mirrors/lua-htmlparser
1
0
Fork 0
mirror of https://github.com/TangentFoxy/lua-htmlparser.git synced 2026-01-10 16:08:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

added test_attr (and fixed one or two edge cases)

Browse Source
This commit is contained in:
Wouter Scherphof
2013-04-03 22:32:23 +02:00
parent b3bbb56d9f
commit 10a5faf192
3 changed files with 65 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/htmlparser.lua
View File

@@ -23,21 +23,19 @@ local function parse(text)
local start, k, eq, quote, v
start, apos, k, eq, quote = string.find(tagst,
"%s+" .. -- some uncaptured space
"([^%s=]+)" .. -- k = an unspaced string up to an optional "="
"([^%s=/>]+)" .. -- k = an unspaced string up to an optional "=" or the "/" or ">"
"(=?)" .. -- eq = the optional; "=", else ""
"(['\"]?)", -- quote = an optional "'" or '"' following the "=", or ""
apos)
if not k or k == "/>" or k == ">" then break end
if eq == "" then
v = ""
else
if eq == "=" then
local pattern = "=([^%s>]*)"
if quote ~= '' then
if quote ~= "" then
pattern = quote .. "([^" .. quote .. "]*)" .. quote
end
start, apos, v = string.find(tagst, pattern, apos)
end
tag:addattribute(k, v)
tag:addattribute(k, v or "")
end
if voidelements[string.lower(tag.name)] then

13
src/htmlparser/ElementNode.lua
View File

@@ -51,9 +51,8 @@ function ElementNode:addattribute(k, v)
self.attributes[k] = v
if string.lower(k) == "id" then
self.id = v
end
-- class attribute contains "space-separated tokens", each of which we'd like quick access to
if string.lower(k) == "class" then
elseif string.lower(k) == "class" then
for class in string.gmatch(v, "%S+") do
table.insert(self.classes, class)
end
@@ -98,18 +97,20 @@ local function select(self, s)
local sets = {[""] = self.deeperelements, ["["] = self.deeperattributes,
["#"] = self.deeperids, ["."] = self.deeperclasses}
local function match(t, w)
local m, v
if t == "[" then w, m, v = string.match(w,
local m, e, v
if t == "[" then w, m, e, v = string.match(w,
"([^=|%*~%$!%^]+)" .. -- w = 1 or more characters up to a possible "=", "|", "*", "~", "$", "!", or "^"
"([|%*~%$!%^]?)" .. -- m = an optional "|", "*", "~", "$", "!", or "^", preceding the optional "="
"=?" .. -- an optional uncaptured "="
"(=?)" .. -- e = the optional "="
"(.*)" -- v = anything following the "=", or else ""
)
end
local matched = Set:new(sets[t][w])
-- attribute value selectors
if v and v ~= "" then
if e == "=" then
if #v < 2 then v = "'" .. v .. "'" end -- values should be quoted
v = string.sub(v, 2, #v - 1) -- strip quotes
if m == "!" then matched = Set:new(self.deepernodes) end -- include those without that attribute
for node in pairs(matched) do
local a = node.attributes[w]
-- equals