fix bug where not checking for attrs != nil

This commit is contained in:
ccppi 2024-07-02 12:11:24 +02:00
parent 605480dbfd
commit f629ff784f
3 changed files with 26 additions and 18 deletions

BIN
main

Binary file not shown.

View File

@ -10,7 +10,7 @@ var html: string
#var node: XmlNode #var node: XmlNode
echo "URL:" echo "URL:"
var url: string = readLine(stdin) var url: string = "https://www.comparis.ch/immobilien/marktplatz/lenzburg/mieten"#readLine(stdin)
echo "given url is: ",url echo "given url is: ",url
try: try:
@ -20,7 +20,7 @@ try:
var htmlnode: XmlNode var htmlnode: XmlNode
var entry : Entry var entry : Entry
add(entry.desc,Descriptor(name : "", html_context_tag : "div",html_context_attrs : "class",html_context_key : "css-1d60yla ehesakb4", html_tag : "a",attrs : "href")) add(entry.desc,Descriptor(name : "test", html_context_tag : "div",html_context_attrs : "class",html_context_key : "css-1wvv8q", html_tag : "a",aattrs : "href"))
#add(entry.desc,Descriptor(name:"key_and_string_search",contains_string: "a",html_context_tag : "div", html_tag : "a",attrs : "href")) #add(entry.desc,Descriptor(name:"key_and_string_search",contains_string: "a",html_context_tag : "div", html_tag : "a",attrs : "href"))
echo entry echo entry

View File

@ -11,7 +11,7 @@ type
html_context_attrs* : string html_context_attrs* : string
html_context_key* : string html_context_key* : string
contains_string* : string contains_string* : string
attrs* : string aattrs* : string
content* : seq[string] content* : seq[string]
type type
Entry* = object Entry* = object
@ -30,26 +30,34 @@ proc getEntryFromHtml*(entry :var Entry, node : XmlNode) : Entry =
context = node.findAll(entry.desc[i].html_context_tag) context = node.findAll(entry.desc[i].html_context_tag)
else: else:
tmpContext = node.findAll(entry.desc[i].html_context_tag) tmpContext = node.findAll(entry.desc[i].html_context_tag)
echo(tmpContext.len)
#filter by attributes #filter by attributes
for i3,c in tmpContext: for c in tmpContext:
echo("c: ",c.attrs, "i3: ", i3, "i: ", i, "len tmpContext: ",tmpContext.len()) echo c.len
if c.attrs.hasKey(entry.desc[i].html_context_attrs): #echo("c: ",c.attrs, " i3: ", i3, " i: ", i, " len tmpContext: ", "entry: ",entry.desc[i])
echo("HAS-Context_KEY: ",c.attrs) if entry.desc[i].html_context_attrs != "":
if entry.desc[i].html_context_key != "": if c.attrs != nil:
if c.attrs[entry.desc[i].html_context_attrs] == entry.desc[i].html_context_key: if c.attrs.hasKey(entry.desc[i].html_context_attrs):
context.add(c) echo("c.attrs[key]: ",c.attrs[entry.desc[i].html_context_attrs])
echo("key_MATCH") if entry.desc[i].html_context_key != "":
if entry.desc[i].html_context_key.contains(c.attrs[entry.desc[i].html_context_attrs]):
context.add(c)
echo("context_key_ MATCH")
else:
context.add(c)
else:
echo("c.attrs does not have the key")
else: else:
context.add(c) echo("no attributes on this tag")
echo("out of for loop") echo("out of for loop")
#end filter context #end filter context
for a in context: for a in context:
let subContext = a.findAll(entry.desc[0].html_tag) let subContext = a.findAll(entry.desc[i].html_tag)
for b in subContext: for b in subContext:
if entry.desc[i].attrs != "": if entry.desc[i].aattrs != "":
if b.attrs.hasKey(entry.desc[i].attrs): if b.attrs.hasKey(entry.desc[i].aattrs):
echo("found key [",entry.desc[i].attrs,"] in : ",b) echo("found key [",entry.desc[i].aattrs,"] in : ",b)
echo("value: ",b.attrs[entry.desc[i].attrs]) echo("value: ",b.attrs[entry.desc[i].aattrs])
if entry.desc[i].contains_string != "": if entry.desc[i].contains_string != "":
var content_txt :string = b.innerText() var content_txt :string = b.innerText()
echo(" and string(innerText): ",content_txt) echo(" and string(innerText): ",content_txt)
@ -57,7 +65,7 @@ proc getEntryFromHtml*(entry :var Entry, node : XmlNode) : Entry =
echo("PATTERN MATCH") echo("PATTERN MATCH")
entry.desc[i].content.add(content_txt) entry.desc[i].content.add(content_txt)
else: else:
entry.desc[i].content.add(b.attrs[entry.desc[i].attrs]) entry.desc[i].content.add(b.attrs[entry.desc[i].aattrs])
elif entry.desc[i].contains_string != "": elif entry.desc[i].contains_string != "":
var content_txt = b.innerText() var content_txt = b.innerText()
echo(" string(innerText): ",content_txt) echo(" string(innerText): ",content_txt)