chore: EventBrite updates

countries/ch.json

@@ -138,8 +138,9 @@
     {
         "name": "Fresque de l'Economie Circulaire",
         "url": "https://www.billetweb.fr/pro/lafresquedeleconomiecirculaire",
-	"language_code": "fr",
+        "language_code": "fr",
         "type": "scraper",
+        "iframe": "event41148",
         "id": 300
     },
     {

src/trouver_une_fresque_scraper/scraper/eventbrite.py

@@ -15,7 +15,7 @@
 from selenium.webdriver.support import expected_conditions as EC
 
 from trouver_une_fresque_scraper.db.records import get_record_dict
-from trouver_une_fresque_scraper.utils.date_and_time import get_dates
+from trouver_une_fresque_scraper.utils.date_and_time import get_dates_from_element
 from trouver_une_fresque_scraper.utils.errors import (
     FreskError,
     FreskDateBadFormat,
@@ -184,12 +184,15 @@ def get_eventbrite_data(sources, service, options):
             ###########################################################
             # Is it an online event?
             ################################################################
-            online = False
-            try:
-                online_el = driver.find_element(By.CSS_SELECTOR, "p.location-info__address-text")
-                online = is_online(online_el.text)
-            except NoSuchElementException:
-                pass
+            online = is_online(title)
+            if not online:
+                try:
+                    short_location_el = driver.find_element(
+                        By.CSS_SELECTOR, "span.start-date-and-location__location"
+                    )
+                    online = is_online(short_location_el.text)
+                except NoSuchElementException:
+                    pass
 
             ################################################################
             # Location data
@@ -205,11 +208,16 @@ def get_eventbrite_data(sources, service, options):
             country_code = ""
 
             if not online:
-                location_el = driver.find_element(By.CSS_SELECTOR, "div.location-info__address")
-                full_location_text = location_el.text.split("\n")
-                location_name = full_location_text[0]
-                address_and_city = full_location_text[1]
-                full_location = f"{location_name}, {address_and_city}"
+                try:
+                    full_location_el = driver.find_element(
+                        By.CSS_SELECTOR, 'div[class^="Location-module__addressWrapper___"'
+                    )
+                except NoSuchElementException:
+                    logging.error(
+                        f"Location element not found for offline event {link}.",
+                    )
+                    continue
+                full_location = full_location_el.text.replace("\n", ", ")
 
                 try:
                     address_dict = get_address(full_location)
@@ -231,7 +239,7 @@ def get_eventbrite_data(sources, service, options):
             # Description
             ################################################################
             try:
-                description_title_el = driver.find_element(By.CSS_SELECTOR, "div.eds-text--left")
+                description_title_el = driver.find_element(By.CSS_SELECTOR, "div.event-description")
                 description = description_title_el.text
             except NoSuchElementException:
                 logging.info("Rejecting record: Description not found.")
@@ -274,14 +282,15 @@ def get_eventbrite_data(sources, service, options):
                         try:
                             date_info_el = driver.find_element(
                                 by=By.CSS_SELECTOR,
-                                value="span.date-info__full-datetime",
+                                value="time.start-date-and-location__date",
                             )
-                            event_time = date_info_el.text
                         except NoSuchElementException:
                             raise FreskDateNotFound
 
                         try:
-                            event_start_datetime, event_end_datetime = get_dates(event_time)
+                            event_start_datetime, event_end_datetime = get_dates_from_element(
+                                date_info_el
+                            )
                         except FreskDateBadFormat as error:
                             logging.info(f"Reject record: {error}")
                             continue
@@ -307,26 +316,42 @@ def get_eventbrite_data(sources, service, options):
 
                         if not already_scanned:
                             event_info.append(
-                                [uuid, event_start_datetime, event_end_datetime, tickets_link]
+                                [
+                                    uuid,
+                                    event_start_datetime,
+                                    event_end_datetime,
+                                    tickets_link,
+                                ]
                             )
 
             # There is only one event on this page.
             except TimeoutException:
+                ################################################################
+                # Single event with multiple dates (a "collection").
+                ################################################################
+                try:
+                    check_availability_btn = driver.find_element(
+                        by=By.CSS_SELECTOR, value="button.check-availability-btn__button"
+                    )
+                    # TODO: add support for this.
+                    logging.error(f"EventBrite collection not supported in event {link}.")
+                    continue
+                except NoSuchElementException:
+                    pass
+
                 ################################################################
                 # Dates
                 ################################################################
                 try:
                     date_info_el = driver.find_element(
                         by=By.CSS_SELECTOR,
-                        value="span.date-info__full-datetime",
+                        value="time.start-date-and-location__date",
                     )
-                    event_time = date_info_el.text
-                except NoSuchElementException as error:
-                    logging.info(f"Reject record: {error}")
-                    continue
+                except NoSuchElementException:
+                    raise FreskDateNotFound
 
                 try:
-                    event_start_datetime, event_end_datetime = get_dates(event_time)
+                    event_start_datetime, event_end_datetime = get_dates_from_element(date_info_el)
                 except FreskDateBadFormat as error:
                     logging.info(f"Reject record: {error}")
                     continue

src/trouver_une_fresque_scraper/utils/date_and_time.py

@@ -1,5 +1,6 @@
 import re
 import traceback
+import logging
 
 from datetime import datetime, timedelta
 from dateutil.parser import parse
@@ -272,7 +273,7 @@ def get_dates(event_time):
                 FRENCH_MONTHS[match.group("month")],
                 int(match.group("day")),
                 int(start_parts[0]),
-                int(start_parts[1]) if len(start_parts) > 1 and len(start_parts[1]) else 0,
+                (int(start_parts[1]) if len(start_parts) > 1 and len(start_parts[1]) else 0),
             )
             end_parts = match.group("end_time").split("h")
             event_end_datetime = datetime(
@@ -290,4 +291,86 @@ def get_dates(event_time):
     except Exception as e:
         if not isinstance(e, FreskError):
             traceback.print_exc()
+        logging.error(f"get_dates: {event_time}")
+        raise FreskDateBadFormat(event_time)
+
+
+def get_dates_from_element(el):
+    """Returns start and end datetime objects extracted from the element.
+
+    The "datetime" attribute of the element is used if present to extract the date, otherwise falls back on get_dates to parse the day and hours from the element text. Returns None, None on failure.
+
+    May throw FreskDateDifferentTimezone, FreskDateBadFormat and any exception thrown by get_dates.
+    """
+    event_day = el.get_attribute("datetime")
+    event_time = el.text
+
+    try:
+        # Leverage the datetime attribute if present.
+        # datetime: 2025-12-05
+        # text: déc. 5 de 9am à 12pm UTC+1
+        if event_day:
+            day_match = re.match(r"(?P<year>\d{4})-(?P<month>\d{1,2})-(?P<day>\d{1,2})", event_day)
+
+            def PATTERN_TIME(hour_name, minute_name, pm_name):
+                return (
+                    r"(?P<"
+                    + hour_name
+                    + r">\d{1,2})(?P<"
+                    + minute_name
+                    + r">:\d{2})?(?P<"
+                    + pm_name
+                    + r">(am|pm|vorm.|nachm.))"
+                )
+
+            def ParseTime(match_object, hour_name, minute_name, pm_name):
+                hour = int(match_object.group(hour_name))
+                PATTERN_PM = ["pm", "nachm."]
+                if match_object.group(pm_name) in PATTERN_PM and hour < 12:
+                    hour += 12
+
+                minute = 0
+                match_minute = hour_match.group(minute_name)
+                if match_minute:
+                    minute = int(match_minute[1:])
+
+                return hour, minute
+
+            # TODO: add proper support for timezone.
+            # We use re.search to skip the text for the date at the beginning of the string.
+            hour_match = re.search(
+                r"(de|von)\s"
+                + PATTERN_TIME("start_hour", "start_minute", "start_am_or_pm")
+                + r"\s"
+                + r"(à|bis)\s"
+                + PATTERN_TIME("end_hour", "end_minute", "end_am_or_pm")
+                + r"\s"
+                + r"((UTC|MEZ)(?P<timezone>.*))",
+                event_time,
+            )
+            if day_match and hour_match:
+                timezone = hour_match.group("timezone")
+                if timezone and timezone not in ("+1", "+2"):
+                    raise FreskDateDifferentTimezone(event_time)
+                dt = datetime(
+                    int(day_match.group("year")),
+                    int(day_match.group("month")),
+                    int(day_match.group("day")),
+                )
+                start_hour, start_minute = ParseTime(
+                    hour_match, "start_hour", "start_minute", "start_am_or_pm"
+                )
+                end_hour, end_minute = ParseTime(
+                    hour_match, "end_hour", "end_minute", "end_am_or_pm"
+                )
+                return datetime(dt.year, dt.month, dt.day, start_hour, start_minute), datetime(
+                    dt.year, dt.month, dt.day, end_hour, end_minute
+                )
+
+        return get_dates(event_time)
+
+    except Exception as e:
+        if not isinstance(e, FreskError):
+            traceback.print_exc()
+        logging.error(f"get_dates_from_element: {event_time}")
         raise FreskDateBadFormat(event_time)

src/trouver_une_fresque_scraper/utils/date_and_time_test.py

@@ -1,11 +1,12 @@
 from datetime import datetime
 import logging
+from attrs import define
 
 
 from trouver_une_fresque_scraper.utils import date_and_time
 
 
-def run_tests():
+def run_get_dates_tests():
     # tuple fields:
     # 1. Test case name or ID
     # 2. Input date string
@@ -80,3 +81,85 @@ def run_tests():
             logging.error(f"{test_case[0]}: expected {test_case[2]} but got {actual_start_time}")
         if actual_end_time != test_case[3]:
             logging.error(f"{test_case[0]}: expected {test_case[3]} but got {actual_end_time}")
+
+
+@define
+class MockWebDriverElement:
+    text: str
+    dt: str | None
+
+    def get_attribute(self, ignored: str) -> str | None:
+        return self.dt
+
+
+def run_get_dates_from_element_tests():
+    # tuple fields:
+    # 1. Test case name or ID
+    # 2. Input date string
+    # 3. Expected output start datetime
+    # 4. Expected output end datetime
+    test_cases = [
+        (
+            "BilletWeb: no datetime, fallback on text parsing",
+            None,
+            "Thu Oct 19, 2023 from 01:00 PM to 02:00 PM",
+            datetime(2023, 10, 19, 13, 0),
+            datetime(2023, 10, 19, 14, 0),
+        ),
+        (
+            "EventBrite: morning",
+            "2025-12-05",
+            "déc. 5 de 8am à 11am UTC",
+            datetime(2025, 12, 5, 8, 0),
+            datetime(2025, 12, 5, 11, 0),
+        ),
+        (
+            "EventBrite: evening",
+            "2025-12-12",
+            "déc. 12 de 6pm à 9pm UTC+1",
+            datetime(2025, 12, 12, 18, 0),
+            datetime(2025, 12, 12, 21, 0),
+        ),
+        (
+            "EventBrite: afternoon in German",
+            "2024-12-16",
+            "Dez. 16 von 5nachm. bis 8nachm. UTC",
+            datetime(2024, 12, 16, 17, 0),
+            datetime(2024, 12, 16, 20, 0),
+        ),
+        (
+            "EventBrite: afternoon with minutes in German",
+            "2024-12-03",
+            "Dez. 3 von 5:30nachm. bis 8:30nachm. MEZ",
+            datetime(2024, 12, 3, 17, 30),
+            datetime(2024, 12, 3, 20, 30),
+        ),
+        (
+            "EventBrite: PM adds 12 to the hours only from 1 PM onwards",
+            "2025-12-14",
+            "déc. 14 de 9:30am à 12:30pm UTC+1",
+            datetime(2025, 12, 14, 9, 30),
+            datetime(2025, 12, 14, 12, 30),
+        ),
+        (
+            "EventBrite: start and end minutes differ",
+            "2026-01-21",
+            "janv. 21 de 9am à 12:30pm UTC+1",
+            datetime(2026, 1, 21, 9, 0),
+            datetime(2026, 1, 21, 12, 30),
+        ),
+    ]
+    for test_case in test_cases:
+        logging.info(f"Running {test_case[0]}")
+        actual_start_time, actual_end_time = date_and_time.get_dates_from_element(
+            MockWebDriverElement(dt=test_case[1], text=test_case[2])
+        )
+        if actual_start_time != test_case[3]:
+            logging.error(f"{test_case[0]}: expected {test_case[3]} but got {actual_start_time}")
+        if actual_end_time != test_case[4]:
+            logging.error(f"{test_case[0]}: expected {test_case[4]} but got {actual_end_time}")
+
+
+def run_tests():
+    run_get_dates_tests()
+    run_get_dates_from_element_tests()