Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,53 +2,50 @@ import gradio as gr
|
|
| 2 |
import requests
|
| 3 |
import xml.etree.ElementTree as ET
|
| 4 |
|
| 5 |
-
|
| 6 |
def collect_urls_from_sitemap(sitemap_content):
|
| 7 |
root = ET.fromstring(sitemap_content)
|
| 8 |
-
|
| 9 |
urls = []
|
| 10 |
for url_element in root.findall(".//{http://www.sitemaps.org/schemas/sitemap/0.9}url"):
|
| 11 |
loc_element = url_element.find("{http://www.sitemaps.org/schemas/sitemap/0.9}loc")
|
| 12 |
if loc_element is not None:
|
| 13 |
urls.append(loc_element.text)
|
| 14 |
-
|
| 15 |
return urls
|
| 16 |
|
| 17 |
-
|
| 18 |
def compare_sitemaps(old_urls, new_urls):
|
| 19 |
added_urls = list(set(new_urls) - set(old_urls))
|
| 20 |
removed_urls = list(set(old_urls) - set(new_urls))
|
| 21 |
return added_urls, removed_urls
|
| 22 |
|
| 23 |
-
|
| 24 |
def compare_sitemaps_ui(old_sitemap_url, new_sitemap_url):
|
| 25 |
old_response = requests.get(old_sitemap_url)
|
| 26 |
new_response = requests.get(new_sitemap_url)
|
| 27 |
-
|
| 28 |
if old_response.status_code == 200 and new_response.status_code == 200:
|
| 29 |
old_sitemap_content = old_response.content
|
| 30 |
new_sitemap_content = new_response.content
|
| 31 |
-
|
| 32 |
old_urls = collect_urls_from_sitemap(old_sitemap_content)
|
| 33 |
new_urls = collect_urls_from_sitemap(new_sitemap_content)
|
| 34 |
-
|
| 35 |
added_urls, removed_urls = compare_sitemaps(old_urls, new_urls)
|
| 36 |
-
|
| 37 |
added_urls_text = "\n".join(added_urls)
|
| 38 |
removed_urls_text = "\n".join(removed_urls)
|
| 39 |
-
|
| 40 |
-
return
|
| 41 |
else:
|
| 42 |
-
return "Failed to fetch sitemaps."
|
| 43 |
-
|
| 44 |
|
| 45 |
iface = gr.Interface(
|
| 46 |
fn=compare_sitemaps_ui,
|
| 47 |
inputs=["text", "text"],
|
| 48 |
-
outputs="
|
| 49 |
title="Sitemap Comparator",
|
| 50 |
description="Enter the URLs of the old and new sitemaps to compare.",
|
| 51 |
examples=[["http://bonrix.net/old-sitemap.xml", "http://bonrix.net/sitemap.xml"]]
|
| 52 |
)
|
| 53 |
|
| 54 |
iface.launch()
|
|
|
|
|
|
| 2 |
import requests
|
| 3 |
import xml.etree.ElementTree as ET
|
| 4 |
|
|
|
|
| 5 |
def collect_urls_from_sitemap(sitemap_content):
|
| 6 |
root = ET.fromstring(sitemap_content)
|
| 7 |
+
|
| 8 |
urls = []
|
| 9 |
for url_element in root.findall(".//{http://www.sitemaps.org/schemas/sitemap/0.9}url"):
|
| 10 |
loc_element = url_element.find("{http://www.sitemaps.org/schemas/sitemap/0.9}loc")
|
| 11 |
if loc_element is not None:
|
| 12 |
urls.append(loc_element.text)
|
| 13 |
+
|
| 14 |
return urls
|
| 15 |
|
|
|
|
| 16 |
def compare_sitemaps(old_urls, new_urls):
|
| 17 |
added_urls = list(set(new_urls) - set(old_urls))
|
| 18 |
removed_urls = list(set(old_urls) - set(new_urls))
|
| 19 |
return added_urls, removed_urls
|
| 20 |
|
|
|
|
| 21 |
def compare_sitemaps_ui(old_sitemap_url, new_sitemap_url):
|
| 22 |
old_response = requests.get(old_sitemap_url)
|
| 23 |
new_response = requests.get(new_sitemap_url)
|
| 24 |
+
|
| 25 |
if old_response.status_code == 200 and new_response.status_code == 200:
|
| 26 |
old_sitemap_content = old_response.content
|
| 27 |
new_sitemap_content = new_response.content
|
| 28 |
+
|
| 29 |
old_urls = collect_urls_from_sitemap(old_sitemap_content)
|
| 30 |
new_urls = collect_urls_from_sitemap(new_sitemap_content)
|
| 31 |
+
|
| 32 |
added_urls, removed_urls = compare_sitemaps(old_urls, new_urls)
|
| 33 |
+
|
| 34 |
added_urls_text = "\n".join(added_urls)
|
| 35 |
removed_urls_text = "\n".join(removed_urls)
|
| 36 |
+
|
| 37 |
+
return {"Newly Added URLs": added_urls_text, "Removed URLs": removed_urls_text}
|
| 38 |
else:
|
| 39 |
+
return {"Newly Added URLs": "Failed to fetch sitemaps.", "Removed URLs": "Failed to fetch sitemaps."}
|
|
|
|
| 40 |
|
| 41 |
iface = gr.Interface(
|
| 42 |
fn=compare_sitemaps_ui,
|
| 43 |
inputs=["text", "text"],
|
| 44 |
+
outputs=[gr.outputs.Textbox(label="Newly Added URLs"), gr.outputs.Textbox(label="Removed URLs")],
|
| 45 |
title="Sitemap Comparator",
|
| 46 |
description="Enter the URLs of the old and new sitemaps to compare.",
|
| 47 |
examples=[["http://bonrix.net/old-sitemap.xml", "http://bonrix.net/sitemap.xml"]]
|
| 48 |
)
|
| 49 |
|
| 50 |
iface.launch()
|
| 51 |
+
|