This commit is contained in:
2024-07-03 14:03:25 +02:00
parent a2a0001aef
commit 42bd90fb7b
4 changed files with 1981 additions and 41 deletions

View File

@@ -14,6 +14,7 @@ type
aattrs* : string
attrs_key*: string
content* : seq[string]
getInline : bool
type
Entry* = object
desc* : seq[Descriptor]
@@ -53,36 +54,42 @@ proc getEntryFromHtml*(entry :var Entry, node : XmlNode) : Entry =
echo("c.attrs does not have the key: ", entry.desc[i].html_context_attrs)
else:
echo("no attributes on this tag")
echo("----------CONTEXT RESULT")
echo(context)
#echo("----------CONTEXT RESULT")
#echo(context)
if entry.desc[i].html_tag == "":
echo("No html_tag specified, return whole context as string")
for a in context:
entry.desc[i].content.add(a.innerText)
return entry
else:
#end filter context
for a in context:
let subContext = a.findAll(entry.desc[i].html_tag)
for b in subContext:
if entry.desc[i].aattrs != "":
if b.attrs != nil:
if b.attrs.hasKey(entry.desc[i].aattrs):
echo("found key [",entry.desc[i].aattrs,"] in : ",b)
echo("value: ",b.attrs[entry.desc[i].aattrs])
if entry.desc[i].contains_string != "":
var content_txt :string = b.innerText()
echo(" and string(innerText): ",content_txt)
if content_txt.contains(entry.desc[i].contains_string):
echo("ATTRS MATCH")
entry.desc[i].content.add(content_txt)
elif entry.desc[i].attrs_key != "":
#if entry.desc[i].attrs_key.contains(b.attrs[entry.desc[i].aattrs]):
if b.attrs[entry.desc[i].aattrs].contains(entry.desc[i].attrs_key):
echo("KEY_MATCH")
entry.desc[i].content.add(b.innerText())
else:
entry.desc[i].content.add(b.attrs[entry.desc[i].aattrs])
elif entry.desc[i].contains_string != "":
var content_txt = b.innerText()
echo(" string(innerText): ",content_txt)
if content_txt.contains(entry.desc[i].contains_string):
echo("PATTERN MATCH")
entry.desc[i].content.add(content_txt)
for a in context:
let subContext = a.findAll(entry.desc[i].html_tag)
for b in subContext:
if entry.desc[i].aattrs != "":
if b.attrs != nil:
if b.attrs.hasKey(entry.desc[i].aattrs):
echo("found key [",entry.desc[i].aattrs,"] in : ",b)
echo("value: ",b.attrs[entry.desc[i].aattrs])
if entry.desc[i].contains_string != "":
var content_txt :string = b.innerText()
echo(" and string(innerText): ",content_txt)
if content_txt.contains(entry.desc[i].contains_string):
echo("ATTRS MATCH")
entry.desc[i].content.add(content_txt)
elif entry.desc[i].attrs_key != "":
#if entry.desc[i].attrs_key.contains(b.attrs[entry.desc[i].aattrs]):
if b.attrs[entry.desc[i].aattrs].contains(entry.desc[i].attrs_key):
echo("KEY_MATCH")
entry.desc[i].content.add(b.innerText())
else:
entry.desc[i].content.add(b.attrs[entry.desc[i].aattrs])
elif entry.desc[i].contains_string != "":
var content_txt = b.innerText()
echo(" string(innerText): ",content_txt)
if content_txt.contains(entry.desc[i].contains_string):
echo("PATTERN MATCH")
entry.desc[i].content.add(content_txt)
return entry
proc test() =