chore: sync to unstract-client 1.7.1 and llmwhisperer-client 2.9.1

pyproject.toml

@@ -29,8 +29,8 @@ dependencies = [
     # Pinned exactly: the CLI derives its flags and help text from these
     # clients, so one that moves changes the CLI's surface. Each release re-pins
     # deliberately, against the specs vendored in `src/unstract_cli/specs`.
-    "unstract-client==1.7.0",
-    "llmwhisperer-client==2.9.0",
+    "unstract-client==1.7.1",
+    "llmwhisperer-client==2.9.1",
     # Both clients pull this in, but the CLI imports its exception classes
     # directly to classify a failure, so it names the dependency itself.
     "requests>=2.32.3",

src/unstract_cli/specs/docstudio.json

@@ -65,6 +65,14 @@
             "readOnly": true,
             "type": "string"
           },
+          "owner_emails": {
+            "description": "Email of each owner, earliest first. Empty if none is a person.",
+            "items": {
+              "type": "string"
+            },
+            "readOnly": true,
+            "type": "array"
+          },
           "run_count": {
             "readOnly": true,
             "type": "integer"
@@ -91,6 +99,7 @@
           "is_owner",
           "last_5_run_statuses",
           "last_run_time",
+          "owner_emails",
           "run_count",
           "workflow",
           "workflow_name"
@@ -230,10 +239,6 @@
             "maximum": 300,
             "minimum": -1,
             "type": "integer"
-          },
-          "use_file_history": {
-            "default": false,
-            "type": "boolean"
           }
         },
         "type": "object"

src/unstract_cli/specs/llmwhisperer.json

@@ -1549,7 +1549,8 @@
             "schema": {
               "default": 10.0,
               "type": "number"
-            }
+            },
+            "x-internal": true
           },
           {
             "in": "query",
@@ -1585,7 +1586,8 @@
             "schema": {
               "default": false,
               "type": "boolean"
-            }
+            },
+            "x-internal": true
           },
           {
             "in": "query",
@@ -1662,7 +1664,8 @@
             "schema": {
               "default": 0.0,
               "type": "number"
-            }
+            },
+            "x-internal": true
           },
           {
             "in": "query",
@@ -1671,8 +1674,6 @@
             "schema": {
               "default": "form",
               "enum": [
-                "document_insights",
-                "excel",
                 "form",
                 "high_quality",
                 "low_cost",
@@ -1689,9 +1690,7 @@
             "schema": {
               "default": "layout_preserving",
               "enum": [
-                "dump-text",
                 "layout_preserving",
-                "line-printer",
                 "text"
               ],
               "type": "string"

src/unstract_cli/specs/provenance.json

@@ -1,16 +1,16 @@
 {
   "docstudio.json": {
-    "client": "unstract-client==1.7.0",
+    "client": "unstract-client==1.7.1",
     "repo": "https://github.com/Zipstack/unstract",
-    "commit": "520b98d7acf5a6d138b24d9787bd788b89150b76",
+    "commit": "5d46c66ccff8ae1ac3c31a3c3d5d64206e329002",
     "path": "specs/docstudio-oss.json",
-    "sha256": "68a31eaf72e54daf8173ae3ef42e4174b257c93b1afc608114582a81be135bd1"
+    "sha256": "c0de3a14254600f436483561ebb737feac4023fdef2fd46d6abf15dcd2182ad9"
   },
   "llmwhisperer.json": {
-    "client": "llmwhisperer-client==2.9.0",
+    "client": "llmwhisperer-client==2.9.1",
     "repo": "https://github.com/Zipstack/unstract-llm-whisperer",
-    "commit": "750f941ee229e12cc05d8bd85edaab6a337a8758",
+    "commit": "fa6cf651a594227f2249dfe0990f80f1ff780398",
     "path": "specs/llmwhisperer.json",
-    "sha256": "88ecc01e92443ba5ba6079db7f57cf3038f97670cb796f13c326268a3d79f366"
+    "sha256": "c015d035b9e06f58e80aabdb5190fbb582dc1b4b8ee48c3a6924fe4392edc650"
   }
 }

tests/derived_flags.json

@@ -33,17 +33,6 @@
       "choices": [],
       "short": null
     },
-    "--derotate-threshold": {
-      "name": "derotate_threshold",
-      "type": "number",
-      "default": null,
-      "description": "The page rotation in degrees beyond which the page is straightened and re-read.",
-      "array": false,
-      "nullable": false,
-      "required": false,
-      "choices": [],
-      "short": null
-    },
     "--file-name": {
       "name": "file_name",
       "type": "string",
@@ -77,17 +66,6 @@
       "choices": [],
       "short": null
     },
-    "--ignore-vertical-text": {
-      "name": "ignore_vertical_text",
-      "type": "boolean",
-      "default": null,
-      "description": "Whether to drop vertically oriented text instead of extracting it.",
-      "array": false,
-      "nullable": false,
-      "required": false,
-      "choices": [],
-      "short": null
-    },
     "--include-line-confidence": {
       "name": "include_line_confidence",
       "type": "boolean",
@@ -169,17 +147,6 @@
       "choices": [],
       "short": null
     },
-    "--min-table-width": {
-      "name": "min_table_width",
-      "type": "number",
-      "default": null,
-      "description": "The minimum width a table must span, as a fraction of the page width, to be extracted as a table.",
-      "array": false,
-      "nullable": false,
-      "required": false,
-      "choices": [],
-      "short": null
-    },
     "--mode": {
       "name": "mode",
       "type": "string",
@@ -189,8 +156,6 @@
       "nullable": false,
       "required": false,
       "choices": [
-        "document_insights",
-        "excel",
         "form",
         "high_quality",
         "low_cost",
@@ -208,9 +173,7 @@
       "nullable": false,
       "required": false,
       "choices": [
-        "dump-text",
         "layout_preserving",
-        "line-printer",
         "text"
       ],
       "short": null
@@ -449,17 +412,6 @@
       "required": false,
       "choices": [],
       "short": null
-    },
-    "--use-file-history": {
-      "name": "use_file_history",
-      "type": "boolean",
-      "default": false,
-      "description": "Reuse a previous result for the same file.",
-      "array": false,
-      "nullable": false,
-      "required": false,
-      "choices": [],
-      "short": null
     }
   },
   "docstudio:status": {

tests/test_commands.py

@@ -191,6 +191,34 @@ def test_the_v1_commands_are_registered():
     assert set(tree["auth"]["commands"]) == {"login", "whoami"}
 
 
+@pytest.mark.parametrize(
+    ("command", "flag"),
+    [
+        (("whisper", "extract"), "--derotate-threshold"),
+        (("whisper", "extract"), "--min-table-width"),
+        (("whisper", "extract"), "--ignore-vertical-text"),
+        (("docstudio", "deployment", "run"), "--use-file-history"),
+    ],
+)
+def test_a_withdrawn_option_is_unknown_to_the_shipped_command(capsys, command, flag):
+    """The services stopped publishing these; a caller still passing one must
+    learn that before the request, not from a silently ignored setting."""
+    code, out, _ = run(capsys, *command, "x.pdf", flag)
+    assert code == int(ExitCode.USAGE)
+    message = envelope(out)["error"]["message"]
+    assert "No such option" in message and flag in message
+
+
+@pytest.mark.parametrize(
+    ("flag", "value"),
+    [("--output-mode", "line-printer"), ("--mode", "excel")],
+)
+def test_a_withdrawn_choice_is_refused_by_the_shipped_command(capsys, flag, value):
+    code, out, _ = run(capsys, "whisper", "extract", "x.pdf", flag, value)
+    assert code == int(ExitCode.USAGE)
+    assert f"'{value}' is not one of" in envelope(out)["error"]["message"]
+
+
 # --------------------------------------------------------------------------- #
 # whisper extract
 # --------------------------------------------------------------------------- #

tests/test_discover.py

@@ -58,8 +58,6 @@ def test_full_carries_enough_to_build_a_call(capsys):
 
     assert params["source"]["kind"] == "argument" and params["source"]["required"]
     assert params["mode"]["choices"] == [
-        "document_insights",
-        "excel",
         "form",
         "high_quality",
         "low_cost",

tests/test_params.py

@@ -273,7 +273,7 @@ def test_choices_come_from_the_spec_unless_the_overlay_narrows_them():
     """A wrong value must fail before the request, not after -- and the list it
     is checked against is the service's own, not a copy that can fall behind."""
     spec_declared = _by_name(operation_params("llmwhisperer", "extract"))["mode"]
-    assert "excel" in spec_declared.choices
+    assert "form" in spec_declared.choices
     assert click_option(spec_declared, {}).type.choices == spec_declared.choices
 
     option = click_option(spec_declared, {"mode": {"choices": ["form", "table"]}})