huawei-cloud-daily-hot-topics

Query and display today's hot topics in one shot: fetch a ranked hot list (rank / title / hot value / source) from public, authentication-free hot-list REST APIs over HTTPS (read-only), normalise the results and print them as a readable Markdown table by default (plain text and JSON are also available). Providers are abstracted and tried in order, degrading to a bundled sample dataset — clearly marked as sample data — when the network or every provider is unavailable, so a single run always yields a readable result without crashing. Implemented with the Python 3 standard library only; it does not call the Huawei Cloud CLI, any Huawei Cloud SDK, or any Huawei Cloud REST API, requires no AK/SK credentials, and touches no cloud resource. Triggers include: "查询今日热点", "今日热点榜单", "热门话题", "今日热搜", "热点榜", "daily hot topics", "today hot list", "trending topics".

Daily Hot Topics

Overview

This skill answers "what is hot today?" in a single call. It fetches a ranked hot-topic list from public, keyless hot-list REST endpoints (read-only over HTTPS), normalises every provider into one shape, and renders a readable list.

• One run, one ranked list — a single execution returns today's hot topics with rank / title / hot value / source, up to --top entries.
• Multiple providers, ordered degradation — the default chain tries several public sources in order; the first success wins. --source pins a preferred provider while keeping the remaining providers as fallbacks.
• Bundled offline fallback — if the network is unavailable or every provider fails, the skill falls back to a bundled sample dataset and clearly labels it as sample data (non-real-time), so the command still exits 0 and stays readable instead of crashing.
• Readable by default, machine-readable on demand — Markdown (default), plain text, and JSON output; results can be written to a file.
• Self-contained — pure Python 3 standard library (only urllib), zero third-party dependencies, no credentials.

Prerequisites

• Python 3.8+ available as python3 (or python) on PATH.
• Outbound HTTPS to the public hot-list endpoints is optional: without it the skill degrades to the bundled scripts/sample_data.json and still succeeds.
• No Huawei Cloud credentials, CLI (KooCLI), SDK, or AK/SK are needed; this skill never accesses Huawei Cloud resources.
• scripts/sample_data.json ships with the skill and is used only as the final fallback.

Workflow

1. Resolve the provider chain — resolve_chain(source) returns the ordered list of providers to try: the default chain for auto, or the requested source first followed by the remaining default providers.
2. Fetch each provider — fetch_source(name, timeout, retries) performs a read-only HTTPS GET with a bounded timeout and a small retry budget, parses the JSON, validates the envelope, and normalises the records.
3. Normalise — normalize(records, label) maps heterogeneous fields into {rank, title, hot, source, url}, skipping records without a usable title.
4. Degrade — fetch(...) walks the chain; the first provider that returns a non-empty list wins. If all fail (or --offline is set), it loads the bundled sample dataset and marks the result as sample data, collecting a warning trail.
5. Render — render_markdown() / render_text() / render_json() turn the result into readable output, constrained to --top entries.
6. Emit — the result goes to standard output, or to a file when --output is given.

Core Commands

bash

# Default: today's hot list as a Markdown table (top 10)
python3 scripts/fetch_hot_topics.py

# Plain text, only the top 5
python3 scripts/fetch_hot_topics.py --format text --top 5

# Machine-readable JSON (parseable), top 3
python3 scripts/fetch_hot_topics.py --format json --top 3

# Pin a preferred provider (falls back to the other providers, then sample data)
python3 scripts/fetch_hot_topics.py --source zhihu --top 20

# Skip the network entirely and use the bundled sample dataset
python3 scripts/fetch_hot_topics.py --offline

# List the available providers
python3 scripts/fetch_hot_topics.py --list-sources

# Write the result to a file, with a longer timeout
python3 scripts/fetch_hot_topics.py --format markdown --timeout 30 --output hot.md

# Run the bundled unit tests (offline; HTTP is mocked)
python3 -m unittest discover -s tests -v

The command-line entry point is scripts/fetch_hot_topics.py. The bundled scripts/sample_data.json is what the offline fallback reads by default.

Parameter Confirmation

Parameter	Required	Description	Default
--top	No	Number of entries to show	10
--source	No	Preferred provider: auto, weibo, zhihu, douyin, vvhan	auto
--format	No	Output format: markdown, text, or json	markdown
--output	No	Write the result to a file instead of stdout	stdout
--timeout	No	Per-request timeout in seconds	20
--retries	No	Retry attempts per provider	2
--offline	No	Skip the network and use the bundled sample dataset	off
--list-sources	No	Print the available providers and exit	off

Reference Documents

Document	Description
references/usage-guide.md	Command recipes, provider chain, offline behaviour, extending providers
references/verification-method.md	How the skill is verified (offline unit tests + live read-only spot checks)
references/acceptance-criteria.md	Acceptance criteria and out-of-scope items
references/dataflow-diagram.md	Mermaid data flow diagram
references/iam-policies.md	Permission notes (none required)

Notes

• The skill is read-only: it fetches public hot lists over HTTPS and writes only to standard output or a user-chosen --output file. It never touches a cloud resource and never persists fetched data.
• Degradation is explicit: when the network or providers fail, the fallback result is flagged as sample data and a warning trail is printed, so a sample list is never mistaken for a live ranking.
• The exit code is 0 on success (including fallback to sample data); a non-zero code is returned only for invalid arguments or when --output cannot be written.
• Third-party public endpoints are not under our control; providers are easy to replace or extend by editing the PROVIDERS registry in the script.