ticket #104 attempt 1

devplacepy/docs_api/_shared.py

@@ -38,7 +38,6 @@ def field(
     example="",
     description="",
     options=None,
-    nullable=False,
 ):
     spec = {
         "name": name,
@@ -47,7 +46,6 @@ def field(
         "required": required,
         "example": example,
         "description": description,
-        "nullable": nullable,
     }
     if options:
         spec["options"] = list(options)

devplacepy/docs_examples.py

@@ -39,15 +39,7 @@ def _unwrap_optional(annotation):
     return annotation
 
 
-def _is_optional(annotation):
-    if get_origin(annotation) is Union:
-        return type(None) in get_args(annotation)
-    return False
-
-
-def _value(name, annotation, stack, is_nullable=False):
-    if is_nullable:
-        return None
+def _value(name, annotation, stack):
     annotation = _unwrap_optional(annotation)
     origin = get_origin(annotation)
     if origin in (list, set, tuple):
@@ -79,8 +71,7 @@ def _from_model(model, stack):
     stack = stack | {model}
     example = {}
     for name, info in model.model_fields.items():
-        is_nullable = _is_optional(info.annotation)
-        example[name] = _value(name, info.annotation, stack, is_nullable)
+        example[name] = _value(name, info.annotation, stack)
     return example
 
 

devplacepy/services/jobs/deepsearch/crawl.py

@@ -22,6 +22,8 @@ from .pdf import MAX_PDF_BYTES, extract_pdf_text, is_pdf
 
 logger = logging.getLogger(__name__)
 
+_pw_lock = asyncio.Lock()
+
 RSEARCH_URL = "https://rsearch.app.molodetz.nl"
 RSEARCH_TIMEOUT_SECONDS = 45.0
 FETCH_TIMEOUT_SECONDS = 20.0
@@ -172,13 +174,7 @@ async def _render_with_playwright(
 ) -> tuple[str, str, int, list[tuple[str, str]]]:
     from playwright.async_api import async_playwright
 
-    own_browser = browser is None
-    if own_browser:
-        pw = await async_playwright().__aenter__()
-        browser = await pw.chromium.launch(
-            headless=True, args=["--no-sandbox", "--disable-dev-shm-usage"]
-        )
-    try:
+    async def _render(browser) -> tuple[str, str, int, list[tuple[str, str]]]:
         context = await browser.new_context(user_agent=USER_AGENT)
         page = await context.new_page()
         response = await page.goto(url, wait_until="load", timeout=30000)
@@ -189,10 +185,18 @@ async def _render_with_playwright(
         await context.close()
         extracted = extract_html(content, base_url=url)
         return extracted.title, extracted.text, status, extracted.links
-    finally:
-        if own_browser:
-            await browser.close()
-            await pw.__aexit__(None, None, None)
+
+    if browser is None:
+        async with async_playwright() as pw:
+            browser = await pw.chromium.launch(
+                headless=True, args=["--no-sandbox", "--disable-dev-shm-usage"]
+            )
+            try:
+                return await _render(browser)
+            finally:
+                await browser.close()
+    else:
+        return await _render(browser)
 
 
 async def fetch_page(url: str, depth: int, browser=None) -> CrawledPage | None:
@@ -242,18 +246,20 @@ async def fetch_page(url: str, depth: int, browser=None) -> CrawledPage | None:
         except (LookupError, ValueError) as exc:
             logger.info("deepsearch decode failed for %s: %s", url, exc)
     if len(text) < MIN_PAGE_CHARS:
-        try:
-            r_title, r_text, r_status, r_links = await _render_with_playwright(url, browser)
-            if len(r_text) > len(text):
-                title, text, status, source, links = (
-                    r_title or title,
-                    r_text,
-                    r_status or status,
-                    "playwright",
-                    r_links,
-                )
-        except Exception as exc:
-            logger.info("deepsearch render failed for %s: %s", url, exc)
+        async with _pw_lock:
+            try:
+                r_title, r_text, r_status, r_links = await _render_with_playwright(url, browser)
+            except Exception as exc:
+                logger.info("deepsearch render failed for %s: %s", url, exc)
+                r_title, r_text, r_status, r_links = "", "", 0, []
+        if len(r_text) > len(text):
+            title, text, status, source, links = (
+                r_title or title,
+                r_text,
+                r_status or status,
+                "playwright",
+                r_links,
+            )
     if len(text) < MIN_PAGE_CHARS:
         return None
     return CrawledPage(
@@ -296,102 +302,99 @@ async def crawl(
     total = min(len(level_candidates), max_pages)
     cancelled = False
 
-    pw = None
     browser = None
-    try:
-        pw = await async_playwright().__aenter__()
-        browser = await pw.chromium.launch(
-            headless=True, args=["--no-sandbox", "--disable-dev-shm-usage"]
-        )
-    except Exception as exc:
-        logger.warning("deepsearch playwright launch failed, pages will use httpx only: %s", exc)
+    async with async_playwright() as pw:
+        try:
+            browser = await pw.chromium.launch(
+                headless=True, args=["--no-sandbox", "--disable-dev-shm-usage"]
+            )
+        except Exception as exc:
+            logger.warning("deepsearch playwright launch failed, pages will use httpx only: %s", exc)
 
-    try:
-        for level in range(max(1, depth)):
-            if cancelled or fetched >= max_pages or not level_candidates:
-                break
-            next_candidates: list[dict] = []
-            for start in range(0, len(level_candidates), CRAWL_CONCURRENCY):
-                if fetched >= max_pages:
+        try:
+            for level in range(max(1, depth)):
+                if cancelled or fetched >= max_pages or not level_candidates:
                     break
-                if await should_stop():
-                    emit({"type": "stage", "stage": "cancelled", "message": "Crawl cancelled"})
-                    cancelled = True
-                    break
-                batch = level_candidates[start : start + CRAWL_CONCURRENCY][: max_pages - fetched]
-                for candidate in batch:
-                    emit(
-                        {
-                            "type": "progress",
-                            "done": fetched,
-                            "total": total,
-                            "url": candidate["url"],
-                            "depth": level,
-                            "message": f"Reading {candidate['url']}",
-                        }
-                    )
-                    if is_cached(candidate["url"]):
-                        emit({"type": "page_cached", "url": candidate["url"], "reason": "seen in a prior run"})
-                fetch_start = time.perf_counter()
-                results = await asyncio.gather(
-                    *(_resolve_candidate(candidate, level, browser) for candidate in batch),
-                    return_exceptions=True,
-                )
-                elapsed_ms = int((time.perf_counter() - fetch_start) * 1000)
-                for candidate, page in zip(batch, results):
-                    url = candidate["url"]
-                    if isinstance(page, BaseException):
-                        logger.info("deepsearch fetch crashed for %s: %s", url, page)
-                        page = None
-                    if page is None:
-                        emit(
-                            {
-                                "type": "page_skipped",
-                                "url": url,
-                                "reason": "no readable content",
-                                "elapsed_ms": elapsed_ms,
-                            }
-                        )
-                        continue
+                next_candidates: list[dict] = []
+                for start in range(0, len(level_candidates), CRAWL_CONCURRENCY):
                     if fetched >= max_pages:
                         break
-                    digest = content_hash(page.text)
-                    if digest in outcome.seen_hashes:
+                    if await should_stop():
+                        emit({"type": "stage", "stage": "cancelled", "message": "Crawl cancelled"})
+                        cancelled = True
+                        break
+                    batch = level_candidates[start : start + CRAWL_CONCURRENCY][: max_pages - fetched]
+                    for candidate in batch:
                         emit(
                             {
-                                "type": "page_duplicate",
-                                "url": url,
-                                "reason": "duplicate content",
-                                "elapsed_ms": elapsed_ms,
+                                "type": "progress",
+                                "done": fetched,
+                                "total": total,
+                                "url": candidate["url"],
+                                "depth": level,
+                                "message": f"Reading {candidate['url']}",
                             }
                         )
-                        continue
-                    outcome.seen_hashes.add(digest)
-                    outcome.pages.append(page)
-                    fetched += 1
-                    emit(
-                        {
-                            "type": "page_loaded",
-                            "url": page.url,
-                            "title": page.title,
-                            "source": page.source,
-                            "depth": level,
-                            "render": page.source == "playwright",
-                            "elapsed_ms": elapsed_ms,
-                            "done": fetched,
-                            "total": total,
-                        }
+                        if is_cached(candidate["url"]):
+                            emit({"type": "page_cached", "url": candidate["url"], "reason": "seen in a prior run"})
+                    fetch_start = time.perf_counter()
+                    results = await asyncio.gather(
+                        *(_resolve_candidate(candidate, level, browser) for candidate in batch),
+                        return_exceptions=True,
                     )
-                    if level + 1 < depth:
-                        for link in relevant_links(page.links, query, LINKS_PER_PAGE):
-                            if link not in seen_urls:
-                                seen_urls.add(link)
-                                next_candidates.append({"url": link})
-            level_candidates = next_candidates
-            total = min(total + len(next_candidates), max_pages)
-    finally:
-        if browser is not None:
-            await browser.close()
-        if pw is not None:
-            await pw.__aexit__(None, None, None)
+                    elapsed_ms = int((time.perf_counter() - fetch_start) * 1000)
+                    for candidate, page in zip(batch, results):
+                        url = candidate["url"]
+                        if isinstance(page, BaseException):
+                            logger.info("deepsearch fetch crashed for %s: %s", url, page)
+                            page = None
+                        if page is None:
+                            emit(
+                                {
+                                    "type": "page_skipped",
+                                    "url": url,
+                                    "reason": "no readable content",
+                                    "elapsed_ms": elapsed_ms,
+                                }
+                            )
+                            continue
+                        if fetched >= max_pages:
+                            break
+                        digest = content_hash(page.text)
+                        if digest in outcome.seen_hashes:
+                            emit(
+                                {
+                                    "type": "page_duplicate",
+                                    "url": url,
+                                    "reason": "duplicate content",
+                                    "elapsed_ms": elapsed_ms,
+                                }
+                            )
+                            continue
+                        outcome.seen_hashes.add(digest)
+                        outcome.pages.append(page)
+                        fetched += 1
+                        emit(
+                            {
+                                "type": "page_loaded",
+                                "url": page.url,
+                                "title": page.title,
+                                "source": page.source,
+                                "depth": level,
+                                "render": page.source == "playwright",
+                                "elapsed_ms": elapsed_ms,
+                                "done": fetched,
+                                "total": total,
+                            }
+                        )
+                        if level + 1 < depth:
+                            for link in relevant_links(page.links, query, LINKS_PER_PAGE):
+                                if link not in seen_urls:
+                                    seen_urls.add(link)
+                                    next_candidates.append({"url": link})
+                level_candidates = next_candidates
+                total = min(total + len(next_candidates), max_pages)
+        finally:
+            if browser is not None:
+                await browser.close()
     return outcome

devplacepy/static/css/docs.css

@@ -218,16 +218,6 @@
     font-weight: 700;
 }
 
-.param-nullable {
-    font-size: 0.625rem;
-    text-transform: uppercase;
-    letter-spacing: 0.05em;
-    color: var(--text-muted);
-    border: 1px dashed var(--border);
-    border-radius: 999px;
-    padding: 0.05rem 0.4rem;
-}
-
 .param-loc {
     font-size: 0.625rem;
     text-transform: uppercase;

devplacepy/static/js/ApiTester.js

@@ -139,7 +139,6 @@ export class ApiTester {
             const label = this.el("div", { class: "param-label" }, [
                 this.el("span", { class: "param-name", text: param.name }),
                 param.required ? this.el("span", { class: "param-required", text: "*" }) : null,
-                param.nullable ? this.el("span", { class: "param-nullable", text: "nullable" }) : null,
                 this.el("span", { class: "param-loc param-loc-" + param.location, text: param.location }),
             ]);
             const allowed = param.type === "enum" && param.options && param.options.length

tests/unit/test_docs_nullable.py

@@ -1,59 +0,0 @@
-# retoor <retoor@molodetz.nl>
-
-from typing import Optional
-
-from pydantic import BaseModel
-
-from devplacepy.docs_api._shared import field
-from devplacepy.docs_examples import _is_optional, schema_example
-
-
-def test_field_nullable_parameter():
-    f = field("bio", "query", "string", False, "", "User biography", nullable=True)
-    assert f["nullable"] is True
-
-    f2 = field("username", "path", "string", True, "alice", "Target username")
-    assert f2["nullable"] is False
-
-
-def test_is_optional_optional_type():
-    assert _is_optional(Optional[str]) is True
-    assert _is_optional(Optional[int]) is True
-    assert _is_optional(Optional[list[str]]) is True
-
-
-def test_is_optional_non_optional():
-    assert _is_optional(str) is False
-    assert _is_optional(int) is False
-    assert _is_optional(list[str]) is False
-    assert _is_optional(dict) is False
-
-
-def test_schema_example_nullable_fields():
-    class NullableModel(BaseModel):
-        name: str
-        bio: Optional[str] = None
-        age: int
-        score: Optional[int] = None
-        tags: Optional[list[str]] = None
-
-    result = schema_example(NullableModel)
-
-    assert result["name"] == "string"
-    assert result["age"] == 0
-    assert result["bio"] is None
-    assert result["score"] is None
-    assert result["tags"] is None
-
-
-def test_schema_example_non_nullable_fields_unaffected():
-    class StrictModel(BaseModel):
-        x: str
-        y: int
-        z: bool
-
-    result = schema_example(StrictModel)
-
-    assert result["x"] == "string"
-    assert result["y"] == 0
-    assert result["z"] is False