Merge pull request #20297 from overleaf/mj-stream-bib-log

[web] Stream .blg files before parsing

GitOrigin-RevId: 2e7a16a19e62fc32bbcb7fe35bed745b938e6963

services/web/frontend/js/features/pdf-preview/util/output-files.js

@@ -9,6 +9,7 @@ import '@/utils/readable-stream-async-iterator-polyfill'
 const TRANSIENT_WARNING_REGEX = /^(Reference|Citation).+undefined on input line/
 
 const MAX_LOG_SIZE = 1024 * 1024 // 1MB
+const MAX_BIB_LOG_SIZE_PER_FILE = MAX_LOG_SIZE
 
 export function handleOutputFiles(outputFiles, projectId, data) {
   const outputFile = outputFiles.get('output.pdf')
@@ -77,31 +78,11 @@ export const handleLogFiles = async (outputFiles, data, signal) => {
   const logFile = outputFiles.get('output.log')
 
   if (logFile) {
-    try {
-      const logFileAbortController = new AbortController()
-
-      // abort fetching the log file if the main signal is aborted
-      signal.addEventListener('abort', () => {
-        logFileAbortController.abort()
-      })
-
-      const response = await fetch(buildURL(logFile, data.pdfDownloadDomain), {
-        signal: logFileAbortController.signal,
-      })
-
-      result.log = ''
-
-      const reader = response.body.pipeThrough(new TextDecoderStream())
-      for await (const chunk of reader) {
-        result.log += chunk
-        if (result.log.length > MAX_LOG_SIZE) {
-          logFileAbortController.abort()
-        }
-      }
-    } catch (e) {
-      debugConsole.warn(e) // ignore failure to fetch the log file, but log a warning
-    }
-
+    result.log = await fetchFileWithSizeLimit(
+      buildURL(logFile, data.pdfDownloadDomain),
+      signal,
+      MAX_LOG_SIZE
+    )
     try {
       let { errors, warnings, typesetting } = HumanReadableLogs.parse(
         result.log,
@@ -129,23 +110,18 @@ export const handleLogFiles = async (outputFiles, data, signal) => {
     }
   }
   for (const blgFile of blgFiles) {
+    const log = await fetchFileWithSizeLimit(
+      buildURL(blgFile, data.pdfDownloadDomain),
+      signal,
+      MAX_BIB_LOG_SIZE_PER_FILE
+    )
     try {
-      const response = await fetch(buildURL(blgFile, data.pdfDownloadDomain), {
-        signal,
-      })
-
-      const log = await response.text()
-
-      try {
-        const { errors, warnings } = new BibLogParser(log, {
-          maxErrors: 100,
-        }).parse()
-        accumulateResults({ errors, warnings }, 'BibTeX:')
-      } catch (e) {
-        // BibLog parsing errors are ignored
-      }
+      const { errors, warnings } = new BibLogParser(log, {
+        maxErrors: 100,
+      }).parse()
+      accumulateResults({ errors, warnings }, 'BibTeX:')
     } catch (e) {
-      debugConsole.warn(e) // ignore failure to fetch/parse the log file, but log a warning
+      // BibLog parsing errors are ignored
     }
   }
 
@@ -260,3 +236,33 @@ function normalizeFilePath(path, rootDocDirname) {
 function isTransientWarning(warning) {
   return TRANSIENT_WARNING_REGEX.test(warning.message)
 }
+
+async function fetchFileWithSizeLimit(url, signal, maxSize) {
+  let result = ''
+  try {
+    const abortController = new AbortController()
+    // abort fetching the log file if the main signal is aborted
+    signal.addEventListener('abort', () => {
+      abortController.abort()
+    })
+
+    const response = await fetch(url, {
+      signal: abortController.signal,
+    })
+
+    if (!response.ok) {
+      throw new Error('Failed to fetch log file')
+    }
+
+    const reader = response.body.pipeThrough(new TextDecoderStream())
+    for await (const chunk of reader) {
+      result += chunk
+      if (result.length > maxSize) {
+        abortController.abort()
+      }
+    }
+  } catch (e) {
+    debugConsole.warn(e) // ignore failure to fetch the log file, but log a warning
+  }
+  return result
+}