mirror of
https://github.com/msva/lua-htmlparser.git
synced 2024-11-04 23:34:20 +00:00
Moved some details to a function
... 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
parent
de746865be
commit
64f3eb4df3
@ -89,12 +89,34 @@ end
|
|||||||
|
|
||||||
local function select(self, s)
|
local function select(self, s)
|
||||||
if not s or type(s) ~= "string" or s == "" then return Set:new() end
|
if not s or type(s) ~= "string" or s == "" then return Set:new() end
|
||||||
|
|
||||||
|
local function match(t, w)
|
||||||
local sets = {
|
local sets = {
|
||||||
[""] = self.deeperelements,
|
[""] = self.deeperelements,
|
||||||
["["] = self.deeperattributes,
|
["["] = self.deeperattributes,
|
||||||
["#"] = self.deeperids,
|
["#"] = self.deeperids,
|
||||||
["."] = self.deeperclasses
|
["."] = 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})
|
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
|
||||||
@ -107,28 +129,17 @@ 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, v in string.gmatch(part,
|
for t, w 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
|
local matched = match(t, w)
|
||||||
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
|
|
||||||
if filter == "not" then
|
if filter == "not" then
|
||||||
excludes = excludes + match
|
excludes = excludes + matched
|
||||||
else
|
else
|
||||||
resultset = resultset * match
|
resultset = resultset * matched
|
||||||
end
|
end
|
||||||
filter = nil
|
filter = nil
|
||||||
::nextw::
|
::nextw::
|
||||||
|
|||||||
@ -9,7 +9,11 @@ local function parse(text)
|
|||||||
local node, descend, tpos, opentags = root, true, 1, {}
|
local node, descend, tpos, opentags = root, true, 1, {}
|
||||||
while true do
|
while true do
|
||||||
local openstart, name
|
local openstart, name
|
||||||
openstart, tpos, name = string.find(root._text, "<(%w+)[^>]*>", tpos)
|
openstart, tpos, name = string.find(root._text,
|
||||||
|
"<" .. -- an uncaptured starting "<"
|
||||||
|
"(%w+)" .. -- name = the first word, directly following the "<"
|
||||||
|
"[^>]*>", -- include, but not capture everything up to the next ">"
|
||||||
|
tpos)
|
||||||
if not name then break end
|
if not name then break end
|
||||||
local tag = ElementNode:new(name, node, descend, openstart, tpos)
|
local tag = ElementNode:new(name, node, descend, openstart, tpos)
|
||||||
node = tag
|
node = tag
|
||||||
@ -17,12 +21,17 @@ local function parse(text)
|
|||||||
local tagst, apos = tag:gettext(), 1
|
local tagst, apos = tag:gettext(), 1
|
||||||
while true do
|
while true do
|
||||||
local start, k, eq, quote, v
|
local start, k, eq, quote, v
|
||||||
start, apos, k, eq, quote = string.find(tagst, "%s+([^%s=]+)(=?)(['\"]?)", apos)
|
start, apos, k, eq, quote = string.find(tagst,
|
||||||
if not k then break end
|
"%s+" .. -- some uncaptured space
|
||||||
|
"([^%s=]+)" .. -- k = an unspaced string up to an optional "="
|
||||||
|
"(=?)" .. -- eq = the optiona; "=", else ""
|
||||||
|
"(['\"]?)", -- quote = an optional "'" or '"' following the "=", or ""
|
||||||
|
apos)
|
||||||
|
if not k or k == "/>" then break end
|
||||||
if eq == "" then
|
if eq == "" then
|
||||||
v = ""
|
v = ""
|
||||||
else
|
else
|
||||||
local pattern = "=([^%s'\">]*)"
|
local pattern = "=([^%s>]*)"
|
||||||
if quote ~= '' then
|
if quote ~= '' then
|
||||||
pattern = quote .. "([^" .. quote .. "]*)" .. quote
|
pattern = quote .. "([^" .. quote .. "]*)" .. quote
|
||||||
end
|
end
|
||||||
|
|||||||
@ -1,5 +1,5 @@
|
|||||||
<!DOCTYPE html>
|
<!DOCTYPE html>
|
||||||
<html lang="en" test1='val1' test2="val='2'" test3='val="3"' test4="val = 4" test5=val5>
|
<html lang="en" test1='val1' test2="val='2'" test3='val="3"' test4="val = 4" test5=val5 test6=val""6>
|
||||||
<head>
|
<head>
|
||||||
<meta charset="utf-8" />
|
<meta charset="utf-8" />
|
||||||
<link rel="stylesheet" href="test.css" />
|
<link rel="stylesheet" href="test.css" />
|
||||||
|
|||||||
10
test.lua
10
test.lua
@ -10,11 +10,8 @@ local root = htmlparser.parse(text)
|
|||||||
local function p(n)
|
local function p(n)
|
||||||
local space = string.rep(" ", n.level)
|
local space = string.rep(" ", n.level)
|
||||||
local s = space .. n.name
|
local s = space .. n.name
|
||||||
for i,v in ipairs(n.nodes) do
|
|
||||||
s = s .. " nodes[" .. i .. "]=" .. v.name
|
|
||||||
end
|
|
||||||
for k,v in pairs(n.attributes) do
|
for k,v in pairs(n.attributes) do
|
||||||
s = s .. " " .. k .. "=[" .. v .. "]"
|
s = s .. " " .. k .. "=[[" .. v .. "]]"
|
||||||
end
|
end
|
||||||
print(s)
|
print(s)
|
||||||
for i,v in ipairs(n.nodes) do
|
for i,v in ipairs(n.nodes) do
|
||||||
@ -57,6 +54,11 @@ select("[itemscope]:not([itemprop])")
|
|||||||
|
|
||||||
select("link[rel='alternate']")
|
select("link[rel='alternate']")
|
||||||
select("[test2=\"val='2'\"]")
|
select("[test2=\"val='2'\"]")
|
||||||
|
select("[test5='val5']")
|
||||||
|
select("[test6='val\"\"6']")
|
||||||
|
select("[itemscope='']")
|
||||||
|
select("[itemscope=]")
|
||||||
|
select("[itemscope]")
|
||||||
|
|
||||||
print("\nchapters")
|
print("\nchapters")
|
||||||
local sel, chapters = root("ol.chapters > li"), {}
|
local sel, chapters = root("ol.chapters > li"), {}
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user