def parseDocument(self, doc):
doc = pq(doc);
wrapparent = self.articleRule.wrapparent
pageparent = self.articleRule.pageparent
content_re = "";
#子页面url
urls = []
#文本数据内容
content = ""
article = doc.find(wrapparent);
#pages
if pageparent:
urls = self.parsePage(article, pageparent)
#need title, tags
extrarules = self.articleRule.extrarules
#只有文章是有content
if len(extrarules):
for key, rule, fetch_all in extrarules:
field = Field(field_id=key, rule=rule);
value = getElementData(doc, rule, self.data["images"], fetch_all)
self.data[field.get('name')] = field
if field.is_article_content():
content_re = field.get("rule")
content = value
elif field.is_gallery_content():
content_re = field.get("rule")
content = []
if (isinstance(value, list)):
content += value
else:
field.value = value
#采集分页内容
if len(urls) > 0 and content_re:
for next_url in urls:
next_page = Fetch(next_url, charset = self.seed["charset"], timeout = self.seed["timeout"]).read()
if next_page is not None:
next_page = self._getContent(next_page, wrapparent, content_re);
if next_page:
if isinstance(content, list):
content.append(next_page)
else:
content += next_page
if content and content_re:
if isinstance(content, list):
self.data['content'].value = content
self.data['images'] += content
else:
content = Readability(content, self.url, self.articleRule.filters)
images = content.getImages();
self.data['content'].value = content.getContent();
self.data['images'] += images
def parseListPage(self, site, doc, listurl):
'''
分析采集回来的页面
@param site Fetch instance
@param doc 页面String stream
@param url link
'''
doc = pq(doc);
list = doc.find(self.listRule.getListParent());
extrarules = self.listRule.extrarules
if list:
def entry(i, e):
#link
urlParent = self.listRule.getContentUrl()
if e.tag == "a":
link = e.get("href")
else:
link = getElementData(e, urlParent)
if link is not None:
link = urlparse.urljoin(listurl, link);
_item = Item({
"type" : self.seed_type,
"images" : []
})
for field_id, _rule, fetch_all in extrarules:
field = Field(field_id = field_id, rule=_rule)
value = getElementData(e, _rule, _item["images"])
#TODO:
# filter HOOK
field.value = value
_item[field["name"]] = field
if (link is not None):
_item['url'] = link
# get item guid
if self.guid_rule:
guid = self.getItemGUID(_item)
elif self.seed_type in self.dont_craw_content:
self.guid_rule = []
for f in _item.fields:
self.guid_rule.append(_item[f]["id"])
guid = self.getItemGUID(_item)
self.guid_rule = None
else:
self.guid_rule = "url"
guid = self.getItemGUID(_item)
self.guid_rule = None
self.items[guid] = _item
if len(self.listRule.getEntryItem()) == 0:
list.children().map(entry)
else:
list.find(self.listRule.getEntryItem()).map(entry)
def _getContent(self, html, wrapparent, content_re): if not html: return html = pq(html).find(wrapparent) _content = getElementData(html, content_re); if _content: return _content
def getHtml(self):
#自动加上html标记
if self.content.find("<html>") == -1:
content = "<html><body>" + self.content + "</body></html>"
self.html = pq(content)
def getElementData(obj, rule, images=None, fetch_all=0):
"""
根据rule对obj的进行解析
obj可以是pq后的对象， 也可以是html页面
images将会把解析过程的image连接插入此表中
规则可以有两种模式：
1. DOM selector
1.1 选择器类似于jquery 比如你要某个a的url
>> a.attr("href")
1.2 需要一个标签内的文本内容
>> div[id="content"].text()
1.3 需要获得某个子元素中的内容
>> li.eq(1).text() #li元素组中的第2个文本内容
2. 正则模式
正则模式需要的内容使用[arg]标签，其余可以使用(*)填充
"""
if not isinstance(obj, pq):
obj = pq(obj);
old_rule = rule
rule = rule.split(".")
#避免有url链接
if len(rule) > 1 and old_rule.find("[arg]") == -1:
#第一个永远是dom选择
selectRule = rule.pop(0)
#移除 ( )
selectRule = selectRule.replace("(", "");
selectRule = selectRule.replace(")", "");
selecteddom = obj.find(selectRule);
for attr in rule:
m = attrParrent.match(attr)
if m:
action, v = m.groups()
if v:
v = v.encode("utf-8")
#去除引号
v = v.strip("\'").strip('\"');
if action == "attr" and hasattr(selecteddom, "attr") and v:
if fetch_all == 1:
values = []
dom_count = len(selecteddom)
for i in range(dom_count):
vv = selecteddom.eq(i).attr(v)
if vv:
values.append(vv)
if is_image(vv):
images.append(vv)
return values
else:
value = selecteddom.attr(v)
if selecteddom and selecteddom[0].tag == "img" and v == "src" and images is not None:
images.append(value)
return value
elif action == "eq" and hasattr(selecteddom, "eq"):
_rules = attr.split(" ")
if len(rule) > 1:
selecteddom = selecteddom.eq(int(v))
if len(_rules) > 1:
'''
假设eq后面还有子元素
eq(1) a
'''
_rules.pop(0)
_dom = " ".join(_rules)
selecteddom = selecteddom.find(_dom)
else:
return selecteddom.eq(int(v))
elif action == "text" and hasattr(selecteddom, "text"):
return safeunicode(selecteddom.text()).strip()
elif action == "html" and hasattr(selecteddom, "html"):
return safeunicode(selecteddom.html()).strip()
elif len(rule) == 1:
rule = rule.pop()
#正则模式
if rule.find('[arg]'):
content = obj.html()
content_text = obj.text()
rule = rule.replace('[arg]', '(.+)?')
rule = rule.replace('(*)', '.+?')
if isinstance(content, unicode):
rule = safeunicode(rule)
else:
rule = safestr(rule)
parrent = re.compile(rule, re.MULTILINE | re.UNICODE)
try:
result = parrent.search(content)
if result is not None:
result = safeunicode(result.group(1)).strip()
return result
else:
result = parrent.search(content_text)
if result is not None:
result = safeunicode(result.group(1)).strip()
return result
except:
return None
return None
