better inference

introduck/inference.py

@@ -1,3 +1,4 @@
+import pandas as pd
 import spacy
 
 
@@ -13,71 +14,19 @@ def _load_spacy_model() -> spacy.Language:
     return _load_spacy_model.nlp
 
 
-def extract_contacts_from_email(payload: str) -> list[list[str]]:
+def extract_contacts_from_text(payload: str) -> dict:
     if not payload:
-        return []
+        return {}
 
     nlp: spacy.Language = _load_spacy_model()
     doc = nlp(payload)
 
-    contacts: list[list[str]] = []
-    for ent in doc.ents:
-        if ent.label_ != "PERSON":
-            continue
-
-        contacts.append([ent.text, "unknown", "unknown", "no notes"])
-
-    return contacts
-
-
-def generate_acceptance_reply(payload: str) -> str:
-    if not payload:
-        return ""
-
-    msg: str = """\
-Hi,
-
-Thanks for reaching out! I will be glad to make an intro to {PERSON or COMPANY}!
-
-Looking forward to talking to you,
-{SIGNATURE}\
-"""
-
-    return msg
-
-
-def generate_rejection_reply(payload: str) -> str:
-    if not payload:
-        return ""
+    contacts_cols: list[str] = ["name", "email"]
+    contacts: pd.DataFrame = pd.DataFrame(columns=contacts_cols)
 
-    msg: str = """\
-Hi,
-
-Thanks for reaching out!
-
-I'm not able to make an intro at this time,
-but I'll keep your company in mind for the future.
-
-Thanks again,
-{SIGNATURE}\
-"""
-
-    return msg
-
-
-def highlight_named_entities(payload: str, labels: list[str] = None) -> dict:
-    if not payload:
-        return {"text": "", "entities": []}
-
-    if labels is None:
-        labels = ["ORG", "PERSON"]
-
-    nlp: spacy.Language = _load_spacy_model()
-    doc = nlp(payload)
-
-    entities: list = []
+    entities: list[dict] = []
     for ent in doc.ents:
-        if ent.label_ not in labels:
+        if ent.label_ != "PERSON":
             continue
 
         entity: dict = {
@@ -87,4 +36,12 @@ def highlight_named_entities(payload: str, labels: list[str] = None) -> dict:
 
         entities.append(entity)
 
-    return {"text": payload, "entities": entities}
+        contact: dict = {
+            "name": ent.text,
+            "email": "unknown"}
+
+        contacts = pd.concat(
+            [contacts, pd.DataFrame([contact])],
+            ignore_index=True)
+
+    return {"contacts": contacts, "entities": entities, "payload": payload}

introduck/routes.py

@@ -1,13 +1,10 @@
 import gradio as gr
+import pandas as pd
 
 from fastapi import FastAPI
 from gradio.routes import App as GradioApp
-
-from introduck.inference import extract_contacts_from_email
-from introduck.inference import generate_acceptance_reply
-from introduck.inference import generate_rejection_reply
-from introduck.inference import highlight_named_entities
-from introduck.utils import dump_email_as_string
+from introduck.inference import extract_contacts_from_text
+from introduck.utils import dump_email_to_string
 
 _INTRO_SUBJECT_EXAMPLE: str = "Could you make an intro?"
 _INTRO_MESSAGE_EXAMPLE: str = """\
@@ -34,27 +31,70 @@ Best,
 """
 
 
-def _analyze_message(sender: str, recipients: str, subject: str, body: str):
+def _analyze_message(
+    sender: str,
+    recipients: str,
+    subject: str,
+    body: str
+) -> (pd.DataFrame, dict):
+    default_outputs = None, None
+
+    # TODO: validate sender and recipients
+
+    if not subject:
+        return default_outputs
+
+    if not body:
+        return default_outputs
+
     msg_data: dict = {
         "from": sender,
         "to": recipients,
         "subject": subject,
         "body": body}
 
-    msg: str = dump_email_as_string(data=msg_data)
+    msg: str = dump_email_to_string(data=msg_data)
 
-    contacts: list = extract_contacts_from_email(payload=msg)
-    acceptance_reply: str = generate_acceptance_reply(payload=msg)
-    rejection_reply: str = generate_rejection_reply(payload=msg)
-    highlighted_text: dict = highlight_named_entities(payload=msg)
+    output: dict = extract_contacts_from_text(payload=msg)
 
-    return contacts, acceptance_reply, rejection_reply, highlighted_text
+    contacts: pd.DataFrame = output.get("contacts", pd.DataFrame())
+    highlighted_text: dict = {
+        "text": output.get("payload", str),
+        "entities": output.get("entities", [])}
+
+    return contacts, highlighted_text
 
 
 def _use_message_template() -> (str, str):
     return _INTRO_SUBJECT_EXAMPLE, _INTRO_MESSAGE_EXAMPLE
 
 
+def _validate_message(
+    sender: str,
+    recipients: str,
+    subject: str,
+    body: str
+) -> dict:
+    errors: list[str] = []
+
+    # ...
+
+    if not subject:
+        errors.append("- subject can't be empty;")
+
+    if not body:
+        errors.append("- message can't be empty;")
+
+    new_value: str = ""
+    if errors:
+        errors_list: str = "\n".join(errors)
+        new_value = f"```\nPlease, fix this errors:\n{errors_list}\n```"
+
+    new_visibility: bool = len(errors) > 0
+
+    return gr.update(value=new_value, visible=new_visibility)
+
+
 def create_playground_route() -> FastAPI:
     # TODO: Fix once resolved https://github.com/gradio-app/gradio/issues/1683
     # TODO: ...and remove elem_id="htxt" from HighlightedText for RF822 output
@@ -109,6 +149,10 @@ def create_playground_route() -> FastAPI:
         #     label="Attachments:",
         #     file_count="multiple")
 
+        email_errors_output: gr.Markdown = gr.Markdown(
+            value="",
+            visible=False)
+
         with gr.Row():
             email_template_button: gr.Button = gr.Button(
                 value="Use template",
@@ -128,30 +172,11 @@ def create_playground_route() -> FastAPI:
 
         with gr.Tabs(selected="default"):
             with gr.TabItem(label="Contacts", id="default"):
-                contacts_table_headers: list[str] = [
-                    "Name",
-                    "Company",
-                    "Email",
-                    "NOTE"]
-
-                contacts_output: gr.Dataframe = gr.Dataframe(
-                    headers=contacts_table_headers,
-                    max_cols=len(contacts_table_headers),
-                    max_rows=16)
-
-            with gr.TabItem(label="Replies"):
-                with gr.Column():
-                    acceptance_template_output: gr.Textbox = gr.Textbox(
-                        label="Use this message to accept an intro request:",
-                        interactive=True)
-
-                    rejection_template_output: gr.Textbox = gr.Textbox(
-                        label="Use this message to decline an intro request:",
-                        interactive=True)
-
-            with gr.TabItem(label="RFC822"):
+                contacts_output: gr.Dataframe = gr.Dataframe(max_rows=16)
+
+            with gr.TabItem(label="RFC822-ish"):
                 rfc822_output: gr.HighlightedText = gr.HighlightedText(
-                    label="RFC822 message (for debugging purposes):",
+                    label="RFC822-ish message (for debugging purposes):",
                     show_legend=False,
                     elem_id="htxt")  # temporary fix, see above for more info
 
@@ -174,11 +199,6 @@ def create_playground_route() -> FastAPI:
 
         # ui end, continue with signals and slots...
 
-        email_template_button.click(
-            fn=_use_message_template,
-            inputs=[],
-            outputs=[email_subject_input, email_body_input])
-
         email_submit_button.click(
             fn=_analyze_message,
             inputs=[
@@ -188,8 +208,20 @@ def create_playground_route() -> FastAPI:
                 email_body_input],
             outputs=[
                 contacts_output,
-                acceptance_template_output,
-                rejection_template_output,
                 rfc822_output])
 
+        email_submit_button.click(
+            fn=_validate_message,
+            inputs=[
+                email_sender_input,
+                email_recipients_input,
+                email_subject_input,
+                email_body_input],
+            outputs=[email_errors_output])
+
+        email_template_button.click(
+            fn=_use_message_template,
+            inputs=[],
+            outputs=[email_subject_input, email_body_input])
+
     return GradioApp.create_app(blocks=base_blocks)

introduck/utils.py

@@ -1,4 +1,4 @@
-def dump_email_as_string(data: dict) -> str:
+def dump_email_to_string(data: dict) -> str:
     msg_from: str = data.get("from", "") or "*"
     msg_to: str = data.get("to", "") or "*"
     msg_subject: str = data.get("subject", "") or "*"
@@ -11,3 +11,8 @@ def dump_email_as_string(data: dict) -> str:
     msg += f"{msg_body}\n"
 
     return msg
+
+
+def validate_email_address(data: str) -> bool:
+    # TODO: https://pypi.org/project/email-validator/
+    return False

requirements.txt

@@ -1,5 +1,6 @@
 fastapi[all]
 gradio==3.4.0  # keep in sync with version from README.md metadata
+pandas==1.5.0
 spacy==3.4.0
 uvicorn[standard]