marcus/pdf-umbenenner
1
0
Fork 0
Code Activity

M4 AP-006 Reihenfolge, Konsistenz und Scope bereinigen

Browse Source
This commit is contained in:
Marcus van Elst
2026-04-03 07:52:21 +02:00
parent 00c4cf1e5c
commit ca91a60cad
3 changed files with 127 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

94
pdf-umbenenner-application/src/main/java/de/gecheckt/pdf/umbenenner/application/service/M4DocumentProcessor.java
View File

@@ -192,6 +192,100 @@ public class M4DocumentProcessor {
}
}
/**
* Applies the full M4 processing logic for one identified document candidate.
* <p>
* The caller must have already computed a valid {@link DocumentFingerprint} for the
* candidate. This method handles the complete M4 processing flow:
* <ol>
* <li>Load document master record.</li>
* <li>Handle terminal SUCCESS / FAILED_FINAL skip cases first.</li>
* <li>Only if not terminal: execute the M3 flow (PDF extraction + pre-checks).</li>
* <li>Map M3 outcome to M4 status, counters and retryable flag.</li>
* <li>Persist exactly one historised processing attempt.</li>
* <li>Persist the updated document master record.</li>
* </ol>
* <p>
* This method never throws. All persistence failures are caught, logged, and
* treated as controlled per-document failures so the batch run can continue.
*
* @param candidate the source document candidate being processed; must not be null
* @param fingerprint the successfully computed fingerprint for this candidate;
* must not be null
* @param context the current batch run context (for run ID and timing);
* must not be null
* @param attemptStart the instant at which processing of this candidate began;
* must not be null
* @param m3Executor functional interface to execute the M3 pipeline when needed;
* must not be null
*/
public void processWithM3Execution(
SourceDocumentCandidate candidate,
DocumentFingerprint fingerprint,
BatchRunContext context,
Instant attemptStart,
java.util.function.Function<SourceDocumentCandidate, DocumentProcessingOutcome> m3Executor) {
Objects.requireNonNull(candidate, "candidate must not be null");
Objects.requireNonNull(fingerprint, "fingerprint must not be null");
Objects.requireNonNull(context, "context must not be null");
Objects.requireNonNull(attemptStart, "attemptStart must not be null");
Objects.requireNonNull(m3Executor, "m3Executor must not be null");
// Step 1: Load the document master record
DocumentRecordLookupResult lookupResult =
documentRecordRepository.findByFingerprint(fingerprint);
// Step 2: Handle persistence lookup failure – cannot safely proceed
if (lookupResult instanceof PersistenceLookupTechnicalFailure failure) {
LOG.error("Cannot process '{}': master record lookup failed: {}",
candidate.uniqueIdentifier(), failure.errorMessage());
return;
}
// Step 3: Determine the action based on the lookup result
switch (lookupResult) {
case DocumentTerminalSuccess terminalSuccess -> {
// Document already successfully processed → skip
LOG.info("Skipping '{}': already successfully processed (fingerprint: {}).",
candidate.uniqueIdentifier(), fingerprint.sha256Hex());
persistSkipAttempt(
candidate, fingerprint, terminalSuccess.record(),
ProcessingStatus.SKIPPED_ALREADY_PROCESSED,
context, attemptStart);
}
case DocumentTerminalFinalFailure terminalFailure -> {
// Document finally failed → skip
LOG.info("Skipping '{}': already finally failed (fingerprint: {}).",
candidate.uniqueIdentifier(), fingerprint.sha256Hex());
persistSkipAttempt(
candidate, fingerprint, terminalFailure.record(),
ProcessingStatus.SKIPPED_FINAL_FAILURE,
context, attemptStart);
}
case DocumentUnknown ignored -> {
// New document – execute M3 pipeline and process
DocumentProcessingOutcome m3Outcome = m3Executor.apply(candidate);
processAndPersistNewDocument(candidate, fingerprint, m3Outcome, context, attemptStart);
}
case DocumentKnownProcessable knownProcessable -> {
// Known but not terminal – execute M3 pipeline and process
DocumentProcessingOutcome m3Outcome = m3Executor.apply(candidate);
processAndPersistKnownDocument(
candidate, fingerprint, m3Outcome, knownProcessable.record(),
context, attemptStart);
}
default ->
// Exhaustive sealed hierarchy; this branch is unreachable
LOG.error("Unexpected lookup result type for '{}': {}",
candidate.uniqueIdentifier(), lookupResult.getClass().getSimpleName());
}
}
// -------------------------------------------------------------------------
// Skip path
// -------------------------------------------------------------------------

30
pdf-umbenenner-application/src/main/java/de/gecheckt/pdf/umbenenner/application/usecase/DefaultBatchRunProcessingUseCase.java
View File

@@ -194,9 +194,15 @@ public class DefaultBatchRunProcessingUseCase implements BatchRunProcessingUseCa
* <li>Compute the SHA-256 fingerprint of the candidate file content.</li>
* <li>If fingerprint computation fails: log as non-identifiable run event and
* return — no SQLite record is created.</li>
* <li>Execute the M3 pipeline (PDF extraction + pre-checks).</li>
* <li>Delegate to {@link M4DocumentProcessor} for idempotency check, status/counter
* mapping, and consistent two-level persistence.</li>
* <li>Load document master record.</li>
* <li>If already {@code SUCCESS} → persist skip attempt with
* {@code SKIPPED_ALREADY_PROCESSED}.</li>
* <li>If already {@code FAILED_FINAL} → persist skip attempt with
* {@code SKIPPED_FINAL_FAILURE}.</li>
* <li>Otherwise execute the M3 pipeline (extraction + pre-checks).</li>
* <li>Map M3 result into M4 status, counters and retryable flag.</li>
* <li>Persist exactly one historised processing attempt.</li>
* <li>Persist the updated document master record.</li>
* </ol>
* <p>
* Per-document errors do not abort the overall batch run. Each candidate ends
@@ -227,15 +233,15 @@ public class DefaultBatchRunProcessingUseCase implements BatchRunProcessingUseCa
LOG.debug("Fingerprint computed for '{}': {}",
candidate.uniqueIdentifier(), fingerprint.sha256Hex());
// Step M4-2..M4-8: Execute M3 pipeline and delegate M4 logic to the processor
// The M3 pipeline runs only if the document is not in a terminal state;
// M4DocumentProcessor handles the terminal check internally.
// We run M3 eagerly here and pass the result; M4DocumentProcessor will
// ignore it for terminal documents.
DocumentProcessingOutcome m3Outcome = runM3Pipeline(candidate);
// Delegate idempotency check, status mapping, and persistence to M4DocumentProcessor
m4DocumentProcessor.process(candidate, fingerprint, m3Outcome, context, attemptStart);
// Delegate the complete M4 processing logic to the processor
// The processor handles loading document master record, checking terminal status,
// executing M3 pipeline only when needed, and persisting results consistently
m4DocumentProcessor.processWithM3Execution(
candidate,
fingerprint,
context,
attemptStart,
this::runM3Pipeline); // Pass the M3 executor as a function
}
}
}

12
pdf-umbenenner-application/src/test/java/de/gecheckt/pdf/umbenenner/application/usecase/BatchRunProcessingUseCaseTest.java
View File

@@ -641,6 +641,18 @@ class BatchRunProcessingUseCaseTest {
super.process(candidate, fingerprint, m3Outcome, context, attemptStart);
}
@Override
public void processWithM3Execution(
de.gecheckt.pdf.umbenenner.domain.model.SourceDocumentCandidate candidate,
de.gecheckt.pdf.umbenenner.domain.model.DocumentFingerprint fingerprint,
de.gecheckt.pdf.umbenenner.domain.model.BatchRunContext context,
java.time.Instant attemptStart,
java.util.function.Function<de.gecheckt.pdf.umbenenner.domain.model.SourceDocumentCandidate, de.gecheckt.pdf.umbenenner.domain.model.DocumentProcessingOutcome> m3Executor) {
processCallCount++;
// Delegate to super so the real logic runs (with no-op repos)
super.processWithM3Execution(candidate, fingerprint, context, attemptStart, m3Executor);
}
int processCallCount() { return processCallCount; }
}