added number of retries

src/cristallina/uscan.py

@@ -8,6 +8,9 @@ from sfdata import SFDataFiles, SFScanInfo
 
 from partialjson.json_parser import JSONParser
 
+from loguru import logger
+
+MAX_STEP_WAIT = 300  # Maximum wait time in seconds for files to appear
 
 class UnfinishedScanInfo(SFScanInfo):
     """
@@ -69,18 +72,26 @@ class UnfinishedScanInfo(SFScanInfo):
         yielded_count = 0
         
         while True:
+            retries = 0
             self._parse_partial_json(self.fname) 
 
             # Check if we have new files to yield
             while self.files and len(self.files) > yielded_count:
+
                 fns = self.files[yielded_count]
-                
+                       
                 if not files_available_on_disk(fns):
                     time.sleep(self.refresh_interval)
-                    continue  # Wait and recheck
+                    retries += 1
+                    if (retries * self.refresh_interval) < MAX_STEP_WAIT:  # Wait up to 5 minutes for files to appear
+                        continue  # Wait and recheck
+                    else:
+                        logger.error(f"Timeout waiting for files to become available for step {yielded_count} {fns}")
+                        # we still yield the remaining files to avoid infinite loop, but log an error and leave it to the caller to handle missing data
 
                 yielded_count += 1
-
+                retries = 0
+                
                 try:
                     with SFDataFiles(*fns) as data:
                         yield data
@@ -117,6 +128,8 @@ def is_finished_scan(fname):
 
 def files_available_on_disk(fnames):
     """Check if all files for this step are available on disk and contain some data."""
+    # fnames = [fn for fn in fnames if not "PVDATA" in fn]  # PVDATA files are not written to disk at the moment!
+    # logger.debug(f"Skipping PVDATA files for availability check as a workaround!")
     if all(Path(fn).exists() for fn in fnames):
        return all(Path(fn).stat().st_size > 0 for fn in fnames)
     return False