shashank created CAMEL-27067:
--------------------------------

             Summary: camel-stream - a consumer with scanStream=true and 
readLine=false reads the stream but never routes the data
                 Key: CAMEL-27067
                 URL: https://issues.apache.org/jira/browse/CAMEL-27067
             Project: Camel
          Issue Type: Bug
          Components: camel-stream
            Reporter: shashank
            Assignee: shashank


h3. What happens

With {{readLine=false}} the stream consumer reads the stream in raw mode 
({{readFromStreamRawMode}}). Without {{scanStream}} it reads the stream once 
and creates one exchange with the bytes. With {{scanStream=true}} it loops and 
calls {{is.readAllBytes()}} on each scan, but the bytes it read are never 
passed to {{processRaw}}: no exchange is created, and the data is dropped. A 
route such as

{code}
from("stream:file?fileName=app.log&scanStream=true&readLine=false")
    .to("mock:result");
{code}

consumes the file (and everything appended to it later) and never sends 
anything.

Also, in this mode a file that does not exist when the consumer starts gives a 
null stream, and the first {{readAllBytes()}} throws a {{NullPointerException}} 
that ends the consumer thread. The line mode handles a null reader.

h3. How to reproduce

{{ScanStreamFileRawModeTest}} (camel-stream): writes "Hello\nWorld\n" to a 
file, starts 
{{stream:file?...&scanStream=true&scanStreamDelay=100&readLine=false}}, and 
expects one exchange with that text, then appends "Bye\n" and expects a second 
exchange with it. On main no exchange arrives and the test fails after 10 
seconds.

h3. Fix

In the scan loop of the raw mode, route the bytes that were read with 
{{processRaw}} (and keep the index), and reopen the stream for {{retry}} only 
at the end of the stream, as the line mode does (reopening after each read 
would read the same data again). A null stream counts as the end of the stream. 
This also ends a busy loop on main: with {{scanStream=true&retry=true}} the raw 
mode reopened the stream after every read and only slept at the end of the 
stream, so a non-empty file (or URL) was read again and again without a pause. 
With {{retry}} a reopen still starts from the beginning of the file, as in the 
line mode, so {{fileWatcher}} is the option to use for a file that grows. Scan 
exchanges carry {{CamelStreamComplete=false}}, as in the line mode. The 
non-scan raw mode now also keeps the index it gets back from {{processRaw}}.

camel-4.18.x has the same code.

_Filed with Claude Code on behalf of allthingssecurity._



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

Reply via email to