"""
Mining sémantique à grande échelle (fictif)
But : détecter des opportunités longue traîne par cluster, intent, et couverture du site.
Entrées : URLs + titles/Hn + Search Console (exports) + logs (pages réellement crawlées).
Sorties : backlog de contenus + recommandations de maillage interne.
"""

from collections import defaultdict
import math

def tokenize(text: str) -> list[str]:
    return [t.lower() for t in text.replace("’","'").split() if len(t) > 2]

def cosine(a: dict, b: dict) -> float:
    # mini TF (fictif)
    inter = set(a) & set(b)
    num = sum(a[t] * b[t] for t in inter)
    den = math.sqrt(sum(v*v for v in a.values())) * math.sqrt(sum(v*v for v in b.values()))
    return (num / den) if den else 0.0

def vectorize(tokens: list[str]) -> dict:
    v = defaultdict(int)
    for t in tokens:
        v[t] += 1
    return dict(v)

def propose_clusters(pages: list[dict], queries: list[str]) -> list[dict]:
    page_vecs = [(p["url"], vectorize(tokenize(p["title"] + " " + p["h1"]))) for p in pages]
    opportunities = []
    for q in queries:
        qv = vectorize(tokenize(q))
        best = sorted(
            [(cosine(qv, pv), url) for url, pv in page_vecs],
            reverse=True
        )[:3]
        opportunities.append({
            "query": q,
            "best_matching_pages": best,
            "intent": "info" if "comment" in q or "guide" in q else "commerciale",
            "suggested_content": f"Créer une page dédiée sur: {q}"
        })
    return opportunities

if __name__ == "__main__":
    pages = [
      {"url":"/services-seo-freelance/", "title":"Services SEO freelance", "h1":"Consultant SEO freelance"},
      {"url":"/audit-seo/", "title":"Audit SEO complet", "h1":"Audit technique + contenu"},
      {"url":"/netlinking/", "title":"Netlinking white-hat", "h1":"Backlinks de qualité"}
    ]
    queries = [
      "comment prioriser un audit SEO technique",
      "analyse logs googlebot pour e-commerce",
      "cluster sémantique relooking intérieur petit budget"
    ]
    for o in propose_clusters(pages, queries):
        print(o)

Prêt. Clique sur “Run (simulé)” pour afficher une sortie d’exemple.

"""
DataViz SEO (fictif) : croiser GSC + conversions
But : attribuer des conversions aux pages / requêtes et sortir un dataset prêt Looker Studio.
Entrées : export Search Console (page, query, clicks, impressions) + export conversions (url, leads, revenue).
Sortie : dataset enrichi avec KPI (CTR, CVR, Rev/Click, etc.).
"""

import csv

def safe_div(a, b):
    return (a / b) if b else 0.0

def enrich(gsc_rows, conv_rows):
    conv_by_url = {r["url"]: r for r in conv_rows}
    out = []
    for r in gsc_rows:
        url = r["page"]
        conv = conv_by_url.get(url, {"leads":0, "revenue":0})
        clicks = int(r["clicks"])
        impressions = int(r["impressions"])
        leads = int(conv.get("leads", 0))
        revenue = float(conv.get("revenue", 0))

        out.append({
          "page": url,
          "query": r["query"],
          "clicks": clicks,
          "impressions": impressions,
          "ctr": round(safe_div(clicks, impressions) * 100, 2),
          "leads": leads,
          "cvr": round(safe_div(leads, clicks) * 100, 2),
          "revenue": revenue,
          "revenue_per_click": round(safe_div(revenue, clicks), 2)
        })
    return out

if __name__ == "__main__":
    gsc = [
      {"page":"/services-seo-freelance/", "query":"consultant seo freelance", "clicks":"120", "impressions":"3400"},
      {"page":"/audit-seo/", "query":"audit seo complet", "clicks":"80", "impressions":"2100"}
    ]
    conv = [
      {"url":"/services-seo-freelance/", "leads":"9", "revenue":"2700"},
      {"url":"/audit-seo/", "leads":"4", "revenue":"1200"}
    ]

    dataset = enrich(gsc, conv)
    print("Dataset prêt Looker Studio (extrait):")
    for row in dataset:
        print(row)

Prêt. Clique sur “Run (simulé)” pour afficher une sortie d’exemple.