Merge branch 'main' of https://github.com/SculptTechProject/GinioCrawler

Author: SculptTechProject

scripts/to_xlsx.py

@@ -1,5 +1,7 @@
-import sys, pandas as pd
-from openpyxl.styles import Font, Alignment
+import sys
+
+import pandas as pd
+from openpyxl.styles import Alignment, Font
 from openpyxl.utils import get_column_letter
 
 

tests/test_crawl_run.py

@@ -1,32 +1,61 @@
+<<<<<<< HEAD
 import httpx, pytest, respx
 from src.main import crawl_one, run
+=======
+import httpx
+import pytest
+import respx
+
+from main import crawl_one, run
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
+
 
 @pytest.mark.asyncio
 @respx.mock
 async def test_crawl_one_merges_contact(monkeypatch):
     monkeypatch.setattr("ginio.in_robots", lambda url: True)
-    respx.get("https://site.test/").mock(return_value=httpx.Response(200, text="""
+    respx.get("https://site.test/").mock(
+        return_value=httpx.Response(
+            200,
+            text="""
         <a href="/contact">Contact</a>
         <p>[email protected]</p>
-    """))
-    respx.get("https://site.test/contact").mock(return_value=httpx.Response(200, text="""
+    """,
+        )
+    )
+    respx.get("https://site.test/contact").mock(
+        return_value=httpx.Response(
+            200,
+            text="""
         <p>[email protected] 123 456 789</p>
-    """))
+    """,
+        )
+    )
     async with httpx.AsyncClient() as client:
         out = await crawl_one("https://site.test/", client)
     assert "[email protected]" in out["emails"] and "[email protected]" in out["emails"]
     assert any("123" in p for p in out["phones"])
 
+
 @pytest.mark.asyncio
 @respx.mock
 async def test_run_full(monkeypatch):
-    respx.get("https://serpapi.com/search").mock(return_value=httpx.Response(
-        200, json={"organic_results":[{"link":"https://a.pl"}, {"link":"https://b.pl"}]}
-    ))
+    respx.get("https://serpapi.com/search").mock(
+        return_value=httpx.Response(
+            200,
+            json={
+                "organic_results": [{"link": "https://a.pl"}, {"link": "https://b.pl"}]
+            },
+        )
+    )
     # stub robots + strony
     monkeypatch.setattr("ginio.in_robots", lambda url: True)
-    respx.get("https://a.pl").mock(return_value=httpx.Response(200, text="<title>A</title>"))
-    respx.get("https://b.pl").mock(return_value=httpx.Response(200, text="<title>B</title>"))
-    monkeypatch.setenv("SERPAPI_KEY","x")
+    respx.get("https://a.pl").mock(
+        return_value=httpx.Response(200, text="<title>A</title>")
+    )
+    respx.get("https://b.pl").mock(
+        return_value=httpx.Response(200, text="<title>B</title>")
+    )
+    monkeypatch.setenv("SERPAPI_KEY", "x")
     results = await run("foo")
-    assert {r["title"] for r in results} == {"A","B"}
+    assert {r["title"] for r in results} == {"A", "B"}

tests/test_excel.py

@@ -1,17 +1,42 @@
 from pathlib import Path
+
 import pandas as pd
+<<<<<<< HEAD
 from src.main import save_results
+=======
+
+from main import save_results
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
+
 
 def test_save_results_creates_files(tmp_path, monkeypatch):
+<<<<<<< HEAD
     root_out = tmp_path/"wyniki"
     data = [{"url":"https://x","title":"X","emails":["a@x"],"phones":["123"],"contact_url":None}]
     from src.main import write_excel as real_write_excel
+=======
+    root_out = tmp_path / "wyniki"
+    data = [
+        {
+            "url": "https://x",
+            "title": "X",
+            "emails": ["a@x"],
+            "phones": ["123"],
+            "contact_url": None,
+        }
+    ]
+    from main import write_excel as real_write_excel
+
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
     xlsx_called = {}
+
     def fake_write(csv, xlsx):
-        xlsx_called["csv"] = csv; xlsx_called["xlsx"] = xlsx
+        xlsx_called["csv"] = csv
+        xlsx_called["xlsx"] = xlsx
         return real_write_excel(csv, xlsx)
+
     monkeypatch.setattr("ginio.write_excel", fake_write)
     csv_p, xlsx_p = save_results(data, "20250101_000000", root_out)
     assert Path(csv_p).exists() and Path(xlsx_p).exists()
     df = pd.read_excel(xlsx_p)
-    assert set(df.columns) >= {"url","title","emails","phones","contact_url"}
+    assert set(df.columns) >= {"url", "title", "emails", "phones", "contact_url"}

tests/test_gui.py

@@ -1,27 +1,58 @@
 from unittest.mock import patch
 
+
 def test_ensure_api_key_env(monkeypatch):
+<<<<<<< HEAD
     monkeypatch.setenv("SERPAPI_KEY","secret")
     from src.main import ensure_api_key
     assert ensure_api_key()=="secret"
 
 def test_ensure_api_key_prompt(tmp_path, monkeypatch):
     monkeypatch.setenv("APPDATA", str(tmp_path))
     from src.main import ensure_api_key
+=======
+    monkeypatch.setenv("SERPAPI_KEY", "secret")
+    from main import ensure_api_key
+
+    assert ensure_api_key() == "secret"
+
+
+def test_ensure_api_key_prompt(tmp_path, monkeypatch):
+    monkeypatch.setenv("APPDATA", str(tmp_path))
+    from main import ensure_api_key
+
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
     with patch("ginio.simpledialog.askstring", return_value="abc"):
         key = ensure_api_key()
-    assert key=="abc"
+    assert key == "abc"
+
 
 def test_start_calls_run_without_threading(monkeypatch):
+<<<<<<< HEAD
     import src.app_gui as g
+=======
+    import app_gui as g
+
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
     g.build_ui()
-    g.entry_query.delete(0,'end'); g.entry_query.insert(0,"kawa")
-    monkeypatch.setenv("SERPAPI_KEY","x")
+    g.entry_query.delete(0, "end")
+    g.entry_query.insert(0, "kawa")
+    monkeypatch.setenv("SERPAPI_KEY", "x")
+
     class DummyThread:
-        def __init__(self, target, daemon): self.target=target
-        def start(self): self.target()
+        def __init__(self, target, daemon):
+            self.target = target
+
+        def start(self):
+            self.target()
+
     monkeypatch.setattr("ginio_gui.threading.Thread", DummyThread)
-    monkeypatch.setattr("ginio_gui.run", lambda q: [{"url":"u","title":"t","emails":[],"phones":[],"contact_url":None}])
+    monkeypatch.setattr(
+        "ginio_gui.run",
+        lambda q: [
+            {"url": "u", "title": "t", "emails": [], "phones": [], "contact_url": None}
+        ],
+    )
     with patch.object(g, "messagebox"):
         g.start()
         assert g.btn_start["state"] == "normal"

tests/test_net.py

@@ -1,15 +1,27 @@
+<<<<<<< HEAD
 import httpx, pytest, respx
 from src.main import fetch
+=======
+import httpx
+import pytest
+import respx
+
+from main import fetch
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
+
 
 @pytest.mark.asyncio
 @respx.mock
 async def test_fetch_ok(monkeypatch):
     monkeypatch.setattr("ginio.in_robots", lambda url: True)
-    respx.get("https://x.test/ok").mock(return_value=httpx.Response(200, text="<h1>ok</h1>"))
+    respx.get("https://x.test/ok").mock(
+        return_value=httpx.Response(200, text="<h1>ok</h1>")
+    )
     async with httpx.AsyncClient() as client:
         html = await fetch("https://x.test/ok", client)
     assert "ok" in html
 
+
 @pytest.mark.asyncio
 @respx.mock
 async def test_fetch_respects_robots(monkeypatch):

tests/test_parse.py

@@ -1,8 +1,14 @@
+<<<<<<< HEAD
 from src.main import parse_info, absolutize
+=======
+from main import absolutize, parse_info
+
+>>>>>>> e772c4167010570baed9b34907b8a8834fe77e81
 
 def test_absolutize():
     assert absolutize("https://ex.com/dir/", "../a") == "https://ex.com/a"
 
+
 def test_parse_info_extracts_contact():
     html = """
     <html><head><title> ACME </title></head>