docfixer.py

reduced python source for embedded apps

            parent.normalize()
            children = parent.childNodes
            if children[-1].nodeType == TEXT:
                children[-1].data = string.rstrip(children[-1].data)


def fixup_trailing_whitespace(doc, wsmap):
    queue = [doc]
    while queue:
        node = queue[0]
        del queue[0]
        if wsmap.has_key(node.get_nodeName()):
            ws = wsmap[node.get_tagName()]
            children = node.childNodes
            children.reverse()
            if children[0].nodeType == TEXT:
                data = string.rstrip(children[0].data) + ws
                children[0].data = data
            children.reverse()
            # hack to get the title in place:
            if node.get_tagName() == "title" \
               and node.parentNode.firstChild.get_nodeType() == ELEMENT:
                node.parentNode.insertBefore(doc.createText("\n  "),
                                             node.parentNode.firstChild)
        for child in node.childNodes:
            if child.nodeType == ELEMENT:
                queue.append(child)


def normalize(doc):
    for node in doc.childNodes:
        if node.nodeType == ELEMENT:
            node.normalize()


def cleanup_trailing_parens(doc, element_names):
    d = {}
    for gi in element_names:
        d[gi] = gi
    rewrite_element = d.has_key
    queue = []
    for node in doc.childNodes:
        if node.nodeType == ELEMENT:
            queue.append(node)
    while queue:
        node = queue[0]
        del queue[0]
        if rewrite_element(node.get_tagName()):
            children = node.childNodes
            if len(children) == 1 \
               and children[0].nodeType == TEXT:
                data = children[0].data
                if data[-2:] == "()":
                    children[0].data = data[:-2]
        else:
            for child in node.childNodes:
                if child.nodeType == ELEMENT:
                    queue.append(child)


def contents_match(left, right):
    left_children = left.childNodes
    right_children = right.childNodes
    if len(left_children) != len(right_children):
        return 0
    for l, r in map(None, left_children, right_children):
        nodeType = l.nodeType
        if nodeType != r.nodeType:
            return 0
        if nodeType == ELEMENT:
            if l.get_tagName() != r.get_tagName():
                return 0
            # should check attributes, but that's not a problem here
            if not contents_match(l, r):
                return 0
        elif nodeType == TEXT:
            if l.data != r.data:
                return 0
        else:
            # not quite right, but good enough
            return 0
    return 1


def create_module_info(doc, section):
    # Heavy.
    node = extract_first_element(section, "modulesynopsis")
    if node is None:
        return
    node._node.name = "synopsis"
    lastchild = node.childNodes[-1]
    if lastchild.nodeType == TEXT \
       and lastchild.data[-1:] == ".":
        lastchild.data = lastchild.data[:-1]
    modauthor = extract_first_element(section, "moduleauthor")
    if modauthor:
        modauthor._node.name = "author"
        modauthor.appendChild(doc.createTextNode(
            modauthor.getAttribute("name")))
        modauthor.removeAttribute("name")
    platform = extract_first_element(section, "platform")
    if section.get_tagName() == "section":
        modinfo_pos = 2
        modinfo = doc.createElement("moduleinfo")
        moddecl = extract_first_element(section, "declaremodule")
        name = None
        if moddecl:
            modinfo.appendChild(doc.createTextNode("\n    "))
            name = moddecl.attributes["name"].value
            namenode = doc.createElement("name")
            namenode.appendChild(doc.createTextNode(name))
            modinfo.appendChild(namenode)
            type = moddecl.attributes.get("type")
            if type:
                type = type.value
                modinfo.appendChild(doc.createTextNode("\n    "))
                typenode = doc.createElement("type")
                typenode.appendChild(doc.createTextNode(type))
                modinfo.appendChild(typenode)
        versionadded = extract_first_element(section, "versionadded")
        if versionadded:
            modinfo.setAttribute("added", versionadded.getAttribute("version"))
        title = get_first_element(section, "title")
        if title:
            children = title.childNodes
            if len(children) >= 2 \
               and children[0].get_nodeName() == "module" \
               and children[0].childNodes[0].data == name:
                # this is it; morph the <title> into <short-synopsis>
                first_data = children[1]
                if first_data.data[:4] == " ---":
                    first_data.data = string.lstrip(first_data.data[4:])
                title._node.name = "short-synopsis"
                if children[-1].nodeType == TEXT \
                   and children[-1].data[-1:] == ".":
                    children[-1].data = children[-1].data[:-1]
                section.removeChild(title)
                section.removeChild(section.childNodes[0])
                title.removeChild(children[0])
                modinfo_pos = 0
            else:
                ewrite("module name in title doesn't match"
                       " <declaremodule/>; no <short-synopsis/>\n")
        else:
            ewrite("Unexpected condition: <section/> without <title/>\n")
        modinfo.appendChild(doc.createTextNode("\n    "))
        modinfo.appendChild(node)
        if title and not contents_match(title, node):
            # The short synopsis is actually different,
            # and needs to be stored:
            modinfo.appendChild(doc.createTextNode("\n    "))
            modinfo.appendChild(title)
        if modauthor:
            modinfo.appendChild(doc.createTextNode("\n    "))
            modinfo.appendChild(modauthor)
        if platform:
            modinfo.appendChild(doc.createTextNode("\n    "))
            modinfo.appendChild(platform)
        modinfo.appendChild(doc.createTextNode("\n  "))
        section.insertBefore(modinfo, section.childNodes[modinfo_pos])
        section.insertBefore(doc.createTextNode("\n  "), modinfo)
        #
        # The rest of this removes extra newlines from where we cut out
        # a lot of elements.  A lot of code for minimal value, but keeps
        # keeps the generated *ML from being too funny looking.
        #
        section.normalize()
        children = section.childNodes
        for i in range(len(children)):
            node = children[i]
            if node.get_nodeName() == "moduleinfo":
                nextnode = children[i+1]
                if nextnode.nodeType == TEXT:
                    data = nextnode.data
                    if len(string.lstrip(data)) < (len(data) - 4):
                        nextnode.data = "\n\n\n" + string.lstrip(data)


def cleanup_synopses(doc, fragment):
    for node in find_all_elements(fragment, "section"):
        create_module_info(doc, node)


def fixup_table_structures(doc, fragment):
    for table in find_all_elements(fragment, "table"):
        fixup_table(doc, table)


def fixup_table(doc, table):
    # create the table head
    thead = doc.createElement("thead")
    row = doc.createElement("row")
    move_elements_by_name(doc, table, row, "entry")
    thead.appendChild(doc.createTextNode("\n    "))
    thead.appendChild(row)
    thead.appendChild(doc.createTextNode("\n    "))
    # create the table body
    tbody = doc.createElement("tbody")
    prev_row = None
    last_was_hline = 0
    children = table.childNodes
    for child in children:
        if child.nodeType == ELEMENT:
            tagName = child.get_tagName()
            if tagName == "hline" and prev_row is not None:
                prev_row.setAttribute("rowsep", "1")
            elif tagName == "row":
                prev_row = child
    # save the rows:
    tbody.appendChild(doc.createTextNode("\n    "))
    move_elements_by_name(doc, table, tbody, "row", sep="\n    ")
    # and toss the rest:
    while children:
        child = children[0]
        nodeType = child.nodeType
        if nodeType == TEXT:
            if string.strip(child.data):
                raise ConversionError("unexpected free data in table")
            table.removeChild(child)
            continue
        if nodeType == ELEMENT:
            if child.get_tagName() != "hline":
                raise ConversionError(
                    "unexpected <%s> in table" % child.get_tagName())
            table.removeChild(child)
            continue
        raise ConversionError(
            "unexpected %s node in table" % child.__class__.__name__)
    # nothing left in the <table>; add the <thead> and <tbody>
    tgroup = doc.createElement("tgroup")
    tgroup.appendChild(doc.createTextNode("\n  "))
    tgroup.appendChild(thead)
    tgroup.appendChild(doc.createTextNode("\n  "))
    tgroup.appendChild(tbody)
    tgroup.appendChild(doc.createTextNode("\n  "))
    table.appendChild(tgroup)
    # now make the <entry>s look nice:
    for row in table.getElementsByTagName("row"):
        fixup_row(doc, row)


def fixup_row(doc, row):
    entries = []
    map(entries.append, row.childNodes[1:])
    for entry in entries:
        row.insertBefore(doc.createTextNode("\n         "), entry)
#    row.appendChild(doc.createTextNode("\n      "))


def move_elements_by_name(doc, source, dest, name, sep=None):
    nodes = []
    for child in source.childNodes:
        if child.get_nodeName() == name:
            nodes.append(child)
    for node in nodes:
        source.removeChild(node)
        dest.appendChild(node)
        if sep:
            dest.appendChild(doc.createTextNode(sep))


RECURSE_INTO_PARA_CONTAINERS = (
    "chapter", "abstract", "enumerate",
    "section", "subsection", "subsubsection",
    "paragraph", "subparagraph", "back-matter",
    "howto", "manual",
    "item", "itemize", "fulllineitems", "enumeration", "descriptionlist",
    "definitionlist", "definition",
    )

PARA_LEVEL_ELEMENTS = (
    "moduleinfo", "title", "verbatim", "enumerate", "item",
    "interpreter-session", "back-matter", "interactive-session",
    "opcodedesc", "classdesc", "datadesc",
    "funcdesc", "methoddesc", "excdesc", "memberdesc", "membderdescni",
    "funcdescni", "methoddescni", "excdescni",
    "tableii", "tableiii", "tableiv", "localmoduletable",
    "sectionauthor", "seealso", "itemize",
    # include <para>, so we can just do it again to get subsequent paras:
    PARA_ELEMENT,
    )

PARA_LEVEL_PRECEEDERS = (
    "setindexsubitem",
    "stindex", "obindex", "COMMENT", "label", "input", "title",
    "versionadded", "versionchanged", "declaremodule", "modulesynopsis",
    "moduleauthor", "indexterm", "leader",
    )


def fixup_paras(doc, fragment):
    for child in fragment.childNodes:
        if child.get_nodeName() in RECURSE_INTO_PARA_CONTAINERS:
            fixup_paras_helper(doc, child)
    descriptions = find_all_elements(fragment, "description")
    for description in descriptions:
        fixup_paras_helper(doc, description)


def fixup_paras_helper(doc, container, depth=0):
    # document is already normalized
    children = container.childNodes
    start = skip_leading_nodes(children)
    while len(children) > start:
        if children[start].get_nodeName() in RECURSE_INTO_PARA_CONTAINERS:
            # Something to recurse into:
            fixup_paras_helper(doc, children[start])
        else:
            # Paragraph material:
            build_para(doc, container, start, len(children))
            if DEBUG_PARA_FIXER and depth == 10:
                sys.exit(1)
        start = skip_leading_nodes(children, start + 1)


def build_para(doc, parent, start, i):
    children = parent.childNodes
    after = start + 1
    have_last = 0
    BREAK_ELEMENTS = PARA_LEVEL_ELEMENTS + RECURSE_INTO_PARA_CONTAINERS
    # Collect all children until \n\n+ is found in a text node or a
    # member of BREAK_ELEMENTS is found.
    for j in range(start, i):
        after = j + 1
        child = children[j]
        nodeType = child.nodeType
        if nodeType == ELEMENT:
            if child.get_tagName() in BREAK_ELEMENTS:
                after = j
                break
        elif nodeType == TEXT:
            pos = string.find(child.data, "\n\n")
            if pos == 0:
                after = j
                break
            if pos >= 1:
                child.splitText(pos)
                break
    else:
        have_last = 1
    if (start + 1) > after:
        raise ConversionError(
            "build_para() could not identify content to turn into a paragraph")
    if children[after - 1].nodeType == TEXT: