mirror of
https://github.com/msva/lua-htmlparser.git
synced 2024-11-04 23:34:20 +00:00
141 lines
3.3 KiB
Lua
141 lines
3.3 KiB
Lua
require("luarocks.loader")
|
|
-- Omit next line in actual module clients; it's only to support development of the module itself
|
|
package.path = "../src/?.lua;" .. package.path
|
|
local htmlparser = require("htmlparser")
|
|
|
|
local io = require("io")
|
|
local file = io.input("./sample.html")
|
|
local text = io.read("*a") file:close()
|
|
|
|
local root = htmlparser.parse(text)
|
|
|
|
-- print the tree
|
|
local function p(n)
|
|
local space = string.rep(" ", n.level)
|
|
local s = space .. n.name
|
|
for k,v in pairs(n.attributes) do
|
|
s = s .. " " .. k .. "=[[" .. v .. "]]"
|
|
end
|
|
print(s)
|
|
for i,v in ipairs(n.nodes) do
|
|
p(v)
|
|
end
|
|
end
|
|
p(root)
|
|
|
|
local function select( s )
|
|
print ""
|
|
print("->", s)
|
|
local sel = root:select(s)
|
|
for element in pairs(sel) do
|
|
print(element.name)
|
|
end
|
|
print(sel:len())
|
|
end
|
|
|
|
select("*")
|
|
select("link")
|
|
select("#/contacts/4711")
|
|
select(".chapters")
|
|
select("[href]")
|
|
select("span.firstname")
|
|
select("ul[id]")
|
|
|
|
select("#/contacts/4711")
|
|
select("#/contacts/4711 *")
|
|
select("#/contacts/4711 .lastname")
|
|
select("body li[id]")
|
|
|
|
select("ul")
|
|
select("ul *")
|
|
select("ul > *")
|
|
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'\"]")
|
|
select("[test5='val5']")
|
|
select("[test6='val\"\"6']")
|
|
select("[itemscope='']")
|
|
select("[itemscope=]")
|
|
select("[itemscope]")
|
|
|
|
select("[itemscope][itemprop='address']")
|
|
select("[itemscope][itemprop!='address']")
|
|
select("[itemscope][itemprop!='adres']")
|
|
select("[itemscope][itemprop!='']")
|
|
select("[hreflang|='en']")
|
|
select("[itemprop*='address']")
|
|
select("[words~='two']")
|
|
select("[words~='three']")
|
|
select("[itemprop$='ion']")
|
|
select("[hreflang^='en']")
|
|
|
|
print("\nchapters")
|
|
local sel, chapters = root("ol.chapters > li"), {}
|
|
for e in pairs(sel) do
|
|
table.insert(chapters, e:getcontent())
|
|
end
|
|
-- print
|
|
for i,v in ipairs(chapters) do
|
|
print(i, v)
|
|
end
|
|
|
|
print("\ncontacts")
|
|
local sel, contacts = root("ul.contacts span[class]"), {}
|
|
for e in pairs(sel) do
|
|
local id = e.parent.parent.id -- li > a > span
|
|
contacts[id] = contacts[id] or {}
|
|
contacts[id][e.classes[1]] = e:getcontent()
|
|
end
|
|
-- print
|
|
for k,v in pairs(contacts) do
|
|
print(k)
|
|
for fk,fv in pairs(v) do
|
|
print(fk, fv)
|
|
end
|
|
end
|
|
|
|
print("\nmicrodata")
|
|
local sel, scopes = root("[itemprop]"), {}
|
|
for prop in pairs(sel) do
|
|
if prop.attributes["itemscope"] then goto nextprop end
|
|
local descendantscopes, scope = {}, prop
|
|
while true do
|
|
repeat
|
|
scope = scope.parent
|
|
until scope.attributes["itemscope"]
|
|
if not scope.attributes["itemprop"] then break end
|
|
table.insert(descendantscopes, 1, scope)
|
|
end
|
|
scopes[scope] = scopes[scope] or {}
|
|
local entry = scopes[scope]
|
|
for _,v in ipairs(descendantscopes) do
|
|
entry[v] = entry[v] or {}
|
|
entry = entry[v]
|
|
end
|
|
local k, v = prop.attributes["itemprop"], prop:getcontent()
|
|
entry[k] = v
|
|
::nextprop::
|
|
end
|
|
-- print
|
|
local function printscope(node, table, level)
|
|
level = level or 1
|
|
local scopeprop = node.attributes["itemprop"] or ""
|
|
print(string.rep(" ", level - 1) .. node.attributes["itemtype"], scopeprop)
|
|
for prop,v in pairs(table) do
|
|
if type(prop) == "table" then
|
|
printscope(prop, v, level + 1)
|
|
else
|
|
print(string.rep(" ", level) .. prop .. "=[" .. v .. "]")
|
|
end
|
|
end
|
|
end
|
|
for node,table in pairs(scopes) do
|
|
printscope(node, table)
|
|
end
|