ssbg04/TIS_RMS
1
0 Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix(ocr): improve SF9 OCR parser with orientation fallback and update settings screen edge fade

Browse Source
This commit is contained in:
ssbg04 committed 2026-08-13 12:14:50 +08:00
1 parent e5b02614cc
commit 98495e91eb
4 files changed
+142 -146
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

backend/src/controllers/ocr.controller.js
+14 -112
View File
@@ -17,121 +17,23 @@ exports.extractOcrData = async (req, res) => {
console.log(`[OCR] Received file: ${req.file.originalname} (Type: ${req.file.mimetype})`);
let text = '';
const isExcel = /\.(xlsx|xls|csv)$/i.test(req.file.originalname) ||
req.file.mimetype.includes('spreadsheet') ||
req.file.mimetype.includes('excel') ||
req.file.mimetype.includes('csv');
const isPdf = req.file.mimetype === 'application/pdf' || /\.(pdf)$/i.test(req.file.originalname);
if (isExcel) {
console.log('[OCR] Excel/Spreadsheet file detected. Reading tabs directly via exceljs...');
try {
text = await ocrParser.extractTextFromFile(req.file.path, req.file.originalname, req.file.mimetype);
console.log(`[OCR] Excel parsed successfully. Extracted text length: ${text.length}`);
if (fs.existsSync(req.file.path)) fs.unlinkSync(req.file.path);
} else {
let imagePathToScan = req.file.path;
// ==========================================
// 1. PDF TO IMAGE CONVERSION (Native Ghostscript)
// ==========================================
if (isPdf) {
console.log('[OCR] PDF detected. Spawning Ghostscript natively to convert to PNG...');
const saveDirectory = path.resolve('./uploads/temp_ocr/');
if (!fs.existsSync(saveDirectory)) {
fs.mkdirSync(saveDirectory, { recursive: true });
}
const outputPngPath = path.join(saveDirectory, `temp_ocr_${Date.now()}.png`);
// Execute Ghostscript directly to avoid GraphicsMagick registry issues on Windows
const gsArgs = [
'-dQUIET', '-dPARANOIDSAFER', '-dBATCH', '-dNOPAUSE', '-dNOPROMPT',
'-sDEVICE=png16m', // Output format
'-dTextAlphaBits=4', '-dGraphicsAlphaBits=4', // Anti-aliasing
'-r300', // 300 DPI resolution
'-dFirstPage=1', '-dLastPage=1', // Only convert the first page
`-sOutputFile=${outputPngPath}`,
req.file.path // Input PDF file
];
const isWindows = process.platform === 'win32';
try {
if (isWindows) {
try {
await execFileAsync('gswin64c', gsArgs);
} catch (err) {
if (err.code === 'ENOENT') {
try {
await execFileAsync('gs', gsArgs);
} catch (fallbackErr) {
await execFileAsync('gswin32c', gsArgs);
}
} else throw err;
}
} else {
// Linux / Ubuntu: Run 'gs' executable directly from system PATH
await execFileAsync('gs', gsArgs);
}
} catch (finalErr) {
if (finalErr.code === 'ENOENT') {
const missingErr = isWindows
? 'Ghostscript is not installed or missing from backend/ghostscript folder.'
: 'Ghostscript (gs) is missing on Linux. Install via: sudo apt-get install -y ghostscript';
return res.status(500).json({
message: 'Missing OCR Program',
error: missingErr
});
}
throw finalErr;
}
console.log('[OCR] Ghostscript successfully generated PNG:', outputPngPath);
imagePathToScan = outputPngPath;
generatedImagePath = outputPngPath;
} catch (extractErr) {
if (extractErr.message && extractErr.message.includes('ENOENT')) {
const missingErr = process.platform === 'win32'
? 'Tesseract/Ghostscript OCR is not installed or missing from backend folders.'
: 'Tesseract/Ghostscript OCR is missing on Linux. Install via: sudo apt-get install -y tesseract-ocr tesseract-ocr-eng ghostscript';
return res.status(500).json({
message: 'Missing OCR Program',
error: missingErr
});
}
// ==========================================
// 2. RUN NATIVE TESSERACT EXECUTABLE
// ==========================================
console.log('[OCR] Starting Native Tesseract Engine...');
// Ensure TESSDATA_PREFIX fallback for both Windows and Linux
const defaultTessData = path.join(__dirname, '..', '..', 'tesseract', 'tessdata');
const tessEnv = {
...process.env,
TESSDATA_PREFIX: process.env.TESSDATA_PREFIX || defaultTessData
};
let stdout;
try {
const result = await execFileAsync('tesseract', [
imagePathToScan,
'stdout', // Output to standard output instead of a file
'-l', 'eng'
], { env: tessEnv, maxBuffer: 1024 * 1024 * 10 }); // 10MB buffer just in case
stdout = result.stdout;
} catch (tessErr) {
if (tessErr.code === 'ENOENT') {
const missingErr = process.platform === 'win32'
? 'Tesseract OCR is not installed or missing from backend/tesseract folder.'
: 'Tesseract OCR is missing on Linux. Install via: sudo apt-get install -y tesseract-ocr tesseract-ocr-eng';
return res.status(500).json({
message: 'Missing OCR Program',
error: missingErr
});
}
throw tessErr;
throw extractErr;
} finally {
if (fs.existsSync(req.file.path)) {
try { fs.unlinkSync(req.file.path); } catch (_) {}
}
text = stdout;
console.log('[OCR] Tesseract scan complete. Processing regex...');
// Clean up immediately after reading
if (fs.existsSync(req.file.path)) fs.unlinkSync(req.file.path);
if (generatedImagePath && fs.existsSync(generatedImagePath)) fs.unlinkSync(generatedImagePath);
}
// ==========================================
backend/src/services/ocrParser.js
+114 -12
View File
@@ -38,6 +38,9 @@ exports.parseSF9 = (text) => {
if (lrnMatch) {
extracted.lrn = lrnMatch[1].replace(/[\s\-\.]/g, '');
} else {
const fallbackLrn = processedText.match(/\b\d{12}\b/);
if (fallbackLrn) extracted.lrn = fallbackLrn[0];
}
// 2. Extract Name fields
@@ -56,19 +59,33 @@ exports.parseSF9 = (text) => {
const gradeMatch = text.match(/(?:Grade|Grade\s*Level)[:\s,]*(\d+)/i);
if (gradeMatch) extracted.gradeLevel = gradeMatch[1];
const sectionMatch = text.match(/Section[:\s,]*([^\n,]+)/i);
if (sectionMatch) extracted.section = sectionMatch[1].trim();
const sectionMatch = text.match(/Section[:\s,]*([^\r\n,]+)/i);
if (sectionMatch) {
let sec = sectionMatch[1].trim();
sec = sec.replace(/^[^A-Za-z0-9]+/, '').trim();
extracted.section = sec;
}
const syMatch = text.match(/(?:School\s*Year|S\.?Y\.?)[:\s,]*(\d{4}\s*-\s*\d{4})/i);
if (syMatch) extracted.schoolYear = syMatch[1].replace(/\s/g, '');
const trackMatch = text.match(/(?:TRACK\s*\/?\s*STRAND|TRACK|STRAND)[:\s,]*([A-Za-z0-9\/\-\s]+?)(?=\s*(?:Admitted|Grade|Section|School\s*Year|S\.?Y\.?|LRN|SEX|DOB|DATE|\r|\n|$))/i);
if (trackMatch) extracted.trackStrand = trackMatch[1].trim();
const advMatch = text.match(/(?:Teacher|Adviser|Name\s*of\s*Adviser\/?Teacher)[:\s,]*([A-Za-z\s\-\.]+?)(?=\s*(?:Principal|Signature|Date|\r|\n|$))/i);
if (advMatch) {
let adv = advMatch[1].replace(/[,:]/g, '').trim();
adv = adv.replace(/\s*(?:Signature|Date).*$/i, '').trim();
extracted.adviserName = adv;
}
extracted.scholasticRecords = exports.extractAllScholasticRecords(text);
if (extracted.scholasticRecords.length > 0) {
const latest = extracted.scholasticRecords[extracted.scholasticRecords.length - 1];
extracted.gradeLevel = latest.gradeLevel;
extracted.section = latest.section;
extracted.schoolYear = latest.schoolYear;
extracted.adviserName = latest.adviserName || '';
if (!extracted.gradeLevel) extracted.gradeLevel = latest.gradeLevel;
if (!extracted.section) extracted.section = latest.section;
if (!extracted.schoolYear) extracted.schoolYear = latest.schoolYear;
if (!extracted.adviserName) extracted.adviserName = latest.adviserName || '';
extracted.semester = latest.semester || '';
}
@@ -257,10 +274,10 @@ function extractStudentName(text) {
let extension = '';
// 1. Try explicit separate headers: LAST NAME, FIRST NAME, MIDDLE NAME, EXTENSION
const lastNameMatch = text.match(/(?:LAST\s*NAME|SURNAME|FAMILY\s*NAME)[:\s,]*([A-Za-z\-\s\.\']+?)(?=\s*(?:FIRST|GIVEN|MIDDLE|M\.?I\.?|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\n|$))/i);
const firstNameMatch = text.match(/(?:FIRST\s*NAME|GIVEN\s*NAME)[:\s,]*([A-Za-z\-\s\.\']+?)(?=\s*(?:LAST|SURNAME|MIDDLE|M\.?I\.?|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\n|$))/i);
const middleNameMatch = text.match(/(?:MIDDLE\s*NAME|MIDDLE\s*INITIAL|MIDDLE|M\.?I\.?)[:\s,]*([A-Za-z\-\s\.\']+?)(?=\s*(?:LAST|SURNAME|FIRST|GIVEN|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\n|$))/i);
const extensionMatch = text.match(/(?:EXTENSION\s*NAME|EXT\.?\s*NAME|NAME\s*EXT\.?|EXTENSION|SUFFIX|EXT)[:\s,]*([A-Za-z0-9\.\-]+?)(?=\s*(?:LAST|FIRST|MIDDLE|M\.?I\.?|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\n|$))/i);
const lastNameMatch = text.match(/\b(?:LAST\s*NAME|SURNAME|FAMILY\s*NAME)[:\s,]*([A-Za-z\-\ \.\']+?)(?=\s*(?:FIRST|GIVEN|MIDDLE|\bM\.?I\.?\b|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\r|\n|$))/i);
const firstNameMatch = text.match(/\b(?:FIRST\s*NAME|GIVEN\s*NAME)[:\s,]*([A-Za-z\-\ \.\']+?)(?=\s*(?:LAST|SURNAME|MIDDLE|\bM\.?I\.?\b|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\r|\n|$))/i);
const middleNameMatch = text.match(/\b(?:MIDDLE\s*NAME|MIDDLE\s*INITIAL|MIDDLE|\bM\.?I\.?\b)[:\s,]*([A-Za-z\-\ \.\']+?)(?=\s*(?:LAST|SURNAME|FIRST|GIVEN|EXT|SUFFIX|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\r|\n|$))/i);
const extensionMatch = text.match(/\b(?:EXTENSION\s*NAME|EXT\.?\s*NAME|NAME\s*EXT\.?|EXTENSION|SUFFIX|EXT)[:\s,]*([A-Za-z0-9\.\-]+?)(?=\s*(?:LAST|FIRST|MIDDLE|\bM\.?I\.?\b|NAME|LRN|SEX|DOB|DATE|GRADE|SECTION|S\.?Y\.?|,|\r|\n|$))/i);
if (lastNameMatch) lastName = lastNameMatch[1].trim();
if (firstNameMatch) firstName = firstNameMatch[1].trim();
@@ -269,7 +286,7 @@ function extractStudentName(text) {
// 2. Try combined format if separate headers didn't yield both lastName and firstName
if (!lastName || !firstName) {
const combinedMatch = text.match(/(?:Name|Learner'?s?\s*Name|Name\s*of\s*Learner|Name\s*of\s*Student)[:\s,]+([A-Za-z\-\s\.\']+?),\s*([A-Za-z\-\s\.\']+?)(?=\s*(?:LRN|SEX|GENDER|DOB|BIRTH|DATE|GRADE|SECTION|SCHOOL|S\.?Y\.?|TRACK|STRAND|\n|$))/i);
const combinedMatch = text.match(/(?:Name|Learner'?s?\s*Name|Name\s*of\s*Learner|Name\s*of\s*Student)[:\s,]+([A-Za-z\-\ \.\']+?),\s*([A-Za-z\-\ \.\']+?)(?=\s*(?:LRN|SEX|GENDER|DOB|BIRTH|DATE|GRADE|SECTION|SCHOOL|S\.?Y\.?|TRACK|STRAND|\r|\n|$))/i);
if (combinedMatch) {
const candLast = combinedMatch[1].trim();
let candRest = combinedMatch[2].trim().split(/\s+/);
@@ -566,6 +583,48 @@ exports.extractTextFromFile = async (filePath, originalName = '', mimeType = '')
'-l', 'eng'
], { env: tessEnv, maxBuffer: 1024 * 1024 * 10 });
text = result.stdout || '';
// Auto-rotation fallback: check if initial text parsed key fields
let bestText = text;
let parsedSF9 = exports.parseSF9(bestText);
let parsedSF10 = exports.parseSF10(bestText);
let bestScore = Math.max(evalParseScore(parsedSF9), evalParseScore(parsedSF10));
if (bestScore < 3) {
console.log(`[ocrParser] Initial OCR score (${bestScore}) is low. Trying image rotations (90, 180, 270)...`);
const saveDirectory = path.resolve('./uploads/temp_ocr/');
if (!fs.existsSync(saveDirectory)) {
fs.mkdirSync(saveDirectory, { recursive: true });
}
for (const angle of [90, 180, 270]) {
const rotPath = path.join(saveDirectory, `rot_${angle}_${Date.now()}_${Math.floor(Math.random() * 1000)}.jpg`);
try {
await rotateImage(imagePathToScan, angle, rotPath);
const rotResult = await execFileAsync('tesseract', [
rotPath,
'stdout',
'-l', 'eng'
], { env: tessEnv, maxBuffer: 1024 * 1024 * 10 });
const rotText = rotResult.stdout || '';
const rSF9 = exports.parseSF9(rotText);
const rSF10 = exports.parseSF10(rotText);
const rotScore = Math.max(evalParseScore(rSF9), evalParseScore(rSF10));
if (rotScore > bestScore) {
bestScore = rotScore;
bestText = rotText;
console.log(`[ocrParser] Found better orientation at ${angle} degrees (score ${rotScore}).`);
}
if (bestScore >= 4) break;
} catch (rotErr) {
console.error(`[ocrParser] Rotation ${angle} failed:`, rotErr.message);
} finally {
if (fs.existsSync(rotPath)) {
try { fs.unlinkSync(rotPath); } catch (_) {}
}
}
}
}
text = bestText;
} catch (err) {
console.error('[ocrParser] Tesseract execution failed:', err.message);
throw new Error('Tesseract OCR execution failed: ' + err.message);
@@ -576,4 +635,47 @@ exports.extractTextFromFile = async (filePath, originalName = '', mimeType = '')
}
return text;
};
};
function evalParseScore(parsed) {
if (!parsed) return 0;
let score = 0;
if (parsed.lrn && parsed.lrn.length === 12) score += 3;
if (parsed.lastName && parsed.firstName) score += 3;
if (parsed.gradeLevel) score += 1;
if (parsed.schoolYear) score += 1;
if (parsed.section) score += 1;
if (parsed.scholasticRecords && parsed.scholasticRecords.length > 0) score += 2;
return score;
}
async function rotateImage(srcPath, angle, dstPath) {
if (process.platform === 'win32') {
const s = '$';
const script = 'param([string]' + s + 'src, [int]' + s + 'angle, [string]' + s + 'dst)\n' +
'Add-Type -AssemblyName System.Drawing\n' +
s + 'img = [System.Drawing.Image]::FromFile(' + s + 'src)\n' +
s + 'bmp = New-Object System.Drawing.Bitmap(' + s + 'img)\n' +
s + 'img.Dispose()\n' +
'if (' + s + 'angle -eq 90) { ' + s + 'bmp.RotateFlip([System.Drawing.RotateFlipType]::Rotate90FlipNone) }\n' +
'elseif (' + s + 'angle -eq 180) { ' + s + 'bmp.RotateFlip([System.Drawing.RotateFlipType]::Rotate180FlipNone) }\n' +
'elseif (' + s + 'angle -eq 270) { ' + s + 'bmp.RotateFlip([System.Drawing.RotateFlipType]::Rotate270FlipNone) }\n' +
s + 'bmp.Save(' + s + 'dst, [System.Drawing.Imaging.ImageFormat]::Jpeg)\n' +
s + 'bmp.Dispose()\n';
const dir = path.dirname(dstPath);
if (!fs.existsSync(dir)) fs.mkdirSync(dir, { recursive: true });
const tempScript = path.join(dir, `rot_${Date.now()}_${Math.floor(Math.random() * 1000)}.ps1`);
fs.writeFileSync(tempScript, script, 'utf8');
try {
await execFileAsync('powershell', ['-NoProfile', '-ExecutionPolicy', 'Bypass', '-File', tempScript, '-src', path.resolve(srcPath), '-angle', String(angle), '-dst', path.resolve(dstPath)]);
} finally {
if (fs.existsSync(tempScript)) {
try { fs.unlinkSync(tempScript); } catch (_) {}
}
}
} else {
try {
await execFileAsync('convert', [srcPath, '-rotate', String(angle), dstPath]);
} catch (_) {}
}
}