implement tag searching

main.nim

@@ -3,6 +3,7 @@ import std/htmlparser
 import std/xmltree
 import std/strtabs
 import std/os
+import scrap
 
 var client = newHttpClient()
 var html: string
@@ -13,17 +14,23 @@ var url: string = readLine(stdin)
 echo "given url is: ",url
 
 try:
-    html = client.getContent(url)
+  html = client.getContent(url)
-    let node =  parseHtml(html)
+  let node =  parseHtml(html)
-    echo node
+  
-    for a in node.findAll("a"):
+  var htmlnode: XmlNode
-        if a.attrs.hasKey "href":
+  var entry : Entry
-            echo "striping"
-            let (dir,filename,ext) = splitFile(a.attrs["href"])
-            echo "found a link!",dir & "/" & filename
-        else:
-            echo "Key has no attribute href"
 
+  entry.name = Descriptor(html_context_tag : "div", html_tag : "a",attrs : "href")
+
+  entry.getEntryFromHtml(node)
+#    echo node
+#    for a in node.findAll("a"):
+#      if a.attrs.hasKey "href":
+#        echo "striping"
+#        let (dir,filename,ext) = splitFile(a.attrs["href"])
+#        echo "found a link!",dir & "/" & filename
+#      else:
+#        echo "Key has no attribute href"
 finally:
     client.close()
 

scrap.nim

@@ -3,51 +3,46 @@ import std/xmltree
 import std/strtabs
 
 type 
-  Descriptor = object
+  Descriptor* = object
-    name : string
+    name* : string
-    html_tag : string
+    html_tag* : string
-    html_context_tag : string
+    html_context_tag* : string
-    contains_string : string
+    contains_string* : string
-    attrs : string
+    attrs* : string
 type
-  Entry = object
+  Entry* = object
-    name, tag, description, link, category : Descriptor
+    name*, tag*, description*, link*, category* : Descriptor
 
-proc setDescriptor(desc : var Descriptor, descToChange :  string, value :  string) =
+proc getEntryFromHtml*(entry : Entry, node : XmlNode) =
-  case descToChange:
-    of "name":
-      desc.name  = value
-    of "html_tag":
-      desc.html_tag = value
-    of "html_context_tag":
-      desc.html_context_tag = value
-    of "contains_string":
-      desc.contains_string = value
-    of "attrs":
-      desc.attrs = value
-
-
-#proc newEntry() : Entry =
-#  newDescriptor()
-
-proc getEntryFromHtml(entry : Entry, node : XmlNode) =
   echo(entry.description)
-  var childrens = node.findAll(entry.name.html_tag)
+
+  let context = node.findAll(entry.name.html_context_tag)
+  for a in context:
+    
+    let subContext = a.findAll(entry.name.html_tag)
+    for b in subContext:
+      echo(b)
+      if entry.name.attrs != "":
+        if b.attrs.hasKey(entry.name.attrs):
+          echo("found key")
+          if entry.name.contains_string != "":
+            echo(" and string")
+            #check if b contains contains_string
+            #return function
+      if entry.name.contains_string != "":
+        echo("found string")
+        #check if b contains contains_string
+        #return function
 
 
 proc test() =
   var htmlnode : XmlNode
   var str_html : string
   var entry: Entry
-#  entry = Entry(name : "testname",description : "testdescription",link : "testlink",category : "testcategory")
+  entry.name = Descriptor(html_tag : "p", contains_string : "test_container")
-  entry.name.setDescriptor("name","testname")
-  entry.name.setDescriptor("html_tag","p")
-  entry.link.setDescriptor("name","inserat_link")
-  entry.link.setDescriptor("html_tag","a")
-  entry.link.setDescriptor("attrs","href")
   echo entry
 
   htmlnode = parseHtml(str_html)
-  #entry.getEntryFromHtml()
+  entry.getEntryFromHtml(htmlnode)
 
-test()
+#test()