feat: add gov source for central-government tenders (mr.gov.il)

CHANGELOG.md

@@ -2,6 +2,14 @@
 
 All notable changes are listed here. Format based on [Keep a Changelog](https://keepachangelog.com/).
 
+## Unreleased
+
+### Added
+- `GovSource` (`gov`): central-government tenders from mr.gov.il, read from the portal's HTML search. It is part of
+  the default sources.
+- Tests use a fixed clock, so open/closed results no longer depend on the date the suite runs.
+- `HttpClient.get_text` for sources that serve HTML.
+
 ## 0.1.0 (2026-11-05)
 
 ### Added

README.md

@@ -7,7 +7,9 @@
 
 Israeli public tenders (מכרזים) in Python. One `Tender` object, whichever site it came from.
 
-Sources today: Israel Land Authority land tenders and Beer Sheva municipal tenders.
+Sources today: Israel Land Authority land tenders, Beer Sheva municipal tenders, and central-government tenders
+from mr.gov.il. The first search reads about 25 pages from mr.gov.il, so it takes around 20 seconds. After that the
+responses are cached for an hour.
 
 > Unofficial client. It can lag or miss tenders, and parsing can break when a source changes. Confirm deadlines on
 > the publisher's page before you act. Not affiliated with the Israel Land Authority or any municipality.

docs/model.md

@@ -48,6 +48,12 @@ Keys are the members of `michraz.RmiExtra`, a string enum: `tender.extra[RmiExtr
 | `khal_yaad_rashi` | `KhalYaadRashi`, kept as is. Observed values are 1, 2 and null, so it is a code and not a bid amount. |
 | `mekuvan` | `Mekuvan`, kept as is. Its meaning is unconfirmed. |
 
+### Government `extra`
+
+Keys are the members of `michraz.GovExtra`: `type` (for example `מכרז פומבי`), `status` (the portal's status, for
+example `עודכן`) and `raw`. `source_id` is the portal's publication number, `number` is its procedure number, and
+`documents` is empty because detail pages are not fetched.
+
 ## `Document`
 
 | Field | Type | Meaning |

docs/sources.md

@@ -75,6 +75,25 @@ correct and not confirmed.
 - 402 rows have a winner, 51 have no files. `קבצים` is HTML that is not always wrapped in `<p>`, and some links
   encode the URL as `http&#58;//...`. It is read with `html.parser`.
 
+## Central government, `gov`
+
+- Portal: the Government Procurement Administration's tender site, `https://mr.gov.il/ilgstorefront/he/search/`.
+  It has no JSON API. The search page is server-rendered HTML, 20 results per page, `page` starts at 0.
+- Attribution: `Government Procurement Administration (מינהל הרכש הממשלתי), mr.gov.il`. The license of the data was
+  not checked.
+- Query: `q=:updateDate:archive:false:status:<status>`, `s=TENDER`. The portal reports 8 835 tenders in total, 7 654
+  of them with status `חלף מועד הגשה` (submission date passed). michraz reads only the statuses `פורסם`, `עודכן` and
+  `חדש`.
+- Verified on 2026-11-05: those three statuses returned 123, 256 and 40 tenders. Reading all of them gave 419
+  tenders and 0 parse errors in about 21 seconds. 183 had a deadline still ahead and 14 had no deadline.
+- Fields read: title, type (`מכרז פומבי`), publisher, publication number (used as `source_id`), procedure number
+  (used as `number`), publication date, update date, deadline (`dd/mm/yyyy HH:MM`) and status.
+- Not verified: statuses `בעדכון` (32) and `עתידי` (15) are not read, because it is unclear whether they accept
+  offers. Detail pages and attached documents are not fetched, so `documents` is empty.
+- `tests/fixtures/gov_search.html` is five real result blocks copied from a live page. The embedded images were
+  removed and the count sentence was edited to say 5 so that it matches the excerpt.
+- The site has no `robots.txt` (the request returns an empty body).
+
 ## Robots and terms
 
 Checked 2026-11-04 from the development machine with the michraz User-Agent.

src/michraz/__init__.py

@@ -12,12 +12,23 @@
 from michraz.errors import MichrazError, SourceFormatError, SourceUnavailableError
 from michraz.export import to_csv, to_dataframe, to_dicts
 from michraz.models import Document, Tender
-from michraz.sources import BeerShevaExtra, BeerShevaSource, RmiExtra, RmiSource, Source, SourceName
+from michraz.sources import (
+    BeerShevaExtra,
+    BeerShevaSource,
+    GovExtra,
+    GovSource,
+    RmiExtra,
+    RmiSource,
+    Source,
+    SourceName,
+)
 
 __all__ = [
     "BeerShevaExtra",
     "BeerShevaSource",
     "Document",
+    "GovExtra",
+    "GovSource",
     "Michraz",
     "MichrazError",
     "RmiExtra",

src/michraz/_http.py

@@ -135,7 +135,7 @@ class HttpClient(Closeable):
         min_interval: Minimum seconds between requests to the same host. Must be at least 0.5.
         client: An ``httpx.Client`` to use. By default one is created and owned by this object.
         clock: Monotonic clock in seconds. Replace it in tests.
-        sleep: Sleep function. Replace it in tests so they do not wait.
+        sleep: Sleep function. Defaults to ``time.sleep``, looked up when the client is created.
         wall_clock: Wall-clock time in seconds since the epoch, used for cache age.
 
     Raises:
@@ -150,14 +150,14 @@ def __init__(
         min_interval: float = DEFAULT_MIN_INTERVAL,
         client: httpx.Client | None = None,
         clock: Callable[[], float] = time.monotonic,
-        sleep: Callable[[float], None] = time.sleep,
+        sleep: Callable[[float], None] | None = None,
         wall_clock: Callable[[], float] = time.time,
     ) -> None:
         if min_interval < MIN_INTERVAL_FLOOR:
             raise ValueError(f"min_interval must be at least {MIN_INTERVAL_FLOOR} seconds")
         self.min_interval = min_interval
         self._clock = clock
-        self._sleep = sleep
+        self._sleep = sleep or time.sleep
         self._wall_clock = wall_clock
         self._owns_client = client is None
         self._client = client or httpx.Client(
@@ -207,16 +207,32 @@ def cached_at(
     def _full_url(url: str, params: Mapping[str, Any] | None) -> str:
         return str(httpx.URL(url, params=params)) if params else url
 
+    def get_text(self, source: str, url: str, params: Mapping[str, Any] | None = None) -> str:
+        """Fetches ``url`` with GET and returns the body as text. Cached and rate-limited like JSON.
+
+        Raises:
+            SourceUnavailableError: If the host cannot be reached or answers with an error status.
+        """
+        text: str = self._fetch(source, HttpMethod.GET, self._full_url(url, params), None, lambda r: r.text)
+        return text
+
     def _json(self, source: str, method: HttpMethod, url: str, payload: Any) -> Any:
+        def decode(response: httpx.Response) -> Any:
+            try:
+                return response.json()
+            except ValueError as exc:
+                raise SourceFormatError(source, f"response is not JSON: {_truncate(response.text)!r}") from exc
+
+        return self._fetch(source, method, url, payload, decode)
+
+    def _fetch(
+        self, source: str, method: HttpMethod, url: str, payload: Any, decode: Callable[[httpx.Response], Any]
+    ) -> Any:
         key = _request_key(method, url, payload)
         if self._cache and (hit := self._cache.fresh(key)) is not None:
             logger.debug("cache hit for %s", url)
             return hit[1]
-        response = self._send(source, method, url, payload)
-        try:
-            data = response.json()
-        except ValueError as exc:
-            raise SourceFormatError(source, f"response is not JSON: {_truncate(response.text)!r}") from exc
+        data = decode(self._send(source, method, url, payload))
         if self._cache:
             self._cache.write(key, data)
         return data

src/michraz/client.py

@@ -16,7 +16,7 @@
 )
 from michraz.filters import TenderFilter, sort_by_deadline
 from michraz.models import Tender
-from michraz.sources import BaseSource, BeerShevaSource, RmiSource, Source
+from michraz.sources import BaseSource, BeerShevaSource, GovSource, RmiSource, Source
 
 logger = logging.getLogger(__name__)
 
@@ -30,7 +30,7 @@ class Michraz(Closeable):
             tenders = m.search(keyword="גינון", open_only=True)
 
     Args:
-        sources: Sources to search. Defaults to RMI and Beer Sheva.
+        sources: Sources to search. Defaults to RMI, Beer Sheva and the central government.
         cache_dir: Directory for cached responses. Defaults to ``~/.cache/michraz``. None disables the cache.
         cache_ttl: How long a cached response stays valid.
         min_interval: Minimum seconds between requests to the same host. Must be at least 0.5.
@@ -51,7 +51,9 @@ def __init__(
         min_interval: float = DEFAULT_MIN_INTERVAL,
     ) -> None:
         self._http = HttpClient(cache_dir=cache_dir, cache_ttl=cache_ttl, min_interval=min_interval)
-        self.sources: tuple[Source, ...] = tuple(sources) if sources is not None else (RmiSource(), BeerShevaSource())
+        self.sources: tuple[Source, ...] = (
+            tuple(sources) if sources is not None else (RmiSource(), BeerShevaSource(), GovSource())
+        )
         names = [s.name for s in self.sources]
         if len(set(names)) != len(names):
             raise ValueError(f"source names must be unique, got {names}")

src/michraz/sources/__init__.py

@@ -2,6 +2,17 @@
 
 from michraz.sources.base import BaseSource, Source, SourceName
 from michraz.sources.beersheva import BeerShevaExtra, BeerShevaSource
+from michraz.sources.gov import GovExtra, GovSource
 from michraz.sources.rmi import RmiExtra, RmiSource
 
-__all__ = ["BaseSource", "BeerShevaExtra", "BeerShevaSource", "RmiExtra", "RmiSource", "Source", "SourceName"]
+__all__ = [
+    "BaseSource",
+    "BeerShevaExtra",
+    "BeerShevaSource",
+    "GovExtra",
+    "GovSource",
+    "RmiExtra",
+    "RmiSource",
+    "Source",
+    "SourceName",
+]

src/michraz/sources/base.py

@@ -22,6 +22,7 @@ class SourceName(StrEnum):
 
     RMI = "rmi"
     BEERSHEVA = "beersheva"
+    GOV = "gov"
 
 
 @runtime_checkable