@tikomni/skills 1.0.4 → 1.0.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tikomni/skills",
-  "version": "1.0.4",
+  "version": "1.0.6",
   "description": "TikOmni skill installer CLI for structured social media crawling in Codex, Claude Code, and OpenClaw",
   "license": "MIT",
   "homepage": "https://github.com/mark-ly-wang/TikOmni-Skills#readme",

package/skills/social-media-crawl/scripts/core/asr_pipeline.py

@@ -1586,22 +1586,8 @@ def run_u2_asr_batch_with_timeout_retry(
         existing = mapped_results.get(file_url)
         if existing is None:
             mapped_results[file_url] = candidate
-            continue
-
-        old_score = (
-            1 if existing.get("ok") else 0,
-            len(str(existing.get("transcript_text") or "")),
-            1 if existing.get("transcription_url") else 0,
-            1 if not existing.get("error_reason") else 0,
-        )
-        new_score = (
-            1 if candidate.get("ok") else 0,
-            len(str(candidate.get("transcript_text") or "")),
-            1 if candidate.get("transcription_url") else 0,
-            1 if not candidate.get("error_reason") else 0,
-        )
-        if new_score > old_score:
-            mapped_results[file_url] = candidate
+        # When the provider returns file_url, treat it as the source of truth.
+        # item_index remains a fallback only for older payloads without file_url.
 
     mapped_results = hydrate_u2_batch_results_from_transcription_urls(
         mapped_results=mapped_results,

package/skills/social-media-crawl/scripts/pipelines/run_douyin_single_work.py

@@ -424,7 +424,6 @@ def _u1_fetch_one_video(
             fallback_reason="" if app_response.get("ok") else (
                 "primary_timeout_retry_exhausted" if app_response.get("timeout_retry_exhausted") else "primary_non_timeout_failure"
             ),
-            extra={"response": app_response},
         )
     )
     if app_response.get("ok"):
@@ -458,7 +457,6 @@ def _u1_fetch_one_video(
             fallback_reason="" if web_response.get("ok") else (
                 "fallback_timeout_retry_exhausted" if web_response.get("timeout_retry_exhausted") else "fallback_non_timeout_failure"
             ),
-            extra={"response": web_response},
         )
     )
     web_response["_attempts"] = attempts
@@ -891,23 +889,24 @@ def run_douyin_single_video(
         all_routes_failed=not bool(one_video_response.get("ok")),
     )
     for index, attempt in enumerate(attempts, start=1):
-        response = attempt.get("response") if isinstance(attempt, dict) else None
+        if not isinstance(attempt, dict):
+            continue
+        response = attempt.get("response") if isinstance(attempt.get("response"), dict) else attempt
         endpoint = attempt.get("endpoint") if isinstance(attempt, dict) else None
         label = attempt.get("route_label") if isinstance(attempt, dict) else None
-        if not isinstance(response, dict):
-            if attempt.get("skipped"):
-                trace.append(
-                    {
-                        "step": f"u1_fetch_one_video_attempt_{index}",
-                        "route_label": label,
-                        "endpoint": endpoint,
-                        "accept_reason": attempt.get("accept_reason"),
-                        "fallback_reason": attempt.get("fallback_reason"),
-                        "param_readiness": attempt.get("param_readiness"),
-                        "param_reason": attempt.get("param_reason"),
-                        "skipped": True,
-                    }
-                )
+        if attempt.get("skipped"):
+            trace.append(
+                {
+                    "step": f"u1_fetch_one_video_attempt_{index}",
+                    "route_label": label,
+                    "endpoint": endpoint,
+                    "accept_reason": attempt.get("accept_reason"),
+                    "fallback_reason": attempt.get("fallback_reason"),
+                    "param_readiness": attempt.get("param_readiness"),
+                    "param_reason": attempt.get("param_reason"),
+                    "skipped": True,
+                }
+            )
             continue
         _emit_http_progress(progress, stage="single_video.fetch", response=response, route_label=str(label or "route"))
         step = "u1_fetch_one_video_effective" if index == len(attempts) else f"u1_fetch_one_video_attempt_{index}"

package/skills/social-media-crawl/scripts/pipelines/run_xiaohongshu_single_work.py

@@ -1125,7 +1125,6 @@ def _fetch_note_info(
                 extra={
                     "method": str(route["method"]).upper(),
                     "field_completeness": response.get("_field_completeness"),
-                    "response": response,
                 },
             )
         )
@@ -1843,23 +1842,24 @@ def run_xiaohongshu_extract(
         all_routes_failed=not bool(note_response.get("ok")),
     )
     for index, attempt in enumerate(attempts, start=1):
-        response = attempt.get("response") if isinstance(attempt, dict) else None
+        if not isinstance(attempt, dict):
+            continue
+        response = attempt.get("response") if isinstance(attempt.get("response"), dict) else attempt
         endpoint = attempt.get("endpoint") if isinstance(attempt, dict) else None
         label = attempt.get("route_label") if isinstance(attempt, dict) else None
-        if not isinstance(response, dict):
-            if attempt.get("skipped"):
-                trace.append(
-                    {
-                        "step": f"u1_get_note_info_attempt_{index}",
-                        "route_label": label,
-                        "endpoint": endpoint,
-                        "accept_reason": attempt.get("accept_reason"),
-                        "fallback_reason": attempt.get("fallback_reason"),
-                        "param_readiness": attempt.get("param_readiness"),
-                        "param_reason": attempt.get("param_reason"),
-                        "skipped": True,
-                    }
-                )
+        if attempt.get("skipped"):
+            trace.append(
+                {
+                    "step": f"u1_get_note_info_attempt_{index}",
+                    "route_label": label,
+                    "endpoint": endpoint,
+                    "accept_reason": attempt.get("accept_reason"),
+                    "fallback_reason": attempt.get("fallback_reason"),
+                    "param_readiness": attempt.get("param_readiness"),
+                    "param_reason": attempt.get("param_reason"),
+                    "skipped": True,
+                }
+            )
             continue
         step = "u1_get_note_info_effective" if index == len(attempts) else f"u1_get_note_info_attempt_{index}"
         trace.append(

package/skills/social-media-crawl/tests/test_fixed_pipeline_fallback.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import json
 import os
 import sys
 import unittest
@@ -12,11 +13,61 @@ SKILL_ROOT = Path(__file__).resolve().parents[1]
 if str(SKILL_ROOT) not in sys.path:
     sys.path.insert(0, str(SKILL_ROOT))
 
+from scripts.core.asr_pipeline import run_u2_asr_batch_with_timeout_retry
 from scripts.pipelines import homepage_collectors
+from scripts.pipelines import run_douyin_single_work
 from scripts.pipelines import run_xiaohongshu_single_work
 
 
 class FixedPipelineFallbackTest(unittest.TestCase):
+    def test_douyin_single_fetch_attempt_trace_is_json_serializable(self) -> None:
+        def fake_call_json_api(**_: object) -> dict:
+            return {
+                "ok": True,
+                "status_code": 200,
+                "request_id": "req-dy-single",
+                "error_reason": None,
+                "data": {"aweme_detail": {"aweme_id": "123"}},
+            }
+
+        with patch.object(run_douyin_single_work, "call_json_api", side_effect=fake_call_json_api):
+            response = run_douyin_single_work._u1_fetch_one_video(
+                base_url="https://api.tikomni.com",
+                token="test-token",
+                share_url="https://v.douyin.com/test-single/",
+                app_timeout_ms=1000,
+                web_timeout_ms=1000,
+            )
+
+        json.dumps(response, ensure_ascii=False)
+        self.assertNotIn("response", response["_attempts"][0])
+
+    def test_xhs_single_fetch_attempt_trace_is_json_serializable(self) -> None:
+        def fake_call_json_api(**_: object) -> dict:
+            return {
+                "ok": True,
+                "status_code": 200,
+                "request_id": "req-xhs-single",
+                "error_reason": None,
+                "data": {
+                    "title": "test title",
+                    "desc": "test content",
+                    "video": {"master_url": "https://example.com/video.mp4"},
+                },
+            }
+
+        with patch.object(run_xiaohongshu_single_work, "call_json_api", side_effect=fake_call_json_api):
+            response = run_xiaohongshu_single_work._fetch_note_info(
+                base_url="https://api.tikomni.com",
+                token="test-token",
+                timeout_ms=1000,
+                source_input={"share_text": "https://xhslink.com/test-single", "note_id": "note123"},
+                progress=None,
+            )
+
+        json.dumps(response, ensure_ascii=False)
+        self.assertNotIn("response", response["_attempts"][0])
+
     def test_xhs_single_route_plan_respects_version_priority_and_cookie_gate(self) -> None:
         source_input = {
             "share_text": (
@@ -164,6 +215,71 @@ class FixedPipelineFallbackTest(unittest.TestCase):
         self.assertEqual(web_attempt.get("param_reason"), "fallback_requires_cookie")
         self.assertEqual(raw.get("error_reason"), "posts_all_routes_failed")
 
+    def test_u2_batch_prefers_file_url_mapping_over_item_index_fallback(self) -> None:
+        file_url = "https://example.com/video.mp4"
+        raw_task = {
+            "data": {
+                "task_status": "SUCCEEDED",
+                "task_metrics": {"TOTAL": 1, "SUCCEEDED": 1, "FAILED": 0},
+                "items": [
+                    {
+                        "item_index": 0,
+                        "task_status": "SUCCEEDED",
+                        "transcript_text": "索引映射文本。",
+                        "transcription_url": "https://example.com/index.json",
+                    },
+                    {
+                        "file_url": file_url,
+                        "task_status": "SUCCEEDED",
+                        "transcript_text": "file_url 映射文本。",
+                        "transcription_url": "https://example.com/file.json",
+                    },
+                ],
+            }
+        }
+
+        submit_bundle = {
+            "submit_response": {"ok": True, "request_id": "req-submit"},
+            "task_id": "task-1",
+            "final_submit_status": "success",
+            "retry_chain": [],
+        }
+        poll_result = {
+            "ok": True,
+            "task_id": "task-1",
+            "task_status": "SUCCEEDED",
+            "request_id": "req-poll",
+            "error_reason": "",
+            "raw_task": raw_task,
+            "task_metrics": {"TOTAL": 1, "SUCCEEDED": 1, "FAILED": 0},
+            "batch_progress": {"expected_total": 1, "complete": True},
+            "batch_complete": True,
+            "trace": [],
+        }
+
+        with patch(
+            "scripts.core.asr_pipeline.submit_u2_asr_batch_with_retry",
+            return_value=submit_bundle,
+        ), patch(
+            "scripts.core.asr_pipeline.poll_u2_task_core",
+            return_value=poll_result,
+        ):
+            bundle = run_u2_asr_batch_with_timeout_retry(
+                base_url="https://api.tikomni.com",
+                token="test-token",
+                timeout_ms=1000,
+                file_urls=[file_url],
+                submit_max_retries=0,
+                submit_backoff_ms=0,
+                poll_interval_sec=0.01,
+                max_polls=1,
+                timeout_retry_enabled=False,
+                timeout_retry_max_retries=0,
+            )
+
+        mapped_item = bundle["mapped_results"][file_url]
+        self.assertEqual(mapped_item["transcript_text"], "file_url 映射文本。")
+
 
 if __name__ == "__main__":
     unittest.main()