Skip to content

feat(didi): BigQuery fixes, ElevenLabs provider, no-call blocklist, agent prompts, full template seeding #21

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
DZDasherKTB wants to merge 8 commits into
base: onboarding_2026
Choose a base branch
from
Open

feat(didi): BigQuery fixes, ElevenLabs provider, no-call blocklist, agent prompts, full template seeding #21

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
59e1af3
fix(didi): bigquery scope readonly to full for job execution
nithun Aug 5, 2026
5b6f4c6
fix(didi): fix sp_assigment_id casing, filter @results and reset in _…
nithun Aug 5, 2026
c73989d
feat(didi): ElevenLabs provider, no-call blocklist, BigQuery scope fix
DZDasherKTB Aug 21, 2026
15f7287
docs(didi): Vocallabs agent prompts for all 4 languages
DZDasherKTB Aug 21, 2026
2526684
fix(didi): apply missing changes - provider field, blocklist, daily d…
DZDasherKTB Aug 21, 2026
122a98d
fix(didi): add fetch_recordings method and JS buttons
DZDasherKTB Aug 21, 2026
c2505f8
feat(didi): seed all 32 ParentCallConfig templates in patch
DZDasherKTB Aug 21, 2026
56a9beb
refactor(didi): split into VoiceAgentSettings, VocalLabsSettings, Ele…
DZDasherKTB Aug 22, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1,078 changes: 1,078 additions & 0 deletions docs/agent_prompts/didi_agent_prompt_english_final.txt
View file
Open in desktop

Large diffs are not rendered by default.

601 changes: 601 additions & 0 deletions docs/agent_prompts/didi_agent_prompt_hindi_final.txt
View file
Open in desktop

Large diffs are not rendered by default.

559 changes: 559 additions & 0 deletions docs/agent_prompts/didi_agent_prompt_marathi_final.txt
View file
Open in desktop

Large diffs are not rendered by default.

559 changes: 559 additions & 0 deletions docs/agent_prompts/didi_agent_prompt_punjabi_final.txt
View file
Open in desktop

Large diffs are not rendered by default.

202 changes: 102 additions & 100 deletions tap_lms/hooks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,36 +9,34 @@
app_email = "tech4dev@gmail.com"
app_license = "MIT"

before_migrate = "tap_lms.migrate.before_migrate"


# Document Events
doc_events = {
"School": {
"before_save": "tap_lms.tap_lms.doctype.school.school.before_save"
},
# "Teacher": {
# "on_update": "tap_lms.glific_webhook.update_glific_contact"
# },
# "StudentStageProgress": {
# "after_insert": "tap_lms.tap_lms.doctype.studentonboardingprogress.studentonboardingprogress.update_student_progress",
# "on_update": "tap_lms.tap_lms.doctype.studentonboardingprogress.studentonboardingprogress.update_student_progress"
# },
"Teacher": {
"on_update": "tap_lms.glific_webhook.update_glific_contact"
},
"StudentStageProgress": {
"after_insert": "tap_lms.tap_lms.doctype.studentonboardingprogress.studentonboardingprogress.update_student_progress",
"on_update": "tap_lms.tap_lms.doctype.studentonboardingprogress.studentonboardingprogress.update_student_progress"
},
# CR-002 v2 gamification (2026-05-13): VideoClass completion via
# StudentContentLog drives the activity-points handler. The handler also
# arms the grace clock on the first VideoClass of each week (CR-003
# follow-up 2: atomic Postgres CASE WHEN on `weekly_video_done`). Without
# this hook the activity-points pipeline AND the grace clock are dead.
# "StudentContentLog": {
# "after_insert": "tap_lms.summer_program.activity_points.handle_content_log"
# },
"StudentContentLog": {
"after_insert": "tap_lms.summer_program.activity_points.handle_content_log"
},
# CR-002 v2 gamification (2026-05-13): quiz attempts award per-question
# points (correct → q.points; wrong → q.failed_points). The handler is
# idempotent via `attempt.points_earned` so re-saves of completed
# attempts are no-ops.
# "StudentQuizAttempt": {
# "on_update": "tap_lms.summer_program.quiz_points.handle_attempt_update"
# }
"StudentQuizAttempt": {
"on_update": "tap_lms.summer_program.quiz_points.handle_attempt_update"
}
}

# Scheduled Tasks
Expand Down Expand Up @@ -67,100 +65,104 @@
# Batch.current_calendar_week and unblocks max_allowed_week
# on each PE
scheduler_events = {
# "daily": [
# "tap_lms.tap_lms.page.onboarding_flow_trigger.onboarding_flow_trigger.update_incomplete_stages",
# "tap_lms.summer_program.scheduler.run_daily_actions",
# "tap_lms.summer_program.batch_activation.check_auto_activate",
# ],
"daily": [
"tap_lms.tap_lms.page.onboarding_flow_trigger.onboarding_flow_trigger.update_incomplete_stages",
"tap_lms.summer_program.scheduler.run_daily_actions",
"tap_lms.summer_program.batch_activation.check_auto_activate",
],
"cron": {
# "*/1 * * * *": [
# "tap_lms.summer_program.pe_dispatcher.process_program_actions",
# ],
# # Retired 2026-05-21 (task #50): the legacy escalation_runner ran in
# # parallel with pe_dispatcher.handle_escalation (system A, post-CR-003)
# # without gating on canonical PE state. It double-escalated students
# # who already submitted and disagreed with the per-PE escalation
# # counter. The per-PE dispatcher fires escalations correctly when
# # next_action_at + next_action_type='escalation' are armed by T1
# # (content_no_response). escalation_runner.py is preserved as dead
# # code for one release cycle; remove the module after launch.
# # "0 */2 * * *": [
# # "tap_lms.summer_program.escalation_runner.run_escalation_check",
# # ],
# "0 0 * * 1": [
# "tap_lms.summer_program.batch_admin.auto_advance_batch_week",
# ],
# # CR-027 (2026-06-09): Monday 06:00 UTC weekly content sweep. Runs
# # AFTER auto_advance_batch_week (Monday 00:00 UTC) bumps
# # Batch.current_calendar_week, so the sweep sees the new calendar week.
# # Phase 1 demotes behind students (wk < calendar_week, no video) to
# # normal_escalation via t2_start_escalation; Phase 2 delivers this
# # week's content to current-week candidates via a temp Glific sweep
# # collection. Requires `bench migrate` to register this entry in
# # tabScheduledJobType (L-049) — `bench restart` alone is not enough.
# "0 6 * * 1": [
# "tap_lms.summer_program.scheduler.weekly_content_sweep",
# ],
# # CR-005 (2026-05-15): Tuesday 03:30 UTC = Tuesday 09:00 IST.
# # Fires SP_Content_Delivery against each active BPR's `main`
# # Glific collection. Membership is maintained continuously by
# # state-machine transitions (Approach B) — this cron just fires.
# "30 3 * * 2": [
# "tap_lms.summer_program.scheduler.weekly_content_delivery_trigger",
# ],
# # Task #56 (2026-05-16): hourly watchdog for PEs stuck in
# # feedback_ready because Glific's F5 callback dropped silently.
# # LOG-only — does NOT auto-transition; operator replays manually.
# # See pre_launch.feedback_ready_watchdog for full rationale.
# # Task #17 (2026-05-28): two more hourly watchers added at the same
# # cadence. Both are read-only — they turn silent async failures
# # (Glific sync DLQ, RQ queue depth) into Error Log entries operators
# # can see in the Frappe Desk Error Log list view.
# "0 * * * *": [
# "tap_lms.summer_program.pre_launch.feedback_ready_watchdog",
# "tap_lms.summer_program.scheduler.glific_sync_dlq_watcher",
# "tap_lms.summer_program.scheduler.rq_queue_depth_watcher",
# # CR-025 (2026-06-09): hourly probe to detect stale Glific tokens.
# # Invalidates the stored token when it gets a 401, so the next real
# # API call triggers a fresh login instead of queuing 6 retries against
# # a dead token. Requires `bench migrate` to register this scheduler
# # entry in tabScheduledJobType (L-049).
# "tap_lms.glific_integration.probe_token_health",
# ],
# # Task #97 / removed 2026-05-26 (L-027 MVP discipline):
# # `periodic_glific_reconcile` was wired here at */10 cadence as a
# # drift safety net. After diagnosing the real root cause of the
# # Himani / ST00051295 rendering bug (missing createContactsField
# # definitions, NOT value drift), the cron stopped being MVP-
# # justified — in production normal operation, students don't have
# # Desk access and code paths fire the sync hook correctly. The
# # `set_value` bypass is operator-driven (console backfills).
# #
# # The function `scheduler.periodic_glific_reconcile` is preserved
# # for MANUAL invocation from bench console when needed. Same for
# # `dev_tools.reconcile_pe_to_glific(pe_name)` and
# # `dev_tools.reconcile_batch_to_glific(batch_name)`.
# #
# # If post-launch logs show silent value drift, re-enable here at a
# # cadence appropriate to the observed frequency (likely daily, not
# # */10) — uncomment and `bench migrate`.
# # "*/10 * * * *": [
# # "tap_lms.summer_program.scheduler.periodic_glific_reconcile",
# # ],
# Didi voice agent: hourly campaign processor + reengagement check
"*/1 * * * *": [
"tap_lms.summer_program.pe_dispatcher.process_program_actions",
],
# Retired 2026-05-21 (task #50): the legacy escalation_runner ran in
# parallel with pe_dispatcher.handle_escalation (system A, post-CR-003)
# without gating on canonical PE state. It double-escalated students
# who already submitted and disagreed with the per-PE escalation
# counter. The per-PE dispatcher fires escalations correctly when
# next_action_at + next_action_type='escalation' are armed by T1
# (content_no_response). escalation_runner.py is preserved as dead
# code for one release cycle; remove the module after launch.
# "0 */2 * * *": [
# "tap_lms.summer_program.escalation_runner.run_escalation_check",
# ],
"0 0 * * 1": [
"tap_lms.summer_program.batch_admin.auto_advance_batch_week",
# Didi: trim VoiceCallHistory to last 50 rows per enrollment
"tap_lms.summer_program.campaign_processor.archive_old_call_history",
],
# CR-027 (2026-06-09): Monday 06:00 UTC weekly content sweep. Runs
# AFTER auto_advance_batch_week (Monday 00:00 UTC) bumps
# Batch.current_calendar_week, so the sweep sees the new calendar week.
# Phase 1 demotes behind students (wk < calendar_week, no video) to
# normal_escalation via t2_start_escalation; Phase 2 delivers this
# week's content to current-week candidates via a temp Glific sweep
# collection. Requires `bench migrate` to register this entry in
# tabScheduledJobType (L-049) — `bench restart` alone is not enough.
"0 6 * * 1": [
"tap_lms.summer_program.scheduler.weekly_content_sweep",
],
# CR-005 (2026-05-15): Tuesday 03:30 UTC = Tuesday 09:00 IST.
# Fires SP_Content_Delivery against each active BPR's `main`
# Glific collection. Membership is maintained continuously by
# state-machine transitions (Approach B) — this cron just fires.
"30 3 * * 2": [
"tap_lms.summer_program.scheduler.weekly_content_delivery_trigger",
],
# Task #56 (2026-05-16): hourly watchdog for PEs stuck in
# feedback_ready because Glific's F5 callback dropped silently.
# LOG-only — does NOT auto-transition; operator replays manually.
# See pre_launch.feedback_ready_watchdog for full rationale.
# Task #17 (2026-05-28): two more hourly watchers added at the same
# cadence. Both are read-only — they turn silent async failures
# (Glific sync DLQ, RQ queue depth) into Error Log entries operators
# can see in the Frappe Desk Error Log list view.
"0 * * * *": [
"tap_lms.summer_program.campaign_processor.trigger_scheduled_campaigns",
"tap_lms.summer_program.campaign_processor.check_reengagement",
"tap_lms.summer_program.pre_launch.feedback_ready_watchdog",
"tap_lms.summer_program.scheduler.glific_sync_dlq_watcher",
"tap_lms.summer_program.scheduler.rq_queue_depth_watcher",
# CR-025 (2026-06-09): hourly probe to detect stale Glific tokens.
# Invalidates the stored token when it gets a 401, so the next real
# API call triggers a fresh login instead of queuing 6 retries against
# a dead token. Requires `bench migrate` to register this scheduler
# entry in tabScheduledJobType (L-049).
"tap_lms.glific_integration.probe_token_health",
],
# Didi voice agent: BigQuery sync Mon-Sat 12:00 UTC (5:30 PM IST)
"0 12 * * 1-6": [
# Task #97 / removed 2026-05-26 (L-027 MVP discipline):
# `periodic_glific_reconcile` was wired here at */10 cadence as a
# drift safety net. After diagnosing the real root cause of the
# Himani / ST00051295 rendering bug (missing createContactsField
# definitions, NOT value drift), the cron stopped being MVP-
# justified — in production normal operation, students don't have
# Desk access and code paths fire the sync hook correctly. The
# `set_value` bypass is operator-driven (console backfills).
#
# The function `scheduler.periodic_glific_reconcile` is preserved
# for MANUAL invocation from bench console when needed. Same for
# `dev_tools.reconcile_pe_to_glific(pe_name)` and
# `dev_tools.reconcile_batch_to_glific(batch_name)`.
#
# If post-launch logs show silent value drift, re-enable here at a
# cadence appropriate to the observed frequency (likely daily, not
# */10) — uncomment and `bench migrate`.
# "*/10 * * * *": [
# "tap_lms.summer_program.scheduler.periodic_glific_reconcile",
# ],
# Didi voice nudge: BigQuery -> Frappe sync for StudentGlificContext.
"0 */6 * * *": [
"tap_lms.summer_program.bigquery_sync.sync_bigquery_glific_context",
],
# Didi voice nudge: check for scheduled campaigns due to start.
# Runs hourly. Triggers generate_queue + start_calls for any
# VoiceCallCampaign with scheduled_at <= now and status Draft/Ready.
"0 * * * *": [
"tap_lms.summer_program.campaign_processor.trigger_scheduled_campaigns",
"tap_lms.summer_program.campaign_processor.check_reengagement",
],
},
}

# Page configurations
# page_js = {"onboarding-flow-trigger": "public/js/onboarding_flow_trigger.js"}
page_js = {"onboarding-flow-trigger": "public/js/onboarding_flow_trigger.js"}

# Reports
report_script_custom_doctypes = ["StudentStageProgress"]
Expand Down
Loading