library-mirrors/lua-htmlparser
1
0
Fork 0
mirror of https://github.com/msva/lua-htmlparser.git synced 2024-11-27 12:44:22 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[attribute="value"] working

Browse Source
This commit is contained in:
Wouter Scherphof 2013-03-25 13:02:41 +01:00
parent 77f24f93be
commit de746865be
2 changed files with 21 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
ElementNode.lua
View File

@ -88,14 +88,14 @@ function ElementNode:close(closestart, closeend)
end end
local function select(self, s) local function select(self, s)
if not s or type(s) ~= "string" then return Set:new() end if not s or type(s) ~= "string" or s == "" then return Set:new() end
local subjects, resultset, childrenonly = Set:new({self})
local sets = { local sets = {
[""] = self.deeperelements, [""] = self.deeperelements,
["["] = self.deeperattributes, ["["] = self.deeperattributes,
["#"] = self.deeperids, ["#"] = self.deeperids,
["."] = self.deeperclasses ["."] = self.deeperclasses
} }
local subjects, resultset, childrenonly = Set:new({self})
for part in string.gmatch(s, "%S+") do for part in string.gmatch(s, "%S+") do
if part == ">" then childrenonly = true goto nextpart end if part == ">" then childrenonly = true goto nextpart end
resultset = Set:new() resultset = Set:new()
@ -107,13 +107,24 @@ local function select(self, s)
end end
if part == "*" then goto nextpart end if part == "*" then goto nextpart end
local excludes, filter = Set:new() local excludes, filter = Set:new()
for t, w in string.gmatch(part, for t, w, v in string.gmatch(part,
"([:%[#.]?)" .. -- t = an optional :, [, #, or . "([:%[#.]?)" .. -- t = an optional :, [, #, or .
"([^:%(%[#.%]%)]+)" .. -- w = 1 or more of anything not :, (, [, #, ., ], or ) "([^:%(%[#.%]%)]+)" .. -- w = 1 or more of anything not :, (, [, #, ., ], or )
"%]?%)?" -- followed by an uncaptured optional ] and/or ) "%]?%)?" -- followed by an uncaptured optional ] and/or )
) do ) do
if t == ":" then filter = w goto nextw end if t == ":" then filter = w goto nextw end
if t == "[" then
w, v = string.match(w, "([^=]+)=?(%S*)")
end
local match = sets[t][w] local match = sets[t][w]
if v and v ~= "" then
v = string.sub(v, 2, #v - 1) -- strip quotes
for node in pairs(match) do
if node.attributes[w] ~= v then
match:remove(node)
end
end
end
if filter == "not" then if filter == "not" then
excludes = excludes + match excludes = excludes + match
else else

25
test.lua
View File

@ -51,6 +51,13 @@ select("ul > *")
select("body [class]") select("body [class]")
select("body > [class]") select("body > [class]")
select(".contacts span:not(.firstname)")
select(":not(a)[href]")
select("[itemscope]:not([itemprop])")
select("link[rel='alternate']")
select("[test2=\"val='2'\"]")
print("\nchapters") print("\nchapters")
local sel, chapters = root("ol.chapters > li"), {} local sel, chapters = root("ol.chapters > li"), {}
for e in pairs(sel) do for e in pairs(sel) do
@ -114,21 +121,3 @@ end
for node,table in pairs(scopes) do for node,table in pairs(scopes) do
printscope(node, table) printscope(node, table)
end end
print("\nnot firstname")
local sel = root(".contacts span:not(.firstname)")
for e in pairs(sel) do
print(e.classes[1], e:getcontent())
end
print("\nnot a hrefs")
local sel = root(":not(a)[href]")
for e in pairs(sel) do
print(e.name, e.attributes["href"])
end
print("\ntop itemscopes")
local sel = root("[itemscope]:not([itemprop])")
for e in pairs(sel) do
print(e.name, e.attributes["itemtype"])
end