From b5baaa38c328b38e753b54cc1f3632c2815445cb Mon Sep 17 00:00:00 2001
From: huangrh <huangrh@chinaweal.com.cn>
Date: Thu, 19 Mar 2026 14:36:51 +0800
Subject: [PATCH] Prefer venv Python and normalize venv extraction

---
 .../common/utils/ResourceExtractor.java       |  25 +++++
 .../ocr/engine/FlaskProcessManager.java       | 105 ++++++++++++------
 2 files changed, 94 insertions(+), 36 deletions(-)

diff --git a/src/main/java/com/chinaweal/youfool/reportdetect/common/utils/ResourceExtractor.java b/src/main/java/com/chinaweal/youfool/reportdetect/common/utils/ResourceExtractor.java
index 0b14894..f1d5b22 100644
--- a/src/main/java/com/chinaweal/youfool/reportdetect/common/utils/ResourceExtractor.java
+++ b/src/main/java/com/chinaweal/youfool/reportdetect/common/utils/ResourceExtractor.java
@@ -298,6 +298,7 @@ public class ResourceExtractor {
             Files.createDirectories(targetDir);
             copyDirectoryFromClasspath("python-runtime/venv-offline", targetDir);
         }
+        normalizeVenvDir(targetDir.getParent(), targetDir);
         makeExecutable(targetDir.resolve("bin/python3.10"));
         makeExecutable(targetDir.resolve("bin/python3"));
         log.info("      Done");
@@ -468,4 +469,28 @@ public class ResourceExtractor {
             }
         }
     }
+
+    private void normalizeVenvDir(Path parentDir, Path expectedDir) throws IOException {
+        if (Files.isDirectory(expectedDir)) {
+            return;
+        }
+        Path altLinux = parentDir.resolve("venv-linux-offline");
+        if (Files.isDirectory(altLinux)) {
+            log.info("      Renaming venv directory: {} -> {}", altLinux.getFileName(), expectedDir.getFileName());
+            Files.move(altLinux, expectedDir, StandardCopyOption.REPLACE_EXISTING);
+            return;
+        }
+        Path altVenv = parentDir.resolve("venv");
+        if (Files.isDirectory(altVenv)) {
+            log.info("      Renaming venv directory: {} -> {}", altVenv.getFileName(), expectedDir.getFileName());
+            Files.move(altVenv, expectedDir, StandardCopyOption.REPLACE_EXISTING);
+            return;
+        }
+        Path altOff = parentDir.resolve("venv-offline");
+        if (Files.isDirectory(altOff)) {
+            // already present under expected name in parent, but expectedDir path differs
+            return;
+        }
+        log.warn("      Venv directory not found after extraction under {}", parentDir);
+    }
 }
diff --git a/src/main/java/com/chinaweal/youfool/reportdetect/modules/ocr/engine/FlaskProcessManager.java b/src/main/java/com/chinaweal/youfool/reportdetect/modules/ocr/engine/FlaskProcessManager.java
index 7bd12e0..601e52b 100644
--- a/src/main/java/com/chinaweal/youfool/reportdetect/modules/ocr/engine/FlaskProcessManager.java
+++ b/src/main/java/com/chinaweal/youfool/reportdetect/modules/ocr/engine/FlaskProcessManager.java
@@ -153,10 +153,20 @@ public class FlaskProcessManager implements ApplicationListener<ContextRefreshed
             if (libPath != null) {
                 env.put("PYTHONPATH", libPath.toString());
             }
-            // Prefer embedded Python home when using standalone runtime
-            Path pythonHome = resolveEmbeddedPythonHome();
-            if (pythonHome != null) {
-                env.put("PYTHONHOME", pythonHome.toString());
+            Path venvRoot = resolveEmbeddedVenvRoot();
+            boolean usingVenv = venvRoot != null && embeddedPython != null &&
+                embeddedPython.toAbsolutePath().startsWith(venvRoot.toAbsolutePath());
+            if (usingVenv) {
+                Path venvBin = venvRoot.resolve(isWindows() ? "Scripts" : "bin");
+                env.put("VIRTUAL_ENV", venvRoot.toString());
+                String oldPath = env.getOrDefault("PATH", "");
+                env.put("PATH", venvBin.toString() + File.pathSeparator + oldPath);
+            } else {
+                // Prefer embedded Python home when using standalone runtime
+                Path pythonHome = resolveEmbeddedPythonHome();
+                if (pythonHome != null) {
+                    env.put("PYTHONHOME", pythonHome.toString());
+                }
             }
 
             // Use bundled models
@@ -338,7 +348,40 @@ public class FlaskProcessManager implements ApplicationListener<ContextRefreshed
     }
 
     private Path resolveEmbeddedPython() {
-        // Prefer embedded standalone runtime first
+        // Prefer venv interpreter first (ensures Flask and deps are available)
+        Path venvRoot = resolveEmbeddedVenvRoot();
+        if (venvRoot != null) {
+            Path linuxPython310 = venvRoot.resolve("bin/python3.10");
+            if (Files.exists(linuxPython310)) {
+                return linuxPython310;
+            }
+            Path linuxPython311 = venvRoot.resolve("bin/python3.11");
+            if (Files.exists(linuxPython311)) {
+                return linuxPython311;
+            }
+            Path linuxPython39 = venvRoot.resolve("bin/python3.9");
+            if (Files.exists(linuxPython39)) {
+                return linuxPython39;
+            }
+            Path linuxPython38 = venvRoot.resolve("bin/python3.8");
+            if (Files.exists(linuxPython38)) {
+                return linuxPython38;
+            }
+            Path linuxPythonVenv = venvRoot.resolve("bin/python3");
+            if (Files.exists(linuxPythonVenv)) {
+                return linuxPythonVenv;
+            }
+            Path linuxPythonAlt = venvRoot.resolve("bin/python");
+            if (Files.exists(linuxPythonAlt)) {
+                return linuxPythonAlt;
+            }
+            Path windowsPython = venvRoot.resolve("Scripts/python.exe");
+            if (Files.exists(windowsPython)) {
+                return windowsPython;
+            }
+        }
+
+        // Fallback to embedded standalone runtime
         Path linuxPython310 = Paths.get(resourceDir, "python-runtime/python/bin/python3.10");
         if (Files.exists(linuxPython310)) {
             return linuxPython310;
@@ -351,35 +394,6 @@ public class FlaskProcessManager implements ApplicationListener<ContextRefreshed
         if (Files.exists(linuxPython)) {
             return linuxPython;
         }
-        // Fallback to venv interpreter if bundled (still embedded)
-        linuxPython310 = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python3.10");
-        if (Files.exists(linuxPython310)) {
-            return linuxPython310;
-        }
-        Path linuxPython311 = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python3.11");
-        if (Files.exists(linuxPython311)) {
-            return linuxPython311;
-        }
-        Path linuxPython39 = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python3.9");
-        if (Files.exists(linuxPython39)) {
-            return linuxPython39;
-        }
-        Path linuxPython38 = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python3.8");
-        if (Files.exists(linuxPython38)) {
-            return linuxPython38;
-        }
-        Path linuxPythonVenv = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python3");
-        if (Files.exists(linuxPythonVenv)) {
-            return linuxPythonVenv;
-        }
-        Path linuxPythonAlt = Paths.get(resourceDir, "python-runtime/venv-offline/bin/python");
-        if (Files.exists(linuxPythonAlt)) {
-            return linuxPythonAlt;
-        }
-        Path windowsPython = Paths.get(resourceDir, "python-runtime/venv-offline/Scripts/python.exe");
-        if (Files.exists(windowsPython)) {
-            return windowsPython;
-        }
         Path windowsPythonAlt = Paths.get(resourceDir, "python-runtime/python.exe");
         if (Files.exists(windowsPythonAlt)) {
             return windowsPythonAlt;
@@ -397,7 +411,8 @@ public class FlaskProcessManager implements ApplicationListener<ContextRefreshed
 
     private Path resolveEmbeddedSitePackages() {
         // Linux venvs use lib/pythonX.Y/site-packages
-        Path libRoot = Paths.get(resourceDir, "python-runtime/venv-offline/lib");
+        Path venvRoot = resolveEmbeddedVenvRoot();
+        Path libRoot = venvRoot != null ? venvRoot.resolve("lib") : Paths.get(resourceDir, "python-runtime/venv-offline/lib");
         if (Files.isDirectory(libRoot)) {
             try {
                 try (java.nio.file.DirectoryStream<Path> stream = Files.newDirectoryStream(libRoot)) {
@@ -415,10 +430,28 @@ public class FlaskProcessManager implements ApplicationListener<ContextRefreshed
             }
         }
         // Windows venvs
-        Path venvLibWin = Paths.get(resourceDir, "python-runtime/venv-offline/Lib/site-packages");
+        Path venvLibWin = venvRoot != null ? venvRoot.resolve("Lib/site-packages") :
+            Paths.get(resourceDir, "python-runtime/venv-offline/Lib/site-packages");
         if (Files.exists(venvLibWin)) {
             return venvLibWin;
         }
         return null;
     }
+
+    private Path resolveEmbeddedVenvRoot() {
+        Path venv = Paths.get(resourceDir, "python-runtime/venv-offline");
+        if (Files.isDirectory(venv)) {
+            return venv;
+        }
+        Path alt = Paths.get(resourceDir, "python-runtime/venv-linux-offline");
+        if (Files.isDirectory(alt)) {
+            return alt;
+        }
+        return null;
+    }
+
+    private boolean isWindows() {
+        String os = System.getProperty("os.name").toLowerCase();
+        return os.contains("windows");
+    }
 }