SimpleChatTCRV:Cleanup: Starting flow, cleanup description, readme

Avoid chat_show getting called twice during starting.

Update description of sys_date_time to be more verbose

Update the readme DONE section, where I had few days back forgotten
and started making entries as parts of a item list or so.
This commit is contained in:
hanishkvc 2025-11-17 01:52:23 +05:30
parent c4cbeba463
commit 735a8b382c
3 changed files with 123 additions and 112 deletions

View File

@ -27,12 +27,12 @@ function startme() {
gMe.debug_disable();
devel_expose()
gMe.toolsMgr.init(gMe).then(()=>{
gMe.multiChat.chat_show(gMe.multiChat.curChatId);
gMe.multiChat.handle_session_switch(gMe.multiChat.curChatId)
})
for (let cid of gMe.defaultChatIds) {
gMe.multiChat.new_chat_session(cid);
}
gMe.multiChat.setup_ui(gMe.defaultChatIds[0], true);
gMe.multiChat.setup_ui(gMe.defaultChatIds[0]);
gMe.multiChat.show_sessions();
}

View File

@ -654,132 +654,141 @@ channel of the handshake.
having ai responses built over multiple steps of tool callings etal. So independent of our client side
sliding window based drop off or even before they kick in, this can help in many cases.
* UI - add ClearChat button and logic. Also add unicode icons for same as well as for Settings.
UI - add ClearChat button and logic. Also add unicode icons for same as well as for Settings.
* renamed pdf_to_text to fetch_pdf_as_text so that ai model can understand the semantic better.
Renamed pdf_to_text to fetch_pdf_as_text so that ai model can understand the semantic better.
* sys_date_time tool call has been added.
sys_date_time tool call has been added.
* Refactor code and flow a bit wrt the client web ui
* Move the main chat related classes into its own js module file, independent of the main
runtime entry point (rather move out the runtime entry point into its own file). This allows
these classes to be referenced from other modules like tools related modules with full access
to these classes's details for developers and static check tools.
* building on same make the Tools management code into a ToolsManager class which is inturn
instantiated and the handle stored in top level Me class. This class also maintains and
manages the web workers as well as routing of the tool calling among others.
* add a common helper for posting results directly to the main thread side web worker callback
handlers. Inturn run the calling through a setTimeout0, so that delayed/missing response
situation rescuer timeout logic etal flow doesnt get messed for now.
Refactor code and flow a bit wrt the client web ui
* Move the main chat related classes into its own js module file, independent of the main
runtime entry point (rather move out the runtime entry point into its own file). This allows
these classes to be referenced from other modules like tools related modules with full access
to these classes's details for developers and static check tools.
* building on same make the Tools management code into a ToolsManager class which is inturn
instantiated and the handle stored in top level Me class. This class also maintains and
manages the web workers as well as routing of the tool calling among others.
* add a common helper for posting results directly to the main thread side web worker callback
handlers. Inturn run the calling through a setTimeout0, so that delayed/missing response
situation rescuer timeout logic etal flow doesnt get messed for now.
* track tool calling and inturn maintain pending tool calls so that only still valid tool call responses
will be accepted when the asynchronous tool call response is recieved. Also take care of clearing
pending tool call tracking in unhappy paths like when exception noticied as part of tool call execution,
or if there is no response within the configured timeout period.
NOTE: Currently the logic supports only 1 pending tool call per chat session.
Track tool calling and inturn maintain pending tool calls so that only still valid tool call responses
will be accepted when the asynchronous tool call response is recieved. Also take care of clearing
pending tool call tracking in unhappy paths like when exception noticied as part of tool call execution,
or if there is no response within the configured timeout period.
NOTE: Currently the logic supports only 1 pending tool call per chat session.
* add support for fetch_xml_as_text tool call, fix importmaps in index.html
Add support for fetch_xml_as_text tool call, fix importmaps in index.html
* renamed and updated logic wrt xml fetching to be fetch_xml_filtered. allow one to use re to identify
the tags to be filtered in a fine grained manner including filtering based on tag heirarchy
* avoid showing empty skipped tag blocks
Renamed and updated logic wrt xml fetching to be fetch_xml_filtered. allow one to use re to identify
the tags to be filtered in a fine grained manner including filtering based on tag heirarchy
* avoid showing empty skipped tag blocks
* logic which shows the generated tool call has been updated to trap errors when parsing the function call
arguments generated by the ai. This ensures that the chat ui itself doesnt get stuck in it. Instead now
the tool call response can inform the ai model that its function call had issues.
Logic which shows the generated tool call has been updated to trap errors when parsing the function call
arguments generated by the ai. This ensures that the chat ui itself doesnt get stuck in it. Instead now
the tool call response can inform the ai model that its function call had issues.
* renamed fetch_web_url_text to fetch_html_text, so that gen ai model wont try to use this to fetch xml or
rss files, because it will return empty content, because there wont be any html content to strip the tags
and unwanted blocks before returning.
Renamed fetch_web_url_text to fetch_html_text, so that gen ai model wont try to use this to fetch xml or
rss files, because it will return empty content, because there wont be any html content to strip the tags
and unwanted blocks before returning.
* capture the body of ai server not ok responses, to help debug as well as to show same to user.
Capture the body of ai server not ok responses, to help debug as well as to show same to user.
* extract and include the outline of titles (along with calculated numbering) in the text output of pdftext
* ensure that one doesnt recurse beyond a predefined limit.
Extract and include the outline of titles (along with calculated numbering) in the text output of pdftext
* ensure that one doesnt recurse beyond a predefined limit.
* Convert NSChatMessage from typedef to Class and update ChatMessageEx, SimpleChat, MultiChatUI classes to
make use of the same.
* helpers consolidated
* helpers to check if given instance contains reasoning or content or toolcall or tool response related
fields/info in them.
* helpers to get the corresponding field values
* some of these helpers where in ChatMessageEx and beyond before
* now the network handshaked fields are declared as undefined by default (instead of empty values).
this ensures that json stringify will automatically discard fields whose values are still undefined.
* add fields wrt tool response and update full flow to directly work with these fields instead of the
xml based serialisation which was previously used for maintaining the different tool response fields
within the content field (and inturn extract from there when sending to server).
* now a dataset based attribute is used to identify when input element contains user input and when
it contains tool call result/response.
* this simplifies the flow wrt showing chat message (also make it appear more cleanly) as well as
identifying not yet accepted tool result and showing in user query input field and related things.
* ALERT: ON-DISK-STORAGE structure of chat sessions have changed wrt tool responses. So old saves will
no longer work wrt tool responses
Convert NSChatMessage from typedef to Class and update ChatMessageEx, SimpleChat, MultiChatUI classes to
make use of the same.
* helpers consolidated
* helpers to check if given instance contains reasoning or content or toolcall or tool response related
fields/info in them.
* helpers to get the corresponding field values
* some of these helpers where in ChatMessageEx and beyond before
* now the network handshaked fields are declared as undefined by default (instead of empty values).
this ensures that json stringify will automatically discard fields whose values are still undefined.
* add fields wrt tool response and update full flow to directly work with these fields instead of the
xml based serialisation which was previously used for maintaining the different tool response fields
within the content field (and inturn extract from there when sending to server).
* now a dataset based attribute is used to identify when input element contains user input and when
it contains tool call result/response.
* this simplifies the flow wrt showing chat message (also make it appear more cleanly) as well as
identifying not yet accepted tool result and showing in user query input field and related things.
* ALERT: ON-DISK-STORAGE structure of chat sessions have changed wrt tool responses. So old saves will
no longer work wrt tool responses
* UI updates
* update logic to allow empty tool results to be sent to ai engine server
* css - when user input textarea is in tool result mode (ie wrt TOOL.TEMP role), change the background
color to match the tool role chat message block color, so that user can easily know that the input
area is being used for submitting tool response or user response, at any given moment in time.
UI updates
* update logic to allow empty tool results to be sent to ai engine server
* css - when user input textarea is in tool result mode (ie wrt TOOL.TEMP role), change the background
color to match the tool role chat message block color, so that user can easily know that the input
area is being used for submitting tool response or user response, at any given moment in time.
* Vision
* Add image_url field. Allow user to load image, which is inturn stored as a dataURL in image_url.
* when user presses submit with a message, if there is some content (image for now) in dataURL,
then initialise image_url field with same.
* when generating chat messages for ai server network handshake, create the mixed content type of
content field which includes both the text (from content field) and image (from image_url field)
ie if a image_url is found wrt a image.
* follow the openai format/template wrt these mixed content messages.
* Usage: specify a mmproj file directly or through -hf, additionally had to set --batch-size to 8k
and ubatch-size to 2k wrt gemma3-4b-it
* when showing chat instantiate img elements to show image_urls.
* limit horizontally to max width and vertically to 20% of the height
* show any image loaded by the user, in the corresponding image button
* consolidate dataurl handling into a bunch of helper functions.
* trap quota errors wrt localStorage etal
* dont forget to reset the file type input's value, so that reselecting the same image still
triggers the input's change event.
Vision
* Add image_url field. Allow user to load image, which is inturn stored as a dataURL in image_url.
* when user presses submit with a message, if there is some content (image for now) in dataURL,
then initialise image_url field with same.
* when generating chat messages for ai server network handshake, create the mixed content type of
content field which includes both the text (from content field) and image (from image_url field)
ie if a image_url is found wrt a image.
* follow the openai format/template wrt these mixed content messages.
* Usage: specify a mmproj file directly or through -hf, additionally had to set --batch-size to 8k
and ubatch-size to 2k wrt gemma3-4b-it
* when showing chat instantiate img elements to show image_urls.
* limit horizontally to max width and vertically to 20% of the height
* show any image loaded by the user, in the corresponding image button
* consolidate dataurl handling into a bunch of helper functions.
* trap quota errors wrt localStorage etal
* dont forget to reset the file type input's value, so that reselecting the same image still
triggers the input's change event.
* SimpleChat class now allows extra fields to be specified while adding, in a generic way using a
object/literal object or equivalent.
SimpleChat class now allows extra fields to be specified while adding, in a generic way using a
object/literal object or equivalent.
* UI Cleanup - msgs spaced out, toolcall edit hr not always, scroll ui only when required,
hide settings/info till user requests, heading gradient
UI Cleanup - msgs spaced out, toolcall edit hr not always, scroll ui only when required,
hide settings/info till user requests, heading gradient
* iDB module
* add open, transact, put and get. Use for chat session save and load
* getKeys used to show Restore/Load button wrt chat sessions.
iDB module
* add open, transact, put and get. Use for chat session save and load
* getKeys used to show Restore/Load button wrt chat sessions.
* ChatMessage
* assign a globally unique (ie across sessions) id to each chat message instance.
* add support for deleting chat message based on its uniquie id in SimpleChat.
* try ensure that adjacent messages remain on screen, after a message is deleted from session.
* add a popover div block in html, which acts as a popup menu containing buttons to work with
individual chat messages.
* experiment and finalise on anchor based relative positioning of the popover menu.
* have a del button, which allows one to delete the currently in focus chat message.
* have a copy button, which allows one to copy the textual content into system clipboard.
ChatMessage
* assign a globally unique (ie across sessions) id to each chat message instance.
* add support for deleting chat message based on its uniquie id in SimpleChat.
* try ensure that adjacent messages remain on screen, after a message is deleted from session.
* add a popover div block in html, which acts as a popup menu containing buttons to work with
individual chat messages.
* experiment and finalise on anchor based relative positioning of the popover menu.
* have a del button, which allows one to delete the currently in focus chat message.
* have a copy button, which allows one to copy the textual content into system clipboard.
* MultiChatUI
* chat_show takes care of showing or clearing tool call edit / trigger as well as tool response
edit / submit. Also show the currently active tool call and its response before it is submitted
was previously only shown in the edit / trigger and edit / submit ui elements, now instead it
also shows as part of the chat session message blocks, so that user can delete or copy these
if needed using the same mechanism as other messages in the chat session.
* use a delete msg helper, which takes care of deleting the msg from chat session as well as
efficiently update ui to any extent by removing the corresponding element directly from existing
chat session ui without recreating the full chat session ui.
* a helper to add a message into specified chat session, as well as show/update in the chat session
ui by appending the chat message, instead of recreating the full chat session ui.
...
MultiChatUI
* chat_show takes care of showing or clearing tool call edit / trigger as well as tool response
edit / submit. Also show the currently active tool call and its response before it is submitted
was previously only shown in the edit / trigger and edit / submit ui elements, now instead it
also shows as part of the chat session message blocks, so that user can delete or copy these
if needed using the same mechanism as other messages in the chat session.
* use a delete msg helper, which takes care of deleting the msg from chat session as well as
efficiently update ui to any extent by removing the corresponding element directly from existing
chat session ui without recreating the full chat session ui.
* a helper to add a message into specified chat session, as well as show/update in the chat session
ui by appending the chat message, instead of recreating the full chat session ui.
...
MultiChatUI+
* both chat_show and chat_uirefresh (if lastN >= 2) both take care of updating tool call edit/trigger
as well as the tool call response edit/submit related ui elements suitably.
* chat_show recreates currently active sliding window of chat session (which could even be full)
* while chat_uirefresh recreates/updates ui only for the lastN messages (prefer in general, as optimal)
* normal user response / query submit as well as tool call response or error submit have been updated
to use the optimal uirefresh logic now.
Cleanup in general
* Inform end user when loading from a saved session.
* Update starting entry point flow to avoid calling chat_show twice indirectly, inturn leading to
two restore previously saved session blocks. Rather when adding tool calls support, and inturn
had to account for delayed collating of available simpleproxy based tool calls, I forgot to clean
this flow up.
* Make the sys_date_time template description bit more verbose, just in case.
* MultiChatUI+
* both chat_show and chat_uirefresh (if lastN >= 2) both take care of updating tool call edit/trigger
as well as the tool call response edit/submit related ui elements suitably.
* chat_show recreates currently active sliding window of chat session (which could even be full)
* while chat_uirefresh recreates/updates ui only for the lastN messages (prefer in general, as optimal)
* normal user response / query submit as well as tool call response or error submit have been updated
to use the optimal uirefresh logic now.
#### ToDo

View File

@ -23,12 +23,14 @@ let sysdatetime_meta = {
"properties": {
"template": {
"type": "string",
"description": `Optional template can be any combination of Y,M,D,h,m,s.
"description": `Template is used to control what is included in the returned date time string.
It can be any combination of Y,M,D,h,m,s. Here
Y - FullYear 4 digits, M - Month 2 digits, D - Day 2 digits,
h - hour 2 digits, m - minutes 2 digits, s - seconds 2 digits,
any other char will be returned as is
any other char will be returned as is.
If no template is given, it defaults to YMDThm
If no template is given, it defaults to YMDThm.
Remember that the template characters are case sensitive.
`
}
},