cli-web-tripadvisor 0.1.1__py3-none-any.whl

cli_web/tripadvisor/README.md

@@ -0,0 +1,110 @@
+# cli-web-tripadvisor
+
+> Generated by [CLI-Anything-Web](../../../../cli-anything-web-plugin/) from [tripadvisor.com](https://www.tripadvisor.com)
+
+Agent-native CLI for TripAdvisor — search destinations, hotels, restaurants,
+and attractions from the terminal. Data is extracted from SSR HTML pages
+(JSON-LD structured data) and the TypeAheadJson REST endpoint; DataDome bot
+protection is bypassed with curl_cffi Safari-iOS impersonation.
+
+## Installation
+
+```bash
+cd tripadvisor/agent-harness
+pip install -e .
+```
+
+Binary: `cli-web-tripadvisor`
+
+## Usage
+
+```bash
+# 1. Resolve a destination to its geo_id
+cli-web-tripadvisor locations search "Paris"
+
+# 2. Search hotels (faster with a known geo_id)
+cli-web-tripadvisor hotels search "Paris" --geo-id 187147
+cli-web-tripadvisor hotels search "Paris" --page 2
+
+# 3. Hotel detail by TripAdvisor URL
+cli-web-tripadvisor hotels get "https://www.tripadvisor.com/Hotel_Review-g187147-d229968-Reviews-..."
+
+# Restaurants
+cli-web-tripadvisor restaurants search "New York City"
+cli-web-tripadvisor restaurants get "https://www.tripadvisor.com/Restaurant_Review-..."
+
+# Attractions / things to do
+cli-web-tripadvisor attractions search "London" --page 2
+cli-web-tripadvisor attractions get "https://www.tripadvisor.com/Attraction_Review-..."
+
+# Interactive REPL (default when no subcommand is given)
+cli-web-tripadvisor
+```
+
+### Commands
+
+| Group | Command | Purpose |
+|-------|---------|---------|
+| `locations` | `search QUERY [--max N]` | Search destinations; returns `geo_id` for use with `--geo-id` (default max: 6) |
+| `hotels` | `search LOCATION [--geo-id ID] [--page N]` | Search hotels in a location (30 per page) |
+| `hotels` | `get URL` | Detailed hotel info from a TripAdvisor URL |
+| `restaurants` | `search LOCATION [--geo-id ID] [--page N]` | Search restaurants in a location (30 per page) |
+| `restaurants` | `get URL` | Detailed restaurant info from a TripAdvisor URL |
+| `attractions` | `search LOCATION [--geo-id ID] [--page N]` | Search attractions/things to do (30 per page) |
+| `attractions` | `get URL` | Detailed attraction info from a TripAdvisor URL |
+| — | `doctor` | Environment / connectivity self-check |
+| — | `mcp` | Run as an MCP server over stdio (every command exposed as a tool) |
+
+Typical workflow: `locations search` → grab the `geo_id` → pass it to
+`hotels`/`restaurants`/`attractions search` via `--geo-id` to skip the
+lookup round-trip.
+
+## Auth
+
+**No authentication required.** All search, listing, and detail operations
+use public TripAdvisor pages. There is no `login` command and no `auth.json`.
+
+If requests start failing with `AUTH_EXPIRED` (HTTP 401/403), DataDome bot
+protection has triggered — the impersonation profile can be overridden via:
+
+```bash
+export CLI_WEB_TRIPADVISOR_IMPERSONATE=safari17_2_ios   # default
+```
+
+## JSON Output
+
+Every command supports `--json` (global or per-command) for structured,
+agent-friendly output:
+
+```bash
+cli-web-tripadvisor --json locations search "Paris"
+cli-web-tripadvisor hotels search "Paris" --geo-id 187147 --json
+```
+
+Errors are also structured in JSON mode:
+
+```json
+{"error": true, "code": "NOT_FOUND", "message": "Page not found: ..."}
+```
+
+## REPL
+
+Running `cli-web-tripadvisor` with no subcommand opens an interactive REPL.
+Type `help` for the command list, `quit`/`exit` to leave. Quoted arguments
+work (`hotels search "New York City"`), and starting the REPL with `--json`
+makes every REPL command emit JSON.
+
+## Testing
+
+```bash
+cd tripadvisor/agent-harness
+pip install -e .
+
+# Unit tests (offline — HTTP layer is mocked)
+python -m pytest cli_web/tripadvisor/tests/test_core.py -v -s
+
+# E2E tests (live site + subprocess CLI tests)
+python -m pytest cli_web/tripadvisor/tests/test_e2e.py -v -s
+```
+
+See `TRIPADVISOR.md` (in `agent-harness/`) for the full API map.

cli_web/tripadvisor/__main__.py

@@ -0,0 +1,6 @@
+"""Allow running as: python -m cli_web.tripadvisor"""
+
+from .tripadvisor_cli import cli
+
+if __name__ == "__main__":
+    cli()

cli_web/tripadvisor/commands/attractions.py

@@ -0,0 +1,153 @@
+"""Attraction commands for cli-web-tripadvisor."""
+
+from __future__ import annotations
+
+import click
+from rich.console import Console
+from rich.table import Table
+
+from ..core.client import TripAdvisorClient
+from ..utils.helpers import (
+    format_rating,
+    handle_errors,
+    print_json,
+    resolve_json_mode,
+    truncate,
+)
+
+console = Console()
+
+
+@click.group("attractions")
+@click.pass_context
+def attractions(ctx):
+    """Search and browse TripAdvisor attractions and things to do."""
+    ctx.ensure_object(dict)
+
+
+@attractions.command("search")
+@click.argument("location")
+@click.option(
+    "--geo-id", default=None, metavar="ID", help="Use known geo_id to skip location lookup."
+)
+@click.option(
+    "--page", default=1, type=int, show_default=True, help="Page number (30 attractions per page)."
+)
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def search_attractions(ctx, location, geo_id, page, json_mode):
+    """Search attractions and things to do in LOCATION.
+
+    LOCATION is a destination name like "Paris" or "New York City".
+    Use --geo-id to skip the location-lookup step (faster).
+
+    Examples:
+
+      cli-web-tripadvisor attractions search "Paris"
+
+      cli-web-tripadvisor attractions search "Paris" --geo-id 187147
+
+      cli-web-tripadvisor attractions search "London" --page 2
+
+      cli-web-tripadvisor attractions search "Tokyo" --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            result = client.search_attractions(location, geo_id=geo_id, page=page)
+
+        attrlist = result.get("attractions", [])
+
+        if json_mode:
+            print_json(
+                {
+                    "success": True,
+                    "data": {
+                        "location": location,
+                        "geo_id": result.get("geo_id"),
+                        "page": page,
+                        "count": len(attrlist),
+                        "attractions": [a.to_dict() for a in attrlist],
+                    },
+                }
+            )
+            return
+
+        if not attrlist:
+            click.echo(
+                f"No attractions found for '{location}' (page {page}). "
+                "Try a different location name or --geo-id."
+            )
+            return
+
+        table = Table(
+            title=f"Attractions in {location} — page {page}",
+            show_lines=False,
+            expand=False,
+        )
+        table.add_column("ID", style="dim", no_wrap=True, max_width=10)
+        table.add_column("Name", max_width=40)
+        table.add_column("Rating", justify="right", max_width=14)
+        table.add_column("City", max_width=16)
+        table.add_column("Phone", max_width=18)
+
+        for a in attrlist:
+            table.add_row(
+                a.id,
+                truncate(a.name, 40),
+                format_rating(a.rating, a.review_count),
+                a.city or "—",
+                a.telephone or "—",
+            )
+
+        console.print(table)
+        click.echo(f"\nShowing {len(attrlist)} attraction(s) on page {page}.")
+        if len(attrlist) >= 30:
+            click.echo(f"Next page: attractions search '{location}' --page {page + 1}")
+        click.echo("Tip: Use 'attractions get URL' with the attraction URL to see full details.")
+
+
+@attractions.command("get")
+@click.argument("url")
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def get_attraction(ctx, url, json_mode):
+    """Get detailed information for an attraction by its TripAdvisor URL.
+
+    The URL is the full TripAdvisor attraction URL from a search result,
+    e.g. https://www.tripadvisor.com/Attraction_Review-g187147-d188151-Reviews-...
+
+    Examples:
+
+      cli-web-tripadvisor attractions get "https://www.tripadvisor.com/Attraction_Review-g187147-d188151-Reviews-Eiffel_Tower-Paris_Ile_de_France.html"
+
+      cli-web-tripadvisor attractions get "https://www.tripadvisor.com/Attraction_Review-..." --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            attr = client.get_attraction(url)
+
+        if json_mode:
+            print_json({"success": True, "data": {"attraction": attr.to_dict()}})
+            return
+
+        click.echo(f"\n{'=' * 60}")
+        click.echo(f"  {attr.name}")
+        click.echo(f"{'=' * 60}")
+        click.echo(f"  ID:          {attr.id or '—'}")
+        click.echo(f"  Rating:      {format_rating(attr.rating, attr.review_count)}")
+        click.echo(f"  Address:     {attr.address or '—'}")
+        click.echo(f"  City:        {attr.city or '—'}")
+        click.echo(f"  Telephone:   {attr.telephone or '—'}")
+        click.echo(f"  Coordinates: {attr.latitude or '?'}, {attr.longitude or '?'}")
+        if attr.opening_hours:
+            click.echo("  Hours:")
+            for h in attr.opening_hours[:7]:
+                click.echo(f"    {h}")
+        if attr.description:
+            click.echo(f"  Description: {truncate(attr.description, 200)}")
+        click.echo(f"  URL:         {attr.url}")
+        click.echo()

cli_web/tripadvisor/commands/hotels.py

@@ -0,0 +1,155 @@
+"""Hotel commands for cli-web-tripadvisor."""
+
+from __future__ import annotations
+
+import click
+from rich.console import Console
+from rich.table import Table
+
+from ..core.client import TripAdvisorClient
+from ..utils.helpers import (
+    format_rating,
+    handle_errors,
+    print_json,
+    resolve_json_mode,
+    truncate,
+)
+
+console = Console()
+
+
+@click.group("hotels")
+@click.pass_context
+def hotels(ctx):
+    """Search and browse TripAdvisor hotels."""
+    ctx.ensure_object(dict)
+
+
+@hotels.command("search")
+@click.argument("location")
+@click.option(
+    "--geo-id", default=None, metavar="ID", help="Use known geo_id to skip location lookup."
+)
+@click.option(
+    "--page", default=1, type=int, show_default=True, help="Page number (30 hotels per page)."
+)
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def search_hotels(ctx, location, geo_id, page, json_mode):
+    """Search hotels in LOCATION.
+
+    LOCATION is a destination name like "Paris" or "New York City".
+    Use --geo-id to skip the location-lookup step (faster).
+
+    Examples:
+
+      cli-web-tripadvisor hotels search "Paris"
+
+      cli-web-tripadvisor hotels search "Paris" --geo-id 187147
+
+      cli-web-tripadvisor hotels search "Tokyo" --page 2
+
+      cli-web-tripadvisor hotels search "London" --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            result = client.search_hotels(location, geo_id=geo_id, page=page)
+
+        hotellist = result.get("hotels", [])
+
+        if json_mode:
+            print_json(
+                {
+                    "success": True,
+                    "data": {
+                        "location": location,
+                        "geo_id": result.get("geo_id"),
+                        "page": page,
+                        "count": len(hotellist),
+                        "hotels": [h.to_dict() for h in hotellist],
+                    },
+                }
+            )
+            return
+
+        if not hotellist:
+            click.echo(
+                f"No hotels found for '{location}' (page {page}). "
+                "Try a different location name or --geo-id."
+            )
+            return
+
+        table = Table(
+            title=f"Hotels in {location} — page {page}",
+            show_lines=False,
+            expand=False,
+        )
+        table.add_column("ID", style="dim", no_wrap=True, max_width=10)
+        table.add_column("Name", max_width=38)
+        table.add_column("Rating", justify="right", max_width=14)
+        table.add_column("Price", justify="center", max_width=8)
+        table.add_column("City", max_width=16)
+        table.add_column("Phone", max_width=18)
+
+        for h in hotellist:
+            table.add_row(
+                h.id,
+                truncate(h.name, 38),
+                format_rating(h.rating, h.review_count),
+                h.price_range or "—",
+                h.city or "—",
+                h.telephone or "—",
+            )
+
+        console.print(table)
+        click.echo(f"\nShowing {len(hotellist)} hotel(s) on page {page}.")
+        if len(hotellist) >= 30:
+            click.echo(f"Next page: hotels search '{location}' --page {page + 1}")
+        click.echo("Tip: Use 'hotels get URL' with the hotel URL to see full details.")
+
+
+@hotels.command("get")
+@click.argument("url")
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def get_hotel(ctx, url, json_mode):
+    """Get detailed information for a hotel by its TripAdvisor URL.
+
+    The URL is the full TripAdvisor hotel URL from a search result,
+    e.g. https://www.tripadvisor.com/Hotel_Review-g187147-d229968-Reviews-...
+
+    Examples:
+
+      cli-web-tripadvisor hotels get "https://www.tripadvisor.com/Hotel_Review-g187147-d229968-Reviews-Hotel_Astra_Opera_Astotel-Paris_Ile_de_France.html"
+
+      cli-web-tripadvisor hotels get "https://www.tripadvisor.com/Hotel_Review-..." --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            hotel = client.get_hotel(url)
+
+        if json_mode:
+            print_json({"success": True, "data": {"hotel": hotel.to_dict()}})
+            return
+
+        click.echo(f"\n{'=' * 60}")
+        click.echo(f"  {hotel.name}")
+        click.echo(f"{'=' * 60}")
+        click.echo(f"  ID:          {hotel.id or '—'}")
+        click.echo(f"  Rating:      {format_rating(hotel.rating, hotel.review_count)}")
+        click.echo(f"  Price range: {hotel.price_range or '—'}")
+        click.echo(f"  Address:     {hotel.address or '—'}")
+        click.echo(f"  City:        {hotel.city or '—'}")
+        click.echo(f"  Country:     {hotel.country or '—'}")
+        click.echo(f"  Telephone:   {hotel.telephone or '—'}")
+        click.echo(f"  Coordinates: {hotel.latitude or '?'}, {hotel.longitude or '?'}")
+        if hotel.amenities:
+            click.echo(f"  Amenities:   {', '.join(hotel.amenities[:10])}")
+            if len(hotel.amenities) > 10:
+                click.echo(f"               ... and {len(hotel.amenities) - 10} more")
+        click.echo(f"  URL:         {hotel.url}")
+        click.echo()

cli_web/tripadvisor/commands/locations.py

@@ -0,0 +1,93 @@
+"""Location search commands for cli-web-tripadvisor."""
+
+from __future__ import annotations
+
+import click
+from rich.console import Console
+from rich.table import Table
+
+from ..core.client import TripAdvisorClient
+from ..utils.helpers import handle_errors, print_json, resolve_json_mode
+
+console = Console()
+
+
+@click.group("locations")
+@click.pass_context
+def locations(ctx):
+    """Search TripAdvisor destinations and locations."""
+    ctx.ensure_object(dict)
+
+
+@locations.command("search")
+@click.argument("query")
+@click.option(
+    "--max",
+    "max_results",
+    default=6,
+    type=int,
+    show_default=True,
+    help="Maximum number of results.",
+)
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def search_locations(ctx, query, max_results, json_mode):
+    """Search for destinations matching QUERY.
+
+    Returns locations with their geo_id, which can be passed to
+    hotels/restaurants/attractions search via --geo-id.
+
+    Examples:
+
+      cli-web-tripadvisor locations search "Paris"
+
+      cli-web-tripadvisor locations search "New York" --max 10
+
+      cli-web-tripadvisor locations search "Tokyo" --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            results = client.search_locations(query, max_results=max_results)
+
+        if json_mode:
+            print_json(
+                {
+                    "success": True,
+                    "data": {
+                        "query": query,
+                        "count": len(results),
+                        "locations": [loc.to_dict() for loc in results],
+                    },
+                }
+            )
+            return
+
+        if not results:
+            click.echo(f"No locations found for '{query}'.")
+            return
+
+        table = Table(
+            title=f"TripAdvisor Locations — {query}",
+            show_lines=False,
+            expand=False,
+        )
+        table.add_column("Geo ID", style="dim", no_wrap=True, max_width=10)
+        table.add_column("Type", max_width=12)
+        table.add_column("Name", max_width=50)
+        table.add_column("Region", max_width=30)
+
+        for loc in results:
+            table.add_row(
+                loc.geo_id,
+                loc.type,
+                loc.name,
+                loc.geo_name or loc.parent_name or "",
+            )
+
+        console.print(table)
+        click.echo(f"\nFound {len(results)} location(s) for '{query}'.")
+        click.echo(
+            "Tip: Use --geo-id GEO_ID with hotels/restaurants/attractions search for faster results."
+        )

cli_web/tripadvisor/commands/restaurants.py

@@ -0,0 +1,157 @@
+"""Restaurant commands for cli-web-tripadvisor."""
+
+from __future__ import annotations
+
+import click
+from rich.console import Console
+from rich.table import Table
+
+from ..core.client import TripAdvisorClient
+from ..utils.helpers import (
+    format_rating,
+    handle_errors,
+    print_json,
+    resolve_json_mode,
+    truncate,
+)
+
+console = Console()
+
+
+@click.group("restaurants")
+@click.pass_context
+def restaurants(ctx):
+    """Search and browse TripAdvisor restaurants."""
+    ctx.ensure_object(dict)
+
+
+@restaurants.command("search")
+@click.argument("location")
+@click.option(
+    "--geo-id", default=None, metavar="ID", help="Use known geo_id to skip location lookup."
+)
+@click.option(
+    "--page", default=1, type=int, show_default=True, help="Page number (30 restaurants per page)."
+)
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def search_restaurants(ctx, location, geo_id, page, json_mode):
+    """Search restaurants in LOCATION.
+
+    LOCATION is a destination name like "Paris" or "New York City".
+    Use --geo-id to skip the location-lookup step (faster).
+
+    Examples:
+
+      cli-web-tripadvisor restaurants search "Paris"
+
+      cli-web-tripadvisor restaurants search "Paris" --geo-id 187147
+
+      cli-web-tripadvisor restaurants search "Barcelona" --page 2
+
+      cli-web-tripadvisor restaurants search "Rome" --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            result = client.search_restaurants(location, geo_id=geo_id, page=page)
+
+        restlist = result.get("restaurants", [])
+
+        if json_mode:
+            print_json(
+                {
+                    "success": True,
+                    "data": {
+                        "location": location,
+                        "geo_id": result.get("geo_id"),
+                        "page": page,
+                        "count": len(restlist),
+                        "restaurants": [r.to_dict() for r in restlist],
+                    },
+                }
+            )
+            return
+
+        if not restlist:
+            click.echo(
+                f"No restaurants found for '{location}' (page {page}). "
+                "Try a different location name or --geo-id."
+            )
+            return
+
+        table = Table(
+            title=f"Restaurants in {location} — page {page}",
+            show_lines=False,
+            expand=False,
+        )
+        table.add_column("ID", style="dim", no_wrap=True, max_width=10)
+        table.add_column("Name", max_width=36)
+        table.add_column("Rating", justify="right", max_width=14)
+        table.add_column("Price", justify="center", max_width=8)
+        table.add_column("Cuisines", max_width=22)
+        table.add_column("Phone", max_width=18)
+
+        for r in restlist:
+            cuisines = ", ".join(r.cuisines[:2]) if r.cuisines else "—"
+            table.add_row(
+                r.id,
+                truncate(r.name, 36),
+                format_rating(r.rating, r.review_count),
+                r.price_range or "—",
+                cuisines,
+                r.telephone or "—",
+            )
+
+        console.print(table)
+        click.echo(f"\nShowing {len(restlist)} restaurant(s) on page {page}.")
+        if len(restlist) >= 30:
+            click.echo(f"Next page: restaurants search '{location}' --page {page + 1}")
+        click.echo("Tip: Use 'restaurants get URL' with the restaurant URL to see full details.")
+
+
+@restaurants.command("get")
+@click.argument("url")
+@click.option("--json", "json_mode", is_flag=True, help="Output as JSON.")
+@click.pass_context
+def get_restaurant(ctx, url, json_mode):
+    """Get detailed information for a restaurant by its TripAdvisor URL.
+
+    The URL is the full TripAdvisor restaurant URL from a search result,
+    e.g. https://www.tripadvisor.com/Restaurant_Review-g187147-d1035679-Reviews-...
+
+    Examples:
+
+      cli-web-tripadvisor restaurants get "https://www.tripadvisor.com/Restaurant_Review-g187147-d1035679-Reviews-Da_Franco-Paris_Ile_de_France.html"
+
+      cli-web-tripadvisor restaurants get "https://www.tripadvisor.com/Restaurant_Review-..." --json
+    """
+    json_mode = resolve_json_mode(json_mode, ctx)
+
+    with handle_errors(json_mode=json_mode):
+        with TripAdvisorClient() as client:
+            rest = client.get_restaurant(url)
+
+        if json_mode:
+            print_json({"success": True, "data": {"restaurant": rest.to_dict()}})
+            return
+
+        click.echo(f"\n{'=' * 60}")
+        click.echo(f"  {rest.name}")
+        click.echo(f"{'=' * 60}")
+        click.echo(f"  ID:          {rest.id or '—'}")
+        click.echo(f"  Rating:      {format_rating(rest.rating, rest.review_count)}")
+        click.echo(f"  Price range: {rest.price_range or '—'}")
+        if rest.cuisines:
+            click.echo(f"  Cuisines:    {', '.join(rest.cuisines)}")
+        click.echo(f"  Address:     {rest.address or '—'}")
+        click.echo(f"  City:        {rest.city or '—'}")
+        click.echo(f"  Telephone:   {rest.telephone or '—'}")
+        click.echo(f"  Coordinates: {rest.latitude or '?'}, {rest.longitude or '?'}")
+        if rest.opening_hours:
+            click.echo("  Hours:")
+            for h in rest.opening_hours[:7]:
+                click.echo(f"    {h}")
+        click.echo(f"  URL:         {rest.url}")
+        click.echo()