replace glossary_input() with st.file_uploader

better enum for translation models
dont return empty string from file_uploader

daras_ai_v2/asr.py

@@ -272,15 +272,21 @@ class AsrOutputFormat(Enum):
     vtt = "VTT"
 
 
-class TranslationModels(Enum):
-    google = "Google Translate"
-    ghana_nlp = "Ghana NLP"
+class TranslationModel(typing.NamedTuple):
+    label: str
+    supports_glossary: bool = False
+    supports_auto_detect: bool = False
 
-    def supports_glossary(self) -> bool:
-        return self in {self.google}
 
-    def supports_auto_detect(self) -> bool:
-        return self in {self.google}
+class TranslationModels(TranslationModel, Enum):
+    google = TranslationModel(
+        label="Google Translate",
+        supports_glossary=True,
+        supports_auto_detect=True,
+    )
+    ghana_nlp = TranslationModel(
+        label="Ghana NLP Translate",
+    )
 
 
 def translation_language_selector(
@@ -305,7 +311,7 @@ def translation_language_selector(
     return st.selectbox(
         label=label,
         key=key,
-        format_func=lambda k: languages[k] if k else BLANK_OPTION,
+        format_func=lang_format_func,
         options=options,
         **kwargs,
     )
@@ -1071,4 +1077,4 @@ def format_timestamp(seconds: float, always_include_hours: bool, decimal_marker:
 
 
 def should_translate_lang(code: str) -> bool:
-    return code and not code.split("-")[0] != "en"
+    return code and code.split("-")[0] != "en"

daras_ai_v2/doc_search_settings_widgets.py

@@ -33,12 +33,11 @@ def is_user_uploaded_url(url: str) -> bool:
     return _user_media_url_prefix in url
 
 
-def document_uploader(
+def bulk_documents_uploader(
     label: str,
     key: str = "documents",
     accept: typing.Iterable[str] = None,
-    accept_multiple_files=True,
-) -> list[str] | str:
+) -> list[str]:
     st.write(label, className="gui-input")
     documents = st.session_state.get(key) or []
     if isinstance(documents, str):
@@ -47,12 +46,8 @@ def document_uploader(
     if st.session_state.get(f"__custom_checkbox_{key}"):
         if not custom_key in st.session_state:
             st.session_state[custom_key] = "\n".join(documents)
-        if accept_multiple_files:
-            widget = st.text_area
-            kwargs = dict(height=150)
-        else:
-            widget = st.text_input
-            kwargs = {}
+        widget = st.text_area
+        kwargs = dict(height=150)
         text_value = widget(
             label,
             key=custom_key,
@@ -66,28 +61,24 @@ def document_uploader(
             },
             **kwargs,
         )
-        if accept_multiple_files:
-            st.session_state[key] = filter(None, text_value.strip().splitlines())
-        else:
-            st.session_state[key] = text_value
+        st.session_state[key] = filter(None, text_value.strip().splitlines())
     else:
         st.session_state.pop(custom_key, None)
         st.file_uploader(
             label,
             label_visibility="collapsed",
             key=key,
             accept=accept,
-            accept_multiple_files=accept_multiple_files,
+            accept_multiple_files=True,
         )
     st.checkbox("Submit Links in Bulk", key=f"__custom_checkbox_{key}")
-    documents = st.session_state.get(key, [])
-    if accept_multiple_files:
-        try:
-            documents = list(_expand_gdrive_folders(documents))
-        except Exception as e:
-            capture_exception(e)
-            st.error(f"Error expanding gdrive folders: {e}")
-    st.session_state[key] = documents
+    documents = st.session_state.setdefault(key, [])
+    try:
+        documents = list(_expand_gdrive_folders(documents))
+        st.session_state[key] = documents
+    except Exception as e:
+        capture_exception(e)
+        st.error(f"Error expanding gdrive folders: {e}")
     st.session_state[custom_key] = "\n".join(documents)
     return documents
 

daras_ai_v2/glossary.py

@@ -1,10 +1,5 @@
 from daras_ai_v2.asr import google_translate_target_languages
 
-from daras_ai_v2.doc_search_settings_widgets import (
-    document_uploader,
-    SUPPORTED_SPREADSHEET_TYPES,
-)
-
 
 def validate_glossary_document(document: str):
     """
@@ -40,18 +35,6 @@ def validate_glossary_document(document: str):
                 )
 
 
-def glossary_input(
-    label: str = "##### Glossary",
-    key: str = "glossary_document",
-) -> str:
-    return document_uploader(
-        label=label,
-        key=key,
-        accept=SUPPORTED_SPREADSHEET_TYPES,
-        accept_multiple_files=False,
-    )  # type: ignore
-
-
 def create_glossary(
     *,
     language_codes: list[str],

daras_ai_v2/text_output_widget.py

@@ -7,7 +7,7 @@ def text_outputs(
     label: str,
     *,
     key: str = None,
-    value: str = None,
+    value: str | list = None,
     height: int = 200,
 ):
     value = value or st.session_state.get(key)

gooey_ui/components/__init__.py

@@ -577,7 +577,7 @@ def file_uploader(
     label_visibility: LabelVisibility = "visible",
     upload_meta: dict = None,
     optional: bool = False,
-) -> str | list[str]:
+) -> str | list[str] | None:
     if label_visibility != "visible":
         label = None
     key = upload_key or key
@@ -595,14 +595,14 @@ def file_uploader(
             label, value=bool(state.session_state.get(key, value)), disabled=disabled
         ):
             state.session_state.pop(key, None)
-            return ""
+            return None
         label = None
     value = state.session_state.setdefault(key, value)
     if not value:
         if accept_multiple_files:
             value = []
         else:
-            value = ""
+            value = None
     state.session_state[key] = value
     state.RenderTreeNode(
         name="input",
@@ -618,7 +618,7 @@ def file_uploader(
             uploadMeta=upload_meta,
         ),
     ).mount()
-    return value or ""
+    return value
 
 
 def json(value: typing.Any, expanded: bool = False, depth: int = 1):

recipes/BulkEval.py

@@ -13,7 +13,7 @@
 from daras_ai.image_input import upload_file_from_bytes
 from daras_ai_v2.base import BasePage
 from daras_ai_v2.doc_search_settings_widgets import (
-    document_uploader,
+    bulk_documents_uploader,
     SUPPORTED_SPREADSHEET_TYPES,
 )
 from daras_ai_v2.field_render import field_title_desc
@@ -199,7 +199,7 @@ class ResponseModel(BaseModel):
         aggregations: list[list[AggFunctionResult]] | None
 
     def render_form_v2(self):
-        files = document_uploader(
+        files = bulk_documents_uploader(
             f"##### {field_title_desc(self.RequestModel, 'documents')}",
             accept=SUPPORTED_SPREADSHEET_TYPES,
         )

recipes/BulkRunner.py

@@ -11,7 +11,7 @@
 from daras_ai_v2.base import BasePage
 from daras_ai_v2.breadcrumbs import get_title_breadcrumbs
 from daras_ai_v2.doc_search_settings_widgets import (
-    document_uploader,
+    bulk_documents_uploader,
     SUPPORTED_SPREADSHEET_TYPES,
 )
 from daras_ai_v2.field_render import field_title_desc
@@ -104,7 +104,7 @@ def render_form_v2(self):
             flatten_dict_key="url",
         )
 
-        files = document_uploader(
+        files = bulk_documents_uploader(
             f"---\n##### {field_title_desc(self.RequestModel, 'documents')}",
             accept=SUPPORTED_SPREADSHEET_TYPES,
         )

recipes/DocExtract.py

@@ -2,12 +2,13 @@
 import threading
 import typing
 
+from daras_ai_v2.field_render import field_title_desc
 from daras_ai_v2.pydantic_validation import FieldHttpUrl
 import requests
 from aifail import retry_if
 from django.db.models import IntegerChoices
 from furl import furl
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 import gooey_ui as st
 from bots.models import Workflow
@@ -25,7 +26,10 @@
     azure_doc_extract_page_num,
 )
 from daras_ai_v2.base import BasePage
-from daras_ai_v2.doc_search_settings_widgets import document_uploader
+from daras_ai_v2.doc_search_settings_widgets import (
+    bulk_documents_uploader,
+    SUPPORTED_SPREADSHEET_TYPES,
+)
 from daras_ai_v2.enum_selector_widget import enum_selector
 from daras_ai_v2.exceptions import raise_for_status
 from daras_ai_v2.fake_user_agents import FAKE_USER_AGENTS
@@ -34,7 +38,6 @@
     flatapply_parallel,
 )
 from daras_ai_v2.gdrive_downloader import is_gdrive_url, gdrive_download
-from daras_ai_v2.glossary import glossary_input
 from daras_ai_v2.language_model import run_language_model, LargeLanguageModels
 from daras_ai_v2.language_model_settings_widgets import language_model_settings
 from daras_ai_v2.loom_video_widget import youtube_video
@@ -81,7 +84,11 @@ class RequestModel(BaseModel):
         selected_asr_model: typing.Literal[tuple(e.name for e in AsrModels)] | None
         # language: str | None
         google_translate_target: str | None
-        glossary_document: FieldHttpUrl | None
+        glossary_document: FieldHttpUrl | None = Field(
+            title="Translation Glossary",
+            description="""Provide a glossary to customize translation and improve accuracy of domain-specific terms.
+If not specified or invalid, no glossary will be used. Read about the expected format [here](https://docs.google.com/document/d/1TwzAvFmFYekloRKql2PXNPIyqCbsHRL8ZtnWkzAYrh8/edit?usp=sharing).""",
+        )
 
         task_instructions: str | None
 
@@ -101,7 +108,7 @@ def preview_image(self, state: dict) -> str | None:
         return DEFAULT_YOUTUBE_BOT_META_IMG
 
     def render_form_v2(self):
-        document_uploader(
+        bulk_documents_uploader(
             "#### 🤖 Youtube/PDF/Drive URLs",
             accept=("audio/*", "application/pdf", "video/*"),
         )
@@ -132,15 +139,17 @@ def render_settings(self):
             height=300,
         )
         language_model_settings()
-        "##### Document AI Model"
+
         enum_selector(AsrModels, label="##### ASR Model", key="selected_asr_model")
         st.write("---")
+
         google_translate_language_selector()
-        glossary_input()
+        st.file_uploader(
+            label=f"###### {field_title_desc(self.RequestModel, 'glossary_document')}",
+            key="glossary_document",
+            accept=SUPPORTED_SPREADSHEET_TYPES,
+        )
         st.write("---")
-        # enum_selector(
-        #     AsrOutputFormat, label="###### Output Format", key="output_format"
-        # )
 
     def related_workflows(self) -> list:
         from recipes.asr_page import AsrPage

recipes/DocSearch.py

@@ -8,7 +8,7 @@
 from bots.models import Workflow
 from daras_ai_v2.base import BasePage
 from daras_ai_v2.doc_search_settings_widgets import (
-    document_uploader,
+    bulk_documents_uploader,
     is_user_uploaded_url,
     citation_style_selector,
     doc_search_advanced_settings,
@@ -93,7 +93,7 @@ def get_example_preferred_fields(self, state: dict) -> list[str]:
 
     def render_form_v2(self):
         st.text_area("#### Search Query", key="search_query")
-        document_uploader("#### Documents")
+        bulk_documents_uploader("#### Documents")
         prompt_vars_widget("task_instructions", "query_instructions")
 
     def validate_form_v2(self):

recipes/DocSummary.py

@@ -9,7 +9,7 @@
 from daras_ai_v2.asr import AsrModels
 from daras_ai_v2.base import BasePage
 from daras_ai_v2.doc_search_settings_widgets import (
-    document_uploader,
+    bulk_documents_uploader,
 )
 from daras_ai_v2.functional import map_parallel
 from daras_ai_v2.language_model import (
@@ -91,7 +91,7 @@ def preview_image(self, state: dict) -> str | None:
         return DEFAULT_DOC_SUMMARY_META_IMG
 
     def render_form_v2(self):
-        document_uploader("#### 📎 Documents")
+        bulk_documents_uploader("#### 📎 Documents")
         st.text_area("#### 👩‍💻 Instructions", key="task_instructions")
 
     def render_settings(self):