Python nodes.extract_messages函数代码示例

def test_extract_messages_without_rawsource():
    """
    Check node.rawsource is fall-backed by using node.astext() value.

    `extract_message` which is used from Sphinx i18n feature drop ``not node.rawsource``
    nodes. So, all nodes which want to translate must have ``rawsource`` value.
    However, sometimes node.rawsource is not set.

    For example: recommonmark-0.2.0 doesn't set rawsource to `paragraph` node.

    refs #1994: Fall back to node's astext() during i18n message extraction.
    """
    p = nodes.paragraph()
    p.append(nodes.Text('test'))
    p.append(nodes.Text('sentence'))
    assert not p.rawsource  # target node must not have rawsource value
    document = create_new_document()
    document.append(p)
    _transform(document)
    assert_node_count(extract_messages(document), nodes.TextElement, 1)
    assert [m for n, m in extract_messages(document)][0], 'text sentence'

    def write_doc(self, docname, doctree):
        # type: (str, nodes.document) -> None
        catalog = self.catalogs[find_catalog(docname, self.config.gettext_compact)]

        for toctree in self.env.tocs[docname].traverse(addnodes.toctree):
            for node, msg in extract_messages(toctree):
                node.uid = ''  # type: ignore  # Hack UUID model
                catalog.add(msg, node)

        for node, msg in extract_messages(doctree):
            catalog.add(msg, node)

        if 'index' in self.env.config.gettext_additional_targets:
            # Extract translatable messages from index entries.
            for node, entries in traverse_translatable_index(doctree):
                for typ, msg, tid, main, key_ in entries:
                    for m in split_index_msg(typ, msg):
                        if typ == 'pair' and m in pairindextypes.values():
                            # avoid built-in translated message was incorporated
                            # in 'sphinx.util.nodes.process_index_entry'
                            continue
                        catalog.add(m, node)

    def write_doc(self, docname, doctree):
        catalog = self.catalogs[find_catalog(docname, self.config.gettext_compact)]

        for node, msg in extract_messages(doctree):
            catalog.add(msg, node)

        if "index" in self.env.config.gettext_additional_targets:
            # Extract translatable messages from index entries.
            for node, entries in traverse_translatable_index(doctree):
                for typ, msg, tid, main in entries:
                    for m in split_index_msg(typ, msg):
                        if typ == "pair" and m in pairindextypes.values():
                            # avoid built-in translated message was incorporated
                            # in 'sphinx.util.nodes.process_index_entry'
                            continue
                        catalog.add(m, node)

def test_extract_messages():
    text = dedent(
        """
        .. admonition:: admonition title

           admonition body
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.title, 1,
    )

    text = dedent(
        """
        .. figure:: foo.jpg

           this is title
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.caption, 1,
    )

    text = dedent(
        """
        .. rubric:: spam
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.rubric, 1,
    )

    text = dedent(
        """
        | spam
        | egg
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.line, 2,
    )

    text = dedent(
        """
        section
        =======

        +----------------+
        | | **Title 1**  |
        | | Message 1    |
        +----------------+
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.line, 2,
    )

    text = dedent(
        """
        * | **Title 1**
          | Message 1
        """
    )
    yield (
        assert_node_count,
        extract_messages(_get_doctree(text)),
        nodes.line, 2,
    )

    def apply(self):
        env = self.document.settings.env
        settings, source = self.document.settings, self.document['source']
        # XXX check if this is reliable
        assert source.startswith(env.srcdir)
        docname = path.splitext(relative_path(path.join(env.srcdir, 'dummy'),
                                              source))[0]
        textdomain = find_catalog(docname,
                                  self.document.settings.gettext_compact)

        # fetch translations
        dirs = [path.join(env.srcdir, directory)
                for directory in env.config.locale_dirs]
        catalog, has_catalog = init_locale(dirs, env.config.language,
                                           textdomain,
                                           charset=env.config.source_encoding)
        if not has_catalog:
            return

        parser = RSTParser()

        # phase1: replace reference ids with translated names
        for node, msg in extract_messages(self.document):
            msgstr = catalog.gettext(msg)
            # XXX add marker to untranslated parts
            if not msgstr or msgstr == msg or not msgstr.strip():
                # as-of-yet untranslated
                continue

            # Avoid "Literal block expected; none found." warnings.
            # If msgstr ends with '::' then it cause warning message at
            # parser.parse() processing.
            # literal-block-warning is only appear in avobe case.
            if msgstr.strip().endswith('::'):
                msgstr += '\n\n   dummy literal'
                # dummy literal node will discard by 'patch = patch[0]'

            # literalblock need literal block notation to avoid it become
            # paragraph.
            if isinstance(node, LITERAL_TYPE_NODES):
                msgstr = '::\n\n' + indent(msgstr, ' '*3)

            patch = new_document(source, settings)
            CustomLocaleReporter(node.source, node.line).set_reporter(patch)
            parser.parse(msgstr, patch)
            try:
                patch = patch[0]
            except IndexError:  # empty node
                pass
            # XXX doctest and other block markup
            if not isinstance(patch, nodes.paragraph):
                continue  # skip for now

            processed = False  # skip flag

            # update title(section) target name-id mapping
            if isinstance(node, nodes.title):
                section_node = node.parent
                new_name = nodes.fully_normalize_name(patch.astext())
                old_name = nodes.fully_normalize_name(node.astext())

                if old_name != new_name:
                    # if name would be changed, replace node names and
                    # document nameids mapping with new name.
                    names = section_node.setdefault('names', [])
                    names.append(new_name)
                    # Original section name (reference target name) should be kept to refer
                    # from other nodes which is still not translated or uses explicit target
                    # name like "`text to display <explicit target name_>`_"..
                    # So, `old_name` is still exist in `names`.

                    _id = self.document.nameids.get(old_name, None)
                    explicit = self.document.nametypes.get(old_name, None)

                    # * if explicit: _id is label. title node need another id.
                    # * if not explicit:
                    #
                    #   * if _id is None:
                    #
                    #     _id is None means:
                    #
                    #     1. _id was not provided yet.
                    #
                    #     2. _id was duplicated.
                    #
                    #        old_name entry still exists in nameids and
                    #        nametypes for another duplicated entry.
                    #
                    #   * if _id is provided: bellow process
                    if _id:
                        if not explicit:
                            # _id was not duplicated.
                            # remove old_name entry from document ids database
                            # to reuse original _id.
                            self.document.nameids.pop(old_name, None)
                            self.document.nametypes.pop(old_name, None)
                            self.document.ids.pop(_id, None)

                        # re-entry with new named section node.
                        #
#.........这里部分代码省略.........

    def apply(self):
        env = self.document.settings.env
        settings, source = self.document.settings, self.document['source']
        # XXX check if this is reliable
        assert source.startswith(env.srcdir)
        docname = path.splitext(relative_path(env.srcdir, source))[0]
        textdomain = find_catalog(docname,
                                  self.document.settings.gettext_compact)

        # fetch translations
        dirs = [path.join(env.srcdir, directory)
                for directory in env.config.locale_dirs]
        catalog, has_catalog = init_locale(dirs, env.config.language,
                                           textdomain)
        if not has_catalog:
            return

        parser = RSTParser()

        #phase1: replace reference ids with translated names
        for node, msg in extract_messages(self.document):
            msgstr = catalog.gettext(msg)
            # XXX add marker to untranslated parts
            if not msgstr or msgstr == msg or not msgstr.strip():
                # as-of-yet untranslated
                continue

            # Avoid "Literal block expected; none found." warnings.
            # If msgstr ends with '::' then it cause warning message at
            # parser.parse() processing.
            # literal-block-warning is only appear in avobe case.
            if msgstr.strip().endswith('::'):
                msgstr += '\n\n   dummy literal'
                # dummy literal node will discard by 'patch = patch[0]'

            patch = new_document(source, settings)
            CustomLocaleReporter(node.source, node.line).set_reporter(patch)
            parser.parse(msgstr, patch)
            patch = patch[0]
            # XXX doctest and other block markup
            if not isinstance(patch, nodes.paragraph):
                continue # skip for now

            processed = False  # skip flag

            # update title(section) target name-id mapping
            if isinstance(node, nodes.title):
                section_node = node.parent
                new_name = nodes.fully_normalize_name(patch.astext())
                old_name = nodes.fully_normalize_name(node.astext())

                if old_name != new_name:
                    # if name would be changed, replace node names and
                    # document nameids mapping with new name.
                    names = section_node.setdefault('names', [])
                    names.append(new_name)
                    if old_name in names:
                        names.remove(old_name)

                    _id = self.document.nameids.get(old_name, None)
                    explicit = self.document.nametypes.get(old_name, None)

                    # * if explicit: _id is label. title node need another id.
                    # * if not explicit:
                    #
                    #   * _id is None:
                    #
                    #     _id is None means _id was duplicated.
                    #     old_name entry still exists in nameids and
                    #     nametypes for another duplicated entry.
                    #
                    #   * _id is provided: bellow process
                    if not explicit and _id:
                        # _id was not duplicated.
                        # remove old_name entry from document ids database
                        # to reuse original _id.
                        self.document.nameids.pop(old_name, None)
                        self.document.nametypes.pop(old_name, None)
                        self.document.ids.pop(_id, None)

                    # re-entry with new named section node.
                    self.document.note_implicit_target(
                            section_node, section_node)

                    # replace target's refname to new target name
                    def is_named_target(node):
                        return isinstance(node, nodes.target) and  \
                            node.get('refname') == old_name
                    for old_target in self.document.traverse(is_named_target):
                        old_target['refname'] = new_name

                    processed = True

            # glossary terms update refid
            if isinstance(node, nodes.term):
                gloss_entries = env.temp_data.setdefault('gloss_entries', set())
                ids = []
                termnodes = []
                for _id in node['names']:
                    if _id in gloss_entries:
#.........这里部分代码省略.........

    def apply(self, **kwargs):
        # type: (Any) -> None
        settings, source = self.document.settings, self.document['source']
        msgstr = u''

        # XXX check if this is reliable
        assert source.startswith(self.env.srcdir)
        docname = path.splitext(relative_path(path.join(self.env.srcdir, 'dummy'),
                                              source))[0]
        textdomain = find_catalog(docname, self.config.gettext_compact)

        # fetch translations
        dirs = [path.join(self.env.srcdir, directory)
                for directory in self.config.locale_dirs]
        catalog, has_catalog = init_locale(dirs, self.config.language, textdomain)
        if not has_catalog:
            return

        # phase1: replace reference ids with translated names
        for node, msg in extract_messages(self.document):
            msgstr = catalog.gettext(msg)
            # XXX add marker to untranslated parts
            if not msgstr or msgstr == msg or not msgstr.strip():
                # as-of-yet untranslated
                continue

            # Avoid "Literal block expected; none found." warnings.
            # If msgstr ends with '::' then it cause warning message at
            # parser.parse() processing.
            # literal-block-warning is only appear in avobe case.
            if msgstr.strip().endswith('::'):
                msgstr += '\n\n   dummy literal'
                # dummy literal node will discard by 'patch = patch[0]'

            # literalblock need literal block notation to avoid it become
            # paragraph.
            if isinstance(node, LITERAL_TYPE_NODES):
                msgstr = '::\n\n' + indent(msgstr, ' ' * 3)

            patch = publish_msgstr(self.app, msgstr, source,
                                   node.line, self.config, settings)
            # XXX doctest and other block markup
            if not isinstance(patch, nodes.paragraph):
                continue  # skip for now

            processed = False  # skip flag

            # update title(section) target name-id mapping
            if isinstance(node, nodes.title):
                section_node = node.parent
                new_name = nodes.fully_normalize_name(patch.astext())
                old_name = nodes.fully_normalize_name(node.astext())

                if old_name != new_name:
                    # if name would be changed, replace node names and
                    # document nameids mapping with new name.
                    names = section_node.setdefault('names', [])
                    names.append(new_name)
                    # Original section name (reference target name) should be kept to refer
                    # from other nodes which is still not translated or uses explicit target
                    # name like "`text to display <explicit target name_>`_"..
                    # So, `old_name` is still exist in `names`.

                    _id = self.document.nameids.get(old_name, None)
                    explicit = self.document.nametypes.get(old_name, None)

                    # * if explicit: _id is label. title node need another id.
                    # * if not explicit:
                    #
                    #   * if _id is None:
                    #
                    #     _id is None means:
                    #
                    #     1. _id was not provided yet.
                    #
                    #     2. _id was duplicated.
                    #
                    #        old_name entry still exists in nameids and
                    #        nametypes for another duplicated entry.
                    #
                    #   * if _id is provided: bellow process
                    if _id:
                        if not explicit:
                            # _id was not duplicated.
                            # remove old_name entry from document ids database
                            # to reuse original _id.
                            self.document.nameids.pop(old_name, None)
                            self.document.nametypes.pop(old_name, None)
                            self.document.ids.pop(_id, None)

                        # re-entry with new named section node.
                        #
                        # Note: msgnode that is a second parameter of the
                        # `note_implicit_target` is not necessary here because
                        # section_node has been noted previously on rst parsing by
                        # `docutils.parsers.rst.states.RSTState.new_subsection()`
                        # and already has `system_message` if needed.
                        self.document.note_implicit_target(section_node)

                    # replace target's refname to new target name
#.........这里部分代码省略.........

def test_extract_messages(rst, node_cls, count):
    msg = extract_messages(_get_doctree(dedent(rst)))
    assert_node_count(msg, node_cls, count)

    def apply(self):
        env = self.document.settings.env
        settings, source = self.document.settings, self.document['source']
        # XXX check if this is reliable
        assert source.startswith(env.srcdir)
        docname = path.splitext(relative_path(env.srcdir, source))[0]
        textdomain = find_catalog(docname,
                                  self.document.settings.gettext_compact)

        # fetch translations
        dirs = [path.join(env.srcdir, directory)
                for directory in env.config.locale_dirs]
        catalog, has_catalog = init_locale(dirs, env.config.language,
                                           textdomain)
        if not has_catalog:
            return

        parser = RSTParser()

        for node, msg in extract_messages(self.document):
            msgstr = catalog.gettext(msg)
            # XXX add marker to untranslated parts
            if not msgstr or msgstr == msg: # as-of-yet untranslated
                continue

            # Avoid "Literal block expected; none found." warnings.
            # If msgstr ends with '::' then it cause warning message at
            # parser.parse() processing.
            # literal-block-warning is only appear in avobe case.
            if msgstr.strip().endswith('::'):
                msgstr += '\n\n   dummy literal'
                # dummy literal node will discard by 'patch = patch[0]'

            patch = new_document(source, settings)
            CustomLocaleReporter(node.source, node.line).set_reporter(patch)
            parser.parse(msgstr, patch)
            patch = patch[0]
            # XXX doctest and other block markup
            if not isinstance(patch, nodes.paragraph):
                continue # skip for now

            # auto-numbered foot note reference should use original 'ids'.
            def is_autonumber_footnote_ref(node):
                return isinstance(node, nodes.footnote_reference) and \
                    node.get('auto') == 1
            old_foot_refs = node.traverse(is_autonumber_footnote_ref)
            new_foot_refs = patch.traverse(is_autonumber_footnote_ref)
            if len(old_foot_refs) != len(new_foot_refs):
                env.warn_node('inconsistent footnote references in '
                              'translated message', node)
            for old, new in zip(old_foot_refs, new_foot_refs):
                new['ids'] = old['ids']
                for id in new['ids']:
                    self.document.ids[id] = new
                self.document.autofootnote_refs.remove(old)
                self.document.note_autofootnote_ref(new)

            # reference should use original 'refname'.
            # * reference target ".. _Python: ..." is not translatable.
            # * section refname is not translatable.
            # * inline reference "`Python <...>`_" has no 'refname'.
            def is_refnamed_ref(node):
                return isinstance(node, nodes.reference) and  \
                    'refname' in node
            old_refs = node.traverse(is_refnamed_ref)
            new_refs = patch.traverse(is_refnamed_ref)
            applied_refname_map = {}
            if len(old_refs) != len(new_refs):
                env.warn_node('inconsistent references in '
                              'translated message', node)
            for new in new_refs:
                if new['refname'] in applied_refname_map:
                    # 2nd appearance of the reference
                    new['refname'] = applied_refname_map[new['refname']]
                elif old_refs:
                    # 1st appearance of the reference in old_refs
                    old = old_refs.pop(0)
                    refname = old['refname']
                    new['refname'] = refname
                    applied_refname_map[new['refname']] = refname
                else:
                    # the reference is not found in old_refs
                    applied_refname_map[new['refname']] = new['refname']

                self.document.note_refname(new)

            # refnamed footnote and citation should use original 'ids'.
            def is_refnamed_footnote_ref(node):
                footnote_ref_classes = (nodes.footnote_reference,
                                        nodes.citation_reference)
                return isinstance(node, footnote_ref_classes) and \
                    'refname' in node
            old_refs = node.traverse(is_refnamed_footnote_ref)
            new_refs = patch.traverse(is_refnamed_footnote_ref)
            refname_ids_map = {}
            if len(old_refs) != len(new_refs):
                env.warn_node('inconsistent references in '
                              'translated message', node)
            for old in old_refs:
                refname_ids_map[old["refname"]] = old["ids"]
#.........这里部分代码省略.........

    def write_doc(self, docname, doctree):
        catalog = self.catalogs[find_catalog(docname,
                                             self.config.gettext_compact)]

        for node, msg in extract_messages(doctree):
            catalog.add(msg, node)