Closes#41
- Add RequireKeywordInAnchor per-provider flag (default true); set false for
ejobs.ro and bestjobs.eu so Stage 2 anchor-text filter is skipped — their
search URL already filters by relevance server-side
- Update AI system prompts (en + ro) to extract concise job-board-friendly
keywords (role title + key tech, not abstract concepts) and candidate location
- Propagate location through JobMatchResponse -> CreateJobSearchTokenRequest ->
JobSearchTokenEntity -> JobSearchSessionEntity
- Add {location} and {location-slug} substitution in HtmlJobSearcher
- Update provider SearchUrlTemplates to include location:
ejobs.ro: /locuri-de-munca/{location-slug}?q={keywords}
bestjobs.eu: /ro/locuri-de-munca-in-{location-slug}?keywords={keywords}
linkedin.com: ?keywords={keywords}&location={location}
- Three new migrations: AddRequireKeywordInAnchorAndLocation,
ImproveKeywordsAndAddLocation, AddLocationToProviders
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
UseInternalApiKeyProtection was registered after UseSwaggerInDevelopment,
allowing unauthenticated access to /swagger. Swapped order to match
rag-api and cv-matcher-api.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
email-data references shared-data but the email-api Dockerfile never
copied it into the build context, causing MSB9008 during Docker build.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Serilog.Settings.Configuration cannot deserialize NetworkCredential or
MailKit's SecureSocketOptions from JSON, causing an InvalidOperationException
in the binder and preventing containers from starting.
Fix: remove Email from the WriteTo JSON array entirely and wire it in code
inside ConfigureJsonSerilog using a dedicated SerilogEmail:* config section.
The sink is skipped when From/To/Host are absent, so local dev is unaffected.
Also renames the docker-compose env vars from the verbose
Serilog__WriteTo__2__Args__* prefix to the clean SerilogEmail__* prefix.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
HTTP and Playwright fetch failures in HtmlJobSearcher are now logged at
Error so that Serilog's email sink triggers an alert when a job provider
is unreachable. Per-URL match failures remain at Warning (expected noise).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Serilog.Sinks.Email v4 renamed all configuration parameters from their
v2 names. The old names were silently ignored, so no error alert emails
were ever sent.
Parameter renames applied across all 6 appsettings.json and docker-compose:
fromEmail → from
toEmail → to
mailServer → host
networkCredential → credentials
enableSsl: true → connectionSecurity: StartTls
emailSubject → subject
outputTemplate → body
batchPostingLimit / period removed (v4 batching uses a separate overload)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
5 requests / 1 min per IP. docker-compose.yml wired with ${VAR:-default}.
Staging and production .env files updated locally (gitignored).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Require captchaToken query param on initial (non-range) download requests
- Range requests (HTTP resume) bypass captcha — they are continuations of an already-validated download
- Add download rate limit policy: 5 requests / 1 min per IP (configured in .env)
- Inject ICaptchaVerifier; action name is file_download
UI change required: execute grecaptcha.execute(siteKey, {action: 'file_download'})
before triggering the download and append ?captchaToken=<token> to the URL.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
These templates belong to the myAi schema (myai-data) and are read by
CvMatcherController via ITemplateService. The email-data copies were
never read by any code — removing them to avoid confusion.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The job-search status page HTML wrapper was baked into a static helper
method in CvMatcherController. Extracted to a new template key
html.job-search.shell (*) with {{title}} and {{message}} placeholders.
Added to AddTemplates seed and a new AddHtmlJobSearchShell migration
for existing DBs. Controller now calls _templates.Render() for all paths.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The LLM JSON shape was missing the keywords array so res.Keywords was
always empty, causing "none detected" in job search emails. Both en/ro
prompts now include "keywords" in the required JSON shape so the LLM
extracts relevant job-search terms from the CV/job pair.
Note: the cvMatcher.CvMatchResults cache must be cleared on existing DBs
so cached responses (which lack keywords) are not served.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace div/CSS-class approach with nested table layout so the 600px
container is enforced via HTML attributes, not a <style> block that
Outlook strips. Also removes border-radius and display:inline-block.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace div-based layouts with table-based HTML throughout (max-width/border-radius/display:inline-block ignored by Outlook)
- email.match.body: width:100% table with per-cell borders and fixed 130px label column
- email.match.job-search-footer: table-based button with bgcolor attribute
- email.search-results.empty: div replaced with full-width table
- email.search-results.body: remove div wrapper around items
- Add email.search-results.scan-summary and email.search-results.item templates
- CvSearchEmailSender: remove all hardcoded HTML; render via IEmailTemplateService
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Changed table width from 100% to max-width: 500px with margin: 0 auto
- Applies to both English and Romanian email.match.body templates
- Table now narrower and centered in email
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Convert email.match.body, email.match.job-search-footer, email.search-results.body, and email.search-results.empty templates from plain text to proper HTML format in InitialSchema migration
- Update EmailApiEmailSender.BuildMatchEmailBody() to work with HTML templates instead of plain text
- Add WebUtility.HtmlEncode() for security when inserting dynamic content (summary)
- Templates now use semantic HTML tags (table, h2, h3, ul, li, p, div, hr, a) instead of plain text with newlines
- All 32 email template variants (16 keys × 2 languages) and 8 html.job-search.* templates seeded via migration
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Remove html-shell entries from InitialSchema Seed() method
- Create new AddHtmlShellTemplates migration to insert html-shell templates
- Prevents duplicate key errors from having same data in two migrations
- InitialSchema seeds 32 templates (16 keys × 2 languages)
- AddHtmlShellTemplates seeds 2 html-shell templates (start, end)
- Total: 34 templates after both migrations run
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Add email.html-shell.start: Opening HTML wrapper with blue header and MyAi.ro branding
- Add email.html-shell.end: Closing HTML wrapper with footer
- These templates wrap HTML email bodies before sending via SmtpEmailDispatcher
- Language key set to '*' (language-agnostic)
- Ensures email shell templates are seeded automatically on fresh database initialization
Fixes the "Email template not found: key='email.html-shell.start'" error that prevented email sending.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Email migration includes seed data for 14 templates (en, ro)
- CV matcher migration includes seed data for 2 AI prompts (en, ro)
- Tables are created successfully by migrations
- Issue: migrationBuilder.Sql() statements not being executed by EF Core
- Workaround needed: Current seeding approach not working automatically
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Changed configuration error handling to throw InvalidOperationException instead of silently using fallback values. This ensures:
1. Missing email templates (critical config) → 500 error to UI
2. Missing AI prompts (critical config) → 500 error to UI
3. Clear error messages indicating config issue
4. Prompts administrators to check database seeding
Services updated:
- EmailTemplateService.Get() throws for missing template
- CvMatcherService.ScorePairAsync() throws for missing AI prompt
This prevents silent failures with degraded service quality and makes it obvious to users that the system has a configuration problem that needs fixing.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Refactored the AI prompt system to use proper language-specific prompts (en and ro) instead of a single wildcard prompt with runtime {{languageName}} placeholder substitution.
Benefits:
- Language-specific instructions optimized for each language
- Better control over LLM behavior per language
- Cleaner code without placeholder substitution
- Easier to maintain and update prompts per language
Changes:
- Updated cvMatcher InitialSchema migration to seed en and ro prompts separately
- Modified CvMatcherService to retrieve language-specific prompts directly
- Removed LanguageName() helper method (no longer needed)
- Added fallback prompts in service for safety
The English and Romanian prompts now include specific JSON examples in their respective languages, ensuring the LLM understands the expected output format for each language variant.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Added seeding of the ai.cv-match.system-prompt to the AiPrompts table. This prompt is retrieved by CvMatcherService when scoring CV-job pairs with the LLM. The {{languageName}} placeholder is substituted at runtime based on the requested language.
The prompt has a fallback in the service code, but seeding it ensures the proper version is used and avoids relying on the hardcoded fallback.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
EF Core migration scaffolding doesn't recognize InsertData calls made through local functions in manually-edited migrations. Changed to use raw SQL INSERT statements with migrationBuilder.Sql() to directly populate the Templates table with all required email.* and html.job-search.* templates (en+ro).
This ensures templates are present when EmailTemplateService loads the cache, preventing 'Email template not found' warnings and enabling proper email rendering for CV match results and job search pages.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
When matching CVs, the system was finding no templates for email rendering because the email.Templates table was empty. The templates were seeded to the myAi schema, not the email schema.
Added seeding of all required email.* and html.job-search.* templates (en+ro) to the email-data InitialSchema migration. This ensures templates are automatically populated when the migration runs.
Templates seeded:
- email.match.subject, .body, .job-search-footer (en+ro)
- email.search-results.subject, .body, .empty (en+ro)
- html.job-search.started.*, .already-used.*, .expired.*, .invalid.*, .error.* (en+ro)
This fixes the issue where EmailTemplateService would log "Email template not found" warnings and return template keys as fallback text, causing match result emails to fail rendering.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Deleted all incremental migrations and regenerated fresh InitialSchema migrations
that contain the complete, correct schema from the start:
- CvMatcher: InitialSchema with 3-column unique constraint on Results
(CvDocumentId, JobDocumentId, Language)
- Email: InitialSchema with Templates table (consolidated from EmailTemplates)
This creates a cleaner migration history and faster fresh deployments. Since there
is no production data, consolidation is safe and improves maintainability.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
This migration renames the EmailTemplates table to Templates in the email schema.
The migration was scaffolded by EF Core with manual RenameTable commands added.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Refactoring:
- Rename EmailApiDbContext class to EmailDbContext for consistency with other DbContext naming
- Rename DbSet property from EmailTemplates to Templates
- Rename table from EmailTemplates to Templates
- Update all references in Program.cs files (email-api, api, cv-search-job)
- Update all migration files and model snapshot
- Fix cv-search-job migrations assembly name: email-api-data → email-data
This improves naming consistency across the solution.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
When a match result is inserted concurrently between the existence check and the
database insert, log a warning to help with diagnostics while gracefully handling
the idempotent duplicate key violation.
The Results table had a unique constraint on (CvDocumentId, JobDocumentId) but the code
expects uniqueness on (CvDocumentId, JobDocumentId, Language). When matching the same CV
against the same job in different languages, this caused duplicate key violations.
Changes:
- Updated CvMatcherDbContext to define 3-column unique index including Language
- Generated proper EF Core migration to drop 2-column index and create 3-column index
- Updated ModelSnapshot to reflect new 3-column index definition
- Added exception handling in SaveMatchAsync to gracefully handle any race conditions
where duplicate key violations could occur between the existence check and insert
The migration will be automatically applied on container startup via db.Database.Migrate().
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
The unique constraint on cvMatcher.Results was defined as (CvDocumentId, JobDocumentId)
but the code checks for (CvDocumentId, JobDocumentId, Language). This mismatch caused
duplicate key violations when matching the same CV+Job in different languages.
Update the constraint to (CvDocumentId, JobDocumentId, Language) to allow different
languages for the same CV+Job pair while preventing true duplicates.
Resolves: Duplicate key constraint violations on concurrent/repeated match requests
ejobs.ro migrated to a Nuxt SPA - plain HTTP GET returns only the JS
bundle. This change equips cv-search-job with a headless Chromium
(Playwright 1.60) so it can fully render SPA pages before extracting
job links.
- Add UseHeadlessBrowser flag to JobProviderEntity, JobProviderConfig,
and CvSearchDbContext; map it in JobTokenService.ToConfig so the flag
is included in the session provider-config snapshot
- Migration: add UseHeadlessBrowser column; fix ejobs.ro search URL
(remove /user/ prefix that caused 404) and set UseHeadlessBrowser=true
- HtmlJobSearcher: detect flag and dispatch to FetchWithPlaywrightAsync;
plain-HTTP path is unchanged; NetworkIdle timeout falls back to partial
content rather than failing outright
- Dockerfile: download Playwright Chromium in the SDK build stage via
npx; copy browser binaries to the final image; install Chromium system
libs (Ubuntu noble t64 variants)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Individual job listings on bestjobs.eu use /loc-de-munca/{slug} URLs.
The seeded JobLinkContains value /ro/locuri-de-munca/ matched only the
category navigation links (Vanzari, Inginerie, Management...), so
zero job URLs passed the stage-1 href filter and the scraper returned
nothing. Migration updates the stored record (Id=2) to /loc-de-munca/.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
string.Join("") produced no whitespace between inline-block spans,
causing keywords to visually merge in email clients that collapse margins.
Switched to string.Join(" ") and zeroed left margin on each badge so
they wrap cleanly without a gap on the first item.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Deleted stale directories and stray .csproj files that were never added
to the solution after project renames:
- Apis/cv-search-models/ (renamed → cv-search-data)
- Apis/myai-models/ (renamed → myai-data)
- Apis/shared-models/ (empty leftover)
- Apis/cv-search-data/cv-search-models.csproj (stray old csproj)
- Apis/myai-data/myai-models.csproj (stray old csproj)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two migration files had literal schema strings that were missed in earlier passes:
- cv-search-data AddJobSearchTables: two CreateIndex calls used "cvSearch"
- rag-data InitialRagSchema: FK principalSchema used "rag"
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Piggybacks keyword extraction onto the existing CV-to-job LLM call —
no extra API calls. The system prompt now instructs the model to return
8-12 English job-search terms (job titles, technologies, skills, domains)
in a new `keywords` field alongside the existing score/summary fields.
Keywords flow: LLM JSON → JobMatchResponse.Keywords → CreateJobSearchTokenRequest →
JobSearchTokenEntity.Keywords (stored comma-separated) → JobSearchSessionEntity.Keywords
(copied at session-creation time, no RAG call needed).
Changes:
- Add Keywords to JobMatchResponse, CreateJobSearchTokenRequest, JobSearchTokenEntity
- IJobTokenService.CreateTokenAsync now accepts IReadOnlyList<string> keywords
- JobTokenService: store keywords on token; TriggerStartAsync reads token.Keywords
instead of fetching CV text from RAG — removes IRagApiClient dependency
- Remove heuristic ExtractKeywords method
- Migration AddKeywordsToJobSearchTokens: adds Keywords column to cvSearch.JobSearchTokens
- Migration UpdateCvMatchSystemPromptKeywords: updates ai.cv-match.system-prompt seed
to include keywords in the JSON shape
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PDF text extraction often stores all content without newlines. The previous
line-based splitter would produce one line > 200 chars which was filtered out,
yielding empty keywords. Replace with word-level sampling of the first 2000
chars, splitting on whitespace and common delimiters, skipping phone fragments,
emails, and URLs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Down migration was referencing "emailApi" literal instead of MigrationConstants.SchemaName,
which would have dropped the wrong schema on rollback. Also fix stale comment in DbContext.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace hardcoded "cvSearch" string literals with MigrationConstants.SchemaName
in the Up, InsertData, and Down methods, consistent with all other migrations.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Provider config is no longer read from appsettings or env vars.
All three providers (ejobs.ro, bestjobs.eu, linkedin.com) are seeded
into cvSearch.JobProviders by the AddJobProviders migration.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
JobTokenService.CreateTokenAsync queries cvSearch.JobProviders for any
enabled row; returns null (no token created) when the table is empty or
all providers are disabled. TriggerStartAsync snapshots enabled providers
from DB at session-start time, preserving the existing snapshot contract.
CvMatcherController guards link-building on a non-null TokenId so the
"Start a job search" CTA is omitted from match emails when no providers
are configured.
JobSearchSettings.Providers list removed — provider config now lives
exclusively in the DB. CvSearchJobTask.GetProviders falls back to an
empty list with a warning (snapshot should always be populated from DB).
Closes#35
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
New JobProviderEntity persists provider config (name, URL template,
link filter, initial keywords, max results, display order) in the DB
instead of appsettings. Migration seeds three disabled defaults:
ejobs.ro, bestjobs.eu, and linkedin.com.
Closes#35
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add funnel-level logging to HtmlJobSearcher (total anchors found,
stage-1 href-filter count, stage-2 keyword-filter count) and warn
when the keyword list is empty. Log the full search URL and response
size to catch silent HTTP failures or bot-block pages.
In CvSearchJobTask, log keywords and active providers at session start,
per-provider URL counts after each scrape, and every scored URL with its
verdict (ACCEPTED / rejected) at Information level.
Add a scan summary block to the results email (both non-empty and
empty-results paths) showing the CV keywords used as chips and the
comma-separated list of providers scanned.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
claude
gelu