Streaming

analyze_stream() returns a generator that yields Server-Sent Events as Python dicts. Use it when you want real-time progress without blocking.

Basic Usage

for event in client.analyze_stream(
    own_page="https://example.com",
    competitors=["https://comp.com"],
    queries=["target query"],
):
    state = event.get("state")

    if state == "PROGRESS":
        pct = event.get("progress", 0)
        msg = event.get("details", {}).get("message", "")
        print(f"[{pct:3d}%] {msg}")

    elif state == "SUCCESS":
        result = event["result"]
        print("Done!", result.keys())
        break

    elif state == "FAILURE":
        print("Error:", event.get("error"))
        break

Event States

State	Description
PENDING	Task queued, not started yet
PROCESSING	Analysis running
PROGRESS	Progress update with progress (0–100) and details.message
SUCCESS	Analysis complete — event["result"] contains the full result
FAILURE	Error — event["error"] contains code and message

Progress Details

event = {
    "state": "PROGRESS",
    "progress": 45,
    "details": {
        "step": "analysis_internal",
        "message": "Computing TF-IDF vectors"
    }
}

Example: Custom Progress Bar

from unihra import UnihraClient

client = UnihraClient(api_key="YOUR_KEY")

total = 0
for event in client.analyze_stream("https://mysite.com", ["https://comp.com"]):
    if event.get("state") == "PROGRESS":
        pct = event.get("progress", 0)
        bar = "█" * (pct // 5) + "░" * (20 - pct // 5)
        print(f"\r[{bar}] {pct}%", end="", flush=True)
    elif event.get("state") == "SUCCESS":
        print("\nComplete!")
        result = event["result"]
        break

Note

analyze() is a convenience wrapper around analyze_stream() that blocks until SUCCESS and optionally shows a tqdm progress bar. Use analyze_stream() directly for custom UIs, async wrappers, or long-running pipelines.