From 2bd73332337bd3f0ec5af06aa247d8a86bb5df6b Mon Sep 17 00:00:00 2001
From: lehel <lehel@upvest.co>
Date: Wed, 8 Oct 2025 15:08:35 +0200
Subject: [PATCH] new table + translate call

---
 config.go                                        |  1 +
 config/config.yaml                               |  1 +
 llm.go                                           |  1 +
 .../0003_create_sentence_embeddings_table.up.sql | 16 ++++++++++++++++
 ollama_client.go                                 | 16 ++++++++++++++++
 openai_client.go                                 | 16 ++++++++++++++++
 6 files changed, 51 insertions(+)
 create mode 100644 migrations/0003_create_sentence_embeddings_table.up.sql

diff --git a/config.go b/config.go
index df22298..0cb9568 100644
--- a/config.go
+++ b/config.go
@@ -8,6 +8,7 @@ type Config struct {
 	LLM struct {
 		ExtractKeywordsPrompt string `yaml:"extract_keywords_prompt"`
 		DisambiguatePrompt    string `yaml:"disambiguate_prompt"`
+		TranslatePrompt       string `yaml:"translate_prompt"`
 	} `yaml:"llm"`
 }
 
diff --git a/config/config.yaml b/config/config.yaml
index 7bd8c92..5d42153 100644
--- a/config/config.yaml
+++ b/config/config.yaml
@@ -1,3 +1,4 @@
 llm:
   extract_keywords_prompt: "You will extract structured data from the user input. Input text: {{.Message}}. Return ONLY valid minified JSON object with keys: translate (English translation of input), keyword (array of 3-5 concise English veterinary-related keywords derived strictly from the input), animal (animal mentioned or 'unknown'). Example: {\"translate\":\"dog has diarrhea\",\"keyword\":[\"diarrhea\",\"digestive\"],\"animal\":\"dog\"}. Do not add extra text, markdown, or quotes outside JSON."
   disambiguate_prompt: "Given candidate visit entries (JSON array): {{.Entries}} and user message: {{.Message}} choose the best matching visit's ID. Return ONLY JSON: {\"visitReason\":\"<one of the candidate IDs or empty string if none>\"}. No other text."
+  translate_prompt: "Translate the following veterinary-related sentence to English. Input: '{{.Message}}'. Return ONLY the English translation, no extra text, no markdown, no quotes. If already English, return as is."
diff --git a/llm.go b/llm.go
index 837de18..7311786 100644
--- a/llm.go
+++ b/llm.go
@@ -15,6 +15,7 @@ type LLMClientAPI interface {
 	ExtractKeywords(ctx context.Context, message string) (map[string]interface{}, error)
 	DisambiguateBestMatch(ctx context.Context, message string, candidates []Visit) (string, error)
 	GetEmbeddings(ctx context.Context, input string) ([]float64, error)
+	TranslateToEnglish(ctx context.Context, message string) (string, error)
 }
 
 // --- Format Utilities ---
diff --git a/migrations/0003_create_sentence_embeddings_table.up.sql b/migrations/0003_create_sentence_embeddings_table.up.sql
new file mode 100644
index 0000000..36c4990
--- /dev/null
+++ b/migrations/0003_create_sentence_embeddings_table.up.sql
@@ -0,0 +1,16 @@
+-- +goose Up
+CREATE EXTENSION IF NOT EXISTS vector;
+
+CREATE TABLE sentence_embeddings (
+    id SERIAL PRIMARY KEY,
+    visit_id INTEGER NOT NULL,
+    sentence TEXT NOT NULL,
+    translated TEXT,
+    embeddings VECTOR(1536) NOT NULL,
+    created_at TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP
+);
+
+-- +goose Down
+DROP TABLE IF EXISTS sentence_embeddings;
+DROP EXTENSION IF EXISTS vector;
+
diff --git a/ollama_client.go b/ollama_client.go
index a455c92..0b4d786 100644
--- a/ollama_client.go
+++ b/ollama_client.go
@@ -166,3 +166,19 @@ func (llm *OllamaClient) GetEmbeddings(ctx context.Context, input string) ([]flo
 	}
 	return nil, fmt.Errorf("unrecognized embedding response: %.200s", string(raw))
 }
+
+func (llm *OllamaClient) TranslateToEnglish(ctx context.Context, message string) (string, error) {
+	prompt, err := renderPrompt(appConfig.LLM.TranslatePrompt, map[string]string{"Message": message})
+	if err != nil {
+		logrus.WithError(err).Error("[CONFIG] Failed to render Translate prompt")
+		return "", err
+	}
+	logrus.WithField("prompt", prompt).Info("[LLM] TranslateToEnglish prompt")
+
+	resp, err := llm.ollamaCompletion(ctx, prompt, nil)
+	logrus.WithFields(logrus.Fields{"response": resp, "err": err}).Info("[LLM] TranslateToEnglish response")
+	if err != nil {
+		return resp, err
+	}
+	return strings.TrimSpace(resp), nil
+}
diff --git a/openai_client.go b/openai_client.go
index 43a9f29..cce9251 100644
--- a/openai_client.go
+++ b/openai_client.go
@@ -198,3 +198,19 @@ func (llm *OpenAIClient) GetEmbeddings(ctx context.Context, input string) ([]flo
 	}
 	return nil, fmt.Errorf("unrecognized embedding response: %.200s", string(raw))
 }
+
+func (llm *OpenAIClient) TranslateToEnglish(ctx context.Context, message string) (string, error) {
+	prompt, err := renderPrompt(appConfig.LLM.TranslatePrompt, map[string]string{"Message": message})
+	if err != nil {
+		logrus.WithError(err).Error("[CONFIG] Failed to render Translate prompt")
+		return "", err
+	}
+	logrus.WithField("prompt", prompt).Info("[LLM] TranslateToEnglish prompt")
+
+	resp, err := llm.openAICompletion(ctx, prompt, nil)
+	logrus.WithFields(logrus.Fields{"response": resp, "err": err}).Info("[LLM] TranslateToEnglish response")
+	if err != nil {
+		return resp, err
+	}
+	return strings.TrimSpace(resp), nil
+}