Merge pull request 'v0.3.0' (#5) from v0.3.0 into main

Reviewed-on: #5

Project.toml

@@ -1,7 +1,7 @@
 name = "YiemAgent"
 uuid = "e012c34b-7f78-48e0-971c-7abb83b6f0a2"
 authors = ["narawat lamaiin <narawat@outlook.com>"]
-version = "0.2.0"
+version = "0.3.0"
 
 [deps]
 CSV = "336ed68f-0bac-5ca0-87d4-7b16caf5d00b"

src/llmfunction.jl

@@ -1,10 +1,10 @@
 module llmfunction
 
-export  virtualWineUserChatbox, jsoncorrection, checkinventory, # recommendbox,
+export  virtualWineUserChatbox, jsoncorrection, checkwine, # recommendbox,
         virtualWineUserRecommendbox, userChatbox, userRecommendbox, extractWineAttributes_1,
         extractWineAttributes_2, paraphrase
 
-using HTTP, JSON3, URIs, Random, PrettyPrinting, UUIDs, Dates
+using HTTP, JSON3, URIs, Random, PrettyPrinting, UUIDs, Dates, DataFrames
 using GeneralUtils, SQLLLM
 using ..type, ..util
 
@@ -288,23 +288,46 @@ julia> result = checkinventory(agent, input)
 
 # Signature
 """
-function checkinventory(a::T1, input::T2
+function checkwine(a::T1, input::T2; maxattempt::Int=3
   ) where {T1<:agent, T2<:AbstractString}
 
   println("\ncheckinventory order: $input ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
   wineattributes_1 = extractWineAttributes_1(a, input)
   wineattributes_2 = extractWineAttributes_2(a, input)
 
-  _inventoryquery = "retailer name: $(a.retailername), $wineattributes_1, $wineattributes_2"
-  inventoryquery = "Retrieves winery, wine_name, wine_id, vintage, region, country, wine_type, grape, serving_temperature, sweetness, intensity, tannin, acidity, tasting_notes, price and currency of wines that match the following criteria - {$_inventoryquery}"
-  println("\ncheckinventory input: $inventoryquery ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-  # add suppport for similarSQLVectorDB
-  textresult, rawresponse = SQLLLM.query(inventoryquery, a.func[:executeSQL], 
-                        a.func[:text2textInstructLLM];
-                        insertSQLVectorDB=a.func[:insertSQLVectorDB], 
-                        similarSQLVectorDB=a.func[:similarSQLVectorDB],
-                        llmFormatName="qwen3")
+  # placeholder
+  textresult = nothing
+  rawresponse = nothing
   
+  for i in 1:maxattempt
+
+    #CHANGE if you want to add retailer name
+    # _inventoryquery = "retailer name: $(a.retailername), $wineattributes_1, $wineattributes_2"
+    _inventoryquery = "$wineattributes_1, $wineattributes_2"
+    
+    retrieve_attributes = ["winery", "wine_name", "wine_id", "vintage", "region", "country", "wine_type", "grape", "serving_temperature", "sweetness", "intensity", "tannin", "acidity", "tasting_notes", "price", "currency"]
+    inventoryquery = "Retrieves $retrieve_attributes of wines that match the following criteria - {$_inventoryquery}"
+    println("\ncheckinventory input: $inventoryquery ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    # add suppport for similarSQLVectorDB
+    textresult, rawresponse = SQLLLM.query(inventoryquery, a.func[:executeSQL], 
+                          a.func[:text2textInstructLLM];
+                          insertSQLVectorDB=a.func[:insertSQLVectorDB], 
+                          similarSQLVectorDB=a.func[:similarSQLVectorDB],
+                          llmFormatName="qwen3")
+    # check if all of retrieve_attributes appears in textresult
+    isin = [occursin(x, textresult) for x in retrieve_attributes]
+    # check if rawresponse type is DataFrame so that I can check for column
+    if typeof(rawresponse) == DataFrame && 
+      !occursin("The resulting table has 0 row", textresult) &&
+      !all(isin)
+      
+      errornote = "Not all of $retrieve_attributes appear in search result"
+      println("\nERROR YiemAgent checkwine() $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+      continue
+    else
+      break
+    end
+  end
   println("\ncheckinventory result ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
   println(textresult)
 
@@ -339,150 +362,210 @@ function extractWineAttributes_1(a::T1, input::T2; maxattempt=10
   As a helpful sommelier, your task is to extract the user information from the user's query as much as possible to fill out user's preference form.
 
   At each round of conversation, the user will give you the following:
-  User's query: ...
+   - The query: the query provided by the user.
 
   You must follow the following guidelines:
-    - If specific information required in the preference form is not available in the query or there isn't any, mark with "NA" to indicate this.
+    - If specific information required in the preference form is not available in the query or there isn't any, mark with "N/A" to indicate this.
       Additionally, words like 'any' or 'unlimited' mean no information is available.
     - Do not generate other comments.
 
   You should then respond to the user with:
-    Thought: state your understanding of the current situation
-    Wine_name: name of the wine
-    Winery: name of the winery
-    Vintage: the year of the wine
-    Region: a region (NOT a country) where the wine is produced, such as Burgundy, Napa Valley, etc
-    Country: a country where the wine is produced. Can be "Austria", "Australia", "France", "Germany", "Italy", "Portugal", "Spain", "United States"
-    Wine_type: can be one of: "red", "white", "sparkling", "rose", "dessert" or "fortified"
-    Grape_varietal: the name of the primary grape used to make the wine
-    Tasting_notes: a brief description of the wine's taste, such as "butter", "oak", "fruity", etc
-    Wine_price: price range of wine.
-    Occasion: the occasion the user is having the wine for
-    Food_to_be_paired_with_wine: food that the user will be served with the wine such as poultry, fish, steak, etc
+    wine_name: name of the wine
+    winery: name of the winery
+    vintage: the year of the wine
+    region: a region, such as Burgundy, Bordeaux, Champagne, Napa Valley, Tuscany, California, Oregon, etc
+    country: a country where wine is produced. Can be "Austria", "Australia", "France", "Germany", "Italy", "Portugal", "Spain", "United States"
+    wine_type: can be one of: "red", "white", "sparkling", "rose", "dessert" or "fortified"
+    grape_varietal: the name of the primary grape used to make the wine
+    tasting_notes: a word describe the wine's flavor, such as "butter", "oak", "fruity", "raspberry", "earthy", "floral", etc
+    wine_price_min: minimum price range of wine. Example: For wine price 20, wine_price_min will be 0. For wine price 10 to 100, wine_price_min will be 10.
+    wine_price_max: maximum price range of wine. Example: For wine price 20, wine_price_max will be 20. For wine price 10 to 100, wine_price_max will be 100.
+    occasion: the occasion the user is having the wine for
+    food_to_be_paired_with_wine: food that the user will be served with the wine such as poultry, fish, steak, etc
 
-    You should only respond in format as described below:
-    Thought: ...
-    Wine_name: ...
-    Winery: ...
-    Vintage: ...
-    Region: ...
-    Country: ...
-    Wine_type:
-    Grape_varietal: ...
-    Tasting_notes: ...
-    Wine_price: ...
-    Occasion: ...
-    Food_to_be_paired_with_wine: ...
+    You should only respond in JSON format as described below:
+    {
+      "wine_name": "...",
+      "winery": "...",
+      "vintage": "...",
+      "region": "...",
+      "country": "...",
+      "wine_type": "...",
+      "grape_varietal": "...",
+      "tasting_notes": "...",
+      "wine_price_min": "...",
+      "wine_price_max": "...",
+      "occasion": "...",
+      "food_to_be_paired_with_wine": "..."
+    }
     
     Here are some example:
-    User's query: red, Chenin Blanc, Riesling, 20 USD
-    {"reasoning": ..., "winery": "NA", "wine_name": "NA", "vintage": "NA", "region": "NA", "country": "NA", "wine_type": "red, white", "grape_varietal": "Chenin Blanc, Riesling", "tasting_notes": "NA", "wine_price": "0-20", "occasion": "NA", "food_to_be_paired_with_wine": "NA"}
+    
+    User's query: red, Chenin Blanc, Riesling, 20 USD from Tuscany, Italy or Napa Valley, USA
+    {
+      "wine_name": "N/A",
+      "winery": "N/A",
+      "vintage": "N/A",
+      "region": "Tuscany or Napa Valley",
+      "country": "Italy or United States",
+      "wine_type": "red or white",
+      "grape_varietal": "Chenin Blanc or Riesling",
+      "tasting_notes": "citrus",
+      "wine_price_min": "0",
+      "wine_price_max": "20",
+      "occasion": "N/A",
+      "food_to_be_paired_with_wine": "N/A"
+    }
     
     User's query: Domaine du Collier Saumur Blanc 2019, France, white, Merlot
-    {"reasoning": ..., "winery": "Domaine du Collier", "wine_name": "Saumur Blanc", "vintage": "2019", "region": "Saumur", "country": "France", "wine_type": "white", "grape_varietal": "Merlot", "tasting_notes": "NA", "wine_price": "NA", "occasion": "NA", "food_to_be_paired_with_wine": "NA"}
+    {
+      "wine_name": "Saumur Blanc",
+      "winery": "Domaine du Collier",
+      "vintage": "2019",
+      "region": "Saumur",
+      "country": "France",
+      "wine_type": "white",
+      "grape_varietal": "Merlot",
+      "tasting_notes": "plum",
+      "wine_price_min": "N/A",
+      "wine_price_max": "N/A",
+      "occasion": "N/A",
+      "food_to_be_paired_with_wine": "N/A"
+    }
     
     Let's begin!
   """
 
-  header = ["Thought:", "Wine_name:", "Winery:", "Vintage:", "Region:", "Country:", "Wine_type:", "Grape_varietal:", "Tasting_notes:", "Wine_price:", "Occasion:", "Food_to_be_paired_with_wine:"]
-  dictkey = ["thought", "wine_name", "winery", "vintage", "region", "country", "wine_type", "grape_varietal", "tasting_notes", "wine_price", "occasion", "food_to_be_paired_with_wine"]
+  requiredKeys = [:wine_name, :winery, :vintage, :region, :country, :wine_type, :grape_varietal, :tasting_notes, :wine_price_min, :wine_price_max, :occasion, :food_to_be_paired_with_wine]
   errornote = "N/A"
 
-  llmkwargs=Dict(
-    :num_ctx => 32768,
-    :temperature => 0.5,
-  )
-
   for attempt in 1:maxattempt
-    #[PENDING] I should add generatequestion()
-
-    if attempt > 1
-      println("\nYiemAgent extractWineAttributes_1() attempt $attempt/$maxattempt ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-    end
-
     usermsg =
-    """
-    User's query: $input
-    P.S. $errornote
-    """
+      """
+      $input
+      """
+    context = 
+      """
+      <context>
+      P.S. $errornote
+      </context>
+      /no_think
+      """
 
-    _prompt = 
+    unformatPrompt = 
     [
       Dict(:name=> "system", :text=> systemmsg),
       Dict(:name=> "user", :text=> usermsg)
     ]
 
     # put in model format
-    prompt = GeneralUtils.formatLLMtext(_prompt, "granite3")
-    response = a.func[:text2textInstructLLM](prompt; 
-      modelsize="medium", llmkwargs=llmkwargs, senderId=a.id)
+    prompt = GeneralUtils.formatLLMtext(unformatPrompt, a.llmFormatName)
+    # add info
+    prompt = prompt * context
+
+    response = a.func[:text2textInstructLLM](prompt; modelsize="medium", senderId=a.id)
+    response = GeneralUtils.deFormatLLMtext(response, a.llmFormatName)
     response = GeneralUtils.remove_french_accents(response)
-    response = GeneralUtils.deFormatLLMtext(response, "granite3")
+    think, response = GeneralUtils.extractthink(response)
 
-    # check wheter all attributes are in the response
-    checkFlag = false
-    for word in header
-      if !occursin(word, response)
-        errornote = "In your previous attempts, the $word attribute is missing. Please try again."
-        println("\nYiemAgent extractWineAttributes_1() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-        checkFlag = true
-        break
-      end
-    end
-    checkFlag == true ? continue : nothing
-    
-    # check whether response has all answer's key points
-    detected_kw = GeneralUtils.detect_keyword(header, response)
-    if 0 ∈ values(detected_kw)
-      errornote = "In your previous attempts, the response does not have all answer's key points"
-      println("\nYiemAgent extractWineAttributes_1() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-      continue
-    elseif sum(values(detected_kw)) > length(header)
-      errornote = "In your previous attempts, the response has duplicated answer's key points"
-      println("\nYiemAgent extractWineAttributes_1() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    responsedict = nothing
+    try
+      responsedict = copy(JSON3.read(response))  
+    catch
+      println("\nERROR YiemAgent extractWineAttributes_1() failed to parse response: $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
       continue
     end
-    responsedict = GeneralUtils.textToDict(response, header; 
-                                            dictKey=dictkey, symbolkey=true)
+
+    # check whether all answer's key points are in responsedict
+    _responsedictKey = keys(responsedict)
+    responsedictKey = [i for i in _responsedictKey] # convert into a list
+    is_requiredKeys_in_responsedictKey = [i ∈ responsedictKey for i in requiredKeys]
+
+    if length(is_requiredKeys_in_responsedictKey) > length(requiredKeys)
+      errornote = "Your previous attempt has more key points than answer's required key points."
+      println("\nERROR YiemAgent extractWineAttributes_1() $errornote --> $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+      continue
+    elseif !all(is_requiredKeys_in_responsedictKey)
+      zeroind = findall(x -> x == 0, is_requiredKeys_in_responsedictKey)
+      missingkeys = [requiredKeys[i] for i in zeroind]
+      errornote = "$missingkeys are missing from your previous response"
+      println("\nERROR YiemAgent extractWineAttributes_1() $errornote --> $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+      continue
+    end
+
+    # # check whether response has all header
+    # detected_kw = GeneralUtils.detect_keyword(header, response)
+    # kwvalue = [i for i in values(detected_kw)]
+    # zeroind = findall(x -> x == 0, kwvalue)
+    # missingkeys = [header[i] for i in zeroind]
+    # if 0 ∈ values(detected_kw)
+    #   errornote = "$missingkeys are missing from your previous response"
+    #   println("\nERROR YiemAgent decisionMaker() $errornote:\n$response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    #   continue
+    # elseif sum(values(detected_kw)) > length(header)
+    #   errornote = "Your previous attempt has duplicated points"
+    #   println("\nERROR YiemAgent decisionMaker() $errornote:\n$response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    #   continue
+    # end
+    
+    # # check whether response has all answer's key points
+    # detected_kw = GeneralUtils.detect_keyword(header, response)
+    # if 0 ∈ values(detected_kw)
+    #   errornote = "In your previous attempts, the response does not have all answer's key points"
+    #   println("\nYiemAgent extractWineAttributes_1() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    #   continue
+    # elseif sum(values(detected_kw)) > length(header)
+    #   errornote = "In your previous attempts, the response has duplicated answer's key points"
+    #   println("\nYiemAgent extractWineAttributes_1() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    #   println(response)
+    #   continue
+    # end
+    # responsedict = GeneralUtils.textToDict(response, header; 
+    #                                         dictKey=dictkey, symbolkey=true)
+
+    removekeys = [:thought, :tasting_notes, :occasion, :food_to_be_paired_with_wine, :vintage]
+    for i in removekeys
+      delete!(responsedict, i)
+    end
+
+
 
     delete!(responsedict, :thought)
     delete!(responsedict, :tasting_notes)
     delete!(responsedict, :occasion)
     delete!(responsedict, :food_to_be_paired_with_wine)
-
-    println(@__FILE__, " ", @__LINE__)
-    pprintln(responsedict)
+    delete!(responsedict, :vintage)
 
     # check if winery, wine_name, region, country, wine_type, grape_varietal's value are in the query because sometime AI halucinates
     checkFlag = false
-    for i in dictkey
+    for i in requiredKeys
       j = Symbol(i)
-      if j ∉ [:thought, :tasting_notes, :occasion, :food_to_be_paired_with_wine]
+      if j ∉ removekeys
         # in case j is wine_price it needs to be checked differently because its value is ranged
         if j == :wine_price
-          if responsedict[:wine_price] != "NA" 
+          if responsedict[:wine_price] != "N/A" 
             # check whether wine_price is in ranged number
-            if !occursin('-', responsedict[:wine_price]) 
-              errornote = "In your previous attempt, the 'wine_price' was not set to a ranged number. Please adjust it accordingly."
+            if !occursin("to", responsedict[:wine_price]) 
+              errornote = "In your previous attempt, the 'wine_price' was set to $(responsedict[:wine_price]) which is not a correct format. Please adjust it accordingly."
               println("\nERROR YiemAgent extractWineAttributes_1() $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
               checkFlag = true
               break
             end
 
-            # check whether max wine_price is in the input
-            pricerange = split(responsedict[:wine_price], '-')
-            minprice = pricerange[1]
-            maxprice = pricerange[end]
-            if !occursin(maxprice, input)
-              responsedict[:wine_price] = "NA"
-            end
-            # price range like 100-100 is not good
-            if minprice == maxprice
-              errornote = "In your previous attempt, you inputted 'wine_price' with a 'minimum' value equaling the 'maximum', which is not valid."
-              println("\nERROR YiemAgent extractWineAttributes_1() $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-              checkFlag = true
-              break
-            end
+            # # check whether max wine_price is in the input
+            # pricerange = split(responsedict[:wine_price], '-')
+            # minprice = pricerange[1]
+            # maxprice = pricerange[end]
+            # if !occursin(maxprice, input)
+            #   responsedict[:wine_price] = "N/A"
+            # end
+            # # price range like 100-100 is not good
+            # if minprice == maxprice
+            #   errornote = "In your previous attempt, you inputted 'wine_price' with a 'minimum' value equaling the 'maximum', which is not valid."
+            #   println("\nERROR YiemAgent extractWineAttributes_1() $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+            #   checkFlag = true
+            #   break
+            # end
           end
         else
           content = responsedict[j]
@@ -517,7 +600,7 @@ function extractWineAttributes_1(a::T1, input::T2; maxattempt=10
     result = ""
     for (k, v) in responsedict
       # some time LLM generate text with "(some comment)". this line removes it
-      if !occursin("NA", v) && v != "" && !occursin("none", v) && !occursin("None", v)
+      if !occursin("N/A", v) && v != "" && !occursin("none", v) && !occursin("None", v)
         result *= "$k: $v, "
       end
     end
@@ -540,7 +623,7 @@ function extractWineAttributes_2(a::T1, input::T2)::String where {T1<:agent, T2<
 
   conversiontable =
   """
-  <Conversion Table>
+  <conversion_table>
   Intensity level:
     1 to 2: May correspond to "light-bodied" or a similar description.
     2 to 3: May correspond to "med light bodied", "medium light" or a similar description.
@@ -565,127 +648,151 @@ function extractWineAttributes_2(a::T1, input::T2)::String where {T1<:agent, T2<
     3 to 4: May correspond to "medium acidity" or a similar description.
     4 to 5: May correspond to "semi high acidity" or a similar description.
     4 to 5: May correspond to "high acidity" or a similar description.
-  </Conversion Table>
+  </conversion_table>
   """
 
   systemmsg =
   """
   As an helpful sommelier, your task is to fill out the user's preference form based on the corresponding words from the user's query.
 
-  At each round of conversation, the user will give you the current situation:
-  Conversion Table: ...
-  User's query: ...
+  At each round of conversation, you will be given the following information:
+  conversion_table: a conversion table that maps descriptive words to their corresponding integer levels
+  query: the words from the user's query that describe their preferences
 
   The preference form requires the following information:
   sweetness, acidity, tannin, intensity
 
   You must follow the following guidelines:
-    1) If specific information required in the preference form is not available in the query or there isn't any, mark with 'NA' to indicate this.
+    1) If specific information required in the preference form is not available in the query or there isn't any, mark with 'N/A' to indicate this.
       Additionally, words like 'any' or 'unlimited' mean no information is available.
     2) Use the conversion table to convert the descriptive word level of sweetness, intensity, tannin, and acidity into a corresponding integer.
     3) Do not generate other comments.
   You should then respond to the user with:
-    Sweetness_keyword: The exact keywords in the user's query describing the sweetness level of the wine.
-    Sweetness: ( S ), where ( S ) represents integers indicating the range of sweetness levels. Example: 1-2
-    Acidity_keyword: The exact keywords in the user's query describing the acidity level of the wine.
-    Acidity: ( A ), where ( A ) represents integers indicating the range of acidity level. Example: 3-5
-    Tannin_keyword: The exact keywords in the user's query describing the tannin level of the wine.
-    Tannin: ( T ), where ( T ) represents integers indicating the range of tannin level. Example: 1-3
-    Intensity_keyword: The exact keywords in the user's query describing the intensity level of the wine.
-    Intensity: ( I ), where ( I ) represents integers indicating the range of intensity level. Example: 2-4
-  You should only respond in format as described below:
-    Sweetness_keyword: ...
-    Sweetness: ...
-    Acidity_keyword: ...
-    Acidity: ...
-    Tannin_keyword: ...
-    Tannin: ...
-    Intensity_keyword: ...
-    Intensity: ...
+    sweetness_keyword: The exact keywords in the user's query describing the sweetness level of the wine.
+    sweetness: ( S ), where ( S ) represents integers indicating the range of sweetness levels. Example: 1-2
+    acidity_keyword: The exact keywords in the user's query describing the acidity level of the wine.
+    acidity: ( A ), where ( A ) represents integers indicating the range of acidity level. Example: 3-5
+    tannin_keyword: The exact keywords in the user's query describing the tannin level of the wine.
+    tannin: ( T ), where ( T ) represents integers indicating the range of tannin level. Example: 1-3
+    intensity_keyword: The exact keywords in the user's query describing the intensity level of the wine.
+    intensity: ( I ), where ( I ) represents integers indicating the range of intensity level. Example: 2-4
+  You should only respond in JSON format as described below:
+  {
+    "sweetness_keyword": "...",
+    "sweetness": "...",
+    "acidity_keyword": "...",
+    "acidity": "...",
+    "tannin_keyword": "...",
+    "tannin": "...",
+    "intensity_keyword": "...",
+    "intensity": "..."
+  }
     
   Here are some examples:
     User's query: I want a wine with a medium-bodied, low acidity, medium tannin.
-    Sweetness_keyword: NA
-    Sweetness: NA
-    Acidity_keyword: low acidity
-    Acidity: 1-2
-    Tannin_keyword: medium tannin
-    Tannin: 3-4
-    Intensity_keyword: medium-bodied
-    Intensity: 3-4
+    {
+    "sweetness_keyword": "N/A",
+    "sweetness": "N/A",
+    "acidity_keyword": "low acidity",
+    "acidity": "1-2",
+    "tannin_keyword": "medium tannin",
+    "tannin": "3-4",
+    "intensity_keyword": "medium-bodied",
+    "intensity": "3-4"
+    }
 
     User's query: German red wine, under 100, pairs with spicy food
-    Sweetness_keyword: NA
-    Sweetness: NA
-    Acidity_keyword: NA
-    Acidity: NA
-    Tannin_keyword: NA
-    Tannin: NA
-    Intensity_keyword: NA
-    Intensity: NA
+    {
+    "sweetness_keyword": "N/A",
+    "sweetness": "N/A",
+    "acidity_keyword": "N/A",
+    "acidity": "N/A",
+    "tannin_keyword": "N/A",
+    "tannin": "N/A",
+    "intensity_keyword": "N/A",
+    "intensity": "N/A"
+    }
 
   Let's begin!
   """
-  header = ["Sweetness_keyword:", "Sweetness:", "Acidity_keyword:", "Acidity:", "Tannin_keyword:", "Tannin:", "Intensity_keyword:", "Intensity:"]
-  dictkey = ["sweetness_keyword", "sweetness", "acidity_keyword", "acidity", "tannin_keyword", "tannin", "intensity_keyword", "intensity"]
-  errornote = ""
+  requiredKeys = [:sweetness_keyword, :sweetness, :acidity_keyword, :acidity, :tannin_keyword, :tannin, :intensity_keyword, :intensity]
+
+  # header = ["Sweetness_keyword:", "Sweetness:", "Acidity_keyword:", "Acidity:", "Tannin_keyword:", "Tannin:", "Intensity_keyword:", "Intensity:"]
+  # dictkey = ["sweetness_keyword", "sweetness", "acidity_keyword", "acidity", "tannin_keyword", "tannin", "intensity_keyword", "intensity"]
+  errornote = "N/A"
 
   for attempt in 1:10
-    usermsg =
+    context =
     """
     $conversiontable
-    User's query: $input
+    <query> 
+    $input
+    </query>
     P.S. $errornote
+    /no_think
     """
 
-    _prompt = 
+    unformatPrompt = 
     [
       Dict(:name=> "system", :text=> systemmsg),
-      Dict(:name=> "user", :text=> usermsg)
     ]
 
     # put in model format
-    prompt = GeneralUtils.formatLLMtext(_prompt, "granite3")
+    prompt = GeneralUtils.formatLLMtext(unformatPrompt, a.llmFormatName)
+    # add info
+    prompt = prompt * context
 
-    response = a.func[:text2textInstructLLM](prompt)
-    response = GeneralUtils.deFormatLLMtext(response, "granite3")
+    response = a.func[:text2textInstructLLM](prompt; modelsize="medium", senderId=a.id)
+    response = GeneralUtils.deFormatLLMtext(response, a.llmFormatName)
+    response = GeneralUtils.remove_french_accents(response)
+    think, response = GeneralUtils.extractthink(response)
 
-    # check whether response has all answer's key points
-    detected_kw = GeneralUtils.detect_keyword(header, response)
-    if 0 ∈ values(detected_kw)
-      errornote = "In your previous attempt does not have all answer's key points"
-      println("\nERROR YiemAgent extractWineAttributes_2() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
-      continue
-    elseif sum(values(detected_kw)) > length(header)
-      errornote = "In your previous attempt has duplicated answer's key points"
-      println("\nERROR YiemAgent extractWineAttributes_2() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+    responsedict = nothing
+    try
+      responsedict = copy(JSON3.read(response))  
+    catch
+      println("\nERROR YiemAgent extractWineAttributes_2() failed to parse response: $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
       continue
     end
 
-    responsedict = GeneralUtils.textToDict(response, header; 
-                                            dictKey=dictkey, symbolkey=true)
+    # check whether all answer's key points are in responsedict
+    _responsedictKey = keys(responsedict)
+    responsedictKey = [i for i in _responsedictKey] # convert into a list
+    is_requiredKeys_in_responsedictKey = [i ∈ responsedictKey for i in requiredKeys]
+
+    if length(is_requiredKeys_in_responsedictKey) > length(requiredKeys)
+      errornote = "Your previous attempt has more key points than answer's required key points."
+      println("\nERROR YiemAgent extractWineAttributes_2() $errornote --> $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+      continue
+    elseif !all(is_requiredKeys_in_responsedictKey)
+      zeroind = findall(x -> x == 0, is_requiredKeys_in_responsedictKey)
+      missingkeys = [requiredKeys[i] for i in zeroind]
+      errornote = "$missingkeys are missing from your previous response"
+      println("\nERROR YiemAgent extractWineAttributes_2() $errornote --> $response ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
+      continue
+    end
 
     # check whether each describing keyword is in the input to prevent halucination
     for i in ["sweetness", "acidity", "tannin", "intensity"]
       keyword = Symbol(i * "_keyword")  # e.g. sweetness_keyword
       value = responsedict[keyword]
-      if value != "NA" && !occursin(value, input)
+      if value != "N/A" && !occursin(value, input)
         errornote = "In your previous attempt, keyword $keyword: $value does not appear in the input. You must use information from the input only"
         println("\nERROR YiemAgent extractWineAttributes_2() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
         continue
       end
 
-      # if value == "NA" then responsedict[i] = "NA" 
-      # e.g. if sweetness_keyword == "NA" then sweetness = "NA"
-      if value == "NA"
-        responsedict[Symbol(i)] = "NA"
+      # if value == "N/A" then responsedict[i] = "N/A" 
+      # e.g. if sweetness_keyword == "N/A" then sweetness = "N/A"
+      if value == "N/A"
+        responsedict[Symbol(i)] = "N/A"
       end
     end
 
     # some time LLM not put integer range
     for (k, v) in responsedict
       if !occursin("keyword", string(k))
-        if v !== "NA" && (!occursin('-', v) || length(v) > 5)
+        if v !== "N/A" && (!occursin('-', v) || length(v) > 5)
           errornote = "WARNING: The non-range value {$k: $v} is not allowed. It should be specified in a range format, i.e. min-max."
           println("\nERROR YiemAgent extractWineAttributes_2() Attempt $attempt $errornote ", @__FILE__, ":", @__LINE__, " $(Dates.now())")
           continue
@@ -693,18 +800,25 @@ function extractWineAttributes_2(a::T1, input::T2)::String where {T1<:agent, T2<
       end
     end
 
-    # some time LLM says NA-2. Need to convert NA to 1
+    # some time LLM says N/A-2. Need to convert N/A to 1
     for (k, v) in responsedict
-      if occursin("NA", v) && occursin("-", v)
-        new_v = replace(v, "NA"=>"1")
+      if occursin("N/A", v) && occursin("-", v)
+        new_v = replace(v, "N/A"=>"1")
         responsedict[k] = new_v
       end
     end
 
+    # delete some key words from responsedict
+    for (k, v) in responsedict
+      if k ∈ [:sweetness_keyword, :acidity_keyword, :tannin_keyword, :intensity_keyword]
+        delete!(responsedict, k)
+      end
+    end
+
     result = ""
     for (k, v) in responsedict
       # some time LLM generate text with "(some comment)". this line removes it
-      if !occursin("NA", v)
+      if !occursin("N/A", v)
         result *= "$k: $v, "
       end
     end
@@ -752,11 +866,12 @@ function paraphrase(text2textInstructLLM::Function, text::String)
 
     Let's begin!
   """
-
+      #[WORKING] use JSON3 the same as extractWineAttributes_1 is better
+  #[WORKING] change this function to use the same format use decisionMater
   header = ["Paraphrase:"]
   dictkey = ["paraphrase"]
 
-  errornote = ""
+  errornote = "N/A"
   response = nothing # placeholder for show when error msg show up
 
 
@@ -773,11 +888,12 @@ function paraphrase(text2textInstructLLM::Function, text::String)
       ]
 
     # put in model format
-    prompt = GeneralUtils.formatLLMtext(_prompt, "granite3")
+    prompt = GeneralUtils.formatLLMtext(_prompt, a.llmFormatName)
 
     try
       response = text2textInstructLLM(prompt)
-      response = GeneralUtils.deFormatLLMtext(response, "granite3")
+      response = GeneralUtils.deFormatLLMtext(response, a.llmFormatName)
+      think, response = GeneralUtils.extractthink(response)
       # sometime the model response like this "here's how I would respond: ..."
       if occursin("respond:", response)
         errornote = "You don't need to intro your response"

src/type.jl

@@ -1,6 +1,6 @@
 module type
   
-export agent, sommelier, companion
+export agent, sommelier, companion, virtualcustomer
 
 using Dates, UUIDs, DataStructures, JSON3
 using GeneralUtils
@@ -24,17 +24,13 @@ end
 function companion(
   func::NamedTuple  # NamedTuple of functions
   ;
-  systemmsg::Union{String, Nothing}= nothing,
   name::String= "Assistant",
   id::String= GeneralUtils.uuid4snakecase(), 
   maxHistoryMsg::Integer= 20,
   chathistory::Vector{Dict{Symbol, String}} = Vector{Dict{Symbol, String}}(),
-  llmFormatName::String= "granite3"
-  )
-
-  if systemmsg === nothing
-    systemmsg = 
-      """
+  llmFormatName::String= "granite3",
+  systemmsg::String=
+    """
       Your name: $name
       Your sex: Female
       Your role: You are a helpful assistant.
@@ -43,8 +39,8 @@ function companion(
       - Your like to be short and concise.
 
       Let's begin!
-      """
-  end
+      """,
+  )
 
   tools = Dict( # update input format
       "CHATBOX"=> Dict(
@@ -197,8 +193,83 @@ function sommelier(
   """
   memory = Dict{Symbol, Any}(
             :shortmem=> OrderedDict{Symbol, Any}(
-              :available_wine=> [],
-              :found_wine=> [], # used by decisionMaker(). This is to prevent decisionMaker() keep presenting the same wines
+              :db_search_result=> Any[],
+              :scratchpad=> "", #[PENDING] should be a dict e.g. Dict(:database_search_result=>Dict(:wines=> "", :search_query=> ""))
+            ),
+            :events=> Vector{Dict{Symbol, Any}}(),
+            :state=> Dict{Symbol, Any}(
+            ),
+            :recap=> OrderedDict{Symbol, Any}(),
+            
+          )
+
+  newAgent = sommelier(
+    name,
+    id,
+    retailername,
+    tools,
+    maxHistoryMsg,
+    chathistory,
+    memory,
+    func,
+    llmFormatName
+    )
+  
+  return newAgent
+end
+
+
+mutable struct virtualcustomer <: agent
+  name::String  # agent name
+  id::String    # agent id
+  systemmsg::String # system message
+  tools::Dict   
+  maxHistoryMsg::Integer  # e.g. 21th and earlier messages will get summarized
+  chathistory::Vector{Dict{Symbol, Any}}
+  memory::Dict{Symbol, Any}
+  func  # NamedTuple of functions 
+  llmFormatName::String
+end
+
+function virtualcustomer(
+  func,  # NamedTuple of functions
+  ;
+  name::String= "Assistant",
+  id::String= string(uuid4()),
+  maxHistoryMsg::Integer= 20,
+  chathistory::Vector{Dict{Symbol, String}} = Vector{Dict{Symbol, String}}(),
+  llmFormatName::String= "granite3",
+  systemmsg::String=
+    """
+      Your name: $name
+      Your sex: Female
+      Your role: You are a helpful assistant.
+      You should follow the following guidelines:
+      - Focus on the latest conversation.
+      - Your like to be short and concise.
+
+      Let's begin!
+      """,
+  )
+
+  tools = Dict( # update input format
+      "chatbox"=> Dict(
+        :description => "<askbox tool description>Useful for when you need to ask the user for more context. Do not ask the user their own question.</askbox tool description>",
+        :input => """<input>Input is a text in JSON format.</input><input example>{\"Q1\": \"How are you doing?\", \"Q2\": \"How may I help you?\"}</input example>""",
+        :output => "" ,
+      ),
+    )
+
+  """ Memory
+    Ref: Chat prompt format https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/discussions/3
+    NO "system" message in chathistory because I want to add it at the inference time
+    chathistory= [
+      Dict(:name=>"user", :text=> "Wassup!", :timestamp=> Dates.now()),
+      Dict(:name=>"assistant", :text=> "Hi I'm your assistant.", :timestamp=> Dates.now()),
+    ]
+  """
+  memory = Dict{Symbol, Any}(
+            :shortmem=> OrderedDict{Symbol, Any}(
             ),
             :events=> Vector{Dict{Symbol, Any}}(),
             :state=> Dict{Symbol, Any}(
@@ -206,10 +277,10 @@ function sommelier(
             :recap=> OrderedDict{Symbol, Any}(),
           )
 
-  newAgent = sommelier(
+  newAgent = virtualcustomer(
     name,
     id,
-    retailername,
+    systemmsg,
     tools,
     maxHistoryMsg,
     chathistory,

src/util.jl

@@ -1,7 +1,7 @@
 module util
 
 export clearhistory, addNewMessage, chatHistoryToText, eventdict, noises, createTimeline,
-        availableWineToText
+        availableWineToText, createEventsLog, createChatLog
 
 using UUIDs, Dates, DataStructures, HTTP, JSON3
 using GeneralUtils
@@ -297,22 +297,25 @@ function createTimeline(events::T1; eventindex::Union{UnitRange, Nothing}=nothin
       1:length(events)
     end
 
-  # Iterate through events and format each one
-  for (i, event) in zip(ind, events)
+  #[WORKING] Iterate through events and format each one
+  for i in ind
+    event = events[i]
     # If no outcome exists, format without outcome
-    if event[:outcome] === nothing
-      timeline *= "Event_$i $(event[:subject])> $(event[:actioninput])\n"
+    # if event[:actionname] == "CHATBOX"
+    #   timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput])\n"
+    # elseif event[:actionname] == "CHECKINVENTORY" && event[:outcome] === nothing
+    #   timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput]), observation: Not done yet.\n"
     # If outcome exists, include it in formatting  
+    if event[:actionname] == "CHECKWINE"
+      timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput]), observation: $(event[:outcome])\n"
     else
-      timeline *= "Event_$i $(event[:subject])> $(event[:actioninput]) $(event[:outcome])\n"
+      timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput])\n"
     end
   end
 
   # Return formatted timeline string
   return timeline
 end
-
-
 # function createTimeline(events::T1; eventindex::Union{UnitRange, Nothing}=nothing
 #   ) where {T1<:AbstractVector}
 #   # Initialize empty timeline string
@@ -327,15 +330,14 @@ end
 #     end
 
 #   # Iterate through events and format each one
-#   for (i, event) in zip(ind, events)
+#   for i in ind
+#     event = events[i]
 #     # If no outcome exists, format without outcome
-#     subject = titlecase(event[:subject])
 #     if event[:outcome] === nothing
-      
-#       timeline *= "Event_$i) Who: $subject Action_name: $(event[:actionname]) Action_input: $(event[:actioninput])\n"
+#       timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput]), observation: Not done yet.\n"
 #     # If outcome exists, include it in formatting  
 #     else
-#       timeline *= "Event_$i) Who: $subject Action_name: $(event[:actionname]) Action_input: $(event[:actioninput]) Action output: $(event[:outcome])\n"
+#       timeline *= "Event_$i $(event[:subject])> action_name: $(event[:actionname]), action_input: $(event[:actioninput]), observation: $(event[:outcome])\n"
 #     end
 #   end
 
@@ -344,6 +346,73 @@ end
 # end
 
 
+function createEventsLog(events::T1; index::Union{UnitRange, Nothing}=nothing
+  ) where {T1<:AbstractVector}
+  # Initialize empty log array
+  log = Dict{Symbol, String}[]
+
+  # Determine which indices to use - either provided range or full length
+  ind = 
+    if index !== nothing
+      [index...]
+    else
+      1:length(events)
+    end
+
+  # Iterate through events and format each one
+  for i in ind
+    event = events[i]
+    # If no outcome exists, format without outcome
+    if event[:outcome] === nothing
+      subject = event[:subject]
+      actioninput = event[:actioninput]
+      d = Dict{Symbol, String}(:name=>subject, :text=>actioninput)
+      push!(log, d)
+    else
+      subject = event[:subject]
+      actioninput = event[:actioninput]
+      outcome = event[:outcome]
+      str = "Action: $actioninput Outcome: $outcome"
+      d = Dict{Symbol, String}(:name=>subject, :text=>str)
+      push!(log, d)
+    end
+  end
+
+  return log
+end
+
+
+function createChatLog(chatdict::T1; index::Union{UnitRange, Nothing}=nothing
+  ) where {T1<:AbstractVector}
+  # Initialize empty log array
+  log = Dict{Symbol, String}[]
+
+  # Determine which indices to use - either provided range or full length
+  ind = 
+    if index !== nothing
+      [index...]
+    else
+      1:length(chatdict)
+    end
+
+  # Iterate through events and format each one
+  for i in ind
+    event = chatdict[i]
+    subject = event[:name]
+    text = event[:text]
+    d = Dict{Symbol, String}(:name=>subject, :text=>text)
+    push!(log, d)
+  end
+
+  return log
+end
+
+
+
+
+
+
+