Added search events by city feature

README.md

@@ -0,0 +1,62 @@
+# MeetupScraper
+
+A lightweight Python library for scraping upcoming events from [meetup.com](https://meetup.com) groups.
+
+## Features
+
+- Fetch upcoming events for any public Meetup group
+- Filter events by name using regular expressions
+- Structured event data with title, date, URL, and venue information
+- Venue details fetched via Meetup's GraphQL API
+- Search events across all groups in a specific city
+- Automatic space-to-hyphen conversion for meetup group names
+- Handles both datetime and date-only event start times
+
+## Installation
+
+```bash
+pip install -e .
+```
+
+Or install dependencies manually:
+
+```bash
+pip install requests lxml python-dateutil icalendar gql aiohttp
+```
+
+## Running the Script
+
+Run the [`search_events.py`](search_events.py) script directly with the desired arguments:
+
+```bash
+# Search by meetup group
+python search_events.py --meetup Hakierspejs-Łodź --regex Radio
+
+# Search by city
+python search_events.py --city Warsaw
+
+# Search by city with regex filter and more groups
+python search_events.py --city "Krakow" --regex Python --max-groups 20
+```
+
+### CLI Arguments
+
+| Flag | Description | Required |
+|------|-------------|----------|
+| `--meetup`, `-m` | Meetup group name or URL slug (e.g., `Hakierspejs-Łodź`) | One of `--meetup` or `--city` |
+| `--city`, `-c` | City name to search for events (e.g., `Warsaw`, `Krakow`) | One of `--meetup` or `--city` |
+| `--regex`, `-r` | Filter events by title using a regular expression | No |
+| `--max-groups` | Maximum number of groups to search when using `--city` (default: 10) | No |
+| `--debug` | Enable debug logging | No |
+
+## Dependencies
+
+- `requests` — HTTP requests for iCal feed
+- `icalendar` — iCal parsing
+- `gql` + `aiohttp` — GraphQL client for venue data
+- `lxml` — XML/HTML parsing (transitive dependency)
+- `python-dateutil` — Date parsing (transitive dependency)
+
+## License
+
+WTFPL

gql_data.py

@@ -1,8 +1,11 @@
 from gql import Client, gql
 from gql.transport.aiohttp import AIOHTTPTransport
 import json
+import requests
+
 
 def get_gql_data(event_id):
+    """Fetch event details from Meetup's GraphQL API."""
     transport = AIOHTTPTransport(url="https://www.meetup.com/gql2")
 
     client = Client(transport=transport)
@@ -29,8 +32,62 @@ def get_gql_data(event_id):
     }
     """)
 
-
     result = client.execute(query, variable_values={"eventId": str(event_id)})
 
-    return(result)
+    return result
+
 
+def geocode_city(city_name):
+    """Convert city name to latitude/longitude using Nominatim."""
+    url = "https://nominatim.openstreetmap.org/search"
+    params = {
+        "q": city_name,
+        "format": "json",
+        "limit": 1,
+    }
+    headers = {
+        "User-Agent": "meetupscraper/1.0",
+    }
+    response = requests.get(url, params=params, headers=headers)
+    data = response.json()
+
+    if not data:
+        raise ValueError(f"Could not geocode city: {city_name}")
+
+    return float(data[0]["lat"]), float(data[0]["lon"])
+
+
+def get_groups_by_city(city_name, limit=20):
+    """Search for meetup groups in a given city using Meetup's GraphQL API."""
+    lat, lon = geocode_city(city_name)
+
+    transport = AIOHTTPTransport(url="https://www.meetup.com/gql2")
+
+    client = Client(transport=transport)
+
+    query = gql("""
+    query SearchGroups($query: String!, $first: Int!, $lat: Float!, $lon: Float!) {
+      groupSearch(first: $first, filter: {query: $query, lat: $lat, lon: $lon}) {
+        edges {
+          node {
+            id
+            name
+            urlname
+            description
+            city
+            country
+          }
+        }
+        totalCount
+      }
+    }
+    """)
+
+    result = client.execute(query, variable_values={
+        "query": city_name,
+        "first": limit,
+        "lat": lat,
+        "lon": lon,
+    })
+
+    return result

meetupscraper.egg-info/PKG-INFO

@@ -0,0 +1,30 @@
+Metadata-Version: 2.4
+Name: meetupscraper
+Version: 0.11.2
+Summary: A web scraper for meetup.com
+Home-page: http://github.com/hakierspejs/meetupscraper
+Author: Hakiespejs Łódź
+Author-email: d33tah@gmail.com
+License: WTFPL
+Requires-Python: >=3.7
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: requests
+Requires-Dist: lxml
+Requires-Dist: python-dateutil
+Requires-Dist: icalendar
+Requires-Dist: gql
+Requires-Dist: aiohttp
+Requires-Dist: json
+Dynamic: author
+Dynamic: author-email
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: license
+Dynamic: license-file
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
+
+Allows you to access events tied to a specific meetup group in a structured way

meetupscraper.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+LICENSE
+README.md
+setup.py
+meetupscraper/__init__.py
+meetupscraper.egg-info/PKG-INFO
+meetupscraper.egg-info/SOURCES.txt
+meetupscraper.egg-info/dependency_links.txt
+meetupscraper.egg-info/entry_points.txt
+meetupscraper.egg-info/requires.txt
+meetupscraper.egg-info/top_level.txt

meetupscraper.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

meetupscraper.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+meetupscraper = search_events:main

meetupscraper.egg-info/requires.txt

@@ -0,0 +1,7 @@
+requests
+lxml
+python-dateutil
+icalendar
+gql
+aiohttp
+json

meetupscraper.egg-info/top_level.txt

@@ -0,0 +1 @@
+meetupscraper

meetupscraper/__init__.py

@@ -6,7 +6,7 @@
 from datetime import timezone
 import icalendar
 import requests
-from gql_data import get_gql_data
+from gql_data import get_gql_data, get_groups_by_city
 import json
 
 
@@ -62,23 +62,9 @@ def get_upcoming_events(meetup_name, name_regex=None):
             LOGGER.info("Skipping event %r", event_name)
             continue
 
-        #location = str(component.get('location', ''))
-        #venue_name = location
-        #venue_street = ''
-        #
-        #if ',' in location:
-        #    venue_parts = location.split(',', 1)
-        #    venue_name = venue_parts[0].strip()
-        #    venue_street = venue_parts[1].strip()
-        #
-        #venue = Venue(name=venue_name, street=venue_street)
-        ## set to None if no address nor name
-        #if not venue_name and not venue_street:
-        #    venue = None
         try:
             event_id = re.search(r".+\/(\d+)\/", event_url).groups()[0]
             gql_data = get_gql_data(event_id)
-            print(gql_data['event']['venue'])
             venue = Venue(
                     name=gql_data['event']['venue']['name'],
                     street=gql_data['event']['venue']['address'] + ', ' + gql_data['event']['venue']['city']
@@ -92,3 +78,50 @@ def get_upcoming_events(meetup_name, name_regex=None):
             url=event_url,
             venue=venue,
         )
+
+
+def get_events_by_city(city_name, name_regex=None, max_groups=10):
+    """
+    Search for upcoming events in a given city by finding groups and aggregating their events.
+
+    Args:
+        city_name: Name of the city to search for events (e.g., 'Warsaw', 'Krakow')
+        name_regex: Optional regex pattern to filter events by title
+        max_groups: Maximum number of groups to search (default 10)
+
+    Yields:
+        Event objects for all upcoming events in the city
+    """
+    LOGGER.info("Searching for groups in city: %s", city_name)
+
+    # Search for groups in the city
+    groups_result = get_groups_by_city(city_name, limit=max_groups)
+
+    groups = groups_result.get('groupSearch', {}).get('edges', [])
+    if not groups:
+        LOGGER.warning("No groups found for city: %s", city_name)
+        return
+
+    LOGGER.info("Found %d groups in %s", len(groups), city_name)
+
+    # Collect events from all groups
+    seen_events = set()  # To avoid duplicates
+
+    for group_edge in groups:
+        group = group_edge.get('node', {})
+        urlname = group.get('urlname')
+
+        if not urlname:
+            continue
+
+        LOGGER.info("Fetching events from group: %s", group.get('name', urlname))
+
+        try:
+            for event in get_upcoming_events(urlname, name_regex=name_regex):
+                # Use URL as unique identifier to avoid duplicates
+                if event.url not in seen_events:
+                    seen_events.add(event.url)
+                    yield event
+        except Exception as e:
+            LOGGER.warning("Failed to fetch events from group %s: %s", urlname, e)
+            continue