This commit is contained in:
2025-09-21 21:35:39 +02:00
parent 422aae12ee
commit 4a7e1039a9
5 changed files with 553 additions and 6 deletions

98
Medevio1_newrid.py Normal file
View File

@@ -0,0 +1,98 @@
# print_patients_first_page_ids.py
from pathlib import Path
import json, time, sys
from playwright.sync_api import sync_playwright, TimeoutError as PWTimeout
STATE_FILE = r"medevio_storage.json"
PATIENTS_URL = "https://my.medevio.cz/mudr-buzalkova/klinika/pacienti"
def harvest_ids_on_page(page):
ids = set()
for sel in ["div[role='row'][data-id]", "div.MuiDataGrid-row[data-id]"]:
for row in page.locator(sel).all():
pid = row.get_attribute("data-id")
if pid:
ids.add(pid)
return ids
def set_page_size(page, value="100"):
# Open the page-size combobox
for loc in [
page.get_by_role("combobox", name="Řádků na stránce:"),
page.get_by_role("combobox", name="Rows per page:"),
page.locator("div.MuiTablePagination-root [role='combobox']"),
]:
if loc.count():
loc.first.click()
break
# Select option "100" (portal-safe)
opt = page.get_by_role("option", name=value)
if not opt.count():
opt = page.locator(f"//li[normalize-space(.)='{value}']")
opt.first.wait_for(state="visible", timeout=5000)
opt.first.click()
# Wait a moment for refresh
try:
page.wait_for_selector("div[role='row'][data-id]", timeout=10000)
except PWTimeout:
time.sleep(0.8)
def main():
sf = Path(STATE_FILE)
if not sf.exists():
print(f"ERROR: storage not found: {sf}")
sys.exit(1)
with sync_playwright() as p:
browser = p.chromium.launch(headless=False) # set False to watch
context = browser.new_context(storage_state=str(sf))
context.set_default_navigation_timeout(30000)
context.set_default_timeout(15000)
page = context.new_page()
try:
page.goto(PATIENTS_URL, wait_until="domcontentloaded")
except PWTimeout:
print("Warning: goto timeout; continuing…")
# Detect redirect to login
if "/prihlaseni" in page.url.lower():
print("You were redirected to the login page → saved session is expired. Re-run the login-save step.")
browser.close()
return
# (Optional) print pagination label before/after
try:
print("Before:", page.locator("p.MuiTablePagination-displayedRows").first.inner_text())
except Exception:
pass
try:
set_page_size(page, "100")
except Exception as e:
print(f"Could not set page size to 100: {e!r}")
try:
print("After :", page.locator("p.MuiTablePagination-displayedRows").first.inner_text())
except Exception:
pass
page.wait_for_selector("div[role='row'][data-id]", timeout=15000)
ids = sorted(harvest_ids_on_page(page))
print(f"\nCollected {len(ids)} IDs on first page:")
for pid in ids:
print(pid)
# Also save if you want
out_json = Path("patient_ids_first_page.json")
out_csv = Path("patient_ids_first_page.csv")
out_json.write_text(json.dumps(ids, ensure_ascii=False, indent=2), encoding="utf-8")
out_csv.write_text("patient_id\n" + "\n".join(ids), encoding="utf-8")
print(f"\nSaved → {out_json.resolve()}")
print(f"Saved → {out_csv.resolve()}")
browser.close()
if __name__ == "__main__":
main()

View File

@@ -10,9 +10,9 @@ from pathlib import Path
import json, sys, time, re
from playwright.sync_api import sync_playwright, TimeoutError as PWTimeout
STATE_FILE = r"/medevio_storage.json"
STATE_FILE = r"medevio_storage.json"
BASE_URL = "https://my.medevio.cz/mudr-buzalkova/klinika/pacienti"
PATIENT_ID = "236b3759-4c2b-4fa8-ab52-ce4ddb2e9064" # <-- put target ID here
PATIENT_ID = "fcb2414b-067b-4ca2-91b2-6c36a86d4cbb" # <-- put target ID here
# ---------- helpers ----------
@@ -211,7 +211,7 @@ def main():
sys.exit(1)
with sync_playwright() as p:
browser = p.chromium.launch(headless=True) # set False to watch
browser = p.chromium.launch(headless=False) # set False to watch
context = browser.new_context(storage_state=str(sf))
context.set_default_navigation_timeout(30000)
context.set_default_timeout(15000)

View File

@@ -1,3 +1,6 @@
#Tento kod se pripoji do kartoteky Medevio, zmeni na 100 pacientu na stranu, nactene
# medevio_dump_patients_html_to_mysql.py
import time
import json
@@ -10,13 +13,13 @@ from mysql.connector import errorcode
from playwright.sync_api import sync_playwright, TimeoutError as PWTimeout
# ---------- CONFIG ----------
STATE_FILE = r"/medevio_storage.json"
STATE_FILE = r"medevio_storage.json"
BASE_LIST_URL = "https://my.medevio.cz/mudr-buzalkova/klinika/pacienti"
SAVE_DELAY_SECONDS = 10 # throttle: 10 sec per patient
# MySQL connection settings (fill in)
MYSQL_CFG = dict(
host="192.168.1.74",
host="192.168.1.76",
port=3307,
user="root",
password="Vlado9674+",
@@ -234,7 +237,7 @@ def main():
print(f"Already in DB: {len(already)} ids")
with sync_playwright() as p:
browser = p.chromium.launch(headless=True) # set False to watch
browser = p.chromium.launch(headless=False) # set False to watch
context = browser.new_context(storage_state=STATE_FILE)
# 1) Collect all IDs from the listing (all pages)

258
Medevio4_newrid.py Normal file
View File

@@ -0,0 +1,258 @@
# medevio_dump_patients_html_to_mysql.py
import time
import json
from pathlib import Path
from datetime import datetime
from typing import Set
import mysql.connector
from mysql.connector import errorcode
from playwright.sync_api import sync_playwright, TimeoutError as PWTimeout
# ---------- CONFIG ----------
STATE_FILE = r"medevio_storage.json"
BASE_LIST_URL = "https://my.medevio.cz/mudr-buzalkova/klinika/pacienti"
SAVE_DELAY_SECONDS = 10 # throttle: 10 sec per patient
# MySQL connection settings (fill in)
MYSQL_CFG = dict(
host="192.168.1.74",
port=3307,
user="root",
password="Vlado9674+",
database="medevio",
)
TABLE_NAME = "patients_html" # schema created automatically
# ---------- DB helpers ----------
def db_connect():
try:
conn = mysql.connector.connect(**MYSQL_CFG)
return conn
except mysql.connector.Error as e:
raise SystemExit(f"MySQL connection failed: {e}")
def db_ensure_table(conn):
ddl = f"""
CREATE TABLE IF NOT EXISTS `{TABLE_NAME}` (
patient_id VARCHAR(64) PRIMARY KEY,
html LONGTEXT NOT NULL,
fetched_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
"""
cur = conn.cursor()
cur.execute(ddl)
conn.commit()
cur.close()
def db_existing_ids(conn) -> Set[str]:
ids = set()
cur = conn.cursor()
cur.execute(f"SELECT patient_id FROM `{TABLE_NAME}`")
for (pid,) in cur.fetchall():
ids.add(pid)
cur.close()
return ids
def db_upsert_html(conn, patient_id: str, html: str):
cur = conn.cursor()
cur.execute(
f"""INSERT INTO `{TABLE_NAME}` (patient_id, html, fetched_at)
VALUES (%s, %s, NOW())
ON DUPLICATE KEY UPDATE html = VALUES(html), fetched_at = VALUES(fetched_at)""",
(patient_id, html),
)
conn.commit()
cur.close()
# ---------- Playwright helpers ----------
def wait_for_grid_ready(page):
# grid present & at least one row (be generous on timeout)
page.wait_for_selector("div[role='rowgroup']", timeout=20000)
page.wait_for_selector("div[role='row'][data-id]", timeout=20000)
def set_page_size_100(page):
# Click the page-size combobox (CZ/EN + generic)
for loc in [
page.get_by_role("combobox", name="Řádků na stránce:"),
page.get_by_role("combobox", name="Rows per page:"),
page.locator("div.MuiTablePagination-root [role='combobox']"),
]:
if loc.count():
loc.first.click()
break
# Select 100 (MUI menu often renders in a portal)
opt = page.get_by_role("option", name="100")
if not opt.count():
opt = page.locator("//li[normalize-space(.)='100']")
opt.first.wait_for(state="visible", timeout=5000)
opt.first.click()
# Wait for rows to refresh
try:
page.wait_for_selector("div[role='row'][data-id]", timeout=10000)
except PWTimeout:
time.sleep(0.8)
def harvest_ids_on_current_page(page) -> Set[str]:
ids = set()
for sel in ["div[role='row'][data-id]", "div.MuiDataGrid-row[data-id]"]:
for row in page.locator(sel).all():
pid = row.get_attribute("data-id")
if pid:
ids.add(pid)
return ids
def click_next_page(page) -> bool:
# Prefer ARIA label
nxt = page.get_by_role("button", name="Go to next page")
if nxt.count():
try:
if nxt.first.is_enabled():
nxt.first.click()
return True
except Exception:
pass
# Fallback (CZ)
nxt2 = page.get_by_role("button", name="Další")
if nxt2.count():
try:
if nxt2.first.is_enabled():
nxt2.first.click()
return True
except Exception:
pass
return False
def ensure_detail_open(page) -> bool:
# Detail drawer/dialog visible?
for sel in ["[role='dialog']", "div.MuiDrawer-paper", "div[aria-modal='true']"]:
loc = page.locator(sel)
if loc.count() and loc.first.is_visible():
return True
return False
# ---------- Main workflow ----------
def collect_all_patient_ids(context) -> Set[str]:
page = context.new_page()
page.set_default_timeout(15000)
page.set_default_navigation_timeout(30000)
# Use domcontentloaded (SPAs often keep network busy)
page.goto(BASE_LIST_URL, wait_until="domcontentloaded")
if "/prihlaseni" in page.url.lower():
raise SystemExit("Session expired → refresh medevio_storage.json via the login script.")
wait_for_grid_ready(page)
# optional: print label like "125 z 1856"
try:
label = page.locator("p.MuiTablePagination-displayedRows").first.inner_text()
print("Pagination label BEFORE:", label)
except Exception:
pass
# Set 100/page
try:
set_page_size_100(page)
try:
label = page.locator("p.MuiTablePagination-displayedRows").first.inner_text()
print("Pagination label AFTER :", label)
except Exception:
pass
except Exception as e:
print(f"Warning: could not set page size to 100: {e!r}")
all_ids: Set[str] = set()
page_index = 1
while True:
wait_for_grid_ready(page)
ids_now = harvest_ids_on_current_page(page)
print(f"Page {page_index}: harvested {len(ids_now)} ids")
all_ids |= ids_now
# Try to go next; if cannot, break
if not click_next_page(page):
break
# Wait for DOM to actually update (new rows)
try:
page.wait_for_load_state("domcontentloaded", timeout=10000)
except PWTimeout:
pass
time.sleep(0.5)
page_index += 1
page.close()
print(f"Total unique IDs collected: {len(all_ids)}")
return all_ids
def fetch_and_store_patient_html(context, conn, patient_id: str):
page = context.new_page()
page.set_default_timeout(15000)
page.set_default_navigation_timeout(30000)
url = f"{BASE_LIST_URL}?pacient={patient_id}"
page.goto(url, wait_until="domcontentloaded")
# If detail didnt open, fallback: go to list, click row
if not ensure_detail_open(page):
page.goto(BASE_LIST_URL, wait_until="domcontentloaded")
try:
page.wait_for_selector(f"div[role='row'][data-id='{patient_id}']", timeout=15000)
page.locator(f"div[role='row'][data-id='{patient_id}']").first.click()
# wait for drawer/dialog
page.wait_for_selector("[role='dialog'], div.MuiDrawer-paper, div[aria-modal='true']", timeout=12000)
except PWTimeout:
print(f"[{patient_id}] detail panel did not open — skipping")
page.close()
return
# Save full HTML of the page (includes the open detail drawer)
html = page.content()
db_upsert_html(conn, patient_id, html)
print(f"[{patient_id}] saved HTML ({len(html)} bytes) at {datetime.now().isoformat(timespec='seconds')}")
page.close()
# Throttle per your requirement
time.sleep(SAVE_DELAY_SECONDS)
def main():
# Check storage exists
if not Path(STATE_FILE).exists():
raise SystemExit(f"Storage not found: {STATE_FILE}")
# DB ready
conn = db_connect()
db_ensure_table(conn)
already = db_existing_ids(conn)
print(f"Already in DB: {len(already)} ids")
with sync_playwright() as p:
browser = p.chromium.launch(headless=True) # set False to watch
context = browser.new_context(storage_state=STATE_FILE)
# 1) Collect all IDs from the listing (all pages)
all_ids = collect_all_patient_ids(context)
# 2) Iterate and store HTML (skip existing)
todo = [pid for pid in sorted(all_ids) if pid not in already]
print(f"To fetch now: {len(todo)} ids (skipping {len(all_ids)-len(todo)} already saved)")
for i, pid in enumerate(todo, 1):
try:
fetch_and_store_patient_html(context, conn, pid)
except Exception as e:
print(f"[{pid}] ERROR: {e!r} — continuing with next")
browser.close()
conn.close()
print("Done.")
if __name__ == "__main__":
main()

View File

@@ -0,0 +1,188 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import time
from pathlib import Path
from datetime import datetime
import pymysql
from pymysql.cursors import DictCursor
from playwright.sync_api import sync_playwright, TimeoutError as PWTimeout
# ========= CONFIG =========
MYSQL_CFG = dict(
host="192.168.1.76",
port=3307,
user="root",
password="Vlado9674+",
database="medevio",
cursorclass=DictCursor,
autocommit=False, # we commit in batches
)
# Column in patients_extracted that stores Medevio UUID used in the URL:
UUID_COLUMN = "rid" # <-- change if your column name differs
# Output columns (will be created if missing; MySQL 8.0+ supports IF NOT EXISTS):
REGISTERED_COL = "medevio_registered" # TINYINT(1) NULL/0/1
CHECKED_AT_COL = "medevio_checked_at" # DATETIME NULL
ERROR_COL = "medevio_check_error" # TEXT NULL (optional)
# Medevio routing
PATIENT_URL_TMPL = "https://my.medevio.cz/mudr-buzalkova/klinika/pacienti?pacient={uuid}"
# Login session (created earlier with your script)
STATE_FILE = Path("medevio_storage.json")
# Batch/pace
BATCH_LIMIT = 5 # how many patients per run
SLEEP_SECONDS = 3 # wait between patients (requested)
NAV_TIMEOUT = 20_000 # ms
TEXT_TIMEOUT = 15_000 # ms (for main area/heading)
# Texts indicating NOT registered:
NOT_REGISTERED_STRINGS = [
"Pacientka zatím nemá Medevio účet.",
"Pacient zatím nemá Medevio účet.",
]
# ==========================
SELECT_SQL = f"""
SELECT {UUID_COLUMN} AS uuid, jmeno, prijmeni, rc
FROM patients_extracted
WHERE {UUID_COLUMN} IS NOT NULL
AND {UUID_COLUMN} <> ''
AND {REGISTERED_COL} IS NULL
LIMIT {BATCH_LIMIT};
"""
UPDATE_OK_SQL = f"""
UPDATE patients_extracted
SET {REGISTERED_COL}=%s, {CHECKED_AT_COL}=NOW(), {ERROR_COL}=NULL
WHERE {UUID_COLUMN}=%s
"""
UPDATE_ERR_SQL = f"""
UPDATE patients_extracted
SET {REGISTERED_COL}=NULL, {CHECKED_AT_COL}=NOW(), {ERROR_COL}=%s
WHERE {UUID_COLUMN}=%s
"""
DDL_SQLS = [
f"ALTER TABLE patients_extracted ADD COLUMN {REGISTERED_COL} TINYINT(1) NULL",
f"ALTER TABLE patients_extracted ADD COLUMN {CHECKED_AT_COL} DATETIME NULL",
f"ALTER TABLE patients_extracted ADD COLUMN {ERROR_COL} TEXT NULL",
]
CHECKS_FOR_DDL_SQLS=[
f"SELECT COUNT(*) AS cnt FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = 'patients_extracted' AND COLUMN_NAME = '{REGISTERED_COL}'",
f"SELECT COUNT(*) AS cnt FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = 'patients_extracted' AND COLUMN_NAME = '{CHECKED_AT_COL}'",
f"SELECT COUNT(*) AS cnt FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = 'patients_extracted' AND COLUMN_NAME = '{ERROR_COL}'",
]
def ensure_columns(conn):
with conn.cursor() as cur:
for ddl,ddlcheck in zip(DDL_SQLS,CHECKS_FOR_DDL_SQLS):
cur.execute(ddlcheck)
row = cur.fetchone()
if row["cnt"] == 0:
print("Column missing")
cur.execute(ddl)
print(f"✓ Executed: {ddl}")
else:
print("Column exists")
conn.commit()
def pick_registered_flag(page_text: str) -> int:
text = page_text or ""
# If any NOT-registered phrase is present → 0; otherwise assume registered → 1
for marker in NOT_REGISTERED_STRINGS:
if marker in text:
return 0
return 1
def main():
# --- DB: fetch a batch to process ---
conn = pymysql.connect(**MYSQL_CFG)
try:
ensure_columns(conn)
with conn.cursor() as cur:
cur.execute("SET NAMES utf8mb4 COLLATE utf8mb4_czech_ci")
cur.execute("SET collation_connection = 'utf8mb4_czech_ci'")
cur.execute(SELECT_SQL)
rows = cur.fetchall()
if not rows:
print("No patients to check (all have medevio_registered filled).")
return
print(f"Will process {len(rows)} patients…")
# --- Playwright session ---
with sync_playwright() as p:
browser = p.chromium.launch(headless=False, slow_mo=0)
context = browser.new_context(storage_state=str(STATE_FILE))
page = context.new_page()
page.set_default_timeout(NAV_TIMEOUT)
processed = ok = errs = 0
for r in rows:
processed += 1
# pid = r["id"]
uuid = r["uuid"]
name = f"{r.get('prijmeni','')}, {r.get('jmeno','')}"
rc = r.get("rc","")
url = PATIENT_URL_TMPL.format(uuid=uuid)
print(f"URL pro otevření pacienta je: {url}0")
print(f"[{processed:>3}] {name} | RC {rc} | {uuid}{url}")
try:
page.goto(url, wait_until="domcontentloaded")
# Optionally wait for a stable anchor; fallback to sleep
try:
# A stable bit we saw earlier
page.get_by_text("Historie požadavků").wait_for(timeout=TEXT_TIMEOUT)
except PWTimeout:
pass
# Wait the requested 3 seconds for the UI to settle
time.sleep(SLEEP_SECONDS)
# Get full text and detect
full_text = page.content() # HTML; safer to check visible text too:
vis_text = page.inner_text("body")
registered = pick_registered_flag(full_text) if full_text else pick_registered_flag(vis_text)
with conn.cursor() as cur:
cur.execute(UPDATE_OK_SQL, (registered, pid))
conn.commit()
ok += 1
state = "REGISTERED" if registered == 1 else "NOT REGISTERED"
print(f"{state}")
except Exception as e:
conn.rollback()
errs += 1
msg = f"{type(e).__name__}: {e}"
with conn.cursor() as cur:
cur.execute(UPDATE_ERR_SQL, (msg[:1000], pid))
conn.commit()
print(f" ! ERROR → {msg}")
browser.close()
print(f"Done. processed={processed}, ok={ok}, errors={errs}")
finally:
conn.close()
if __name__ == "__main__":
main()