Spaces:

Vibeaxis
/

Fake-Fast-Lite

Runtime error

Update app.py

9df2916 verified about 1 year ago

9.16 kB

import time, re, socket
from contextlib import closing
import httpx
import gradio as gr
from bs4 import BeautifulSoup
import tldextract

TIMEOUT = 8.0
UA = "VAX-FakeFastLite/1.0 (+https://vibeaxis.com)"

EDGE_HINT_HEADERS = [
    "cf-cache-status","x-cache","x-cache-hits","x-ser","server-timing",
    "x-worker-cache","x-akamai-transformed","x-cdn","x-amz-cf-pop","via"
]

CACHE_HEADERS = [
    "cache-control","age","etag","last-modified","vary","expires","pragma","surrogate-control"
]

def first_byte_timing(url: str) -> float | None:
    """
    Approximate TTFB by timing a streamed GET until first chunk.
    This is not a lab-grade metric but it’s consistent enough for “fake-fast” sniffing.
    """
    try:
        with httpx.Client(http2=False, timeout=TIMEOUT, headers={"User-Agent": UA}, follow_redirects=True) as c:
            start = time.perf_counter()
            with c.stream("GET", url) as r:
                for chunk in r.iter_bytes():
                    if chunk:
                        return (time.perf_counter() - start) * 1000.0
    except Exception:
        return None

def fetch(url: str):
    if not re.match(r'^https?://', url, flags=re.I):
        url = "https://" + url

    errors = []
    try:
        with httpx.Client(http2=False, timeout=TIMEOUT, headers={"User-Agent": UA}, follow_redirects=True) as c:
            head = c.head(url)
            # fallback if HEAD blocked
            if head.status_code >= 400 or int(head.headers.get("content-length", "0")) == 0:
                head = c.get(url)
            html = c.get(url).text
    except Exception as e:
        return {"error": f"Request failed: {e}"}

    # Parse DOM counts
    scripts = re.findall(r"<script\b", html, flags=re.I)
    inlines = re.findall(r"<script\b[^>]*>(?!\s*</script>)", html, flags=re.I)
    css_links = re.findall(r"<link\b[^>]*rel=[\"']stylesheet[\"']", html, flags=re.I)

    soup = BeautifulSoup(html, "html.parser")
    title = (soup.title.string.strip() if soup.title and soup.title.string else "")

    # Edge vs origin hints
    edge = {}
    for h in EDGE_HINT_HEADERS:
        if h in head.headers:
            edge[h] = head.headers.get(h)

    cache = {}
    for h in CACHE_HEADERS:
        if h in head.headers:
            cache[h] = head.headers.get(h)

    # CDN-ish tell via "via" or known server banners
    via = head.headers.get("via", "") + " " + head.headers.get("server", "")
    cdnish = any(k in via.lower() for k in ["cloudflare","akamai","fastly","cache","cdn","cloudfront","varnish"])

    # TTFB approx
    ttfb_ms = first_byte_timing(url)

    # Heuristics
    script_count = len(scripts)
    inline_count = len(inlines)
    css_count = len(css_links)

    cache_control = head.headers.get("cache-control","").lower()
    max_age = None
    if "max-age=" in cache_control:
        try:
            max_age = int(re.search(r"max-age=(\d+)", cache_control).group(1))
        except Exception:
            max_age = None

    age_hdr = head.headers.get("age")
    try:
        age_val = int(age_hdr) if age_hdr is not None else None
    except:
        age_val = None

    cf_status = head.headers.get("cf-cache-status","").lower()

    # Verdict: “fake-fast-ish” if CDN says HIT but TTFB still chunky and DOM heavy
    flags = []
    if ttfb_ms is not None and ttfb_ms > 800:
        flags.append(f"High TTFB ({ttfb_ms:.0f} ms)")
    if cdnish and ("hit" in cf_status or (age_val and age_val > 0)):
        flags.append("Edge cache present")
    if script_count > 20:
        flags.append(f"Script buffet ({script_count})")
    if inline_count > 3:
        flags.append(f"Inline scripts ({inline_count})")
    if css_count > 6:
        flags.append(f"Many stylesheets ({css_count})")

    # “Fake-fast smell” score (0–100)
    score = 0
    if ttfb_ms is None:
        score += 15
    else:
        if ttfb_ms > 1500: score += 45
        elif ttfb_ms > 800: score += 30
        elif ttfb_ms > 400: score += 15
    if cdnish: score += 10
    if "hit" in cf_status: score += 10
    if script_count > 20: score += 15
    if inline_count > 3: score += 10
    if css_count > 6: score += 10
    if max_age is not None and max_age < 60: score += 10

    verdict = (
        "Likely Fake-Fast (edge cache masking slow origin)"
        if score >= 60 else
        "Borderline (mixed signals)"
        if score >= 35 else
        "Looks Clean (at a glance)"
    )

    return {
        "url": url,
        "title": title,
        "ttfb_ms": None if ttfb_ms is None else round(ttfb_ms, 1),
        "headers_edge": edge,
        "headers_cache": cache,
        "counts": {
            "scripts_total": script_count,
            "scripts_inline": inline_count,
            "stylesheets": css_count
        },
        "flags": flags,
        "score_0_100": score,
        "verdict": verdict,
    }

def ui_analyze(url):
    data = fetch(url)
    if "error" in data:
        return data["error"], None, None, None, None

    summary = f"**{data['verdict']}** — score {data['score_0_100']}/100"
    meta = f"**TTFB:** {data['ttfb_ms']} ms • **Scripts:** {data['counts']['scripts_total']} (inline {data['counts']['scripts_inline']}) • **CSS:** {data['counts']['stylesheets']}"
    edge = "\n".join([f"- **{k}:** {v}" for k,v in (data['headers_edge'] or {}).items()])
    cache = "\n".join([f"- **{k}:** {v}" for k,v in (data['headers_cache'] or {}).items()])
    flags = ", ".join(data["flags"]) if data["flags"] else "—"

    md_summary = f"{summary}\n\n{meta}\n\n**Signals:** {flags}"
    md_edge = edge or "—"
    md_cache = cache or "—"

    return md_summary, md_edge, md_cache, data["url"], data.get("title","")

with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="blue")) as demo:
    gr.Markdown("# Fake-Fast Lite\n**Sniffs edge-cache lipstick on a slow origin.** Not a lab test, just fast receipts.")
    with gr.Row():
        url = gr.Textbox(label="URL", placeholder="https://example.com")
        go = gr.Button("Analyze", variant="primary")
    out_summary = gr.Markdown()
    with gr.Row():
        out_edge = gr.Markdown(label="Edge / CDN Hints")
        out_cache = gr.Markdown(label="Cache Headers")
    with gr.Row():
        out_url = gr.Textbox(label="Final URL", interactive=False)
        out_title = gr.Textbox(label="Page Title", interactive=False)

    go.click(fn=ui_analyze, inputs=url, outputs=[out_summary, out_edge, out_cache, out_url, out_title])

if __name__ == "__main__":
    demo.launch()
# Add near imports
import urllib.parse as urlparse
import gradio as gr

# Modify your ui_analyze signature to accept the request object
def ui_analyze(url, request: gr.Request):
    # Prefill from ?url= if textbox empty
    if (not url) and request and hasattr(request, "query_params"):
        url = request.query_params.get("url", "") or ""
    data = fetch(url)
    if "error" in data:
        return data["error"], None, None, url, "", "", ""
    summary = f"**{data['verdict']}** — score {data['score_0_100']}/100"
    meta = f"**TTFB:** {data['ttfb_ms']} ms • **Scripts:** {data['counts']['scripts_total']} (inline {data['counts']['scripts_inline']}) • **CSS:** {data['counts']['stylesheets']}"
    edge = "\n".join([f"- **{k}:** {v}" for k,v in (data['headers_edge'] or {}).items()]) or "—"
    cache = "\n".join([f"- **{k}:** {v}" for k,v in (data['headers_cache'] or {}).items()]) or "—"
    flags = ", ".join(data["flags"]) if data["flags"] else "—"
    md_summary = f"{summary}\n\n{meta}\n\n**Signals:** {flags}"

    # Build shareable link to this Space with ?url=
    base = "https://huggingface.co/spaces/vibeaxis/fake-fast-lite"
    share = f"{base}?url={urlparse.quote(data['url'] or '')}"

    # return: out_summary, out_edge, out_cache, out_url, out_title, hidden_copy_src, share_link
    return md_summary, edge, cache, data["url"], data.get("title",""), md_summary, share

with gr.Blocks(theme=gr.themes.Soft(primary_hue="teal", secondary_hue="sky")) as demo:
    gr.Markdown("# Fake-Fast Lite\n**Sniffs edge-cache lipstick on a slow origin.** Not a lab test, just fast receipts.")

    with gr.Row():
        url = gr.Textbox(label="URL", placeholder="https://example.com")
        go = gr.Button("Analyze", variant="primary")

    out_summary = gr.Markdown()
    with gr.Row():
        out_edge = gr.Markdown(label="Edge / CDN Hints")
        out_cache = gr.Markdown(label="Cache Headers")
    with gr.Row():
        out_url = gr.Textbox(label="Final URL", interactive=False)
        out_title = gr.Textbox(label="Page Title", interactive=False)

    # Hidden textbox mirrors summary so we can copy via JS
    copy_src = gr.Textbox(visible=False)
    share_link = gr.Textbox(label="Share this result", interactive=False)

    # Copy button using front-end JS
    copy_btn = gr.Button("Copy summary")
    copy_btn.click(
        fn=None,
        inputs=copy_src,
        outputs=None,
        _js="(txt) => { navigator.clipboard.writeText(txt); return; }"
    )

    go.click(
        fn=ui_analyze,
        inputs=url,
        outputs=[out_summary, out_edge, out_cache, out_url, out_title, copy_src, share_link]
    )

	import time, re, socket
	from contextlib import closing
	import httpx
	import gradio as gr
	from bs4 import BeautifulSoup
	import tldextract

	TIMEOUT = 8.0
	UA = "VAX-FakeFastLite/1.0 (+https://vibeaxis.com)"

	EDGE_HINT_HEADERS = [
	"cf-cache-status","x-cache","x-cache-hits","x-ser","server-timing",
	"x-worker-cache","x-akamai-transformed","x-cdn","x-amz-cf-pop","via"
	]

	CACHE_HEADERS = [
	"cache-control","age","etag","last-modified","vary","expires","pragma","surrogate-control"
	]

	def first_byte_timing(url: str) -> float \| None:
	"""
	Approximate TTFB by timing a streamed GET until first chunk.
	This is not a lab-grade metric but it’s consistent enough for “fake-fast” sniffing.
	"""
	try:
	with httpx.Client(http2=False, timeout=TIMEOUT, headers={"User-Agent": UA}, follow_redirects=True) as c:
	start = time.perf_counter()
	with c.stream("GET", url) as r:
	for chunk in r.iter_bytes():
	if chunk:
	return (time.perf_counter() - start) * 1000.0
	except Exception:
	return None

	def fetch(url: str):
	if not re.match(r'^https?://', url, flags=re.I):
	url = "https://" + url

	errors = []
	try:
	with httpx.Client(http2=False, timeout=TIMEOUT, headers={"User-Agent": UA}, follow_redirects=True) as c:
	head = c.head(url)
	# fallback if HEAD blocked
	if head.status_code >= 400 or int(head.headers.get("content-length", "0")) == 0:
	head = c.get(url)
	html = c.get(url).text
	except Exception as e:
	return {"error": f"Request failed: {e}"}

	# Parse DOM counts
	scripts = re.findall(r"<script\b", html, flags=re.I)
	inlines = re.findall(r"<script\b[^>]>(?!\s</script>)", html, flags=re.I)
	css_links = re.findall(r"<link\b[^>]*rel=[\"']stylesheet[\"']", html, flags=re.I)

	soup = BeautifulSoup(html, "html.parser")
	title = (soup.title.string.strip() if soup.title and soup.title.string else "")

	# Edge vs origin hints
	edge = {}
	for h in EDGE_HINT_HEADERS:
	if h in head.headers:
	edge[h] = head.headers.get(h)

	cache = {}
	for h in CACHE_HEADERS:
	if h in head.headers:
	cache[h] = head.headers.get(h)

	# CDN-ish tell via "via" or known server banners
	via = head.headers.get("via", "") + " " + head.headers.get("server", "")
	cdnish = any(k in via.lower() for k in ["cloudflare","akamai","fastly","cache","cdn","cloudfront","varnish"])

	# TTFB approx
	ttfb_ms = first_byte_timing(url)

	# Heuristics
	script_count = len(scripts)
	inline_count = len(inlines)
	css_count = len(css_links)

	cache_control = head.headers.get("cache-control","").lower()
	max_age = None
	if "max-age=" in cache_control:
	try:
	max_age = int(re.search(r"max-age=(\d+)", cache_control).group(1))
	except Exception:
	max_age = None

	age_hdr = head.headers.get("age")
	try:
	age_val = int(age_hdr) if age_hdr is not None else None
	except:
	age_val = None

	cf_status = head.headers.get("cf-cache-status","").lower()

	# Verdict: “fake-fast-ish” if CDN says HIT but TTFB still chunky and DOM heavy
	flags = []
	if ttfb_ms is not None and ttfb_ms > 800:
	flags.append(f"High TTFB ({ttfb_ms:.0f} ms)")
	if cdnish and ("hit" in cf_status or (age_val and age_val > 0)):
	flags.append("Edge cache present")
	if script_count > 20:
	flags.append(f"Script buffet ({script_count})")
	if inline_count > 3:
	flags.append(f"Inline scripts ({inline_count})")
	if css_count > 6:
	flags.append(f"Many stylesheets ({css_count})")

	# “Fake-fast smell” score (0–100)
	score = 0
	if ttfb_ms is None:
	score += 15
	else:
	if ttfb_ms > 1500: score += 45
	elif ttfb_ms > 800: score += 30
	elif ttfb_ms > 400: score += 15
	if cdnish: score += 10
	if "hit" in cf_status: score += 10
	if script_count > 20: score += 15
	if inline_count > 3: score += 10
	if css_count > 6: score += 10
	if max_age is not None and max_age < 60: score += 10

	verdict = (
	"Likely Fake-Fast (edge cache masking slow origin)"
	if score >= 60 else
	"Borderline (mixed signals)"
	if score >= 35 else
	"Looks Clean (at a glance)"
	)

	return {
	"url": url,
	"title": title,
	"ttfb_ms": None if ttfb_ms is None else round(ttfb_ms, 1),
	"headers_edge": edge,
	"headers_cache": cache,
	"counts": {
	"scripts_total": script_count,
	"scripts_inline": inline_count,
	"stylesheets": css_count
	},
	"flags": flags,
	"score_0_100": score,
	"verdict": verdict,
	}

	def ui_analyze(url):
	data = fetch(url)
	if "error" in data:
	return data["error"], None, None, None, None

	summary = f"{data['verdict']} — score {data['score_0_100']}/100"
	meta = f"TTFB: {data['ttfb_ms']} ms • Scripts: {data['counts']['scripts_total']} (inline {data['counts']['scripts_inline']}) • CSS: {data['counts']['stylesheets']}"
	edge = "\n".join([f"- {k}: {v}" for k,v in (data['headers_edge'] or {}).items()])
	cache = "\n".join([f"- {k}: {v}" for k,v in (data['headers_cache'] or {}).items()])
	flags = ", ".join(data["flags"]) if data["flags"] else "—"

	md_summary = f"{summary}\n\n{meta}\n\nSignals: {flags}"
	md_edge = edge or "—"
	md_cache = cache or "—"

	return md_summary, md_edge, md_cache, data["url"], data.get("title","")

	with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="blue")) as demo:
	gr.Markdown("# Fake-Fast Lite\nSniffs edge-cache lipstick on a slow origin. Not a lab test, just fast receipts.")
	with gr.Row():
	url = gr.Textbox(label="URL", placeholder="https://example.com")
	go = gr.Button("Analyze", variant="primary")
	out_summary = gr.Markdown()
	with gr.Row():
	out_edge = gr.Markdown(label="Edge / CDN Hints")
	out_cache = gr.Markdown(label="Cache Headers")
	with gr.Row():
	out_url = gr.Textbox(label="Final URL", interactive=False)
	out_title = gr.Textbox(label="Page Title", interactive=False)

	go.click(fn=ui_analyze, inputs=url, outputs=[out_summary, out_edge, out_cache, out_url, out_title])

	if __name__ == "__main__":
	demo.launch()
	# Add near imports
	import urllib.parse as urlparse
	import gradio as gr

	# Modify your ui_analyze signature to accept the request object
	def ui_analyze(url, request: gr.Request):
	# Prefill from ?url= if textbox empty
	if (not url) and request and hasattr(request, "query_params"):
	url = request.query_params.get("url", "") or ""
	data = fetch(url)
	if "error" in data:
	return data["error"], None, None, url, "", "", ""
	summary = f"{data['verdict']} — score {data['score_0_100']}/100"
	meta = f"TTFB: {data['ttfb_ms']} ms • Scripts: {data['counts']['scripts_total']} (inline {data['counts']['scripts_inline']}) • CSS: {data['counts']['stylesheets']}"
	edge = "\n".join([f"- {k}: {v}" for k,v in (data['headers_edge'] or {}).items()]) or "—"
	cache = "\n".join([f"- {k}: {v}" for k,v in (data['headers_cache'] or {}).items()]) or "—"
	flags = ", ".join(data["flags"]) if data["flags"] else "—"
	md_summary = f"{summary}\n\n{meta}\n\nSignals: {flags}"

	# Build shareable link to this Space with ?url=
	base = "https://huggingface.co/spaces/vibeaxis/fake-fast-lite"
	share = f"{base}?url={urlparse.quote(data['url'] or '')}"

	# return: out_summary, out_edge, out_cache, out_url, out_title, hidden_copy_src, share_link
	return md_summary, edge, cache, data["url"], data.get("title",""), md_summary, share

	with gr.Blocks(theme=gr.themes.Soft(primary_hue="teal", secondary_hue="sky")) as demo:
	gr.Markdown("# Fake-Fast Lite\nSniffs edge-cache lipstick on a slow origin. Not a lab test, just fast receipts.")

	with gr.Row():
	url = gr.Textbox(label="URL", placeholder="https://example.com")
	go = gr.Button("Analyze", variant="primary")

	out_summary = gr.Markdown()
	with gr.Row():
	out_edge = gr.Markdown(label="Edge / CDN Hints")
	out_cache = gr.Markdown(label="Cache Headers")
	with gr.Row():
	out_url = gr.Textbox(label="Final URL", interactive=False)
	out_title = gr.Textbox(label="Page Title", interactive=False)

	# Hidden textbox mirrors summary so we can copy via JS
	copy_src = gr.Textbox(visible=False)
	share_link = gr.Textbox(label="Share this result", interactive=False)

	# Copy button using front-end JS
	copy_btn = gr.Button("Copy summary")
	copy_btn.click(
	fn=None,
	inputs=copy_src,
	outputs=None,
	_js="(txt) => { navigator.clipboard.writeText(txt); return; }"
	)

	go.click(
	fn=ui_analyze,
	inputs=url,
	outputs=[out_summary, out_edge, out_cache, out_url, out_title, copy_src, share_link]
	)