refactor: extract gmail matching service

This commit is contained in:
2026-04-01 16:59:29 +02:00
parent 61c12d3479
commit 69e78d8951
5 changed files with 180 additions and 172 deletions
@@ -0,0 +1,167 @@
using JobTrackerApi.Controllers;
using JobTrackerApi.Models;
namespace JobTrackerApi.Services;
public sealed record GmailMatchReason(string Label, string Value, int Points);
public sealed record GmailScoredMessageResult(
GmailMessageSummary Message,
bool AlreadyImported,
int Score,
string Confidence,
IReadOnlyList<string> MatchedQueries,
IReadOnlyList<GmailMatchReason> Reasons);
public interface IGmailJobMatchingService
{
IReadOnlyList<string> BuildJobQueries(JobApplication job, string? queryOverride);
GmailScoredMessageResult ScoreMessage(JobApplication job, GmailQueryMatchedMessage candidate, bool alreadyImported, bool threadAlreadyImported);
}
public sealed class GmailJobMatchingService : IGmailJobMatchingService
{
public IReadOnlyList<string> BuildJobQueries(JobApplication job, string? queryOverride)
{
var queries = new List<string>();
void Add(string? query)
{
if (!string.IsNullOrWhiteSpace(query))
{
queries.Add(query.Trim());
}
}
Add(queryOverride);
if (!string.IsNullOrWhiteSpace(job.Company?.RecruiterEmail))
Add($"(from:{job.Company.RecruiterEmail.Trim()} OR to:{job.Company.RecruiterEmail.Trim()}) newer_than:365d");
if (!string.IsNullOrWhiteSpace(job.Company?.RecruiterName))
Add($"\"{job.Company.RecruiterName.Trim()}\" newer_than:365d");
if (!string.IsNullOrWhiteSpace(job.Company?.Name) && !string.IsNullOrWhiteSpace(job.JobTitle))
Add($"\"{job.Company.Name.Trim()}\" \"{job.JobTitle.Trim()}\" newer_than:365d");
if (!string.IsNullOrWhiteSpace(job.Company?.Name))
Add($"\"{job.Company.Name.Trim()}\" (application OR interview OR recruiter OR role OR position) newer_than:365d");
if (!string.IsNullOrWhiteSpace(job.JobTitle))
Add($"subject:\"{job.JobTitle.Trim()}\" newer_than:365d");
foreach (var subject in job.Messages
.Select(message => message.Subject)
.Where(subject => !string.IsNullOrWhiteSpace(subject))
.Distinct(StringComparer.OrdinalIgnoreCase)
.Take(2))
{
Add($"subject:\"{subject!.Trim()}\" newer_than:365d");
}
if (queries.Count == 0)
Add("newer_than:365d (application OR interview OR recruiter OR role OR position)");
return queries.Distinct(StringComparer.OrdinalIgnoreCase).ToList();
}
public GmailScoredMessageResult ScoreMessage(JobApplication job, GmailQueryMatchedMessage candidate, bool alreadyImported, bool threadAlreadyImported)
{
var reasons = new List<GmailMatchReason>();
var score = 0;
var message = candidate.Message;
var subject = message.Subject ?? string.Empty;
var from = message.From ?? string.Empty;
var to = message.To ?? string.Empty;
var snippet = message.Snippet ?? string.Empty;
var haystack = $"{subject} {from} {to} {snippet}";
if (candidate.MatchedQueries.Count > 0)
{
var queryHitPoints = Math.Min(12, candidate.MatchedQueries.Count * 4);
score += queryHitPoints;
reasons.Add(new GmailMatchReason("queryHits", candidate.MatchedQueries.Count.ToString(), queryHitPoints));
}
if (!string.IsNullOrWhiteSpace(job.Company?.Name) && ContainsValue(haystack, job.Company.Name))
{
score += 18;
reasons.Add(new GmailMatchReason("company", job.Company.Name.Trim(), 18));
}
if (!string.IsNullOrWhiteSpace(job.Company?.RecruiterEmail) && (ContainsValue(from, job.Company.RecruiterEmail) || ContainsValue(to, job.Company.RecruiterEmail)))
{
score += 20;
reasons.Add(new GmailMatchReason("recruiterEmail", job.Company.RecruiterEmail.Trim(), 20));
}
if (!string.IsNullOrWhiteSpace(job.Company?.RecruiterName) && ContainsValue(haystack, job.Company.RecruiterName))
{
score += 12;
reasons.Add(new GmailMatchReason("recruiter", job.Company.RecruiterName.Trim(), 12));
}
foreach (var token in SplitTerms(job.JobTitle).Take(4))
{
if (!ContainsValue(haystack, token)) continue;
score += 5;
reasons.Add(new GmailMatchReason("jobTitle", token, 5));
}
foreach (var subjectLine in job.Messages
.Select(existing => existing.Subject)
.Where(existing => !string.IsNullOrWhiteSpace(existing))
.Distinct(StringComparer.OrdinalIgnoreCase)
.Take(2))
{
if (!ContainsValue(subject, subjectLine!)) continue;
score += 8;
reasons.Add(new GmailMatchReason("existingSubject", subjectLine!.Trim(), 8));
}
if (message.Date is { } messageDate)
{
var ageDays = Math.Abs((DateTimeOffset.UtcNow - messageDate).TotalDays);
if (ageDays <= 45)
{
score += 4;
reasons.Add(new GmailMatchReason("recency", "45d", 4));
}
else if (ageDays <= 180)
{
score += 2;
reasons.Add(new GmailMatchReason("recency", "180d", 2));
}
}
if (threadAlreadyImported && !alreadyImported)
reasons.Add(new GmailMatchReason("status", "thread-already-imported", 0));
if (alreadyImported)
reasons.Add(new GmailMatchReason("status", "already-imported", 0));
reasons = reasons
.GroupBy(reason => new { reason.Label, reason.Value, reason.Points })
.Select(group => group.First())
.OrderByDescending(reason => reason.Points)
.ThenBy(reason => reason.Label, StringComparer.Ordinal)
.ThenBy(reason => reason.Value, StringComparer.Ordinal)
.ToList();
return new GmailScoredMessageResult(message, alreadyImported, score, ToConfidence(score), candidate.MatchedQueries, reasons);
}
private static bool ContainsValue(string haystack, string? value)
=> !string.IsNullOrWhiteSpace(value) && haystack.Contains(value.Trim(), StringComparison.OrdinalIgnoreCase);
private static IEnumerable<string> SplitTerms(string? value)
{
if (string.IsNullOrWhiteSpace(value)) yield break;
foreach (var token in value.Split(new[] { ' ', '/', '-', ',', '.', '(', ')', ':' }, StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries)
.Where(token => token.Length >= 3)
.Distinct(StringComparer.OrdinalIgnoreCase))
{
yield return token;
}
}
private static string ToConfidence(int score) => score switch
{
>= 30 => "high",
>= 16 => "medium",
_ => "low"
};
}