 2)
             else:
                 raise ValueError(f'Unhandled special {name}')
         else:
             for param in arg.params:
                 self.apply_value(code, param)
             if name in custom_templates:
                 custom_templates[name](self, code, arg)
             elif (
                 (name in self.state.page_parser.expand_template
                  and not self.state.page_parser.expand_template[name](arg))
                     or name in self.state.page_parser.ignore_templates):
                 # self.warn(f"Template {name} should not be expanded")
                 return
             else:
                 template_page = self.state.get_template(name)
                 if template_page:
                     sub_template_params = params_to_dict(arg.params)
                     self.state.add_result('_' + name, sub_template_params)
                     new_text = TemplateParser(
                         f'{self.template_name}.{name}', self.word,
                         template_page.content, sub_template_params,
                         self.state).run()
                     new_arg = mw_parse(str(new_text).strip())
                     code.replace(arg, new_arg)
                     return new_arg
                 else:
                     self.warn(f"Template {name} is not known")
     elif typ == Parameter:
         self.apply_wikitext(arg.name)
         self.apply_wikitext(arg.value)
     elif typ == Tag:
         if str(arg.tag).strip() == 'noinclude':
             code.remove(arg)
         else:
             self.apply_wikitext(arg.contents)
     elif typ == Wikilink:
         self.apply_wikitext(arg.title)
         self.apply_wikitext(arg.text)
     elif typ == Heading:
         self.apply_wikitext(arg.title)
     elif typ == HTMLEntity:
         code.replace(arg, unescape(str(arg)))
     elif typ == Comment:
         code.remove(arg)
     elif typ == ExternalLink:
         return
     else:
         raise ValueError(f'Unknown type {typ} in {arg}')