def transform_header(self, obj):
check(obj, "content", -1, "type") == "template"
check(obj, "content", -1, "name") == "anker"
heading = text_rstrip(obj["content"][:-1])
anchor = obj["content"][-1]["params"]["1"]
return merge(obj, {"content": heading, "anchor": anchor})
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name") == "table"
content = obj["children"]
if lookup(content, 0, "name") == "tbody":
content = content[0]["children"]
return {"type": "table", "content": self(content)}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name").of(("ul", "ol"))
items = [{
"type": "listitem",
"content": self(li["children"])
} for li in obj["children"]]
return {
"type": "list",
"ordered": obj["name"] == "ol",
"items": items
}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name") == "dl"
items = [{
"type": "definitionlistitem",
"definition": self(dt["children"]),
"explanation": self(dd["children"])
} for dt, dd in zip(obj["children"][::2], obj["children"][1::2])]
if not items:
message = "A definition list must not be empty!"
log_parser_error(message, obj, position=self.current_section)
return {"type": "error", "message": message}
return {"type": "definitionlist", "items": items}
def transform_element(self, obj):
if lookup(obj, "attrs", "about") in self._template_ids:
return None
check(obj, "attrs", "typeof").of([
"mw:Transclusion", "mw:Transclusion mw:Video/Thumb",
"mw:Transclusion mw:Image"
])
template = json.loads(obj["attrs"]["data-mw"])["parts"][0]
try:
template = template["template"]
except (TypeError, KeyError):
return {
"type": "error",
"message": "Template spans over several HTML elements."
}
name = template["target"]["wt"].strip()
# labeled section transclusion needs unchanged case.
if not name.startswith("#lst:"):
name = name.lower()
if name != "(!":
# Template includes a table afterwards
self._template_ids.add(obj["attrs"]["about"])
name = remove_prefix(name, ":mathe für nicht-freaks: vorlage:")
params = template["params"]
params = {k: v["wt"] for k, v in params.items()}
params = {key: self.parse_parameter_value(name, key, value) \
for key, value in params.items()
if not params.get(key + "-noprint", False)}
# TODO: Find better solution
if params.get("noprint", False):
return None
return {"type": "template", "name": name, "params": params}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name").of(("figure", "span", "figure-inline"))
check(obj, "attrs", "typeof").of(("mw:Image", "mw:Image/Thumb"))
caption = [
child for child in obj["children"]
if child["name"] == "figcaption"
]
try:
caption = caption[0]["children"]
except IndexError:
caption = []
img = obj["children"][0]["children"][0]
name = canonical_image_name(img["attrs"]["resource"])
license = self.api.get_image_license(name)
return {
"type": "image",
"caption": self(caption),
"name": name,
"thumbnail": obj["attrs"]["typeof"] == "mw:Image/Thumb",
"inline": obj["name"] in ("span", "figure-inline"),
"license": license,
"noprint": "noprint" in obj["attrs"].get("class", "")
}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name") == "span"
check(obj, "attrs", "typeof") == "mw:Image"
message = "Inline images are not allowed"
log_parser_error(message, obj, position=self.current_section)
return {"type": "error", "message": message}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name") == "ul"
check(obj, "attrs", "typeof") == "mw:Extension/gallery"
data_mw = json.loads(obj["attrs"]["data-mw"])
spec = data_mw["body"]["extsrc"].strip()
items = [self.parse_gallery_item(x) for x in spec.splitlines()]
return {
"type": "gallery",
"widths": int(data_mw["attrs"].get("widths", 120)),
"heights": int(data_mw["attrs"].get("heights", 120)),
"items": items
}
def transform_dict(self, obj):
check(obj, "type").of(DEFAULT_VALUES)
return merge(DEFAULT_VALUES[obj["type"]],
super(NodeTransformation, self).act_on_dict(obj))
def transform_dict(self, obj):
check(obj, "attrs", "typeof") == "mw:Extension/math"
formula = json.loads(obj["attrs"]["data-mw"])["body"]["extsrc"]
return {"type": "inlinemath", "formula": formula.strip()}
def transform_dict(self, obj):
check(obj, "type") == "element"
check(obj, "name").of(("span", ))
check(obj, "attrs", "typeof") == "mw:FallbackId"
return None