fixing the tesseract path issues for full page ocr

app.py

@@ -13,12 +13,83 @@ try:
 except ImportError:
     print("python-dotenv not installed, skipping .env file loading")
 
-# Set TESSDATA_PREFIX if not already set
-if not os.environ.get('TESSDATA_PREFIX'):
-    tessdata_dir = "/usr/share/tesseract-ocr/4.00/tessdata"
-    if os.path.exists(tessdata_dir):
-        os.environ['TESSDATA_PREFIX'] = tessdata_dir
-        print(f"Set TESSDATA_PREFIX to {tessdata_dir}")
+# Get the current directory
+current_dir = os.path.dirname(os.path.abspath(__file__))
+
+# Function to setup Tesseract
+def setup_tesseract():
+    """Setup Tesseract OCR environment."""
+    print("Setting up Tesseract OCR environment...")
+    
+    # Create tessdata directory if it doesn't exist
+    tessdata_dir = os.path.join(current_dir, "tessdata")
+    os.makedirs(tessdata_dir, exist_ok=True)
+    
+    # Set TESSDATA_PREFIX environment variable if not already set
+    if not os.environ.get('TESSDATA_PREFIX'):
+        # Check multiple possible locations
+        possible_tessdata_dirs = [
+            tessdata_dir,  # Our local tessdata directory
+            "/usr/share/tesseract-ocr/4.00/tessdata",  # Common location in Hugging Face
+            "/usr/share/tesseract-ocr/tessdata",  # Another common location
+            "/usr/local/share/tessdata",  # Standard installation location
+        ]
+        
+        # Use the first directory that exists
+        for dir_path in possible_tessdata_dirs:
+            if os.path.exists(dir_path):
+                os.environ['TESSDATA_PREFIX'] = dir_path
+                print(f"Set TESSDATA_PREFIX to {dir_path}")
+                break
+        else:
+            # If none exist, use our local directory
+            os.environ['TESSDATA_PREFIX'] = tessdata_dir
+            print(f"No existing tessdata directory found, set TESSDATA_PREFIX to {tessdata_dir}")
+    
+    # Download eng.traineddata if it doesn't exist in our local tessdata
+    eng_traineddata = os.path.join(tessdata_dir, "eng.traineddata")
+    if not os.path.exists(eng_traineddata):
+        try:
+            print("Downloading eng.traineddata...")
+            url = "https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata"
+            urllib.request.urlretrieve(url, eng_traineddata)
+            print("Downloaded eng.traineddata")
+        except Exception as e:
+            print(f"Error downloading eng.traineddata: {e}")
+    
+    # Configure pytesseract
+    try:
+        import pytesseract
+        # Check if tesseract is in PATH
+        tesseract_cmd = shutil.which("tesseract")
+        if tesseract_cmd:
+            pytesseract.pytesseract.tesseract_cmd = tesseract_cmd
+            print(f"Set pytesseract.tesseract_cmd to {tesseract_cmd}")
+        else:
+            # Try common locations
+            common_locations = [
+                "/usr/bin/tesseract",
+                "/usr/local/bin/tesseract",
+                "/app/tesseract/tesseract"
+            ]
+            for location in common_locations:
+                if os.path.isfile(location) and os.access(location, os.X_OK):
+                    pytesseract.pytesseract.tesseract_cmd = location
+                    print(f"Set pytesseract.tesseract_cmd to {location}")
+                    break
+            else:
+                print("Warning: Could not find tesseract executable")
+    except ImportError:
+        print("pytesseract not installed")
+    
+    # Try to import tesserocr to verify it's working
+    try:
+        import tesserocr
+        print(f"tesserocr imported successfully, version: {tesserocr.tesseract_version()}")
+    except ImportError:
+        print("tesserocr not installed or not working")
+    except Exception as e:
+        print(f"Error importing tesserocr: {e}")
 
 # Load Gemini API key from environment variable
 gemini_api_key = os.getenv("GOOGLE_API_KEY")
@@ -29,9 +100,6 @@ if not gemini_api_key:
 else:
     print(f"Found Gemini API key: {gemini_api_key[:5]}...{gemini_api_key[-5:] if len(gemini_api_key) > 10 else ''}")
 
-# Get the current directory
-current_dir = os.path.dirname(os.path.abspath(__file__))
-
 # Add the current directory to the Python path
 sys.path.append(current_dir)
 
@@ -53,29 +121,6 @@ except ModuleNotFoundError:
         # Try import again
         from src.main import main
 
-# Function to setup Tesseract
-def setup_tesseract():
-    """Setup Tesseract OCR environment."""
-    # Create tessdata directory if it doesn't exist
-    tessdata_dir = os.path.join(current_dir, "tessdata")
-    os.makedirs(tessdata_dir, exist_ok=True)
-    
-    # Set TESSDATA_PREFIX environment variable if not already set
-    if not os.environ.get('TESSDATA_PREFIX'):
-        os.environ['TESSDATA_PREFIX'] = tessdata_dir
-        print(f"Set TESSDATA_PREFIX to {tessdata_dir}")
-    
-    # Download eng.traineddata if it doesn't exist
-    eng_traineddata = os.path.join(tessdata_dir, "eng.traineddata")
-    if not os.path.exists(eng_traineddata):
-        try:
-            print("Downloading eng.traineddata...")
-            url = "https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata"
-            urllib.request.urlretrieve(url, eng_traineddata)
-            print("Downloaded eng.traineddata")
-        except Exception as e:
-            print(f"Error downloading eng.traineddata: {e}")
-
 # Call setup function at import time
 setup_tesseract()
 

fix_tesseract_huggingface.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python
+"""
+Script to diagnose and fix Tesseract issues in Hugging Face environments.
+"""
+
+import os
+import sys
+import shutil
+import subprocess
+import platform
+from pathlib import Path
+import urllib.request
+
+def diagnose_tesseract():
+    """Diagnose Tesseract installation and configuration issues."""
+    print("=== Tesseract Diagnostics ===")
+    
+    # Check OS
+    print(f"Operating System: {platform.system()} {platform.release()}")
+    
+    # Check if tesseract is in PATH
+    tesseract_path = shutil.which("tesseract")
+    if tesseract_path:
+        print(f"✅ Tesseract found in PATH: {tesseract_path}")
+        try:
+            version = subprocess.check_output(["tesseract", "--version"], 
+                                             stderr=subprocess.STDOUT, 
+                                             universal_newlines=True)
+            print(f"✅ Tesseract version info:\n{version.splitlines()[0]}")
+        except (subprocess.SubprocessError, FileNotFoundError) as e:
+            print(f"❌ Error running tesseract: {e}")
+    else:
+        print("❌ Tesseract not found in PATH")
+    
+    # Check common installation locations
+    common_locations = [
+        "/usr/bin/tesseract",
+        "/usr/local/bin/tesseract",
+        "/opt/conda/bin/tesseract",
+        "/app/tesseract/tesseract",
+        r"C:\Program Files\Tesseract-OCR\tesseract.exe"
+    ]
+    
+    for location in common_locations:
+        if os.path.isfile(location) and os.access(location, os.X_OK):
+            print(f"✅ Tesseract executable found at: {location}")
+    
+    # Check TESSDATA_PREFIX
+    tessdata_prefix = os.environ.get('TESSDATA_PREFIX')
+    if tessdata_prefix:
+        print(f"✅ TESSDATA_PREFIX is set to: {tessdata_prefix}")
+        if os.path.exists(tessdata_prefix):
+            print(f"✅ TESSDATA_PREFIX directory exists")
+            eng_traineddata = os.path.join(tessdata_prefix, "eng.traineddata")
+            if os.path.exists(eng_traineddata):
+                print(f"✅ eng.traineddata found at: {eng_traineddata}")
+            else:
+                print(f"❌ eng.traineddata not found at: {eng_traineddata}")
+        else:
+            print(f"❌ TESSDATA_PREFIX directory does not exist: {tessdata_prefix}")
+    else:
+        print("❌ TESSDATA_PREFIX environment variable not set")
+    
+    # Check pytesseract
+    try:
+        import pytesseract
+        print(f"✅ pytesseract is installed")
+        print(f"✅ pytesseract.tesseract_cmd = {pytesseract.pytesseract.tesseract_cmd}")
+    except ImportError:
+        print("❌ pytesseract is not installed")
+    
+    # Check tesserocr
+    try:
+        import tesserocr
+        print(f"✅ tesserocr is installed")
+        print(f"✅ tesserocr version: {tesserocr.tesseract_version()}")
+    except ImportError:
+        print("❌ tesserocr is not installed")
+    except Exception as e:
+        print(f"❌ Error importing tesserocr: {e}")
+
+def fix_tesseract():
+    """Fix common Tesseract issues."""
+    print("\n=== Fixing Tesseract Issues ===")
+    
+    # Create local tessdata directory
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    tessdata_dir = os.path.join(current_dir, "tessdata")
+    os.makedirs(tessdata_dir, exist_ok=True)
+    print(f"✅ Created local tessdata directory: {tessdata_dir}")
+    
+    # Set TESSDATA_PREFIX to our local directory
+    os.environ['TESSDATA_PREFIX'] = tessdata_dir
+    print(f"✅ Set TESSDATA_PREFIX to: {tessdata_dir}")
+    
+    # Download eng.traineddata
+    eng_traineddata = os.path.join(tessdata_dir, "eng.traineddata")
+    if not os.path.exists(eng_traineddata):
+        try:
+            print("Downloading eng.traineddata...")
+            url = "https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata"
+            urllib.request.urlretrieve(url, eng_traineddata)
+            print("✅ Downloaded eng.traineddata")
+        except Exception as e:
+            print(f"❌ Error downloading eng.traineddata: {e}")
+    else:
+        print("✅ eng.traineddata already exists")
+    
+    # Configure pytesseract
+    try:
+        import pytesseract
+        tesseract_path = shutil.which("tesseract")
+        if tesseract_path:
+            pytesseract.pytesseract.tesseract_cmd = tesseract_path
+            print(f"✅ Set pytesseract.tesseract_cmd to {tesseract_path}")
+        else:
+            # Try common locations
+            common_locations = [
+                "/usr/bin/tesseract",
+                "/usr/local/bin/tesseract",
+                "/app/tesseract/tesseract"
+            ]
+            for location in common_locations:
+                if os.path.isfile(location) and os.access(location, os.X_OK):
+                    pytesseract.pytesseract.tesseract_cmd = location
+                    print(f"✅ Set pytesseract.tesseract_cmd to {location}")
+                    break
+    except ImportError:
+        print("❌ pytesseract not installed, please install it with: pip install pytesseract")
+    
+    # Add TESSDATA_PREFIX to .env file for persistence
+    try:
+        with open(".env", "a") as f:
+            f.write(f"\nTESSDATAFIX_PREFIX={tessdata_dir}\n")
+        print("✅ Added TESSDATA_PREFIX to .env file")
+    except Exception as e:
+        print(f"❌ Error adding TESSDATA_PREFIX to .env file: {e}")
+    
+    print("\n=== Tesseract Fix Complete ===")
+    print("Please restart your application for changes to take effect.")
+
+if __name__ == "__main__":
+    diagnose_tesseract()
+    fix_tesseract() 

packages.txt

@@ -0,0 +1,6 @@
+tesseract-ocr
+tesseract-ocr-eng
+libtesseract-dev
+libleptonica-dev
+imagemagick
+poppler-utils 

setup.sh

@@ -5,10 +5,28 @@ set -e
 
 echo "Setting up Tesseract OCR environment..."
 
-# Install google-genai package
-echo "Installing Google Gemini API client..."
+# Install required packages if not already installed
+if ! command -v tesseract &> /dev/null; then
+    echo "Tesseract not found, attempting to install..."
+    apt-get update -y || echo "Failed to update apt, continuing anyway"
+    apt-get install -y tesseract-ocr tesseract-ocr-eng libtesseract-dev libleptonica-dev || echo "Failed to install tesseract via apt, continuing anyway"
+fi
+
+# Install Python dependencies
+echo "Installing Python dependencies..."
+pip install -q -U pytesseract pillow opencv-python-headless pdf2image
 pip install -q -U google-genai
-echo "Google Gemini API client installed successfully"
+echo "Python dependencies installed successfully"
+
+# Install tesserocr with pip
+echo "Installing tesserocr..."
+pip install -q -U tesserocr || echo "Failed to install tesserocr with pip, trying with specific compiler flags..."
+
+# If tesserocr installation failed, try with specific compiler flags
+if ! python -c "import tesserocr" &> /dev/null; then
+    echo "Trying alternative tesserocr installation..."
+    CPPFLAGS="-I/usr/local/include -I/usr/include" LDFLAGS="-L/usr/local/lib -L/usr/lib" pip install -q -U tesserocr || echo "Failed to install tesserocr with compiler flags, continuing anyway"
+fi
 
 # Create tessdata directory if it doesn't exist
 mkdir -p tessdata
@@ -20,33 +38,34 @@ echo "TESSDATA_PREFIX set to: $TESSDATA_PREFIX"
 # Download eng.traineddata if it doesn't exist
 if [ ! -f "tessdata/eng.traineddata" ]; then
   echo "Downloading eng.traineddata..."
-  wget -O tessdata/eng.traineddata https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata
+  wget -O tessdata/eng.traineddata https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata || \
+  curl -o tessdata/eng.traineddata https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata
   echo "Downloaded eng.traineddata"
 else
   echo "eng.traineddata already exists"
 fi
 
-# Also copy to system location
-if [ -d "/usr/local/share/tessdata" ]; then
-  echo "Copying eng.traineddata to system location..."
-  sudo cp -f tessdata/eng.traineddata /usr/local/share/tessdata/ || echo "Failed to copy to system location, continuing anyway"
-fi
+# Try to copy to system locations (may fail in restricted environments)
+for tessdata_dir in "/usr/share/tesseract-ocr/4.00/tessdata" "/usr/share/tesseract-ocr/tessdata" "/usr/local/share/tessdata"; do
+  if [ -d "$tessdata_dir" ]; then
+    echo "Copying eng.traineddata to $tessdata_dir..."
+    cp -f tessdata/eng.traineddata "$tessdata_dir/" 2>/dev/null || echo "Failed to copy to $tessdata_dir, continuing anyway"
+  fi
+done
 
 # Verify Tesseract installation
 echo "Verifying Tesseract installation..."
-tesseract --version || echo "Tesseract not found in PATH"
-
-# Test Tesseract functionality
-echo "Testing Tesseract functionality..."
-echo "Hello World" > test.txt
-convert -size 100x30 xc:white -font Arial -pointsize 12 -fill black -annotate +10+20 "Hello World" test.png || echo "ImageMagick convert not available, skipping test image creation"
-
-if [ -f "test.png" ]; then
-  tesseract test.png test_output || echo "Tesseract test failed, but continuing"
-  if [ -f "test_output.txt" ]; then
-    echo "Tesseract test output:"
-    cat test_output.txt
-  fi
-fi
+tesseract --version || echo "Tesseract not found in PATH, but may still be available to Python"
+
+# Test tesserocr if installed
+echo "Testing tesserocr..."
+python -c "import tesserocr; print(f'tesserocr version: {tesserocr.tesseract_version()}')" || echo "tesserocr not working, but may still be able to use pytesseract"
+
+# Test pytesseract
+echo "Testing pytesseract..."
+python -c "import pytesseract; print(f'pytesseract path: {pytesseract.tesseract_cmd}')" || echo "pytesseract not working"
+
+echo "Setup completed"
 
-echo "Setup completed" 
+# Add TESSDATA_PREFIX to .env file for persistence
+echo "TESSDATA_PREFIX=$(pwd)/tessdata" >> .env