class NestedOuterPage(PageModel):
model_class = dict
page_tree = Html(
StrictNode("div.outer")(
Node("> span")(outertxt=Text()),
Node("div.inner")(nested=NestedInnerPage()),
))
class AttrPage(PageModel):
model_class = dict
page_tree = Html(
Node("a.mylink")(
href=Attr("href"),
title=Attr("title"),
text=Text(),
))
class PostprocPage(PageModel):
model_class = dict
page_tree = Html(Node("div.lower")(lower=Text()))
@classmethod
def postproc(cls, dic):
dic['upper'] = dic.pop('lower', '').upper()
return dic
class SimplePage(PageModel):
model_class = dict
page_tree = Html(
Node("body")(
Node("div.div_1")(div1=Text()),
Node("#div_2", "#asdf")(div2=Text()),
Node("span"),
body=Text(),
), )
class PhrasalVerbLink(PageModel):
model_class = models.Link
page_tree = Html(
Node("a")(
url=Attr("href"),
key=Attr("title"),
link_type=Constant("phrasal verbs"),
part_of_speech=Constant("phrasal verb"),
)
)
class Entry(PageModel):
model_class = models.Entry
page_tree = Html(
Node("div#headword div#headwordleft span.BASE")(
original_key=Text()
),
Node("div#headbar")(
Node.optional("span.STYLE-LEVEL")(
style_level=Text()
),
Node.optional("span.PRON")(
pron=Text()
),
Node.optional("span.PART-OF-SPEECH")(
part_of_speech=Text()
)
),
Node.optional("div.SUMMARY div.p")(
intro_paragraph=Text()
),
Node.list("div.SENSE-BODY")(
senses=Sense()
),
Node.optional("div#phrases_container > ul")(
Node.list("li")(
phrs=PhraseLink()
)
),
Node.optional("div#phrasal_verbs_container > ul")(
Node.list("li")(
phrvbs=PhrasalVerbLink()
)
),
Node.optional("div.entrylist > ul")(
Node.list("li")(
relwrds=RelatedWordLink()
)
)
)
@classmethod
def postproc(cls, dic):
dic['links'] = dic.pop('relwrds', [])
dic['links'] += dic.pop('phrvbs', []) + dic.pop('phrs', [])
return dic
class RelatedWordLink(PageModel):
model_class = models.Link
page_tree = Html(
Node("a")(
Node.optional("span.PART-OF-SPEECH")(
part_of_speech=Text()
),
key=Attr("title"),
url=Attr("href"),
link_type=Constant("related words"),
)
)
@classmethod
def postproc(cls, dic):
k = dic["key"]
p = dic.get("part_of_speech", "")
k = k[:-len(p)]
k = k.strip()
dic["key"] = k
class NestedInnerPage(PageModel):
model_class = dict
page_tree = Html(Node("span")(innertxt=Text()))
class ThisClassElem(PageModel):
model_class = dict
page_tree = Html(
Node("> div.head")(head=Text()),
Node.optional("> div.tail")(tail=ThisClass()),
)
class OptionalNodePage(PageModel):
model_class = dict
page_tree = Html(Node.optional("div.missing")(x=Text()))
class StrictPage(PageModel):
model_class = dict
page_tree = Html(StrictNode("div.strict")(Node("span"), ))
class MissingNodePage(PageModel):
model_class = dict
page_tree = Html(Node("div.missing")(x=Text()))
class ThisClassPage(PageModel):
model_class = dict
page_tree = Html(Node("div.list")(li=ThisClassElem()))
class InvalidPageTwo(PageModel):
model_class = dict
page_tree = Html(Node("div")(Text()))
class ConcatPage(PageModel):
model_class = dict
page_tree = Html(
Node("div.list")(
Node.list("span.elem").concat(", ")(concatenated=Text())))
class ConstantPage(PageModel):
model_class = dict
page_tree = Html(Node("div.doesnotmatter")(const=Constant("myconstant")))
class TakefirstPage(PageModel):
model_class = dict
page_tree = Html(Node.list("div.listelem").take_first()(firstelem=Text()))
class InvalidPage(PageModel):
model_class = dict
page_tree = Html(
Node("div.one")(dupfield=Text()),
Node("div.two")(dupfield=Text()))
class ListPage(PageModel):
model_class = dict
page_tree = Html(Node(".list")(Node.list(".listelem")(mylistfield=Text())))
