mirrors/lua-htmlparser
1
0
Fork 0
mirror of https://github.com/TangentFoxy/lua-htmlparser.git synced 2025-07-28 11:02:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Moved some details to a function

Browse Source 
... regarding [attribute=value] and other matching.
Also explained (in comments) the different matching patterns.
And fixed a bug where /> would be listed as an attribute.
And added a few more tests.
This commit is contained in:
Wouter Scherphof
2013-03-25 13:55:35 +01:00
parent de746865be
commit 64f3eb4df3
4 changed files with 52 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
ElementNode.lua
View File

@@ -89,12 +89,34 @@ end
local function select(self, s)
if not s or type(s) ~= "string" or s == "" then return Set:new() end
local sets = {
[""] = self.deeperelements,
["["] = self.deeperattributes,
["#"] = self.deeperids,
["."] = self.deeperclasses
}
local function match(t, w)
local sets = {
[""] = self.deeperelements,
["["] = self.deeperattributes,
["#"] = self.deeperids,
["."] = self.deeperclasses
}
local v
if t == "[" then
w, v = string.match(w,
"([^=]+)" .. -- w = 1 or more characters up to a possible "="
"=?" .. -- an optional uncaptured "="
"(.*)" -- v = anything following the "=", or else ""
)
end
local matched = sets[t][w]
if v and v ~= "" then
v = string.sub(v, 2, #v - 1) -- strip quotes
for node in pairs(matched) do
if node.attributes[w] ~= v then
matched:remove(node)
end
end
end
return matched
end
local subjects, resultset, childrenonly = Set:new({self})
for part in string.gmatch(s, "%S+") do
if part == ">" then childrenonly = true goto nextpart end
@@ -107,28 +129,17 @@ local function select(self, s)
end
if part == "*" then goto nextpart end
local excludes, filter = Set:new()
for t, w, v in string.gmatch(part,
for t, w in string.gmatch(part,
"([:%[#.]?)" .. -- t = an optional :, [, #, or .
"([^:%(%[#.%]%)]+)" .. -- w = 1 or more of anything not :, (, [, #, ., ], or )
"%]?%)?" -- followed by an uncaptured optional ] and/or )
) do
if t == ":" then filter = w goto nextw end
if t == "[" then
w, v = string.match(w, "([^=]+)=?(%S*)")
end
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
local matched = match(t, w)
if filter == "not" then
excludes = excludes + match
excludes = excludes + matched
else
resultset = resultset * match
resultset = resultset * matched
end
filter = nil
::nextw::